Bumped tantivy version to 0.15.3 in Cargo.toml

Closes #1101

* fix delete documents with sort by field

Co-authored-by: Andre-Philippe Paquet <appaquet@gmail.com>

.github/FUNDING.yml

@@ -0,0 +1,12 @@
+# These are supported funding model platforms
+
+github: fulmicoton
+patreon: # Replace with a single Patreon username
+open_collective: # Replace with a single Open Collective username
+ko_fi: # Replace with a single Ko-fi username
+tidelift: # Replace with a single Tidelift platform-name/package-name e.g., npm/babel
+community_bridge: # Replace with a single Community Bridge project-name e.g., cloud-foundry
+liberapay: # Replace with a single Liberapay username
+issuehunt: # Replace with a single IssueHunt username
+otechie: # Replace with a single Otechie username
+custom: # Replace with up to 4 custom sponsorship URLs e.g., ['link1', 'link2']

.github/ISSUE_TEMPLATE/actions.md

@@ -0,0 +1,13 @@
+---
+name: Actions
+about: Actions not directly related to producing code.
+
+---
+
+# Actions title
+
+Action description. 
+e.g. 
+- benchmark
+- investigate and report
+- etc.

.github/dependabot.yml

@@ -0,0 +1,8 @@
+version: 2
+updates:
+- package-ecosystem: cargo
+  directory: "/"
+  schedule:
+    interval: daily
+    time: "20:00"
+  open-pull-requests-limit: 10

.github/workflows/test.yml

@@ -0,0 +1,24 @@
+name: Rust
+
+on:
+  push:
+    branches: [ main ]
+  pull_request:
+    branches: [ main ]
+
+env:
+  CARGO_TERM_COLOR: always
+
+jobs:
+  build:
+
+    runs-on: ubuntu-latest
+
+    steps:
+    - uses: actions/checkout@v2
+    - name: Build
+      run: cargo build --verbose --workspace
+    - name: Run tests
+      run: cargo test --verbose --workspace
+    - name: Check Formatting
+      run: cargo fmt --all -- --check

.gitignore

@@ -1,4 +1,5 @@
 tantivy.iml
+proptest-regressions
 *.swp
 target
 target/debug
@@ -11,3 +12,4 @@ cpp/simdcomp/bitpackingbenchmark
 *.bk
 .idea
 trace.dat
+cargo-timing*

.travis.yml

@@ -47,6 +47,7 @@ matrix:
 before_install:
   - set -e
   - rustup self update
+  - rustup component add rustfmt
 
 install:
   - sh ci/install.sh
@@ -60,6 +61,7 @@ before_script:
 
 script:
   - bash ci/script.sh
+  - cargo fmt --all -- --check
 
 before_deploy:
   - sh ci/before_deploy.sh

ARCHITECTURE.md

@@ -0,0 +1,295 @@
+# Tantivy
+
+## What is tantivy?
+
+Tantivy is a library that is meant to build search engines. Although it is by no means a port of Lucene, its architecture is strongly inspired by it. If you are familiar with Lucene, you may be struck by the overlapping vocabulary.
+This is not fortuitous.
+
+Tantivy's bread and butter is to address the problem of full-text search :
+
+Given a large set of textual documents, and a text query, return the K-most relevant documents in a very efficient way. To execute these queries rapidly, the tantivy needs to build an index beforehand. The relevance score implemented in the tantivy is not configurable. Tantivy uses the same score as the default similarity used in Lucene / Elasticsearch, called [BM25](https://en.wikipedia.org/wiki/Okapi_BM25).
+
+But tantivy's scope does not stop there. Numerous features are required to power rich-search applications. For instance, one may want to:
+- compute the count of documents matching a query in the different section of an e-commerce website,
+- display an average price per meter square for a real estate search engine,
+- take into account historical user data to rank documents in a specific way,
+- or even use tantivy to power an OLAP database.
+
+A more abstract description of the problem space tantivy is trying to address is the following.
+
+Ingest a large set of documents, create an index that makes it possible to
+rapidly select all documents matching a given predicate (also known as a query) and
+collect some information about them ([See collector](#collector-define-what-to-do-with-matched-documents)).
+
+Roughly speaking the design is following these guiding principles:
+- Search should be O(1) in memory.
+- Indexing should be O(1) in memory. (In practice it is just sublinear)
+- Search should be as fast as possible
+
+This comes at the cost of the dynamicity of the index: while it is possible to add, and delete documents from our corpus, the tantivy is designed to handle these updates in large batches.
+
+## [core/](src/core): Index, segments, searchers.
+
+Core contains all of the high-level code to make it possible to create an index, add documents, delete documents and commit.
+
+This is both the most high-level part of tantivy, the least performance-sensitive one, the seemingly most mundane code... And paradoxically the most complicated part.
+
+### Index and Segments...
+
+A tantivy index is a collection of smaller independent immutable segments. 
+Each segment contains its own independent set of data structures.
+
+A segment is identified by a segment id that is in fact a UUID.
+The file of a segment has the format
+
+ ```segment-id . ext ```
+
+The extension signals which data structure (or [`SegmentComponent`](src/core/segment_component.rs)) is stored in the file.
+
+A small `meta.json` file is in charge of keeping track of the list of segments, as well as the schema.
+
+On commit, one segment per indexing thread is written to disk, and the `meta.json` is then updated atomically.
+
+For a better idea of how indexing works, you may read the [following blog post](https://fulmicoton.com/posts/behold-tantivy-part2/).
+
+
+### Deletes
+
+Deletes happen by deleting a "term". Tantivy does not offer any notion of primary id, so it is up to the user to use a field in their schema as if it was a primary id, and delete the associated term if they want to delete only one specific document.
+
+On commit, tantivy will find all of the segments with documents matching this existing term and create a [tombstone file](src/fastfield/delete.rs) that represents the bitset of the document that are deleted.
+Like all segment files, this file is immutable. Because it is possible to have more than one tombstone file at a given instant, the tombstone filename has the format ``` segment_id . commit_opstamp . del```.
+
+An opstamp is simply an incremental id that identifies any operation applied to the index. For instance, performing a commit or adding a document.
+
+
+### DocId
+
+Within a segment, all documents are identified by a DocId that ranges within `[0, max_doc)`.
+where `max_doc` is the number of documents in the segment, (deleted or not). Having such a compact `DocId` space is key to the compression of our data structures.
+
+The DocIds are simply allocated in the order documents are added to the index.
+
+### Merges
+
+In separate threads, tantivy's index writer search for opportunities to merge segments.
+The point of segment merge is to:
+- eventually get rid of tombstoned documents
+- reduce the otherwise ever-growing number of segments.
+
+Indeed, while having several segments instead of one does not hurt search too much, having hundreds can have a measurable impact on the search performance.
+
+### Searcher
+
+The user of the library usually does not need to know about the existence of Segments.
+Searching is done through an object called a [`Searcher`](src/core/searcher.rs), that captures a
+snapshot of the index at one point of time, by holding a list of [SegmentReader](src/core/segment_reader.rs).
+
+In other words, regardless of commits, file garbage collection, or segment merge that might happen, as long as the user holds and reuse the same [Searcher](src/core/searcher.rs), search will happen on an immutable snapshot of the index.
+
+## [directory/](src/directory): Where should the data be stored?
+
+Tantivy, like Lucene, abstracts the place where the data should be stored in a key-trait
+called [`Directory`](src/directory/directory.rs).
+Contrary to Lucene however, "files" are quite different from some kind of `io::Read` object.
+Check out [`src/directory/directory.rs`](src/directory/directory.rs) trait for more details.
+
+Tantivy ships two main directory implementation: the `MMapDirectory` and the `RAMDirectory`,
+but users can extend tantivy with their own implementation.
+
+## [schema/](src/schema): What are documents?
+
+Tantivy's document follows a very strict schema, decided before building any index.
+
+The schema defines all of the fields that the indexes [`Document`](src/schema/document.rs) may and should contain, their types (`text`, `i64`, `u64`, `Date`, ...) as well as how it should be indexed / represented in tantivy.
+
+Depending on the type of the field, you can decide to
+- put it in the docstore
+- store it as a fast field
+- index it
+
+Practically, tantivy will push values associated with this type to up to 3 respective
+data structures.
+
+*Limitations*
+
+As of today, tantivy's schema imposes a 1:1 relationship between a field that is being ingested and a field represented in the search index. In sophisticated search application, it is fairly common to want to index a field twice using different tokenizers, or to index the concatenation of several fields together into one field.
+
+This is not something tantivy supports, and it is up to the user to duplicate field / concatenate fields before feeding them to tantivy.
+
+## General information about these data structures.
+
+All data structures in tantivy, have:
+- a writer
+- a serializer
+- a reader
+
+The writer builds an in-memory representation of a batch of documents. This representation is not searchable. It is just meant as an intermediary mutable representation, to which we can sequentially add
+the document of a batch. At the end of the batch (or if a memory limit is reached), this representation
+is then converted into an on-disk immutable representation, that is extremely compact.
+This conversion is done by the serializer.
+
+Finally, the reader is in charge of offering an API to read on this on-disk read-only representation.
+In tantivy, readers are designed to require very little anonymous memory. The data is read straight from an mmapped file, and loading an index is as fast as mmapping its files.
+
+## [store/](src/store): Here is my DocId, Gimme my document!
+
+The docstore is a row-oriented storage that, for each document, stores a subset of the fields
+that are marked as stored in the schema. The docstore is compressed using a general-purpose algorithm
+like LZ4.
+
+**Useful for**
+
+In search engines, it is often used to display search results.
+Once the top 10 documents have been identified, we fetch them from the store, and display them or their snippet on the search result page (aka SERP).
+
+**Not useful for**
+
+Fetching a document from the store is typically a "slow" operation. It usually consists in
+- searching into a compact tree-like data structure to find the position of the right block.
+- decompressing a small block
+- returning the document from this block.
+
+It is NOT meant to be called for every document matching a query.
+
+As a rule of thumb, if you hit the docstore more than 100 times per search query, you are probably misusing tantivy.
+
+
+## [fastfield/](src/fastfield): Here is my DocId, Gimme my value!
+
+Fast fields are stored in a column-oriented storage that allows for random access.
+The only compression applied is bitpacking. The column comes with two meta data.
+The minimum value in the column and the number of bits per doc.
+
+Fetching a value for a `DocId` is then as simple as computing
+
+```
+min_value + fetch_bits(num_bits * doc_id..num_bits * (doc_id+1))
+```
+
+This operation just requires one memory fetch.
+Because, DocSets are scanned through in order (DocId are iterated in a sorted manner) which
+also help locality.
+
+In Lucene's jargon, fast fields are called DocValues.
+
+**Useful for**
+
+They are typically integer values that are useful to either rank or compute aggregate over
+all of the documents matching a query (aka [DocSet](src/docset.rs)).
+
+For instance, one could define a function to combine upvotes with tantivy's internal relevancy score.
+This can be done by fetching a fast field during scoring.
+One could also compute the mean price of the items matching a query in an e-commerce website.
+This can be done by fetching a fast field in a collector.
+Finally one could decide to post-filter a docset to remove docset with a price within a specific range.
+If the ratio of filtered out documents is not too low, an efficient way to do this is to fetch the price and apply the filter on the collector side.
+
+Aside from integer values, it is also possible to store an actual byte payload.
+For advanced search engine, it is possible to store all of the features required for learning-to-rank in a byte payload, access it during search, and apply the learning-to-rank model.
+
+Finally facets are a specific kind of fast field, and the associated source code is in [`fastfield/facet_reader.rs`](src/fastfield/facet_reader.rs).
+
+# The inverted search index.
+
+The inverted index is the core part of full-text search.
+When presented a new document with the text field "Hello, happy tax payer!", tantivy breaks it into a list of so-called tokens. In addition to just splitting these strings into tokens, it might also do different kinds of operations like dropping the punctuation, converting the character to lowercase, apply stemming, etc. Tantivy makes it possible to configure the operations to be applied in the schema (tokenizer/ is the place where these operations are implemented).
+
+For instance, the default tokenizer of tantivy would break our text into: `[hello, happy, tax, payer]`.
+The document will therefore be registered in the inverted index as containing the terms
+`[text:hello, text:happy, text:tax, text:payer]`.
+
+The role of the inverted index is, when given a term, gives us in return a very fast iterator over the sorted doc ids that match the term.
+
+Such an iterator is called a posting list. In addition to giving us `DocId`, they can also give us optionally the number of occurrence of the term for each document, also called term frequency or TF.
+
+These iterators being sorted by DocId, one can create an iterator over the document containing `text:tax AND text:payer`, `(text:tax AND text:payer) OR (text:contribuable)` or any boolean expression.
+
+In order to represent the function
+```Term ⟶ Posting```
+
+The inverted index actually consists of two data structures chained together.
+
+- [Term](src/schema/term.rs) ⟶ [TermInfo](src/postings/term_info.rs) is addressed by the term dictionary.
+- [TermInfo](src/postings/term_info.rs) ⟶ [Posting](src/postings/postings.rs) is addressed by the posting lists.
+
+Where [TermInfo](src/postings/term_info.rs) is an object containing some meta data about a term.
+
+
+## [termdict/](src/termdict): Here is a term, give me the [TermInfo](src/postings/term_info.rs)!
+
+Tantivy's term dictionary is mainly in charge of supplying the function
+
+[Term](src/schema/term.rs) ⟶ [TermInfo](src/postings/term_info.rs)
+
+It is itself broken into two parts.
+- [Term](src/schema/term.rs) ⟶ [TermOrdinal](src/termdict/mod.rs) is addressed by a finite state transducer, implemented by the fst crate.
+- [TermOrdinal](src/termdict/mod.rs) ⟶ [TermInfo](src/postings/term_info.rs) is addressed by the term info store.
+
+
+## [postings/](src/postings): Iterate over documents... very fast!
+
+A posting list makes it possible to store a sorted list of doc ids and for each doc store
+a term frequency as well.
+
+The posting lists are stored in a separate file. The [TermInfo](src/postings/term_info.rs) contains an offset into that file and a number of documents for the given posting list. Both are required and sufficient to read the posting list.
+
+The posting list is organized in block of 128 documents.
+One block of doc ids is followed by one block of term frequencies.
+
+The doc ids are delta encoded and bitpacked.
+The term frequencies are bitpacked.
+
+Because the number of docs is rarely a multiple of 128, the last block may contain an arbitrary number of docs between 1 and 127 documents. We then use variable int encoding instead of bitpacking.
+
+## [positions/](src/positions): Where are my terms within the documents?
+
+Phrase queries make it possible to search for documents containing a specific sequence of terms.
+For instance, when the phrase query "the art of war" does not match "the war of art".
+To make it possible, it is possible to specify in the schema that a field should store positions in addition to being indexed.
+
+The token positions of all of the terms are then stored in a separate file with the extension `.pos`.
+The [TermInfo](src/postings/term_info.rs) gives an offset (expressed in position this time) in this file. As we iterate throught the docset,
+we advance the position reader by the number of term frequencies of the current document.
+
+## [fieldnorms/](src/fieldnorms): Here is my doc, how many tokens in this field?
+
+The [BM25](https://en.wikipedia.org/wiki/Okapi_BM25) formula also requires to know the number of tokens stored in a specific field for a given document. We store this information on one byte per document in the fieldnorm.
+The fieldnorm is therefore compressed. Values up to 40 are encoded unchanged.
+
+
+## [tokenizer/](src/tokenizer): How should we process text?
+
+Text processing is key to a good search experience.
+Splits or normalize your text too much, and the search results will have a less precision and a higher recall.
+Do not normalize, or under split your text, you will end up with a higher precision and a lesser recall.
+
+Text processing can be configured by selecting an off-the-shelf [`Tokenizer`](./src/tokenizer/tokenizer.rs) or implementing your own to first split the text into tokens, and then chain different [`TokenFilter`](src/tokenizer/tokenizer.rs)'s to it.
+
+Tantivy's comes with few tokenizers, but external crates are offering advanced tokenizers, such as [Lindera](https://crates.io/crates/lindera) for Japanese.
+
+
+## [query/](src/query): Define and compose queries
+
+The [Query](src/query/query.rs) trait defines what a query is.
+Due to the necessity for some queries to compute some statistics over the entire index, and because the
+index is composed of several `SegmentReader`, the path from transforming a `Query` to an iterator over documents is slightly convoluted, but fundamentally, this is what a Query is.
+
+The iterator over a document comes with some scoring function. The resulting trait is called a
+[Scorer](src/query/scorer.rs) and is specific to a segment.
+
+Different queries can be combined using the [BooleanQuery](src/query/boolean_query/).
+Tantivy comes with different types of queries and can be extended by implementing
+the `Query`, `Weight`, and `Scorer` traits.
+
+## [collector](src/collector): Define what to do with matched documents
+
+Collectors define how to aggregate the documents matching a query, in the broadest sense possible.
+The search will push matched documents one by one, calling their
+`fn collect(doc: DocId, score: Score);` method.
+
+Users may implement their own collectors by implementing the [Collector](src/collector/mod.rs) trait.
+
+## [query-grammar](query-grammar): Defines the grammar of the query parser
+
+While the [QueryParser](src/query/query_parser/query_parser.rs) struct is located in the `query/` directory, the actual parser combinator used to convert user queries into an AST is in an external crate called `query-grammar`. This part was externalized to lighten the work of the compiler.

CHANGELOG.md

@@ -1,7 +1,155 @@
+Tantivy 0.15.3
+=========================
+- Major bugfix. Deleting documents was broken when the index was sorted by a field. (@appaquet, @fulmicoton) #1101
+
+
+Tantivy 0.15.2
+========================
+- Major bugfix. DocStore still panics when a deleted doc is at the beginning of a block. (@appaquet) #1088
+
+Tantivy 0.15.1
+=========================
+- Major bugfix. DocStore panics when first block is deleted. (@appaquet) #1077
+
+Tantivy 0.15.0
+=========================
+- API Changes. Using Range instead of (start, end) in the API and internals (`FileSlice`, `OwnedBytes`, `Snippets`, ...)
+  This change is breaking but migration is trivial.
+- Added an Histogram collector. (@fulmicoton) #994
+- Added support for Option<TCollector>.  (@fulmicoton)
+- DocAddress is now a struct (@scampi) #987
+- Bugfix consistent tie break handling in facet's topk (@hardikpnsp) #357
+- Date field support for range queries (@rihardsk) #516
+- Added lz4-flex as the default compression scheme in tantivy (@PSeitz) #1009
+- Renamed a lot of symbols to avoid all uppercasing on acronyms, as per new clippy recommendation. For instance, RAMDirectory -> RamDirectory. (@fulmicoton)
+- Simplified positions index format (@fulmicoton) #1022
+- Moved bitpacking to bitpacker subcrate and add BlockedBitpacker, which bitpacks blocks of 128 elements (@PSeitz) #1030
+- Added support for more-like-this query in tantivy (@evanxg852000) #1011
+- Added support for sorting an index, e.g presorting documents in an index by a timestamp field. This can heavily improve performance for certain scenarios, by utilizing the sorted data (Top-n optimizations)(@PSeitz). #1026
+- Add iterator over documents in doc store (@PSeitz). #1044
+- Fix log merge policy (@PSeitz). #1043
+- Add detection to avoid small doc store blocks on merge (@PSeitz). #1054
+- Make doc store compression dynamic (@PSeitz). #1060
+- Switch to json for footer version handling (@PSeitz). #1060
+- Updated TermMerger implementation to rely on the union feature of the FST (@scampi) #469
+- Add boolean marking whether position is required in the query_terms API call (@fulmicoton). #1070
+
+
+Tantivy 0.14.0
+=========================
+- Remove dependency to atomicwrites #833 .Implemented by @fulmicoton upon suggestion and research from @asafigan).
+- Migrated tantivy error from the now deprecated `failure` crate to `thiserror` #760. (@hirevo)
+- API Change. Accessing the typed value off a `Schema::Value` now returns an Option instead of panicking if the type does not match.
+- Large API Change in the Directory API. Tantivy used to assume that all files could be somehow memory mapped. After this change, Directory return a `FileSlice` that can be reduced and eventually read into an `OwnedBytes` object. Long and blocking io operation are still required by they do not span over the entire file.
+- Added support for Brotli compression in the DocStore. (@ppodolsky)
+- Added helper for building intersections and unions in BooleanQuery (@guilload)
+- Bugfix in `Query::explain`
+- Removed dependency on `notify` #924. Replaced with `FileWatcher` struct that polls meta file every 500ms in background thread. (@halvorboe @guilload)
+- Added `FilterCollector`, which wraps another collector and filters docs using a predicate over a fast field (@barrotsteindev)
+- Simplified the encoding of the skip reader struct. BlockWAND max tf is now encoded over a single byte. (@fulmicoton)
+- `FilterCollector` now supports all Fast Field value types (@barrotsteindev)
+- FastField are not all loaded when opening the segment reader. (@fulmicoton)
+- Added an API to merge segments, see `tantivy::merge_segments` #1005. (@evanxg852000)
+
+This version breaks compatibility and requires users to reindex everything.
+
+Tantivy 0.13.2
+===================
+Bugfix. Acquiring a facet reader on a segment that does not contain any
+doc with this facet returns `None`. (#896)
+
+Tantivy 0.13.1
+===================
+Made `Query` and `Collector` `Send + Sync`.
+Updated misc dependency versions.
+
+Tantivy 0.13.0
+======================
+Tantivy 0.13 introduce a change in the index format that will require
+you to reindex your index (BlockWAND information are added in the skiplist).
+The index size increase is minor as this information is only added for
+full blocks.
+If you have a massive index for which reindexing is not an option, please contact me
+so that we can discuss possible solutions.
+
+- Bugfix in `FuzzyTermQuery` not matching terms by prefix when it should (@Peachball)
+- Relaxed constraints on the custom/tweak score functions. At the segment level, they can be mut, and they are not required to be Sync + Send.
+- `MMapDirectory::open` does not return a `Result` anymore.
+- Change in the DocSet and Scorer API. (@fulmicoton).
+A freshly created DocSet point directly to their first doc. A sentinel value called TERMINATED marks the end of a DocSet.
+`.advance()` returns the new DocId. `Scorer::skip(target)` has been replaced by `Scorer::seek(target)` and returns the resulting DocId.
+As a result, iterating through DocSet now looks as follows
+```rust
+let mut doc = docset.doc();
+while doc != TERMINATED {
+   // ...
+   doc = docset.advance();
+}
+```
+The change made it possible to greatly simplify a lot of the docset's code.
+- Misc internal optimization and introduction of the `Scorer::for_each_pruning` function. (@fulmicoton)
+- Added an offset option to the Top(.*)Collectors. (@robyoung)
+- Added Block WAND. Performance on TOP-K on term-unions should be greatly increased. (@fulmicoton, and special thanks
+to the PISA team for answering all my questions!)
+
+Tantivy 0.12.0
+======================
+- Removing static dispatch in tokenizers for simplicity. (#762)
+- Added backward iteration for `TermDictionary` stream. (@halvorboe)
+- Fixed a performance issue when searching for the posting lists of a missing term (@audunhalland)
+- Added a configurable maximum number of docs (10M by default) for a segment to be considered for merge (@hntd187, landed by @halvorboe #713)
+- Important Bugfix #777, causing tantivy to retain memory mapping. (diagnosed by @poljar)
+- Added support for field boosting. (#547, @fulmicoton)
+
+## How to update?
+
+Crates relying on custom tokenizer, or registering tokenizer in the manager will require some
+minor changes. Check https://github.com/tantivy-search/tantivy/blob/main/examples/custom_tokenizer.rs
+to check for some code sample.
+
+Tantivy 0.11.3
+=======================
+- Fixed DateTime as a fast field (#735)
+
+Tantivy 0.11.2
+=======================
+- The future returned by `IndexWriter::merge` does not borrow `self` mutably anymore (#732)
+- Exposing a constructor for `WatchHandle` (#731)
+
+Tantivy 0.11.1
+=====================
+- Bug fix #729
+
+
 Tantivy 0.11.0
 =====================
 
 - Added f64 field. Internally reuse u64 code the same way i64 does (@fdb-hiroshima)
+- Various bugfixes in the query parser.
+    - Better handling of hyphens in query parser. (#609)
+    - Better handling of whitespaces.
+- Closes #498 - add support for Elastic-style unbounded range queries for alphanumeric types eg. "title:>hello", "weight:>=70.5", "height:<200" (@petr-tik)
+- API change around `Box<BoxableTokenizer>`. See detail in #629
+- Avoid rebuilding Regex automaton whenever a regex query is reused. #639 (@brainlock)
+- Add footer with some metadata to index files. #605 (@fdb-hiroshima)
+- Add a method to check the compatibility of the footer in the index with the running version of tantivy (@petr-tik)
+- TopDocs collector: ensure stable sorting on equal score. #671 (@brainlock)
+- Added handling of pre-tokenized text fields (#642), which will enable users to
+  load tokens created outside tantivy. See usage in examples/pre_tokenized_text. (@kkoziara)
+- Fix crash when committing multiple times with deleted documents. #681 (@brainlock)
+
+## How to update?
+
+- The index format is changed. You are required to reindex your data to use tantivy 0.11.
+- `Box<dyn BoxableTokenizer>` has been replaced by a `BoxedTokenizer` struct.
+- Regex are now compiled when the `RegexQuery` instance is built. As a result, it can now return
+an error and handling the `Result` is required.
+- `tantivy::version()` now returns a `Version` object. This object implements `ToString()`
+
+Tantivy 0.10.2
+=====================
+
+- Closes #656. Solving memory leak.
 
 Tantivy 0.10.1
 =====================
@@ -17,10 +165,10 @@ Tantivy 0.10.0
 *Tantivy 0.10.0 index format is compatible with the index format in 0.9.0.*
 
 - Added an API to easily tweak or entirely replace the
- default score. See `TopDocs::tweak_score`and `TopScore::custom_score` (@pmasurel)
+ default score. See `TopDocs::tweak_score`and `TopScore::custom_score` (@fulmicoton)
 - Added an ASCII folding filter (@drusellers)
-- Bugfix in `query.count` in presence of deletes (@pmasurel)
+- Bugfix in `query.count` in presence of deletes (@fulmicoton)
-- Added `.explain(...)` in `Query` and `Weight` to (@pmasurel)
+- Added `.explain(...)` in `Query` and `Weight` to (@fulmicoton)
 - Added an efficient way to `delete_all_documents` in `IndexWriter` (@petr-tik).
   All segments are simply removed.
 
@@ -35,7 +183,7 @@ on segment postings should panic from now on.
 - `IndexMeta` is now public.  (@hntd187)
 - `IndexWriter` `add_document`, `delete_term`. `IndexWriter` is `Sync`, making it possible to use it with a `
 Arc<RwLock<IndexWriter>>`. `add_document` and `delete_term` can
-only require a read lock. (@pmasurel)
+only require a read lock. (@fulmicoton)
 - Introducing `Opstamp` as an expressive type alias for `u64`. (@petr-tik)
 - Stamper now relies on `AtomicU64` on all platforms (@petr-tik)
 - Bugfix - Files get deleted slightly earlier
@@ -159,10 +307,10 @@ Tantivy 0.6
 Special thanks to @drusellers and @jason-wolfe for their contributions
 to this release!
 
-- Removed C code. Tantivy is now pure Rust. (@pmasurel)
+- Removed C code. Tantivy is now pure Rust. (@fulmicoton)
-- BM25 (@pmasurel)
+- BM25 (@fulmicoton)
-- Approximate field norms encoded over 1 byte. (@pmasurel)
+- Approximate field norms encoded over 1 byte. (@fulmicoton)
-- Compiles on stable rust (@pmasurel)
+- Compiles on stable rust (@fulmicoton)
 - Add &[u8] fastfield for associating arbitrary bytes to each document (@jason-wolfe) (#270)
     - Completely uncompressed
     - Internally: One u64 fast field for indexes, one fast field for the bytes themselves.
@@ -170,7 +318,7 @@ to this release!
 - Add Stopword Filter support (@drusellers)
 - Add a FuzzyTermQuery (@drusellers)
 - Add a RegexQuery (@drusellers)
-- Various performance improvements (@pmasurel)_
+- Various performance improvements (@fulmicoton)_
 
 
 Tantivy 0.5.2

Cargo.toml

@@ -1,11 +1,11 @@
 [package]
 name = "tantivy"
-version = "0.10.1"
+version = "0.15.3"
 authors = ["Paul Masurel <paul.masurel@gmail.com>"]
 license = "MIT"
 categories = ["database-implementations", "data-structures"]
 description = """Search engine library"""
-documentation = "https://tantivy-search.github.io/tantivy/tantivy/index.html"
+documentation = "https://docs.rs/tantivy/"
 homepage = "https://github.com/tantivy-search/tantivy"
 repository = "https://github.com/tantivy-search/tantivy"
 readme = "README.md"
@@ -13,56 +13,58 @@ keywords = ["search", "information", "retrieval"]
 edition = "2018"
 
 [dependencies]
-base64 = "0.10.0"
+base64 = "0.13"
-byteorder = "1.0"
+byteorder = "1.4.3"
-once_cell = "0.2"
+crc32fast = "1.2.1"
-regex = "1.0"
+once_cell = "1.7.2"
-tantivy-fst = "0.1"
+regex ={ version = "1.5.4", default-features = false, features = ["std"] }
+tantivy-fst = "0.3"
 memmap = {version = "0.7", optional=true}
-lz4 = {version="1.20", optional=true}
+lz4_flex = { version = "0.8.0", default-features = false, features = ["checked-decode"], optional = true }
-snap = {version="0.2"}
+brotli = { version = "3.3", optional = true }
-atomicwrites = {version="0.2.2", optional=true}
+snap = { version = "1.0.5", optional = true }
-tempfile = "3.0"
+tempfile = { version = "3.2", optional = true }
-log = "0.4"
+log = "0.4.14"
-combine = ">=3.6.0,<4.0.0"
+serde = { version = "1.0.126", features = ["derive"] }
-tempdir = "0.3"
+serde_json = "1.0.64"
-serde = "1.0"
+num_cpus = "1.13"
-serde_derive = "1.0"
+fs2={ version = "0.4.3", optional = true }
-serde_json = "1.0"
+levenshtein_automata = "0.2"
-num_cpus = "1.2"
+uuid = { version = "0.8.2", features = ["v4", "serde"] }
-fs2={version="0.4", optional=true}
+crossbeam = "0.8"
-itertools = "0.8"
+futures = { version = "0.3.15", features = ["thread-pool"] }
-levenshtein_automata = {version="0.1", features=["fst_automaton"]}
+tantivy-query-grammar = { version="0.15.0", path="./query-grammar" }
-notify = {version="4", optional=true}
+tantivy-bitpacker = { version="0.1", path="./bitpacker" }
-bit-set = "0.5"
+stable_deref_trait = "1.2"
-uuid = { version = "0.7.2", features = ["v4", "serde"] }
+rust-stemmers = "1.2"
-crossbeam = "0.5"
+downcast-rs = "1.2"
-futures = "0.1"
+bitpacking = { version = "0.8.4", default-features = false, features = ["bitpacker4x"] }
-futures-cpupool = "0.1"
+census = "0.4"
-owning_ref = "0.4"
+fnv = "1.0.7"
-stable_deref_trait = "1.0.0"
+thiserror = "1.0.24"
-rust-stemmers = "1.1"
-downcast-rs = { version="1.0" }
-bitpacking = {version="0.8", default-features = false, features=["bitpacker4x"]}
-census = "0.2"
-fnv = "1.0.6"
-owned-read = "0.4"
-failure = "0.1"
 htmlescape = "0.3.1"
-fail = "0.3"
+fail = "0.4"
-scoped-pool = "1.0"
 murmurhash32 = "0.2"
-chrono = "0.4"
+chrono = "0.4.19"
-smallvec = "0.6"
+smallvec = "1.6.1"
+rayon = "1.5"
+lru = "0.6.5"
+fastdivide = "0.3"
+itertools = "0.10.0"
 
 [target.'cfg(windows)'.dependencies]
-winapi = "0.3"
+winapi = "0.3.9"
 
 [dev-dependencies]
-rand = "0.7"
+rand = "0.8.3"
-maplit = "1"
+maplit = "1.0.2"
 matches = "0.1.8"
-time = "0.1.42"
+proptest = "1.0"
+criterion = "0.3.4"
+
+[dev-dependencies.fail]
+version = "0.4"
+features = ["failpoints"]
 
 [profile.release]
 opt-level = 3
@@ -74,20 +76,23 @@ debug-assertions = true
 overflow-checks = true
 
 [features]
-default = ["mmap"]
+default = ["mmap", "lz4-compression" ]
-mmap = ["atomicwrites", "fs2", "memmap", "notify"]
+mmap = ["fs2", "tempfile", "memmap"]
-lz4-compression = ["lz4"]
+
+brotli-compression = ["brotli"]
+lz4-compression = ["lz4_flex"]
+snappy-compression = ["snap"]
+
 failpoints = ["fail/failpoints"]
 unstable = [] # useful for benches.
 wasm-bindgen = ["uuid/wasm-bindgen"]
 
+[workspace]
+members = ["query-grammar", "bitpacker"]
+
 [badges]
 travis-ci = { repository = "tantivy-search/tantivy" }
 
-[dev-dependencies.fail]
-features = ["failpoints"]
-
-
 # Following the "fail" crate best practises, we isolate
 # tests that define specific behavior in fail check points
 # in a different binary.
@@ -99,3 +104,7 @@ features = ["failpoints"]
 name = "failpoints"
 path = "tests/failpoints/mod.rs"
 required-features = ["fail/failpoints"]
+
+[[bench]]
+name = "analyzer"
+harness = false

Makefile

@@ -0,0 +1,3 @@
+test:
+	echo "Run test only... No examples."
+	cargo test --tests --lib

README.md

@@ -1,11 +1,10 @@
 
-[![Build Status](https://travis-ci.org/tantivy-search/tantivy.svg?branch=master)](https://travis-ci.org/tantivy-search/tantivy)
+[![Build Status](https://travis-ci.org/tantivy-search/tantivy.svg?branch=main)](https://travis-ci.org/tantivy-search/tantivy)
-[![codecov](https://codecov.io/gh/tantivy-search/tantivy/branch/master/graph/badge.svg)](https://codecov.io/gh/tantivy-search/tantivy)
+[![codecov](https://codecov.io/gh/tantivy-search/tantivy/branch/main/graph/badge.svg)](https://codecov.io/gh/tantivy-search/tantivy)
 [![Join the chat at https://gitter.im/tantivy-search/tantivy](https://badges.gitter.im/tantivy-search/tantivy.svg)](https://gitter.im/tantivy-search/tantivy?utm_source=badge&utm_medium=badge&utm_campaign=pr-badge&utm_content=badge)
 [![License: MIT](https://img.shields.io/badge/License-MIT-yellow.svg)](https://opensource.org/licenses/MIT)
-[![Build status](https://ci.appveyor.com/api/projects/status/r7nb13kj23u8m9pj/branch/master?svg=true)](https://ci.appveyor.com/project/fulmicoton/tantivy/branch/master)
+[![Build status](https://ci.appveyor.com/api/projects/status/r7nb13kj23u8m9pj/branch/main?svg=true)](https://ci.appveyor.com/project/fulmicoton/tantivy/branch/main)
 [![Crates.io](https://img.shields.io/crates/v/tantivy.svg)](https://crates.io/crates/tantivy)
-[![Say Thanks!](https://img.shields.io/badge/Say%20Thanks-!-1EAEDB.svg)](https://saythanks.io/to/fulmicoton)
 
 ![Tantivy](https://tantivy-search.github.io/logo/tantivy-logo.png)
 
@@ -21,9 +20,9 @@
 [![Become a patron](https://c5.patreon.com/external/logo/become_a_patron_button.png)](https://www.patreon.com/fulmicoton)
 
 
-**Tantivy** is a **full text search engine library** written in rust.
+**Tantivy** is a **full text search engine library** written in Rust.
 
-It is closer to [Apache Lucene](https://lucene.apache.org/) than to [Elasticsearch](https://www.elastic.co/products/elasticsearch) and [Apache Solr](https://lucene.apache.org/solr/) in the sense it is not
+It is closer to [Apache Lucene](https://lucene.apache.org/) than to [Elasticsearch](https://www.elastic.co/products/elasticsearch) or [Apache Solr](https://lucene.apache.org/solr/) in the sense it is not
 an off-the-shelf search engine server, but rather a crate that can be used
 to build such a search engine.
 
@@ -31,73 +30,70 @@ Tantivy is, in fact, strongly inspired by Lucene's design.
 
 # Benchmark
 
-Tantivy is typically faster than Lucene, but the results will depend on 
-the nature of the queries in your workload.
-
 The following [benchmark](https://tantivy-search.github.io/bench/) break downs 
 performance for different type of queries / collection.
 
+Your mileage WILL vary depending on the nature of queries and their load.
+
 # Features
 
 - Full-text search
-- Configurable tokenizer. (stemming available for 17 latin languages. Third party support for Chinese ([tantivy-jieba](https://crates.io/crates/tantivy-jieba) and [cang-jie](https://crates.io/crates/cang-jie)) and [Japanese](https://crates.io/crates/tantivy-tokenizer-tiny-segmenter)
+- Configurable tokenizer (stemming available for 17 Latin languages with third party support for Chinese ([tantivy-jieba](https://crates.io/crates/tantivy-jieba) and [cang-jie](https://crates.io/crates/cang-jie)), Japanese ([lindera](https://github.com/lindera-morphology/lindera-tantivy) and [tantivy-tokenizer-tiny-segmenter](https://crates.io/crates/tantivy-tokenizer-tiny-segmenter)) and Korean ([lindera](https://github.com/lindera-morphology/lindera-tantivy) + [lindera-ko-dic-builder](https://github.com/lindera-morphology/lindera-ko-dic-builder))
 - Fast (check out the :racehorse: :sparkles: [benchmark](https://tantivy-search.github.io/bench/) :sparkles: :racehorse:)
 - Tiny startup time (<10ms), perfect for command line tools
-- BM25 scoring (the same as lucene)
+- BM25 scoring (the same as Lucene)
-- Natural query language `(michael AND jackson) OR "king of pop"`
+- Natural query language (e.g. `(michael AND jackson) OR "king of pop"`)
-- Phrase queries search (`"michael jackson"`)
+- Phrase queries search (e.g. `"michael jackson"`)
 - Incremental indexing
 - Multithreaded indexing (indexing English Wikipedia takes < 3 minutes on my desktop)
 - Mmap directory
-- SIMD integer compression when the platform/CPU includes the SSE2 instruction set.
+- SIMD integer compression when the platform/CPU includes the SSE2 instruction set
-- Single valued and multivalued u64, i64 and f64 fast fields (equivalent of doc values in Lucene)
+- Single valued and multivalued u64, i64, and f64 fast fields (equivalent of doc values in Lucene)
 - `&[u8]` fast fields
-- Text, i64, u64, f64, dates and hierarchical facet fields
+- Text, i64, u64, f64, dates, and hierarchical facet fields
 - LZ4 compressed document store
 - Range queries
 - Faceted search
 - Configurable indexing (optional term frequency and position indexing)
 - Cheesy logo with a horse
 
-# Non-features
+## Non-features
 
-- Distributed search is out of the scope of tantivy. That being said, tantivy is meant as a
+- Distributed search is out of the scope of Tantivy. That being said, Tantivy is a
 library upon which one could build a distributed search. Serializable/mergeable collector state for instance, 
-are within the scope of tantivy.
+are within the scope of Tantivy.
 
-# Supported OS and compiler
-
-Tantivy works on stable rust (>= 1.27) and supports Linux, MacOS and Windows.
 
 # Getting started
 
-- [tantivy's simple search example](https://tantivy-search.github.io/examples/basic_search.html)
+Tantivy works on stable Rust (>= 1.27) and supports Linux, MacOS, and Windows.
-- [tantivy-cli and its tutorial](https://github.com/tantivy-search/tantivy-cli).
+
-`tantivy-cli` is an actual command line interface that makes it easy for you to create a search engine,
+- [Tantivy's simple search example](https://tantivy-search.github.io/examples/basic_search.html)
-index documents and search via the CLI or a small server with a REST API.
+- [tantivy-cli and its tutorial](https://github.com/tantivy-search/tantivy-cli) - `tantivy-cli` is an actual command line interface that makes it easy for you to create a search engine,
-It will walk you through getting a wikipedia search engine up and running in a few minutes.
+index documents, and search via the CLI or a small server with a REST API.
-- [reference doc for the last released version](https://docs.rs/tantivy/)
+It walks you through getting a wikipedia search engine up and running in a few minutes.
+- [Reference doc for the last released version](https://docs.rs/tantivy/)
 
 # How can I support this project?
 
 There are many ways to support this project. 
 
-- Use tantivy and tell us about your experience on [gitter](https://gitter.im/tantivy-search/tantivy) or by email (paul.masurel@gmail.com)
+- Use Tantivy and tell us about your experience on [Gitter](https://gitter.im/tantivy-search/tantivy) or by email (paul.masurel@gmail.com)
 - Report bugs
 - Write a blog post
 - Help with documentation by asking questions or submitting PRs
-- Contribute code (you can join [our gitter](https://gitter.im/tantivy-search/tantivy) )
+- Contribute code (you can join [our Gitter](https://gitter.im/tantivy-search/tantivy))
-- Talk about tantivy around you
+- Talk about Tantivy around you
-- Drop a word on on [![Say Thanks!](https://img.shields.io/badge/Say%20Thanks-!-1EAEDB.svg)](https://saythanks.io/to/fulmicoton) or even [![Become a patron](https://c5.patreon.com/external/logo/become_a_patron_button.png)](https://www.patreon.com/fulmicoton)
+- [![Become a patron](https://c5.patreon.com/external/logo/become_a_patron_button.png)](https://www.patreon.com/fulmicoton)
 
 # Contributing code
 
-We use the GitHub Pull Request workflow - reference a GitHub ticket and/or include a comprehensive commit message when opening a PR.
+We use the GitHub Pull Request workflow: reference a GitHub ticket and/or include a comprehensive commit message when opening a PR.
 
 ## Clone and build locally
 
-Tantivy compiles on stable rust but requires `Rust >= 1.27`.
+Tantivy compiles on stable Rust but requires `Rust >= 1.27`.
-To check out and run tests, you can simply run :
+To check out and run tests, you can simply run:
 
 ```bash
     git clone https://github.com/tantivy-search/tantivy.git
@@ -108,7 +104,7 @@ To check out and run tests, you can simply run :
 ## Run tests
 
 Some tests will not run with just `cargo test` because of `fail-rs`.
-To run the tests exhaustively, run `./run-tests.sh`
+To run the tests exhaustively, run `./run-tests.sh`.
 
 ## Debug
 
@@ -116,13 +112,13 @@ You might find it useful to step through the programme with a debugger.
 
 ### A failing test
 
-Make sure you haven't run `cargo clean` after the most recent `cargo test` or `cargo build` to guarantee that `target/` dir exists. Use this bash script to find the most name of the most recent debug build of tantivy and run it under rust-gdb.
+Make sure you haven't run `cargo clean` after the most recent `cargo test` or `cargo build` to guarantee that the `target/` directory exists. Use this bash script to find the name of the most recent debug build of Tantivy and run it under `rust-gdb`:
 
 ```bash
 find target/debug/ -maxdepth 1 -executable -type f -name "tantivy*" -printf '%TY-%Tm-%Td %TT %p\n' | sort -r | cut -d " " -f 3 | xargs -I RECENT_DBG_TANTIVY rust-gdb RECENT_DBG_TANTIVY
 ```
 
-Now that you are in rust-gdb, you can set breakpoints on lines and methods that match your source-code and run the debug executable with flags that you normally pass to `cargo test` to like this
+Now that you are in `rust-gdb`, you can set breakpoints on lines and methods that match your source code and run the debug executable with flags that you normally pass to `cargo test` like this:
 
 ```bash
 $gdb run --test-threads 1 --test $NAME_OF_TEST
@@ -130,7 +126,7 @@ $gdb run --test-threads 1 --test $NAME_OF_TEST
 
 ### An example
 
-By default, rustc compiles everything in the `examples/` dir in debug mode. This makes it easy for you to make examples to reproduce bugs.
+By default, `rustc` compiles everything in the `examples/` directory in debug mode. This makes it easy for you to make examples to reproduce bugs:
 
 ```bash
 rust-gdb target/debug/examples/$EXAMPLE_NAME

appveyor.yml

@@ -18,5 +18,6 @@ install:
 build: false
 
 test_script:
-  - REM SET RUST_LOG=tantivy,test & cargo test --verbose --no-default-features --features mmap
+  - REM SET RUST_LOG=tantivy,test & cargo test --all --verbose --no-default-features --features lz4-compression --features mmap
+  - REM SET RUST_LOG=tantivy,test & cargo test test_store --verbose --no-default-features --features lz4-compression --features snappy-compression --features brotli-compression --features mmap
   - REM SET RUST_BACKTRACE=1 & cargo build --examples

benches/analyzer.rs

@@ -0,0 +1,22 @@
+use criterion::{criterion_group, criterion_main, Criterion};
+use tantivy::tokenizer::TokenizerManager;
+
+const ALICE_TXT: &'static str = include_str!("alice.txt");
+
+pub fn criterion_benchmark(c: &mut Criterion) {
+    let tokenizer_manager = TokenizerManager::default();
+    let tokenizer = tokenizer_manager.get("default").unwrap();
+    c.bench_function("default-tokenize-alice", |b| {
+        b.iter(|| {
+            let mut word_count = 0;
+            let mut token_stream = tokenizer.token_stream(ALICE_TXT);
+            while token_stream.advance() {
+                word_count += 1;
+            }
+            assert_eq!(word_count, 30_731);
+        })
+    });
+}
+
+criterion_group!(benches, criterion_benchmark);
+criterion_main!(benches);

bitpacker/Cargo.toml

@@ -0,0 +1,15 @@
+[package]
+name = "tantivy-bitpacker"
+version = "0.1.0"
+edition = "2018"
+authors = ["Paul Masurel <paul.masurel@gmail.com>"]
+license = "MIT"
+categories = []
+description = """Tantivy-sub crate: bitpacking"""
+repository = "https://github.com/tantivy-search/tantivy"
+keywords = []
+
+
+# See more keys and their definitions at https://doc.rust-lang.org/cargo/reference/manifest.html
+
+[dependencies]

bitpacker/benches/bench.rs

@@ -0,0 +1,33 @@
+#![feature(test)]
+
+extern crate test;
+
+#[cfg(test)]
+mod tests {
+    use tantivy_bitpacker::BlockedBitpacker;
+    use test::Bencher;
+    #[bench]
+    fn bench_blockedbitp_read(b: &mut Bencher) {
+        let mut blocked_bitpacker = BlockedBitpacker::new();
+        for val in 0..=21500 {
+            blocked_bitpacker.add(val * val);
+        }
+        b.iter(|| {
+            let mut out = 0;
+            for val in 0..=21500 {
+                out = blocked_bitpacker.get(val);
+            }
+            out
+        });
+    }
+    #[bench]
+    fn bench_blockedbitp_create(b: &mut Bencher) {
+        b.iter(|| {
+            let mut blocked_bitpacker = BlockedBitpacker::new();
+            for val in 0..=21500 {
+                blocked_bitpacker.add(val * val);
+            }
+            blocked_bitpacker
+        });
+    }
+}

bitpacker/src/bitpacker.rs

@@ -1,12 +1,14 @@
-use byteorder::{ByteOrder, LittleEndian, WriteBytesExt};
+use std::{convert::TryInto, io};
-use std::io;
-use std::ops::Deref;
 
-pub(crate) struct BitPacker {
+pub struct BitPacker {
     mini_buffer: u64,
     mini_buffer_written: usize,
 }
-
+impl Default for BitPacker {
+    fn default() -> Self {
+        BitPacker::new()
+    }
+}
 impl BitPacker {
     pub fn new() -> BitPacker {
         BitPacker {
@@ -15,6 +17,7 @@ impl BitPacker {
         }
     }
 
+    #[inline]
     pub fn write<TWrite: io::Write>(
         &mut self,
         val: u64,
@@ -25,14 +28,14 @@ impl BitPacker {
         let num_bits = num_bits as usize;
         if self.mini_buffer_written + num_bits > 64 {
             self.mini_buffer |= val_u64.wrapping_shl(self.mini_buffer_written as u32);
-            output.write_u64::<LittleEndian>(self.mini_buffer)?;
+            output.write_all(self.mini_buffer.to_le_bytes().as_ref())?;
             self.mini_buffer = val_u64.wrapping_shr((64 - self.mini_buffer_written) as u32);
             self.mini_buffer_written = self.mini_buffer_written + num_bits - 64;
         } else {
             self.mini_buffer |= val_u64 << self.mini_buffer_written;
             self.mini_buffer_written += num_bits;
             if self.mini_buffer_written == 64 {
-                output.write_u64::<LittleEndian>(self.mini_buffer)?;
+                output.write_all(self.mini_buffer.to_le_bytes().as_ref())?;
                 self.mini_buffer_written = 0;
                 self.mini_buffer = 0u64;
             }
@@ -43,9 +46,8 @@ impl BitPacker {
     pub fn flush<TWrite: io::Write>(&mut self, output: &mut TWrite) -> io::Result<()> {
         if self.mini_buffer_written > 0 {
             let num_bytes = (self.mini_buffer_written + 7) / 8;
-            let mut arr: [u8; 8] = [0u8; 8];
+            let bytes = self.mini_buffer.to_le_bytes();
-            LittleEndian::write_u64(&mut arr, self.mini_buffer);
+            output.write_all(&bytes[..num_bytes])?;
-            output.write_all(&arr[..num_bytes])?;
             self.mini_buffer_written = 0;
         }
         Ok(())
@@ -60,20 +62,13 @@ impl BitPacker {
 }
 
 #[derive(Clone)]
-pub struct BitUnpacker<Data>
+pub struct BitUnpacker {
-where
-    Data: Deref<Target = [u8]>,
-{
     num_bits: u64,
     mask: u64,
-    data: Data,
 }
 
-impl<Data> BitUnpacker<Data>
+impl BitUnpacker {
-where
+    pub fn new(num_bits: u8) -> BitUnpacker {
-    Data: Deref<Target = [u8]>,
-{
-    pub fn new(data: Data, num_bits: u8) -> BitUnpacker<Data> {
         let mask: u64 = if num_bits == 64 {
             !0u64
         } else {
@@ -82,15 +77,14 @@ where
         BitUnpacker {
             num_bits: u64::from(num_bits),
             mask,
-            data,
         }
     }
 
-    pub fn get(&self, idx: u64) -> u64 {
+    #[inline]
+    pub fn get(&self, idx: u64, data: &[u8]) -> u64 {
         if self.num_bits == 0 {
             return 0u64;
         }
-        let data: &[u8] = &*self.data;
         let num_bits = self.num_bits;
         let mask = self.mask;
         let addr_in_bits = idx * num_bits;
@@ -100,7 +94,10 @@ where
             addr + 8 <= data.len() as u64,
             "The fast field field should have been padded with 7 bytes."
         );
-        let val_unshifted_unmasked: u64 = LittleEndian::read_u64(&data[(addr as usize)..]);
+        let bytes: [u8; 8] = (&data[(addr as usize)..(addr as usize) + 8])
+            .try_into()
+            .unwrap();
+        let val_unshifted_unmasked: u64 = u64::from_le_bytes(bytes);
         let val_shifted = (val_unshifted_unmasked >> bit_shift) as u64;
         val_shifted & mask
     }
@@ -110,7 +107,7 @@ where
 mod test {
     use super::{BitPacker, BitUnpacker};
 
-    fn create_fastfield_bitpacker(len: usize, num_bits: u8) -> (BitUnpacker<Vec<u8>>, Vec<u64>) {
+    fn create_fastfield_bitpacker(len: usize, num_bits: u8) -> (BitUnpacker, Vec<u64>, Vec<u8>) {
         let mut data = Vec::new();
         let mut bitpacker = BitPacker::new();
         let max_val: u64 = (1u64 << num_bits as u64) - 1u64;
@@ -122,14 +119,14 @@ mod test {
         }
         bitpacker.close(&mut data).unwrap();
         assert_eq!(data.len(), ((num_bits as usize) * len + 7) / 8 + 7);
-        let bitunpacker = BitUnpacker::new(data, num_bits);
+        let bitunpacker = BitUnpacker::new(num_bits);
-        (bitunpacker, vals)
+        (bitunpacker, vals, data)
     }
 
     fn test_bitpacker_util(len: usize, num_bits: u8) {
-        let (bitunpacker, vals) = create_fastfield_bitpacker(len, num_bits);
+        let (bitunpacker, vals, data) = create_fastfield_bitpacker(len, num_bits);
         for (i, val) in vals.iter().enumerate() {
-            assert_eq!(bitunpacker.get(i as u64), *val);
+            assert_eq!(bitunpacker.get(i as u64, &data), *val);
         }
     }
 

bitpacker/src/blocked_bitpacker.rs

@@ -0,0 +1,178 @@
+use crate::{minmax, BitUnpacker};
+
+use super::{bitpacker::BitPacker, compute_num_bits};
+
+const BLOCK_SIZE: usize = 128;
+
+/// `BlockedBitpacker` compresses data in blocks of
+/// 128 elements, while keeping an index on it
+///
+#[derive(Debug, Clone)]
+pub struct BlockedBitpacker {
+    // bitpacked blocks
+    compressed_blocks: Vec<u8>,
+    // uncompressed data, collected until BLOCK_SIZE
+    buffer: Vec<u64>,
+    offset_and_bits: Vec<BlockedBitpackerEntryMetaData>,
+}
+impl Default for BlockedBitpacker {
+    fn default() -> Self {
+        BlockedBitpacker::new()
+    }
+}
+
+/// `BlockedBitpackerEntryMetaData` encodes the
+/// offset and bit_width into a u64 bit field
+///
+/// This saves some space, since 7byte is more
+/// than enough and also keeps the access fast
+/// because of alignment
+#[derive(Debug, Clone, Default)]
+struct BlockedBitpackerEntryMetaData {
+    encoded: u64,
+    base_value: u64,
+}
+
+impl BlockedBitpackerEntryMetaData {
+    fn new(offset: u64, num_bits: u8, base_value: u64) -> Self {
+        let encoded = offset | (num_bits as u64) << (64 - 8);
+        Self {
+            encoded,
+            base_value,
+        }
+    }
+    fn offset(&self) -> u64 {
+        (self.encoded << 8) >> 8
+    }
+    fn num_bits(&self) -> u8 {
+        (self.encoded >> 56) as u8
+    }
+    fn base_value(&self) -> u64 {
+        self.base_value
+    }
+}
+
+#[test]
+fn metadata_test() {
+    let meta = BlockedBitpackerEntryMetaData::new(50000, 6, 40000);
+    assert_eq!(meta.offset(), 50000);
+    assert_eq!(meta.num_bits(), 6);
+}
+
+impl BlockedBitpacker {
+    pub fn new() -> Self {
+        let mut compressed_blocks = vec![];
+        compressed_blocks.resize(8, 0);
+        Self {
+            compressed_blocks,
+            buffer: vec![],
+            offset_and_bits: vec![],
+        }
+    }
+
+    /// The memory used (inclusive childs)
+    pub fn mem_usage(&self) -> usize {
+        std::mem::size_of::<BlockedBitpacker>()
+            + self.compressed_blocks.capacity()
+            + self.offset_and_bits.capacity()
+                * std::mem::size_of_val(&self.offset_and_bits.get(0).cloned().unwrap_or_default())
+            + self.buffer.capacity()
+                * std::mem::size_of_val(&self.buffer.get(0).cloned().unwrap_or_default())
+    }
+
+    #[inline]
+    pub fn add(&mut self, val: u64) {
+        self.buffer.push(val);
+        if self.buffer.len() == BLOCK_SIZE as usize {
+            self.flush();
+        }
+    }
+
+    pub fn flush(&mut self) {
+        if let Some((min_value, max_value)) = minmax(self.buffer.iter()) {
+            let mut bit_packer = BitPacker::new();
+            let num_bits_block = compute_num_bits(*max_value - min_value);
+            // todo performance: the padding handling could be done better, e.g. use a slice and
+            // return num_bytes written from bitpacker
+            self.compressed_blocks
+                .resize(self.compressed_blocks.len() - 8, 0); // remove padding for bitpacker
+            let offset = self.compressed_blocks.len() as u64;
+            // todo performance: for some bit_width we
+            // can encode multiple vals into the
+            // mini_buffer before checking to flush
+            // (to be done in BitPacker)
+            for val in self.buffer.iter() {
+                bit_packer
+                    .write(
+                        *val - min_value,
+                        num_bits_block,
+                        &mut self.compressed_blocks,
+                    )
+                    .expect("cannot write bitpacking to output"); // write to in memory can't fail
+            }
+            bit_packer.flush(&mut self.compressed_blocks).unwrap();
+            self.offset_and_bits
+                .push(BlockedBitpackerEntryMetaData::new(
+                    offset,
+                    num_bits_block,
+                    *min_value,
+                ));
+
+            self.buffer.clear();
+            self.compressed_blocks
+                .resize(self.compressed_blocks.len() + 8, 0); // add padding for bitpacker
+        }
+    }
+    #[inline]
+    pub fn get(&self, idx: usize) -> u64 {
+        let metadata_pos = idx / BLOCK_SIZE as usize;
+        let pos_in_block = idx % BLOCK_SIZE as usize;
+        if let Some(metadata) = self.offset_and_bits.get(metadata_pos) {
+            let unpacked = BitUnpacker::new(metadata.num_bits()).get(
+                pos_in_block as u64,
+                &self.compressed_blocks[metadata.offset() as usize..],
+            );
+            unpacked + metadata.base_value()
+        } else {
+            self.buffer[pos_in_block]
+        }
+    }
+
+    pub fn iter(&self) -> impl Iterator<Item = u64> + '_ {
+        // todo performance: we could decompress a whole block and cache it instead
+        let bitpacked_elems = self.offset_and_bits.len() * BLOCK_SIZE;
+        let iter = (0..bitpacked_elems)
+            .map(move |idx| self.get(idx))
+            .chain(self.buffer.iter().cloned());
+        iter
+    }
+}
+
+#[cfg(test)]
+mod tests {
+    use super::*;
+    #[test]
+    fn blocked_bitpacker_empty() {
+        let blocked_bitpacker = BlockedBitpacker::new();
+        assert_eq!(blocked_bitpacker.iter().collect::<Vec<u64>>(), vec![]);
+    }
+    #[test]
+    fn blocked_bitpacker_one() {
+        let mut blocked_bitpacker = BlockedBitpacker::new();
+        blocked_bitpacker.add(50000);
+        assert_eq!(blocked_bitpacker.get(0), 50000);
+        assert_eq!(blocked_bitpacker.iter().collect::<Vec<u64>>(), vec![50000]);
+    }
+    #[test]
+    fn blocked_bitpacker_test() {
+        let mut blocked_bitpacker = BlockedBitpacker::new();
+        for val in 0..21500 {
+            blocked_bitpacker.add(val);
+        }
+        for val in 0..21500 {
+            assert_eq!(blocked_bitpacker.get(val as usize), val);
+        }
+        assert_eq!(blocked_bitpacker.iter().count(), 21500);
+        assert_eq!(blocked_bitpacker.iter().last().unwrap(), 21499);
+    }
+}

bitpacker/src/lib.rs

@@ -0,0 +1,52 @@
+mod bitpacker;
+mod blocked_bitpacker;
+
+pub use crate::bitpacker::BitPacker;
+pub use crate::bitpacker::BitUnpacker;
+pub use crate::blocked_bitpacker::BlockedBitpacker;
+
+/// Computes the number of bits that will be used for bitpacking.
+///
+/// In general the target is the minimum number of bits
+/// required to express the amplitude given in argument.
+///
+/// e.g. If the amplitude is 10, we can store all ints on simply 4bits.
+///
+/// The logic is slightly more convoluted here as for optimization
+/// reasons, we want to ensure that a value spawns over at most 8 bytes
+/// of aligned bytes.
+///
+/// Spanning over 9 bytes is possible for instance, if we do
+/// bitpacking with an amplitude of 63 bits.
+/// In this case, the second int will start on bit
+/// 63 (which belongs to byte 7) and ends at byte 15;
+/// Hence 9 bytes (from byte 7 to byte 15 included).
+///
+/// To avoid this, we force the number of bits to 64bits
+/// when the result is greater than `64-8 = 56 bits`.
+///
+/// Note that this only affects rare use cases spawning over
+/// a very large range of values. Even in this case, it results
+/// in an extra cost of at most 12% compared to the optimal
+/// number of bits.
+pub fn compute_num_bits(n: u64) -> u8 {
+    let amplitude = (64u32 - n.leading_zeros()) as u8;
+    if amplitude <= 64 - 8 {
+        amplitude
+    } else {
+        64
+    }
+}
+
+pub fn minmax<I, T>(mut vals: I) -> Option<(T, T)>
+where
+    I: Iterator<Item = T>,
+    T: Copy + Ord,
+{
+    if let Some(first_el) = vals.next() {
+        return Some(vals.fold((first_el, first_el), |(min_val, max_val), el| {
+            (min_val.min(el), max_val.max(el))
+        }));
+    }
+    None
+}

ci/script.sh

@@ -7,7 +7,7 @@ set -ex
 main() {
     if [ ! -z $CODECOV ]; then
         echo "Codecov"
-        cargo build --verbose && cargo coverage --verbose && bash <(curl -s https://codecov.io/bash) -s target/kcov
+        cargo build --verbose && cargo coverage --verbose --all && bash <(curl -s https://codecov.io/bash) -s target/kcov
     else
         echo "Build"
         cross build --target $TARGET
@@ -15,7 +15,8 @@ main() {
             return
         fi
         echo "Test"
-        cross test --target $TARGET --no-default-features --features mmap -- --test-threads 1
+        cross test --target $TARGET --no-default-features --features mmap
+        cross test --target $TARGET --no-default-features --features mmap query-grammar
     fi
     for example in $(ls examples/*.rs)
     do

examples/basic_search.rs

@@ -5,26 +5,23 @@
 //
 // We will :
 // - define our schema
-// = create an index in a directory
+// - create an index in a directory
-// - index few documents in our index
+// - index a few documents into our index
-// - search for the best document matchings "sea whale"
+// - search for the best document matching a basic query
-// - retrieve the best document original content.
+// - retrieve the best document's original content.
 
 // ---
 // Importing tantivy...
-#[macro_use]
-extern crate tantivy;
 use tantivy::collector::TopDocs;
 use tantivy::query::QueryParser;
 use tantivy::schema::*;
-use tantivy::Index;
+use tantivy::{doc, Index, ReloadPolicy};
-use tantivy::ReloadPolicy;
+use tempfile::TempDir;
-use tempdir::TempDir;
 
 fn main() -> tantivy::Result<()> {
     // Let's create a temporary directory for the
     // sake of this example
-    let index_path = TempDir::new("tantivy_example_dir")?;
+    let index_path = TempDir::new()?;
 
     // # Defining the schema
     //
@@ -33,7 +30,7 @@ fn main() -> tantivy::Result<()> {
     // and for each field, its type and "the way it should
     // be indexed".
 
-    // first we need to define a schema ...
+    // First we need to define a schema ...
     let mut schema_builder = Schema::builder();
 
     // Our first field is title.
@@ -48,7 +45,7 @@ fn main() -> tantivy::Result<()> {
     //
     // `STORED` means that the field will also be saved
     // in a compressed, row-oriented key-value store.
-    // This store is useful to reconstruct the
+    // This store is useful for reconstructing the
     // documents that were selected during the search phase.
     schema_builder.add_text_field("title", TEXT | STORED);
 
@@ -57,8 +54,7 @@ fn main() -> tantivy::Result<()> {
     // need to be able to be able to retrieve it
     // for our application.
     //
-    // We can make our index lighter and
+    // We can make our index lighter by omitting the `STORED` flag.
-    // by omitting `STORED` flag.
     schema_builder.add_text_field("body", TEXT);
 
     let schema = schema_builder.build();
@@ -71,7 +67,7 @@ fn main() -> tantivy::Result<()> {
     // with our schema in the directory.
     let index = Index::create_in_dir(&index_path, schema.clone())?;
 
-    // To insert document we need an index writer.
+    // To insert a document we will need an index writer.
     // There must be only one writer at a time.
     // This single `IndexWriter` is already
     // multithreaded.
@@ -116,18 +112,6 @@ fn main() -> tantivy::Result<()> {
             limbs and branches that arch over the pool"
     ));
 
-    index_writer.add_document(doc!(
-    title => "Of Mice and Men",
-    body => "A few miles south of Soledad, the Salinas River drops in close to the hillside \
-            bank and runs deep and green. The water is warm too, for it has slipped twinkling \
-            over the yellow sands in the sunlight before reaching the narrow pool. On one \
-            side of the river the golden foothill slopes curve up to the strong and rocky \
-            Gabilan Mountains, but on the valley side the water is lined with trees—willows \
-            fresh and green with every spring, carrying in their lower leaf junctures the \
-            debris of the winter’s flooding; and sycamores with mottled, white, recumbent \
-            limbs and branches that arch over the pool"
-    ));
-
     // Multivalued field just need to be repeated.
     index_writer.add_document(doc!(
     title => "Frankenstein",
@@ -149,8 +133,8 @@ fn main() -> tantivy::Result<()> {
     // At this point our documents are not searchable.
     //
     //
-    // We need to call .commit() explicitly to force the
+    // We need to call `.commit()` explicitly to force the
-    // index_writer to finish processing the documents in the queue,
+    // `index_writer` to finish processing the documents in the queue,
     // flush the current index to the disk, and advertise
     // the existence of new documents.
     //
@@ -162,14 +146,14 @@ fn main() -> tantivy::Result<()> {
     // persistently indexed.
     //
     // In the scenario of a crash or a power failure,
-    // tantivy behaves as if has rolled back to its last
+    // tantivy behaves as if it has rolled back to its last
     // commit.
 
     // # Searching
     //
     // ### Searcher
     //
-    // A reader is required to get search the index.
+    // A reader is required first in order to search an index.
     // It acts as a `Searcher` pool that reloads itself,
     // depending on a `ReloadPolicy`.
     //
@@ -185,7 +169,7 @@ fn main() -> tantivy::Result<()> {
 
     // We now need to acquire a searcher.
     //
-    // A searcher points to snapshotted, immutable version of the index.
+    // A searcher points to a snapshotted, immutable version of the index.
     //
     // Some search experience might require more than
     // one query. Using the same searcher ensures that all of these queries will run on the
@@ -205,7 +189,7 @@ fn main() -> tantivy::Result<()> {
     // in both title and body.
     let query_parser = QueryParser::for_index(&index, vec![title, body]);
 
-    // QueryParser may fail if the query is not in the right
+    // `QueryParser` may fail if the query is not in the right
     // format. For user facing applications, this can be a problem.
     // A ticket has been opened regarding this problem.
     let query = query_parser.parse_query("sea whale")?;
@@ -221,7 +205,7 @@ fn main() -> tantivy::Result<()> {
     //
     // We are not interested in all of the documents but
     // only in the top 10. Keeping track of our top 10 best documents
-    // is the role of the TopDocs.
+    // is the role of the `TopDocs` collector.
 
     // We can now perform our query.
     let top_docs = searcher.search(&query, &TopDocs::with_limit(10))?;

examples/custom_collector.rs

@@ -9,15 +9,12 @@
 
 // ---
 // Importing tantivy...
-#[macro_use]
-extern crate tantivy;
 use tantivy::collector::{Collector, SegmentCollector};
-use tantivy::fastfield::FastFieldReader;
+use tantivy::fastfield::{DynamicFastFieldReader, FastFieldReader};
 use tantivy::query::QueryParser;
 use tantivy::schema::Field;
 use tantivy::schema::{Schema, FAST, INDEXED, TEXT};
-use tantivy::SegmentReader;
+use tantivy::{doc, Index, Score, SegmentReader};
-use tantivy::{Index, TantivyError};
 
 #[derive(Default)]
 struct Stats {
@@ -75,16 +72,7 @@ impl Collector for StatsCollector {
         _segment_local_id: u32,
         segment_reader: &SegmentReader,
     ) -> tantivy::Result<StatsSegmentCollector> {
-        let fast_field_reader = segment_reader
+        let fast_field_reader = segment_reader.fast_fields().u64(self.field)?;
-            .fast_fields()
-            .u64(self.field)
-            .ok_or_else(|| {
-                let field_name = segment_reader.schema().get_field_name(self.field);
-                TantivyError::SchemaError(format!(
-                    "Field {:?} is not a u64 fast field.",
-                    field_name
-                ))
-            })?;
         Ok(StatsSegmentCollector {
             fast_field_reader,
             stats: Stats::default(),
@@ -110,14 +98,14 @@ impl Collector for StatsCollector {
 }
 
 struct StatsSegmentCollector {
-    fast_field_reader: FastFieldReader<u64>,
+    fast_field_reader: DynamicFastFieldReader<u64>,
     stats: Stats,
 }
 
 impl SegmentCollector for StatsSegmentCollector {
     type Fruit = Option<Stats>;
 
-    fn collect(&mut self, doc: u32, _score: f32) {
+    fn collect(&mut self, doc: u32, _score: Score) {
         let value = self.fast_field_reader.get(doc) as f64;
         self.stats.count += 1;
         self.stats.sum += value;

examples/custom_tokenizer.rs

@@ -2,14 +2,11 @@
 //
 // In this example, we'll see how to define a tokenizer pipeline
 // by aligning a bunch of `TokenFilter`.
-
-#[macro_use]
-extern crate tantivy;
 use tantivy::collector::TopDocs;
 use tantivy::query::QueryParser;
 use tantivy::schema::*;
 use tantivy::tokenizer::NgramTokenizer;
-use tantivy::Index;
+use tantivy::{doc, Index};
 
 fn main() -> tantivy::Result<()> {
     // # Defining the schema

examples/deleting_updating_documents.rs

@@ -8,13 +8,10 @@
 //
 // ---
 // Importing tantivy...
-#[macro_use]
-extern crate tantivy;
 use tantivy::collector::TopDocs;
 use tantivy::query::TermQuery;
 use tantivy::schema::*;
-use tantivy::Index;
+use tantivy::{doc, Index, IndexReader};
-use tantivy::IndexReader;
 
 // A simple helper function to fetch a single document
 // given its id from our index.
@@ -93,7 +90,7 @@ fn main() -> tantivy::Result<()> {
 
     let frankenstein_isbn = Term::from_field_text(isbn, "978-9176370711");
 
-    // Oops our frankenstein doc seems mispelled
+    // Oops our frankenstein doc seems misspelled
     let frankenstein_doc_misspelled = extract_doc_given_isbn(&reader, &frankenstein_isbn)?.unwrap();
     assert_eq!(
         schema.to_json(&frankenstein_doc_misspelled),

examples/faceted_search.rs

@@ -12,67 +12,101 @@
 
 // ---
 // Importing tantivy...
-#[macro_use]
-extern crate tantivy;
 use tantivy::collector::FacetCollector;
-use tantivy::query::AllQuery;
+use tantivy::query::{AllQuery, TermQuery};
 use tantivy::schema::*;
-use tantivy::Index;
+use tantivy::{doc, Index};
 
 fn main() -> tantivy::Result<()> {
-    // Let's create a temporary directory for the
+    // Let's create a temporary directory for the sake of this example
-    // sake of this example
-    let index_path = TempDir::new("tantivy_facet_example_dir")?;
     let mut schema_builder = Schema::builder();
 
-    schema_builder.add_text_field("name", TEXT | STORED);
+    let name = schema_builder.add_text_field("felin_name", TEXT | STORED);
-
+    // this is our faceted field: its scientific classification
-    // this is our faceted field
+    let classification = schema_builder.add_facet_field("classification", INDEXED);
-    schema_builder.add_facet_field("tags");
 
     let schema = schema_builder.build();
+    let index = Index::create_in_ram(schema);
 
-    let index = Index::create_in_dir(&index_path, schema.clone())?;
+    let mut index_writer = index.writer(30_000_000)?;
-
-    let mut index_writer = index.writer(50_000_000)?;
-
-    let name = schema.get_field("name").unwrap();
-    let tags = schema.get_field("tags").unwrap();
 
     // For convenience, tantivy also comes with a macro to
     // reduce the boilerplate above.
     index_writer.add_document(doc!(
-        name => "the ditch",
+        name => "Cat",
-        tags => Facet::from("/pools/north")
+        classification => Facet::from("/Felidae/Felinae/Felis")
     ));
-
     index_writer.add_document(doc!(
-        name => "little stacey",
+        name => "Canada lynx",
-        tags => Facet::from("/pools/south")
+        classification => Facet::from("/Felidae/Felinae/Lynx")
+    ));
+    index_writer.add_document(doc!(
+        name => "Cheetah",
+        classification => Facet::from("/Felidae/Felinae/Acinonyx")
+    ));
+    index_writer.add_document(doc!(
+        name => "Tiger",
+        classification => Facet::from("/Felidae/Pantherinae/Panthera")
+    ));
+    index_writer.add_document(doc!(
+        name => "Lion",
+        classification => Facet::from("/Felidae/Pantherinae/Panthera")
+    ));
+    index_writer.add_document(doc!(
+        name => "Jaguar",
+        classification => Facet::from("/Felidae/Pantherinae/Panthera")
+    ));
+    index_writer.add_document(doc!(
+        name => "Sunda clouded leopard",
+        classification => Facet::from("/Felidae/Pantherinae/Neofelis")
+    ));
+    index_writer.add_document(doc!(
+        name => "Fossa",
+        classification => Facet::from("/Eupleridae/Cryptoprocta")
     ));
-
     index_writer.commit()?;
 
     let reader = index.reader()?;
-
     let searcher = reader.searcher();
+    {
+        let mut facet_collector = FacetCollector::for_field(classification);
+        facet_collector.add_facet("/Felidae");
+        let facet_counts = searcher.search(&AllQuery, &facet_collector)?;
+        // This lists all of the facet counts, right below "/Felidae".
+        let facets: Vec<(&Facet, u64)> = facet_counts.get("/Felidae").collect();
+        assert_eq!(
+            facets,
+            vec![
+                (&Facet::from("/Felidae/Felinae"), 3),
+                (&Facet::from("/Felidae/Pantherinae"), 4),
+            ]
+        );
+    }
 
-    let mut facet_collector = FacetCollector::for_field(tags);
+    // Facets are also searchable.
-    facet_collector.add_facet("/pools");
+    //
+    // For instance a common UI pattern is to allow the user someone to click on a facet link
+    // (e.g: `Pantherinae`) to drill down and filter the current result set with this subfacet.
+    //
+    // The search would then look as follows.
 
-    let facet_counts = searcher.search(&AllQuery, &facet_collector).unwrap();
+    // Check the reference doc for different ways to create a `Facet` object.
-
+    {
-    // This lists all of the facet counts
+        let facet = Facet::from("/Felidae/Pantherinae");
-    let facets: Vec<(&Facet, u64)> = facet_counts.get("/pools").collect();
+        let facet_term = Term::from_facet(classification, &facet);
-    assert_eq!(
+        let facet_term_query = TermQuery::new(facet_term, IndexRecordOption::Basic);
-        facets,
+        let mut facet_collector = FacetCollector::for_field(classification);
-        vec![
+        facet_collector.add_facet("/Felidae/Pantherinae");
-            (&Facet::from("/pools/north"), 1),
+        let facet_counts = searcher.search(&facet_term_query, &facet_collector)?;
-            (&Facet::from("/pools/south"), 1),
+        let facets: Vec<(&Facet, u64)> = facet_counts.get("/Felidae/Pantherinae").collect();
-        ]
+        assert_eq!(
-    );
+            facets,
+            vec![
+                (&Facet::from("/Felidae/Pantherinae/Neofelis"), 1),
+                (&Facet::from("/Felidae/Pantherinae/Panthera"), 3),
+            ]
+        );
+    }
 
     Ok(())
 }
-
-use tempdir::TempDir;

examples/faceted_search_with_tweaked_score.rs

@@ -0,0 +1,98 @@
+use std::collections::HashSet;
+use tantivy::collector::TopDocs;
+use tantivy::doc;
+use tantivy::query::BooleanQuery;
+use tantivy::schema::*;
+use tantivy::{DocId, Index, Score, SegmentReader};
+
+fn main() -> tantivy::Result<()> {
+    let mut schema_builder = Schema::builder();
+
+    let title = schema_builder.add_text_field("title", STORED);
+    let ingredient = schema_builder.add_facet_field("ingredient", INDEXED);
+
+    let schema = schema_builder.build();
+    let index = Index::create_in_ram(schema.clone());
+
+    let mut index_writer = index.writer(30_000_000)?;
+
+    index_writer.add_document(doc!(
+        title => "Fried egg",
+        ingredient => Facet::from("/ingredient/egg"),
+        ingredient => Facet::from("/ingredient/oil"),
+    ));
+    index_writer.add_document(doc!(
+        title => "Scrambled egg",
+        ingredient => Facet::from("/ingredient/egg"),
+        ingredient => Facet::from("/ingredient/butter"),
+        ingredient => Facet::from("/ingredient/milk"),
+        ingredient => Facet::from("/ingredient/salt"),
+    ));
+    index_writer.add_document(doc!(
+        title => "Egg rolls",
+        ingredient => Facet::from("/ingredient/egg"),
+        ingredient => Facet::from("/ingredient/garlic"),
+        ingredient => Facet::from("/ingredient/salt"),
+        ingredient => Facet::from("/ingredient/oil"),
+        ingredient => Facet::from("/ingredient/tortilla-wrap"),
+        ingredient => Facet::from("/ingredient/mushroom"),
+    ));
+    index_writer.commit()?;
+
+    let reader = index.reader()?;
+    let searcher = reader.searcher();
+    {
+        let facets = vec![
+            Facet::from("/ingredient/egg"),
+            Facet::from("/ingredient/oil"),
+            Facet::from("/ingredient/garlic"),
+            Facet::from("/ingredient/mushroom"),
+        ];
+        let query = BooleanQuery::new_multiterms_query(
+            facets
+                .iter()
+                .map(|key| Term::from_facet(ingredient, &key))
+                .collect(),
+        );
+        let top_docs_by_custom_score =
+            TopDocs::with_limit(2).tweak_score(move |segment_reader: &SegmentReader| {
+                let ingredient_reader = segment_reader.facet_reader(ingredient).unwrap();
+                let facet_dict = ingredient_reader.facet_dict();
+
+                let query_ords: HashSet<u64> = facets
+                    .iter()
+                    .filter_map(|key| facet_dict.term_ord(key.encoded_str()).unwrap())
+                    .collect();
+
+                let mut facet_ords_buffer: Vec<u64> = Vec::with_capacity(20);
+
+                move |doc: DocId, original_score: Score| {
+                    ingredient_reader.facet_ords(doc, &mut facet_ords_buffer);
+                    let missing_ingredients = facet_ords_buffer
+                        .iter()
+                        .filter(|ord| !query_ords.contains(ord))
+                        .count();
+                    let tweak = 1.0 / 4_f32.powi(missing_ingredients as i32);
+
+                    original_score * tweak
+                }
+            });
+        let top_docs = searcher.search(&query, &top_docs_by_custom_score)?;
+
+        let titles: Vec<String> = top_docs
+            .iter()
+            .map(|(_, doc_id)| {
+                searcher
+                    .doc(*doc_id)
+                    .unwrap()
+                    .get_first(title)
+                    .unwrap()
+                    .text()
+                    .unwrap()
+                    .to_owned()
+            })
+            .collect();
+        assert_eq!(titles, vec!["Fried egg", "Egg rolls"]);
+    }
+    Ok(())
+}

examples/integer_range_search.rs

@@ -2,14 +2,10 @@
 //
 // Below is an example of creating an indexed integer field in your schema
 // You can use RangeQuery to get a Count of all occurrences in a given range.
-
-#[macro_use]
-extern crate tantivy;
 use tantivy::collector::Count;
 use tantivy::query::RangeQuery;
 use tantivy::schema::{Schema, INDEXED};
-use tantivy::Index;
+use tantivy::{doc, Index, Result};
-use tantivy::Result;
 
 fn run() -> Result<()> {
     // For the sake of simplicity, this schema will only have 1 field

examples/iterating_docs_and_positions.rs

@@ -1,4 +1,4 @@
-// # Iterating docs and positioms.
+// # Iterating docs and positions.
 //
 // At its core of tantivy, relies on a data structure
 // called an inverted index.
@@ -9,11 +9,8 @@
 
 // ---
 // Importing tantivy...
-#[macro_use]
-extern crate tantivy;
 use tantivy::schema::*;
-use tantivy::Index;
+use tantivy::{doc, DocSet, Index, Postings, TERMINATED};
-use tantivy::{DocId, DocSet, Postings};
 
 fn main() -> tantivy::Result<()> {
     // We first create a schema for the sake of the
@@ -48,7 +45,7 @@ fn main() -> tantivy::Result<()> {
         // Inverted index stands for the combination of
         // - the term dictionary
         // - the inverted lists associated to each terms and their positions
-        let inverted_index = segment_reader.inverted_index(title);
+        let inverted_index = segment_reader.inverted_index(title)?;
 
         // A `Term` is a text token associated with a field.
         // Let's go through all docs containing the term `title:the` and access their position
@@ -61,16 +58,15 @@ fn main() -> tantivy::Result<()> {
         // If you don't need all this information, you may get better performance by decompressing less
         // information.
         if let Some(mut segment_postings) =
-            inverted_index.read_postings(&term_the, IndexRecordOption::WithFreqsAndPositions)
+            inverted_index.read_postings(&term_the, IndexRecordOption::WithFreqsAndPositions)?
         {
             // this buffer will be used to request for positions
             let mut positions: Vec<u32> = Vec::with_capacity(100);
-            while segment_postings.advance() {
+            let mut doc_id = segment_postings.doc();
-                // the number of time the term appears in the document.
+            while doc_id != TERMINATED {
-                let doc_id: DocId = segment_postings.doc(); //< do not try to access this before calling advance once.
-
                 // This MAY contains deleted documents as well.
                 if segment_reader.is_deleted(doc_id) {
+                    doc_id = segment_postings.advance();
                     continue;
                 }
 
@@ -89,6 +85,7 @@ fn main() -> tantivy::Result<()> {
                 // Doc 2: TermFreq 1: [0]
                 // ```
                 println!("Doc {}: TermFreq {}: {:?}", doc_id, term_freq, positions);
+                doc_id = segment_postings.advance();
             }
         }
     }
@@ -109,7 +106,7 @@ fn main() -> tantivy::Result<()> {
         // Inverted index stands for the combination of
         // - the term dictionary
         // - the inverted lists associated to each terms and their positions
-        let inverted_index = segment_reader.inverted_index(title);
+        let inverted_index = segment_reader.inverted_index(title)?;
 
         // This segment posting object is like a cursor over the documents matching the term.
         // The `IndexRecordOption` arguments tells tantivy we will be interested in both term frequencies
@@ -118,13 +115,18 @@ fn main() -> tantivy::Result<()> {
         // If you don't need all this information, you may get better performance by decompressing less
         // information.
         if let Some(mut block_segment_postings) =
-            inverted_index.read_block_postings(&term_the, IndexRecordOption::Basic)
+            inverted_index.read_block_postings(&term_the, IndexRecordOption::Basic)?
         {
-            while block_segment_postings.advance() {
+            loop {
+                let docs = block_segment_postings.docs();
+                if docs.is_empty() {
+                    break;
+                }
                 // Once again these docs MAY contains deleted documents as well.
                 let docs = block_segment_postings.docs();
                 // Prints `Docs [0, 2].`
                 println!("Docs {:?}", docs);
+                block_segment_postings.advance();
             }
         }
     }

examples/multiple_producer.rs

@@ -25,14 +25,11 @@
 
 // ---
 // Importing tantivy...
-#[macro_use]
-extern crate tantivy;
 use std::sync::{Arc, RwLock};
 use std::thread;
 use std::time::Duration;
 use tantivy::schema::{Schema, STORED, TEXT};
-use tantivy::Opstamp;
+use tantivy::{doc, Index, IndexWriter, Opstamp};
-use tantivy::{Index, IndexWriter};
 
 fn main() -> tantivy::Result<()> {
     // # Defining the schema
@@ -49,10 +46,9 @@ fn main() -> tantivy::Result<()> {
     thread::spawn(move || {
         // we index 100 times the document... for the sake of the example.
         for i in 0..100 {
-            let opstamp = {
+            let opstamp = index_writer_clone_1
-                // A read lock is sufficient here.
+                .read().unwrap() //< A read lock is sufficient here.
-                let index_writer_rlock = index_writer_clone_1.read().unwrap();
+                .add_document(
-                index_writer_rlock.add_document(
                     doc!(
                         title => "Of Mice and Men",
                         body => "A few miles south of Soledad, the Salinas River drops in close to the hillside \
@@ -63,8 +59,7 @@ fn main() -> tantivy::Result<()> {
                         fresh and green with every spring, carrying in their lower leaf junctures the \
                         debris of the winter’s flooding; and sycamores with mottled, white, recumbent \
                         limbs and branches that arch over the pool"
-                    ))
+                    ));
-            };
             println!("add doc {} from thread 1 - opstamp {}", i, opstamp);
             thread::sleep(Duration::from_millis(20));
         }

examples/pre_tokenized_text.rs

@@ -0,0 +1,139 @@
+// # Pre-tokenized text example
+//
+// This example shows how to use pre-tokenized text. Sometimes you might
+// want to index and search through text which is already split into
+// tokens by some external tool.
+//
+// In this example we will:
+// - use tantivy tokenizer to create tokens and load them directly into tantivy,
+// - import tokenized text straight from json,
+// - perform a search on documents with pre-tokenized text
+
+use tantivy::collector::{Count, TopDocs};
+use tantivy::query::TermQuery;
+use tantivy::schema::*;
+use tantivy::tokenizer::{PreTokenizedString, SimpleTokenizer, Token, Tokenizer};
+use tantivy::{doc, Index, ReloadPolicy};
+use tempfile::TempDir;
+
+fn pre_tokenize_text(text: &str) -> Vec<Token> {
+    let mut token_stream = SimpleTokenizer.token_stream(text);
+    let mut tokens = vec![];
+    while token_stream.advance() {
+        tokens.push(token_stream.token().clone());
+    }
+    tokens
+}
+
+fn main() -> tantivy::Result<()> {
+    let index_path = TempDir::new()?;
+
+    let mut schema_builder = Schema::builder();
+
+    schema_builder.add_text_field("title", TEXT | STORED);
+    schema_builder.add_text_field("body", TEXT);
+
+    let schema = schema_builder.build();
+
+    let index = Index::create_in_dir(&index_path, schema.clone())?;
+
+    let mut index_writer = index.writer(50_000_000)?;
+
+    // We can create a document manually, by setting the fields
+    // one by one in a Document object.
+    let title = schema.get_field("title").unwrap();
+    let body = schema.get_field("body").unwrap();
+
+    let title_text = "The Old Man and the Sea";
+    let body_text = "He was an old man who fished alone in a skiff in the Gulf Stream";
+
+    // Content of our first document
+    // We create `PreTokenizedString` which contains original text and vector of tokens
+    let title_tok = PreTokenizedString {
+        text: String::from(title_text),
+        tokens: pre_tokenize_text(title_text),
+    };
+
+    println!(
+        "Original text: \"{}\" and tokens: {:?}",
+        title_tok.text, title_tok.tokens
+    );
+
+    let body_tok = PreTokenizedString {
+        text: String::from(body_text),
+        tokens: pre_tokenize_text(body_text),
+    };
+
+    // Now lets create a document and add our `PreTokenizedString`
+    let old_man_doc = doc!(title => title_tok, body => body_tok);
+
+    // ... now let's just add it to the IndexWriter
+    index_writer.add_document(old_man_doc);
+
+    // Pretokenized text can also be fed as JSON
+    let short_man_json = r#"{
+        "title":[{
+            "text":"The Old Man",
+            "tokens":[
+                {"offset_from":0,"offset_to":3,"position":0,"text":"The","position_length":1},
+                {"offset_from":4,"offset_to":7,"position":1,"text":"Old","position_length":1},
+                {"offset_from":8,"offset_to":11,"position":2,"text":"Man","position_length":1}
+            ]
+        }]
+    }"#;
+
+    let short_man_doc = schema.parse_document(&short_man_json)?;
+
+    index_writer.add_document(short_man_doc);
+
+    // Let's commit changes
+    index_writer.commit()?;
+
+    // ... and now is the time to query our index
+
+    let reader = index
+        .reader_builder()
+        .reload_policy(ReloadPolicy::OnCommit)
+        .try_into()?;
+
+    let searcher = reader.searcher();
+
+    // We want to get documents with token "Man", we will use TermQuery to do it
+    // Using PreTokenizedString means the tokens are stored as is avoiding stemming
+    // and lowercasing, which preserves full words in their original form
+    let query = TermQuery::new(
+        Term::from_field_text(title, "Man"),
+        IndexRecordOption::Basic,
+    );
+
+    let (top_docs, count) = searcher
+        .search(&query, &(TopDocs::with_limit(2), Count))
+        .unwrap();
+
+    assert_eq!(count, 2);
+
+    // Now let's print out the results.
+    // Note that the tokens are not stored along with the original text
+    // in the document store
+    for (_score, doc_address) in top_docs {
+        let retrieved_doc = searcher.doc(doc_address)?;
+        println!("Document: {}", schema.to_json(&retrieved_doc));
+    }
+
+    // In contrary to the previous query, when we search for the "man" term we
+    // should get no results, as it's not one of the indexed tokens. SimpleTokenizer
+    // only splits text on whitespace / punctuation.
+
+    let query = TermQuery::new(
+        Term::from_field_text(title, "man"),
+        IndexRecordOption::Basic,
+    );
+
+    let (_top_docs, count) = searcher
+        .search(&query, &(TopDocs::with_limit(2), Count))
+        .unwrap();
+
+    assert_eq!(count, 0);
+
+    Ok(())
+}

examples/snippet.rs

@@ -7,19 +7,16 @@
 
 // ---
 // Importing tantivy...
-#[macro_use]
-extern crate tantivy;
 use tantivy::collector::TopDocs;
 use tantivy::query::QueryParser;
 use tantivy::schema::*;
-use tantivy::Index;
+use tantivy::{doc, Index, Snippet, SnippetGenerator};
-use tantivy::{Snippet, SnippetGenerator};
+use tempfile::TempDir;
-use tempdir::TempDir;
 
 fn main() -> tantivy::Result<()> {
     // Let's create a temporary directory for the
     // sake of this example
-    let index_path = TempDir::new("tantivy_example_dir")?;
+    let index_path = TempDir::new()?;
 
     // # Defining the schema
     let mut schema_builder = Schema::builder();
@@ -72,12 +69,12 @@ fn highlight(snippet: Snippet) -> String {
     let mut result = String::new();
     let mut start_from = 0;
 
-    for (start, end) in snippet.highlighted().iter().map(|h| h.bounds()) {
+    for fragment_range in snippet.highlighted() {
-        result.push_str(&snippet.fragments()[start_from..start]);
+        result.push_str(&snippet.fragments()[start_from..fragment_range.start]);
         result.push_str(" --> ");
-        result.push_str(&snippet.fragments()[start..end]);
+        result.push_str(&snippet.fragments()[fragment_range.clone()]);
         result.push_str(" <-- ");
-        start_from = end;
+        start_from = fragment_range.end;
     }
 
     result.push_str(&snippet.fragments()[start_from..]);

examples/stop_words.rs

@@ -11,13 +11,11 @@
 
 // ---
 // Importing tantivy...
-#[macro_use]
-extern crate tantivy;
 use tantivy::collector::TopDocs;
 use tantivy::query::QueryParser;
 use tantivy::schema::*;
 use tantivy::tokenizer::*;
-use tantivy::Index;
+use tantivy::{doc, Index};
 
 fn main() -> tantivy::Result<()> {
     // this example assumes you understand the content in `basic_search`
@@ -52,7 +50,7 @@ fn main() -> tantivy::Result<()> {
 
     // This tokenizer lowers all of the text (to help with stop word matching)
     // then removes all instances of `the` and `and` from the corpus
-    let tokenizer = SimpleTokenizer
+    let tokenizer = TextAnalyzer::from(SimpleTokenizer)
         .filter(LowerCaser)
         .filter(StopWordFilter::remove(vec![
             "the".to_string(),

query-grammar/Cargo.toml

@@ -0,0 +1,16 @@
+[package]
+name = "tantivy-query-grammar"
+version = "0.15.0"
+authors = ["Paul Masurel <paul.masurel@gmail.com>"]
+license = "MIT"
+categories = ["database-implementations", "data-structures"]
+description = """Search engine library"""
+documentation = "https://tantivy-search.github.io/tantivy/tantivy/index.html"
+homepage = "https://github.com/tantivy-search/tantivy"
+repository = "https://github.com/tantivy-search/tantivy"
+readme = "README.md"
+keywords = ["search", "information", "retrieval"]
+edition = "2018"
+
+[dependencies]
+combine = {version="4", default-features=false, features=[] }

query-grammar/README.md

@@ -0,0 +1,3 @@
+# Tantivy Query Grammar
+
+This crate is used by tantivy to parse queries.

query-grammar/src/lib.rs

@@ -0,0 +1,15 @@
+mod occur;
+mod query_grammar;
+mod user_input_ast;
+use combine::parser::Parser;
+
+pub use crate::occur::Occur;
+use crate::query_grammar::parse_to_ast;
+pub use crate::user_input_ast::{UserInputAst, UserInputBound, UserInputLeaf, UserInputLiteral};
+
+pub struct Error;
+
+pub fn parse_query(query: &str) -> Result<UserInputAst, Error> {
+    let (user_input_ast, _remaining) = parse_to_ast().parse(query).map_err(|_| Error)?;
+    Ok(user_input_ast)
+}

query-grammar/src/occur.rs

@@ -0,0 +1,72 @@
+use std::fmt;
+use std::fmt::Write;
+
+/// Defines whether a term in a query must be present,
+/// should be present or must be not present.
+#[derive(Debug, Clone, Hash, Copy, Eq, PartialEq)]
+pub enum Occur {
+    /// For a given document to be considered for scoring,
+    /// at least one of the document with the Should or the Must
+    /// Occur constraint must be within the document.
+    Should,
+    /// Document without the term are excluded from the search.
+    Must,
+    /// Document that contain the term are excluded from the
+    /// search.
+    MustNot,
+}
+
+impl Occur {
+    /// Returns the one-char prefix symbol for this `Occur`.
+    /// - `Should` => '?',
+    /// - `Must` => '+'
+    /// - `Not` => '-'
+    fn to_char(self) -> char {
+        match self {
+            Occur::Should => '?',
+            Occur::Must => '+',
+            Occur::MustNot => '-',
+        }
+    }
+
+    /// Compose two occur values.
+    pub fn compose(left: Occur, right: Occur) -> Occur {
+        match (left, right) {
+            (Occur::Should, _) => right,
+            (Occur::Must, Occur::MustNot) => Occur::MustNot,
+            (Occur::Must, _) => Occur::Must,
+            (Occur::MustNot, Occur::MustNot) => Occur::Must,
+            (Occur::MustNot, _) => Occur::MustNot,
+        }
+    }
+}
+
+impl fmt::Display for Occur {
+    fn fmt(&self, f: &mut fmt::Formatter) -> fmt::Result {
+        f.write_char(self.to_char())
+    }
+}
+
+#[cfg(test)]
+mod test {
+    use crate::Occur;
+
+    #[test]
+    fn test_occur_compose() {
+        assert_eq!(Occur::compose(Occur::Should, Occur::Should), Occur::Should);
+        assert_eq!(Occur::compose(Occur::Should, Occur::Must), Occur::Must);
+        assert_eq!(
+            Occur::compose(Occur::Should, Occur::MustNot),
+            Occur::MustNot
+        );
+        assert_eq!(Occur::compose(Occur::Must, Occur::Should), Occur::Must);
+        assert_eq!(Occur::compose(Occur::Must, Occur::Must), Occur::Must);
+        assert_eq!(Occur::compose(Occur::Must, Occur::MustNot), Occur::MustNot);
+        assert_eq!(
+            Occur::compose(Occur::MustNot, Occur::Should),
+            Occur::MustNot
+        );
+        assert_eq!(Occur::compose(Occur::MustNot, Occur::Must), Occur::MustNot);
+        assert_eq!(Occur::compose(Occur::MustNot, Occur::MustNot), Occur::Must);
+    }
+}

query-grammar/src/query_grammar.rs

@@ -0,0 +1,642 @@
+use super::user_input_ast::{UserInputAst, UserInputBound, UserInputLeaf, UserInputLiteral};
+use crate::Occur;
+use combine::parser::char::{char, digit, letter, space, spaces, string};
+use combine::parser::Parser;
+use combine::{
+    attempt, choice, eof, many, many1, one_of, optional, parser, satisfy, skip_many1, value,
+};
+use combine::{error::StringStreamError, parser::combinator::recognize};
+
+fn field<'a>() -> impl Parser<&'a str, Output = String> {
+    (
+        (letter().or(char('_'))),
+        many(satisfy(|c: char| {
+            c.is_alphanumeric() || c == '_' || c == '-'
+        })),
+    )
+        .skip(char(':'))
+        .map(|(s1, s2): (char, String)| format!("{}{}", s1, s2))
+}
+
+fn word<'a>() -> impl Parser<&'a str, Output = String> {
+    (
+        satisfy(|c: char| {
+            !c.is_whitespace()
+                && !['-', '^', '`', ':', '{', '}', '"', '[', ']', '(', ')'].contains(&c)
+        }),
+        many(satisfy(|c: char| {
+            !c.is_whitespace() && ![':', '^', '{', '}', '"', '[', ']', '(', ')'].contains(&c)
+        })),
+    )
+        .map(|(s1, s2): (char, String)| format!("{}{}", s1, s2))
+        .and_then(|s: String| match s.as_str() {
+            "OR" | "AND " | "NOT" => Err(StringStreamError::UnexpectedParse),
+            _ => Ok(s),
+        })
+}
+
+/// Parses a date time according to rfc3339
+/// 2015-08-02T18:54:42+02
+/// 2021-04-13T19:46:26.266051969+00:00
+///
+/// NOTE: also accepts 999999-99-99T99:99:99.266051969+99:99
+/// We delegate rejecting such invalid dates to the logical AST compuation code
+/// which invokes chrono::DateTime::parse_from_rfc3339 on the value to actually parse
+/// it (instead of merely extracting the datetime value as string as done here).
+fn date_time<'a>() -> impl Parser<&'a str, Output = String> {
+    let two_digits = || recognize::<String, _, _>((digit(), digit()));
+
+    // Parses a time zone
+    // -06:30
+    // Z
+    let time_zone = {
+        let utc = recognize::<String, _, _>(char('Z'));
+        let offset = recognize((
+            choice([char('-'), char('+')]),
+            two_digits(),
+            char(':'),
+            two_digits(),
+        ));
+
+        utc.or(offset)
+    };
+
+    // Parses a date
+    // 2010-01-30
+    let date = {
+        recognize::<String, _, _>((
+            many1::<String, _, _>(digit()),
+            char('-'),
+            two_digits(),
+            char('-'),
+            two_digits(),
+        ))
+    };
+
+    // Parses a time
+    // 12:30:02
+    // 19:46:26.266051969
+    let time = {
+        recognize::<String, _, _>((
+            two_digits(),
+            char(':'),
+            two_digits(),
+            char(':'),
+            two_digits(),
+            optional((char('.'), many1::<String, _, _>(digit()))),
+            time_zone,
+        ))
+    };
+
+    recognize((date, char('T'), time))
+}
+
+fn term_val<'a>() -> impl Parser<&'a str, Output = String> {
+    let phrase = char('"').with(many1(satisfy(|c| c != '"'))).skip(char('"'));
+    phrase.or(word())
+}
+
+fn term_query<'a>() -> impl Parser<&'a str, Output = UserInputLiteral> {
+    let term_val_with_field = negative_number().or(term_val());
+    (field(), term_val_with_field).map(|(field_name, phrase)| UserInputLiteral {
+        field_name: Some(field_name),
+        phrase,
+    })
+}
+
+fn literal<'a>() -> impl Parser<&'a str, Output = UserInputLeaf> {
+    let term_default_field = term_val().map(|phrase| UserInputLiteral {
+        field_name: None,
+        phrase,
+    });
+    attempt(term_query())
+        .or(term_default_field)
+        .map(UserInputLeaf::from)
+}
+
+fn negative_number<'a>() -> impl Parser<&'a str, Output = String> {
+    (
+        char('-'),
+        many1(digit()),
+        optional((char('.'), many1(digit()))),
+    )
+        .map(|(s1, s2, s3): (char, String, Option<(char, String)>)| {
+            if let Some(('.', s3)) = s3 {
+                format!("{}{}.{}", s1, s2, s3)
+            } else {
+                format!("{}{}", s1, s2)
+            }
+        })
+}
+
+fn spaces1<'a>() -> impl Parser<&'a str, Output = ()> {
+    skip_many1(space())
+}
+
+/// Function that parses a range out of a Stream
+/// Supports ranges like:
+/// [5 TO 10], {5 TO 10}, [* TO 10], [10 TO *], {10 TO *], >5, <=10
+/// [a TO *], [a TO c], [abc TO bcd}
+fn range<'a>() -> impl Parser<&'a str, Output = UserInputLeaf> {
+    let range_term_val = || {
+        attempt(date_time())
+            .or(word())
+            .or(negative_number())
+            .or(char('*').with(value("*".to_string())))
+    };
+
+    // check for unbounded range in the form of <5, <=10, >5, >=5
+    let elastic_unbounded_range = (
+        choice([
+            attempt(string(">=")),
+            attempt(string("<=")),
+            attempt(string("<")),
+            attempt(string(">")),
+        ])
+        .skip(spaces()),
+        range_term_val(),
+    )
+        .map(
+            |(comparison_sign, bound): (&str, String)| match comparison_sign {
+                ">=" => (UserInputBound::Inclusive(bound), UserInputBound::Unbounded),
+                "<=" => (UserInputBound::Unbounded, UserInputBound::Inclusive(bound)),
+                "<" => (UserInputBound::Unbounded, UserInputBound::Exclusive(bound)),
+                ">" => (UserInputBound::Exclusive(bound), UserInputBound::Unbounded),
+                // default case
+                _ => (UserInputBound::Unbounded, UserInputBound::Unbounded),
+            },
+        );
+    let lower_bound = (one_of("{[".chars()), range_term_val()).map(
+        |(boundary_char, lower_bound): (char, String)| {
+            if lower_bound == "*" {
+                UserInputBound::Unbounded
+            } else if boundary_char == '{' {
+                UserInputBound::Exclusive(lower_bound)
+            } else {
+                UserInputBound::Inclusive(lower_bound)
+            }
+        },
+    );
+    let upper_bound = (range_term_val(), one_of("}]".chars())).map(
+        |(higher_bound, boundary_char): (String, char)| {
+            if higher_bound == "*" {
+                UserInputBound::Unbounded
+            } else if boundary_char == '}' {
+                UserInputBound::Exclusive(higher_bound)
+            } else {
+                UserInputBound::Inclusive(higher_bound)
+            }
+        },
+    );
+    // return only lower and upper
+    let lower_to_upper = (
+        lower_bound.skip((spaces(), string("TO"), spaces())),
+        upper_bound,
+    );
+
+    (
+        optional(field()).skip(spaces()),
+        // try elastic first, if it matches, the range is unbounded
+        attempt(elastic_unbounded_range).or(lower_to_upper),
+    )
+        .map(|(field, (lower, upper))|
+             // Construct the leaf from extracted field (optional)
+             // and bounds
+             UserInputLeaf::Range {
+                 field,
+                 lower,
+                 upper
+    })
+}
+
+fn negate(expr: UserInputAst) -> UserInputAst {
+    expr.unary(Occur::MustNot)
+}
+
+fn leaf<'a>() -> impl Parser<&'a str, Output = UserInputAst> {
+    parser(|input| {
+        char('(')
+            .with(ast())
+            .skip(char(')'))
+            .or(char('*').map(|_| UserInputAst::from(UserInputLeaf::All)))
+            .or(attempt(
+                string("NOT").skip(spaces1()).with(leaf()).map(negate),
+            ))
+            .or(attempt(range().map(UserInputAst::from)))
+            .or(literal().map(UserInputAst::from))
+            .parse_stream(input)
+            .into_result()
+    })
+}
+
+fn occur_symbol<'a>() -> impl Parser<&'a str, Output = Occur> {
+    char('-')
+        .map(|_| Occur::MustNot)
+        .or(char('+').map(|_| Occur::Must))
+}
+
+fn occur_leaf<'a>() -> impl Parser<&'a str, Output = (Option<Occur>, UserInputAst)> {
+    (optional(occur_symbol()), boosted_leaf())
+}
+
+fn positive_float_number<'a>() -> impl Parser<&'a str, Output = f64> {
+    (many1(digit()), optional((char('.'), many1(digit())))).map(
+        |(int_part, decimal_part_opt): (String, Option<(char, String)>)| {
+            let mut float_str = int_part;
+            if let Some((chr, decimal_str)) = decimal_part_opt {
+                float_str.push(chr);
+                float_str.push_str(&decimal_str);
+            }
+            float_str.parse::<f64>().unwrap()
+        },
+    )
+}
+
+fn boost<'a>() -> impl Parser<&'a str, Output = f64> {
+    (char('^'), positive_float_number()).map(|(_, boost)| boost)
+}
+
+fn boosted_leaf<'a>() -> impl Parser<&'a str, Output = UserInputAst> {
+    (leaf(), optional(boost())).map(|(leaf, boost_opt)| match boost_opt {
+        Some(boost) if (boost - 1.0).abs() > std::f64::EPSILON => {
+            UserInputAst::Boost(Box::new(leaf), boost)
+        }
+        _ => leaf,
+    })
+}
+
+#[derive(Clone, Copy)]
+enum BinaryOperand {
+    Or,
+    And,
+}
+
+fn binary_operand<'a>() -> impl Parser<&'a str, Output = BinaryOperand> {
+    string("AND")
+        .with(value(BinaryOperand::And))
+        .or(string("OR").with(value(BinaryOperand::Or)))
+}
+
+fn aggregate_binary_expressions(
+    left: UserInputAst,
+    others: Vec<(BinaryOperand, UserInputAst)>,
+) -> UserInputAst {
+    let mut dnf: Vec<Vec<UserInputAst>> = vec![vec![left]];
+    for (operator, operand_ast) in others {
+        match operator {
+            BinaryOperand::And => {
+                if let Some(last) = dnf.last_mut() {
+                    last.push(operand_ast);
+                }
+            }
+            BinaryOperand::Or => {
+                dnf.push(vec![operand_ast]);
+            }
+        }
+    }
+    if dnf.len() == 1 {
+        UserInputAst::and(dnf.into_iter().next().unwrap()) //< safe
+    } else {
+        let conjunctions = dnf.into_iter().map(UserInputAst::and).collect();
+        UserInputAst::or(conjunctions)
+    }
+}
+
+fn operand_leaf<'a>() -> impl Parser<&'a str, Output = (BinaryOperand, UserInputAst)> {
+    (
+        binary_operand().skip(spaces()),
+        boosted_leaf().skip(spaces()),
+    )
+}
+
+pub fn ast<'a>() -> impl Parser<&'a str, Output = UserInputAst> {
+    let boolean_expr = (boosted_leaf().skip(spaces()), many1(operand_leaf()))
+        .map(|(left, right)| aggregate_binary_expressions(left, right));
+    let whitespace_separated_leaves = many1(occur_leaf().skip(spaces().silent())).map(
+        |subqueries: Vec<(Option<Occur>, UserInputAst)>| {
+            if subqueries.len() == 1 {
+                let (occur_opt, ast) = subqueries.into_iter().next().unwrap();
+                match occur_opt.unwrap_or(Occur::Should) {
+                    Occur::Must | Occur::Should => ast,
+                    Occur::MustNot => UserInputAst::Clause(vec![(Some(Occur::MustNot), ast)]),
+                }
+            } else {
+                UserInputAst::Clause(subqueries.into_iter().collect())
+            }
+        },
+    );
+    let expr = attempt(boolean_expr).or(whitespace_separated_leaves);
+    spaces().with(expr).skip(spaces())
+}
+
+pub fn parse_to_ast<'a>() -> impl Parser<&'a str, Output = UserInputAst> {
+    spaces()
+        .with(optional(ast()).skip(eof()))
+        .map(|opt_ast| opt_ast.unwrap_or_else(UserInputAst::empty_query))
+}
+
+#[cfg(test)]
+mod test {
+
+    type TestParseResult = Result<(), StringStreamError>;
+
+    use super::*;
+    use combine::parser::Parser;
+
+    pub fn nearly_equals(a: f64, b: f64) -> bool {
+        (a - b).abs() < 0.0005 * (a + b).abs()
+    }
+
+    fn assert_nearly_equals(expected: f64, val: f64) {
+        assert!(
+            nearly_equals(val, expected),
+            "Got {}, expected {}.",
+            val,
+            expected
+        );
+    }
+
+    #[test]
+    fn test_occur_symbol() -> TestParseResult {
+        assert_eq!(super::occur_symbol().parse("-")?, (Occur::MustNot, ""));
+        assert_eq!(super::occur_symbol().parse("+")?, (Occur::Must, ""));
+        Ok(())
+    }
+
+    #[test]
+    fn test_positive_float_number() {
+        fn valid_parse(float_str: &str, expected_val: f64, expected_remaining: &str) {
+            let (val, remaining) = positive_float_number().parse(float_str).unwrap();
+            assert_eq!(remaining, expected_remaining);
+            assert_nearly_equals(val, expected_val);
+        }
+        fn error_parse(float_str: &str) {
+            assert!(positive_float_number().parse(float_str).is_err());
+        }
+        valid_parse("1.0", 1.0, "");
+        valid_parse("1", 1.0, "");
+        valid_parse("0.234234 aaa", 0.234234f64, " aaa");
+        error_parse(".3332");
+        error_parse("1.");
+        error_parse("-1.");
+    }
+
+    #[test]
+    fn test_date_time() {
+        let (val, remaining) = date_time()
+            .parse("2015-08-02T18:54:42+02:30")
+            .expect("cannot parse date");
+        assert_eq!(val, "2015-08-02T18:54:42+02:30");
+        assert_eq!(remaining, "");
+        assert!(date_time().parse("2015-08-02T18:54:42+02").is_err());
+
+        let (val, remaining) = date_time()
+            .parse("2021-04-13T19:46:26.266051969+00:00")
+            .expect("cannot parse fractional date");
+        assert_eq!(val, "2021-04-13T19:46:26.266051969+00:00");
+        assert_eq!(remaining, "");
+    }
+
+    fn test_parse_query_to_ast_helper(query: &str, expected: &str) {
+        let query = parse_to_ast().parse(query).unwrap().0;
+        let query_str = format!("{:?}", query);
+        assert_eq!(query_str, expected);
+    }
+
+    fn test_is_parse_err(query: &str) {
+        assert!(parse_to_ast().parse(query).is_err());
+    }
+
+    #[test]
+    fn test_parse_empty_to_ast() {
+        test_parse_query_to_ast_helper("", "<emptyclause>");
+    }
+
+    #[test]
+    fn test_parse_query_to_ast_hyphen() {
+        test_parse_query_to_ast_helper("\"www-form-encoded\"", "\"www-form-encoded\"");
+        test_parse_query_to_ast_helper("www-form-encoded", "\"www-form-encoded\"");
+        test_parse_query_to_ast_helper("www-form-encoded", "\"www-form-encoded\"");
+    }
+
+    #[test]
+    fn test_parse_query_to_ast_not_op() {
+        assert_eq!(
+            format!("{:?}", parse_to_ast().parse("NOT")),
+            "Err(UnexpectedParse)"
+        );
+        test_parse_query_to_ast_helper("NOTa", "\"NOTa\"");
+        test_parse_query_to_ast_helper("NOT a", "(-\"a\")");
+    }
+
+    #[test]
+    fn test_boosting() {
+        assert!(parse_to_ast().parse("a^2^3").is_err());
+        assert!(parse_to_ast().parse("a^2^").is_err());
+        test_parse_query_to_ast_helper("a^3", "(\"a\")^3");
+        test_parse_query_to_ast_helper("a^3 b^2", "(*(\"a\")^3 *(\"b\")^2)");
+        test_parse_query_to_ast_helper("a^1", "\"a\"");
+    }
+
+    #[test]
+    fn test_parse_query_to_ast_binary_op() {
+        test_parse_query_to_ast_helper("a AND b", "(+\"a\" +\"b\")");
+        test_parse_query_to_ast_helper("a OR b", "(?\"a\" ?\"b\")");
+        test_parse_query_to_ast_helper("a OR b AND c", "(?\"a\" ?(+\"b\" +\"c\"))");
+        test_parse_query_to_ast_helper("a AND b         AND c", "(+\"a\" +\"b\" +\"c\")");
+        assert_eq!(
+            format!("{:?}", parse_to_ast().parse("a OR b aaa")),
+            "Err(UnexpectedParse)"
+        );
+        assert_eq!(
+            format!("{:?}", parse_to_ast().parse("a AND b aaa")),
+            "Err(UnexpectedParse)"
+        );
+        assert_eq!(
+            format!("{:?}", parse_to_ast().parse("aaa a OR b ")),
+            "Err(UnexpectedParse)"
+        );
+        assert_eq!(
+            format!("{:?}", parse_to_ast().parse("aaa ccc a OR b ")),
+            "Err(UnexpectedParse)"
+        );
+    }
+
+    #[test]
+    fn test_parse_elastic_query_ranges() {
+        test_parse_query_to_ast_helper("title: >a", "title:{\"a\" TO \"*\"}");
+        test_parse_query_to_ast_helper("title:>=a", "title:[\"a\" TO \"*\"}");
+        test_parse_query_to_ast_helper("title: <a", "title:{\"*\" TO \"a\"}");
+        test_parse_query_to_ast_helper("title:<=a", "title:{\"*\" TO \"a\"]");
+        test_parse_query_to_ast_helper("title:<=bsd", "title:{\"*\" TO \"bsd\"]");
+
+        test_parse_query_to_ast_helper("weight: >70", "weight:{\"70\" TO \"*\"}");
+        test_parse_query_to_ast_helper("weight:>=70", "weight:[\"70\" TO \"*\"}");
+        test_parse_query_to_ast_helper("weight: <70", "weight:{\"*\" TO \"70\"}");
+        test_parse_query_to_ast_helper("weight:<=70", "weight:{\"*\" TO \"70\"]");
+        test_parse_query_to_ast_helper("weight: >60.7", "weight:{\"60.7\" TO \"*\"}");
+
+        test_parse_query_to_ast_helper("weight: <= 70", "weight:{\"*\" TO \"70\"]");
+
+        test_parse_query_to_ast_helper("weight: <= 70.5", "weight:{\"*\" TO \"70.5\"]");
+    }
+
+    #[test]
+    fn test_occur_leaf() {
+        let ((occur, ast), _) = super::occur_leaf().parse("+abc").unwrap();
+        assert_eq!(occur, Some(Occur::Must));
+        assert_eq!(format!("{:?}", ast), "\"abc\"");
+    }
+
+    #[test]
+    fn test_field_name() -> TestParseResult {
+        assert_eq!(
+            super::field().parse("my-field-name:a")?,
+            ("my-field-name".to_string(), "a")
+        );
+        assert_eq!(
+            super::field().parse("my_field_name:a")?,
+            ("my_field_name".to_string(), "a")
+        );
+        assert!(super::field().parse(":a").is_err());
+        assert!(super::field().parse("-my_field:a").is_err());
+        assert_eq!(
+            super::field().parse("_my_field:a")?,
+            ("_my_field".to_string(), "a")
+        );
+        Ok(())
+    }
+
+    #[test]
+    fn test_range_parser() {
+        // testing the range() parser separately
+        let res = range()
+            .parse("title: <hello")
+            .expect("Cannot parse felxible bound word")
+            .0;
+        let expected = UserInputLeaf::Range {
+            field: Some("title".to_string()),
+            lower: UserInputBound::Unbounded,
+            upper: UserInputBound::Exclusive("hello".to_string()),
+        };
+        let res2 = range()
+            .parse("title:{* TO hello}")
+            .expect("Cannot parse ununbounded to word")
+            .0;
+        assert_eq!(res, expected);
+        assert_eq!(res2, expected);
+
+        let expected_weight = UserInputLeaf::Range {
+            field: Some("weight".to_string()),
+            lower: UserInputBound::Inclusive("71.2".to_string()),
+            upper: UserInputBound::Unbounded,
+        };
+        let res3 = range()
+            .parse("weight: >=71.2")
+            .expect("Cannot parse flexible bound float")
+            .0;
+        let res4 = range()
+            .parse("weight:[71.2 TO *}")
+            .expect("Cannot parse float to unbounded")
+            .0;
+        assert_eq!(res3, expected_weight);
+        assert_eq!(res4, expected_weight);
+
+        let expected_dates = UserInputLeaf::Range {
+            field: Some("date_field".to_string()),
+            lower: UserInputBound::Exclusive("2015-08-02T18:54:42Z".to_string()),
+            upper: UserInputBound::Inclusive("2021-08-02T18:54:42+02:30".to_string()),
+        };
+        let res5 = range()
+            .parse("date_field:{2015-08-02T18:54:42Z TO 2021-08-02T18:54:42+02:30]")
+            .expect("Cannot parse date range")
+            .0;
+        assert_eq!(res5, expected_dates);
+
+        let expected_flexible_dates = UserInputLeaf::Range {
+            field: Some("date_field".to_string()),
+            lower: UserInputBound::Unbounded,
+            upper: UserInputBound::Inclusive("2021-08-02T18:54:42.12345+02:30".to_string()),
+        };
+
+        let res6 = range()
+            .parse("date_field: <=2021-08-02T18:54:42.12345+02:30")
+            .expect("Cannot parse date range")
+            .0;
+        assert_eq!(res6, expected_flexible_dates);
+    }
+
+    #[test]
+    fn test_parse_query_to_triming_spaces() {
+        test_parse_query_to_ast_helper("   abc", "\"abc\"");
+        test_parse_query_to_ast_helper("abc ", "\"abc\"");
+        test_parse_query_to_ast_helper("(  a OR abc)", "(?\"a\" ?\"abc\")");
+        test_parse_query_to_ast_helper("(a  OR abc)", "(?\"a\" ?\"abc\")");
+        test_parse_query_to_ast_helper("(a OR  abc)", "(?\"a\" ?\"abc\")");
+        test_parse_query_to_ast_helper("a OR abc ", "(?\"a\" ?\"abc\")");
+        test_parse_query_to_ast_helper("(a OR abc )", "(?\"a\" ?\"abc\")");
+        test_parse_query_to_ast_helper("(a OR  abc) ", "(?\"a\" ?\"abc\")");
+    }
+
+    #[test]
+    fn test_parse_query_single_term() {
+        test_parse_query_to_ast_helper("abc", "\"abc\"");
+    }
+
+    #[test]
+    fn test_parse_query_default_clause() {
+        test_parse_query_to_ast_helper("a b", "(*\"a\" *\"b\")");
+    }
+
+    #[test]
+    fn test_parse_query_must_default_clause() {
+        test_parse_query_to_ast_helper("+(a b)", "(*\"a\" *\"b\")");
+    }
+
+    #[test]
+    fn test_parse_query_must_single_term() {
+        test_parse_query_to_ast_helper("+d", "\"d\"");
+    }
+
+    #[test]
+    fn test_single_term_with_field() {
+        test_parse_query_to_ast_helper("abc:toto", "abc:\"toto\"");
+    }
+
+    #[test]
+    fn test_single_term_with_float() {
+        test_parse_query_to_ast_helper("abc:1.1", "abc:\"1.1\"");
+    }
+
+    #[test]
+    fn test_must_clause() {
+        test_parse_query_to_ast_helper("(+a +b)", "(+\"a\" +\"b\")");
+    }
+
+    #[test]
+    fn test_parse_test_query_plus_a_b_plus_d() {
+        test_parse_query_to_ast_helper("+(a b) +d", "(+(*\"a\" *\"b\") +\"d\")");
+    }
+
+    #[test]
+    fn test_parse_test_query_other() {
+        test_parse_query_to_ast_helper("(+a +b) d", "(*(+\"a\" +\"b\") *\"d\")");
+        test_parse_query_to_ast_helper("+abc:toto", "abc:\"toto\"");
+        test_parse_query_to_ast_helper("(+abc:toto -titi)", "(+abc:\"toto\" -\"titi\")");
+        test_parse_query_to_ast_helper("-abc:toto", "(-abc:\"toto\")");
+        test_parse_query_to_ast_helper("abc:a b", "(*abc:\"a\" *\"b\")");
+        test_parse_query_to_ast_helper("abc:\"a b\"", "abc:\"a b\"");
+        test_parse_query_to_ast_helper("foo:[1 TO 5]", "foo:[\"1\" TO \"5\"]");
+    }
+
+    #[test]
+    fn test_parse_query_with_range() {
+        test_parse_query_to_ast_helper("[1 TO 5]", "[\"1\" TO \"5\"]");
+        test_parse_query_to_ast_helper("foo:{a TO z}", "foo:{\"a\" TO \"z\"}");
+        test_parse_query_to_ast_helper("foo:[1 TO toto}", "foo:[\"1\" TO \"toto\"}");
+        test_parse_query_to_ast_helper("foo:[* TO toto}", "foo:{\"*\" TO \"toto\"}");
+        test_parse_query_to_ast_helper("foo:[1 TO *}", "foo:[\"1\" TO \"*\"}");
+        test_parse_query_to_ast_helper("foo:[1.1 TO *}", "foo:[\"1.1\" TO \"*\"}");
+        test_is_parse_err("abc +    ");
+    }
+}

query-grammar/src/user_input_ast.rs

@@ -1,8 +1,9 @@
 use std::fmt;
 use std::fmt::{Debug, Formatter};
 
-use crate::query::Occur;
+use crate::Occur;
 
+#[derive(PartialEq)]
 pub enum UserInputLeaf {
     Literal(UserInputLiteral),
     All,
@@ -35,6 +36,7 @@ impl Debug for UserInputLeaf {
     }
 }
 
+#[derive(PartialEq)]
 pub struct UserInputLiteral {
     pub field_name: Option<String>,
     pub phrase: String,
@@ -49,9 +51,11 @@ impl fmt::Debug for UserInputLiteral {
     }
 }
 
+#[derive(PartialEq)]
 pub enum UserInputBound {
     Inclusive(String),
     Exclusive(String),
+    Unbounded,
 }
 
 impl UserInputBound {
@@ -59,6 +63,7 @@ impl UserInputBound {
         match *self {
             UserInputBound::Inclusive(ref word) => write!(formatter, "[\"{}\"", word),
             UserInputBound::Exclusive(ref word) => write!(formatter, "{{\"{}\"", word),
+            UserInputBound::Unbounded => write!(formatter, "{{\"*\""),
         }
     }
 
@@ -66,6 +71,7 @@ impl UserInputBound {
         match *self {
             UserInputBound::Inclusive(ref word) => write!(formatter, "\"{}\"]", word),
             UserInputBound::Exclusive(ref word) => write!(formatter, "\"{}\"}}", word),
+            UserInputBound::Unbounded => write!(formatter, "\"*\"}}"),
         }
     }
 
@@ -73,115 +79,93 @@ impl UserInputBound {
         match *self {
             UserInputBound::Inclusive(ref contents) => contents,
             UserInputBound::Exclusive(ref contents) => contents,
+            UserInputBound::Unbounded => &"*",
         }
     }
 }
 
-pub enum UserInputAST {
+pub enum UserInputAst {
-    Clause(Vec<UserInputAST>),
+    Clause(Vec<(Option<Occur>, UserInputAst)>),
-    Unary(Occur, Box<UserInputAST>),
-    //    Not(Box<UserInputAST>),
-    //    Should(Box<UserInputAST>),
-    //    Must(Box<UserInputAST>),
     Leaf(Box<UserInputLeaf>),
+    Boost(Box<UserInputAst>, f64),
 }
 
-impl UserInputAST {
+impl UserInputAst {
-    pub fn unary(self, occur: Occur) -> UserInputAST {
+    pub fn unary(self, occur: Occur) -> UserInputAst {
-        UserInputAST::Unary(occur, Box::new(self))
+        UserInputAst::Clause(vec![(Some(occur), self)])
     }
 
-    fn compose(occur: Occur, asts: Vec<UserInputAST>) -> UserInputAST {
+    fn compose(occur: Occur, asts: Vec<UserInputAst>) -> UserInputAst {
-        assert!(occur != Occur::MustNot);
+        assert_ne!(occur, Occur::MustNot);
         assert!(!asts.is_empty());
         if asts.len() == 1 {
             asts.into_iter().next().unwrap() //< safe
         } else {
-            UserInputAST::Clause(
+            UserInputAst::Clause(
                 asts.into_iter()
-                    .map(|ast: UserInputAST| ast.unary(occur))
+                    .map(|ast: UserInputAst| (Some(occur), ast))
                     .collect::<Vec<_>>(),
             )
         }
     }
 
-    pub fn and(asts: Vec<UserInputAST>) -> UserInputAST {
+    pub fn empty_query() -> UserInputAst {
-        UserInputAST::compose(Occur::Must, asts)
+        UserInputAst::Clause(Vec::default())
     }
 
-    pub fn or(asts: Vec<UserInputAST>) -> UserInputAST {
+    pub fn and(asts: Vec<UserInputAst>) -> UserInputAst {
-        UserInputAST::compose(Occur::Should, asts)
+        UserInputAst::compose(Occur::Must, asts)
+    }
+
+    pub fn or(asts: Vec<UserInputAst>) -> UserInputAst {
+        UserInputAst::compose(Occur::Should, asts)
     }
 }
 
-/*
-impl UserInputAST {
-
-    fn compose_occur(self, occur: Occur) -> UserInputAST {
-        match self {
-            UserInputAST::Not(other) => {
-                let new_occur = compose_occur(Occur::MustNot, occur);
-                other.simplify()
-            }
-            _ => {
-                self
-            }
-        }
-    }
-
-    pub fn simplify(self) -> UserInputAST {
-        match self {
-            UserInputAST::Clause(els) => {
-                if els.len() == 1 {
-                    return els.into_iter().next().unwrap();
-                } else {
-                    return self;
-                }
-            }
-            UserInputAST::Not(els) => {
-                if els.len() == 1 {
-                    return els.into_iter().next().unwrap();
-                } else {
-                    return self;
-                }
-            }
-        }
-    }
-}
-*/
-
 impl From<UserInputLiteral> for UserInputLeaf {
     fn from(literal: UserInputLiteral) -> UserInputLeaf {
         UserInputLeaf::Literal(literal)
     }
 }
 
-impl From<UserInputLeaf> for UserInputAST {
+impl From<UserInputLeaf> for UserInputAst {
-    fn from(leaf: UserInputLeaf) -> UserInputAST {
+    fn from(leaf: UserInputLeaf) -> UserInputAst {
-        UserInputAST::Leaf(Box::new(leaf))
+        UserInputAst::Leaf(Box::new(leaf))
     }
 }
 
-impl fmt::Debug for UserInputAST {
+fn print_occur_ast(
-    fn fmt(&self, formatter: &mut fmt::Formatter<'_>) -> Result<(), fmt::Error> {
+    occur_opt: Option<Occur>,
+    ast: &UserInputAst,
+    formatter: &mut fmt::Formatter,
+) -> fmt::Result {
+    if let Some(occur) = occur_opt {
+        write!(formatter, "{}{:?}", occur, ast)?;
+    } else {
+        write!(formatter, "*{:?}", ast)?;
+    }
+    Ok(())
+}
+
+impl fmt::Debug for UserInputAst {
+    fn fmt(&self, formatter: &mut fmt::Formatter) -> fmt::Result {
         match *self {
-            UserInputAST::Clause(ref subqueries) => {
+            UserInputAst::Clause(ref subqueries) => {
                 if subqueries.is_empty() {
                     write!(formatter, "<emptyclause>")?;
                 } else {
                     write!(formatter, "(")?;
-                    write!(formatter, "{:?}", &subqueries[0])?;
+                    print_occur_ast(subqueries[0].0, &subqueries[0].1, formatter)?;
                     for subquery in &subqueries[1..] {
-                        write!(formatter, " {:?}", subquery)?;
+                        write!(formatter, " ")?;
+                        print_occur_ast(subquery.0, &subquery.1, formatter)?;
                     }
                     write!(formatter, ")")?;
                 }
                 Ok(())
             }
-            UserInputAST::Unary(ref occur, ref subquery) => {
+            UserInputAst::Leaf(ref subquery) => write!(formatter, "{:?}", subquery),
-                write!(formatter, "{}({:?})", occur.to_char(), subquery)
+            UserInputAst::Boost(ref leaf, boost) => write!(formatter, "({:?})^{}", leaf, boost),
-            }
-            UserInputAST::Leaf(ref subquery) => write!(formatter, "{:?}", subquery),
         }
     }
 }

src/collector/count_collector.rs

@@ -1,58 +1,40 @@
 use super::Collector;
 use crate::collector::SegmentCollector;
 use crate::DocId;
-use crate::Result;
 use crate::Score;
-use crate::SegmentLocalId;
+use crate::SegmentOrdinal;
 use crate::SegmentReader;
 
 /// `CountCollector` collector only counts how many
 /// documents match the query.
 ///
 /// ```rust
-/// #[macro_use]
-/// extern crate tantivy;
-/// use tantivy::schema::{Schema, TEXT};
-/// use tantivy::{Index, Result};
 /// use tantivy::collector::Count;
 /// use tantivy::query::QueryParser;
+/// use tantivy::schema::{Schema, TEXT};
+/// use tantivy::{doc, Index};
 ///
-/// # fn main() { example().unwrap(); }
+/// let mut schema_builder = Schema::builder();
-/// fn example() -> Result<()> {
+/// let title = schema_builder.add_text_field("title", TEXT);
-///     let mut schema_builder = Schema::builder();
+/// let schema = schema_builder.build();
-///     let title = schema_builder.add_text_field("title", TEXT);
+/// let index = Index::create_in_ram(schema);
-///     let schema = schema_builder.build();
-///     let index = Index::create_in_ram(schema);
-///     {
-///         let mut index_writer = index.writer(3_000_000)?;
-///         index_writer.add_document(doc!(
-///             title => "The Name of the Wind",
-///         ));
-///         index_writer.add_document(doc!(
-///             title => "The Diary of Muadib",
-///         ));
-///         index_writer.add_document(doc!(
-///             title => "A Dairy Cow",
-///         ));
-///         index_writer.add_document(doc!(
-///             title => "The Diary of a Young Girl",
-///         ));
-///         index_writer.commit().unwrap();
-///     }
 ///
-///     let reader = index.reader()?;
+/// let mut index_writer = index.writer(3_000_000).unwrap();
-///     let searcher = reader.searcher();
+/// index_writer.add_document(doc!(title => "The Name of the Wind"));
+/// index_writer.add_document(doc!(title => "The Diary of Muadib"));
+/// index_writer.add_document(doc!(title => "A Dairy Cow"));
+/// index_writer.add_document(doc!(title => "The Diary of a Young Girl"));
+/// assert!(index_writer.commit().is_ok());
 ///
-///     {
+/// let reader = index.reader().unwrap();
-///         let query_parser = QueryParser::for_index(&index, vec![title]);
+/// let searcher = reader.searcher();
-///         let query = query_parser.parse_query("diary")?;
-///         let count = searcher.search(&query, &Count).unwrap();
 ///
-///         assert_eq!(count, 2);
+/// // Here comes the important part
-///     }
+/// let query_parser = QueryParser::for_index(&index, vec![title]);
+/// let query = query_parser.parse_query("diary").unwrap();
+/// let count = searcher.search(&query, &Count).unwrap();
 ///
-///     Ok(())
+/// assert_eq!(count, 2);
-/// }
 /// ```
 pub struct Count;
 
@@ -61,7 +43,11 @@ impl Collector for Count {
 
     type Child = SegmentCountCollector;
 
-    fn for_segment(&self, _: SegmentLocalId, _: &SegmentReader) -> Result<SegmentCountCollector> {
+    fn for_segment(
+        &self,
+        _: SegmentOrdinal,
+        _: &SegmentReader,
+    ) -> crate::Result<SegmentCountCollector> {
         Ok(SegmentCountCollector::default())
     }
 
@@ -69,7 +55,7 @@ impl Collector for Count {
         false
     }
 
-    fn merge_fruits(&self, segment_counts: Vec<usize>) -> Result<usize> {
+    fn merge_fruits(&self, segment_counts: Vec<usize>) -> crate::Result<usize> {
         Ok(segment_counts.into_iter().sum())
     }
 }
@@ -110,20 +96,19 @@ mod tests {
         }
         {
             let mut count_collector = SegmentCountCollector::default();
-            count_collector.collect(0u32, 1f32);
+            count_collector.collect(0u32, 1.0);
             assert_eq!(count_collector.harvest(), 1);
         }
         {
             let mut count_collector = SegmentCountCollector::default();
-            count_collector.collect(0u32, 1f32);
+            count_collector.collect(0u32, 1.0);
             assert_eq!(count_collector.harvest(), 1);
         }
         {
             let mut count_collector = SegmentCountCollector::default();
-            count_collector.collect(0u32, 1f32);
+            count_collector.collect(0u32, 1.0);
-            count_collector.collect(1u32, 1f32);
+            count_collector.collect(1u32, 1.0);
             assert_eq!(count_collector.harvest(), 2);
         }
     }
-
 }

src/collector/custom_score_top_collector.rs

@@ -1,6 +1,5 @@
 use crate::collector::top_collector::{TopCollector, TopSegmentCollector};
 use crate::collector::{Collector, SegmentCollector};
-use crate::Result;
 use crate::{DocAddress, DocId, Score, SegmentReader};
 
 pub(crate) struct CustomScoreTopCollector<TCustomScorer, TScore = Score> {
@@ -12,13 +11,13 @@ impl<TCustomScorer, TScore> CustomScoreTopCollector<TCustomScorer, TScore>
 where
     TScore: Clone + PartialOrd,
 {
-    pub fn new(
+    pub(crate) fn new(
         custom_scorer: TCustomScorer,
-        limit: usize,
+        collector: TopCollector<TScore>,
     ) -> CustomScoreTopCollector<TCustomScorer, TScore> {
         CustomScoreTopCollector {
             custom_scorer,
-            collector: TopCollector::with_limit(limit),
+            collector,
         }
     }
 }
@@ -29,7 +28,7 @@ where
 /// It is the segment local version of the [`CustomScorer`](./trait.CustomScorer.html).
 pub trait CustomSegmentScorer<TScore>: 'static {
     /// Computes the score of a specific `doc`.
-    fn score(&self, doc: DocId) -> TScore;
+    fn score(&mut self, doc: DocId) -> TScore;
 }
 
 /// `CustomScorer` makes it possible to define any kind of score.
@@ -42,12 +41,12 @@ pub trait CustomScorer<TScore>: Sync {
     type Child: CustomSegmentScorer<TScore>;
     /// Builds a child scorer for a specific segment. The child scorer is associated to
     /// a specific segment.
-    fn segment_scorer(&self, segment_reader: &SegmentReader) -> Result<Self::Child>;
+    fn segment_scorer(&self, segment_reader: &SegmentReader) -> crate::Result<Self::Child>;
 }
 
 impl<TCustomScorer, TScore> Collector for CustomScoreTopCollector<TCustomScorer, TScore>
 where
-    TCustomScorer: CustomScorer<TScore>,
+    TCustomScorer: CustomScorer<TScore> + Send + Sync,
     TScore: 'static + PartialOrd + Clone + Send + Sync,
 {
     type Fruit = Vec<(TScore, DocAddress)>;
@@ -58,11 +57,9 @@ where
         &self,
         segment_local_id: u32,
         segment_reader: &SegmentReader,
-    ) -> Result<Self::Child> {
+    ) -> crate::Result<Self::Child> {
+        let segment_collector = self.collector.for_segment(segment_local_id, segment_reader);
         let segment_scorer = self.custom_scorer.segment_scorer(segment_reader)?;
-        let segment_collector = self
-            .collector
-            .for_segment(segment_local_id, segment_reader)?;
         Ok(CustomScoreTopSegmentCollector {
             segment_collector,
             segment_scorer,
@@ -73,7 +70,7 @@ where
         false
     }
 
-    fn merge_fruits(&self, segment_fruits: Vec<Self::Fruit>) -> Result<Self::Fruit> {
+    fn merge_fruits(&self, segment_fruits: Vec<Self::Fruit>) -> crate::Result<Self::Fruit> {
         self.collector.merge_fruits(segment_fruits)
     }
 }
@@ -111,16 +108,16 @@ where
 {
     type Child = T;
 
-    fn segment_scorer(&self, segment_reader: &SegmentReader) -> Result<Self::Child> {
+    fn segment_scorer(&self, segment_reader: &SegmentReader) -> crate::Result<Self::Child> {
         Ok((self)(segment_reader))
     }
 }
 
 impl<F, TScore> CustomSegmentScorer<TScore> for F
 where
-    F: 'static + Sync + Send + Fn(DocId) -> TScore,
+    F: 'static + FnMut(DocId) -> TScore,
 {
-    fn score(&self, doc: DocId) -> TScore {
+    fn score(&mut self, doc: DocId) -> TScore {
         (self)(doc)
     }
 }

src/collector/docset_collector.rs

@@ -0,0 +1,61 @@
+use std::collections::HashSet;
+
+use crate::{DocAddress, DocId, Score};
+
+use super::{Collector, SegmentCollector};
+
+/// Collectors that returns the set of DocAddress that matches the query.
+///
+/// This collector is mostly useful for tests.
+pub struct DocSetCollector;
+
+impl Collector for DocSetCollector {
+    type Fruit = HashSet<DocAddress>;
+    type Child = DocSetChildCollector;
+
+    fn for_segment(
+        &self,
+        segment_local_id: crate::SegmentOrdinal,
+        _segment: &crate::SegmentReader,
+    ) -> crate::Result<Self::Child> {
+        Ok(DocSetChildCollector {
+            segment_local_id,
+            docs: HashSet::new(),
+        })
+    }
+
+    fn requires_scoring(&self) -> bool {
+        false
+    }
+
+    fn merge_fruits(
+        &self,
+        segment_fruits: Vec<(u32, HashSet<DocId>)>,
+    ) -> crate::Result<Self::Fruit> {
+        let len: usize = segment_fruits.iter().map(|(_, docset)| docset.len()).sum();
+        let mut result = HashSet::with_capacity(len);
+        for (segment_local_id, docs) in segment_fruits {
+            for doc in docs {
+                result.insert(DocAddress::new(segment_local_id, doc));
+            }
+        }
+        Ok(result)
+    }
+}
+
+pub struct DocSetChildCollector {
+    segment_local_id: u32,
+    docs: HashSet<DocId>,
+}
+
+impl SegmentCollector for DocSetChildCollector {
+    type Fruit = (u32, HashSet<DocId>);
+
+    fn collect(&mut self, doc: crate::DocId, _score: Score) {
+        self.docs.insert(doc);
+    }
+
+    fn harvest(self) -> (u32, HashSet<DocId>) {
+        (self.segment_local_id, self.docs)
+    }
+}

src/collector/facet_collector.rs

@@ -1,22 +1,19 @@
 use crate::collector::Collector;
 use crate::collector::SegmentCollector;
-use crate::docset::SkipResult;
 use crate::fastfield::FacetReader;
 use crate::schema::Facet;
 use crate::schema::Field;
 use crate::DocId;
-use crate::Result;
 use crate::Score;
-use crate::SegmentLocalId;
+use crate::SegmentOrdinal;
 use crate::SegmentReader;
-use crate::TantivyError;
 use std::cmp::Ordering;
 use std::collections::btree_map;
 use std::collections::BTreeMap;
 use std::collections::BTreeSet;
 use std::collections::BinaryHeap;
-use std::collections::Bound;
 use std::iter::Peekable;
+use std::ops::Bound;
 use std::{u64, usize};
 
 struct Hit<'a> {
@@ -40,7 +37,10 @@ impl<'a> PartialOrd<Hit<'a>> for Hit<'a> {
 
 impl<'a> Ord for Hit<'a> {
     fn cmp(&self, other: &Self) -> Ordering {
-        other.count.cmp(&self.count)
+        other
+            .count
+            .cmp(&self.count)
+            .then(self.facet.cmp(other.facet))
     }
 }
 
@@ -81,21 +81,18 @@ fn facet_depth(facet_bytes: &[u8]) -> usize {
 ///
 ///
 /// ```rust
-/// #[macro_use]
-/// extern crate tantivy;
-/// use tantivy::schema::{Facet, Schema, TEXT};
-/// use tantivy::{Index, Result};
 /// use tantivy::collector::FacetCollector;
 /// use tantivy::query::AllQuery;
+/// use tantivy::schema::{Facet, Schema, INDEXED, TEXT};
+/// use tantivy::{doc, Index};
 ///
-/// # fn main() { example().unwrap(); }
+/// fn example() -> tantivy::Result<()> {
-/// fn example() -> Result<()> {
 ///     let mut schema_builder = Schema::builder();
 ///
 ///     // Facet have their own specific type.
 ///     // It is not a bad practise to put all of your
 ///     // facet information in the same field.
-///     let facet = schema_builder.add_facet_field("facet");
+///     let facet = schema_builder.add_facet_field("facet", INDEXED);
 ///     let title = schema_builder.add_text_field("title", TEXT);
 ///     let schema = schema_builder.build();
 ///     let index = Index::create_in_ram(schema);
@@ -129,7 +126,7 @@ fn facet_depth(facet_bytes: &[u8]) -> usize {
 ///     let searcher = reader.searcher();
 ///
 ///     {
-///			let mut facet_collector = FacetCollector::for_field(facet);
+///         let mut facet_collector = FacetCollector::for_field(facet);
 ///         facet_collector.add_facet("/lang");
 ///         facet_collector.add_facet("/category");
 ///         let facet_counts = searcher.search(&AllQuery, &facet_collector)?;
@@ -145,7 +142,7 @@ fn facet_depth(facet_bytes: &[u8]) -> usize {
 ///     }
 ///
 ///     {
-///			let mut facet_collector = FacetCollector::for_field(facet);
+///         let mut facet_collector = FacetCollector::for_field(facet);
 ///         facet_collector.add_facet("/category/fiction");
 ///         let facet_counts = searcher.search(&AllQuery, &facet_collector)?;
 ///
@@ -160,8 +157,8 @@ fn facet_depth(facet_bytes: &[u8]) -> usize {
 ///         ]);
 ///     }
 ///
-///    {
+///     {
-///			let mut facet_collector = FacetCollector::for_field(facet);
+///         let mut facet_collector = FacetCollector::for_field(facet);
 ///         facet_collector.add_facet("/category/fiction");
 ///         let facet_counts = searcher.search(&AllQuery, &facet_collector)?;
 ///
@@ -174,6 +171,7 @@ fn facet_depth(facet_bytes: &[u8]) -> usize {
 ///
 ///     Ok(())
 /// }
+/// # assert!(example().is_ok());
 /// ```
 pub struct FacetCollector {
     field: Field,
@@ -191,6 +189,11 @@ pub struct FacetSegmentCollector {
     collapse_facet_ords: Vec<u64>,
 }
 
+enum SkipResult {
+    Found,
+    NotFound,
+}
+
 fn skip<'a, I: Iterator<Item = &'a Facet>>(
     target: &[u8],
     collapse_it: &mut Peekable<I>,
@@ -200,14 +203,14 @@ fn skip<'a, I: Iterator<Item = &'a Facet>>(
             Some(facet_bytes) => match facet_bytes.encoded_str().as_bytes().cmp(target) {
                 Ordering::Less => {}
                 Ordering::Greater => {
-                    return SkipResult::OverStep;
+                    return SkipResult::NotFound;
                 }
                 Ordering::Equal => {
-                    return SkipResult::Reached;
+                    return SkipResult::Found;
                 }
             },
             None => {
-                return SkipResult::End;
+                return SkipResult::NotFound;
             }
         }
         collapse_it.next();
@@ -262,13 +265,10 @@ impl Collector for FacetCollector {
 
     fn for_segment(
         &self,
-        _: SegmentLocalId,
+        _: SegmentOrdinal,
         reader: &SegmentReader,
-    ) -> Result<FacetSegmentCollector> {
+    ) -> crate::Result<FacetSegmentCollector> {
-        let field_name = reader.schema().get_field_name(self.field);
+        let facet_reader = reader.facet_reader(self.field)?;
-        let facet_reader = reader.facet_reader(self.field).ok_or_else(|| {
-            TantivyError::SchemaError(format!("Field {:?} is not a facet field.", field_name))
-        })?;
 
         let mut collapse_mapping = Vec::new();
         let mut counts = Vec::new();
@@ -277,14 +277,14 @@ impl Collector for FacetCollector {
         let mut collapse_facet_it = self.facets.iter().peekable();
         collapse_facet_ords.push(0);
         {
-            let mut facet_streamer = facet_reader.facet_dict().range().into_stream();
+            let mut facet_streamer = facet_reader.facet_dict().range().into_stream()?;
             if facet_streamer.advance() {
                 'outer: loop {
                     // at the begining of this loop, facet_streamer
                     // is positionned on a term that has not been processed yet.
                     let skip_result = skip(facet_streamer.key(), &mut collapse_facet_it);
                     match skip_result {
-                        SkipResult::Reached => {
+                        SkipResult::Found => {
                             // we reach a facet we decided to collapse.
                             let collapse_depth = facet_depth(facet_streamer.key());
                             let mut collapsed_id = 0;
@@ -304,7 +304,7 @@ impl Collector for FacetCollector {
                             }
                             break;
                         }
-                        SkipResult::End | SkipResult::OverStep => {
+                        SkipResult::NotFound => {
                             collapse_mapping.push(0);
                             if !facet_streamer.advance() {
                                 break;
@@ -330,7 +330,7 @@ impl Collector for FacetCollector {
         false
     }
 
-    fn merge_fruits(&self, segments_facet_counts: Vec<FacetCounts>) -> Result<FacetCounts> {
+    fn merge_fruits(&self, segments_facet_counts: Vec<FacetCounts>) -> crate::Result<FacetCounts> {
         let mut facet_counts: BTreeMap<Facet, u64> = BTreeMap::new();
         for segment_facet_counts in segments_facet_counts {
             for (facet, count) in segment_facet_counts.facet_counts {
@@ -371,9 +371,12 @@ impl SegmentCollector for FacetSegmentCollector {
             }
             let mut facet = vec![];
             let facet_ord = self.collapse_facet_ords[collapsed_facet_ord];
-            facet_dict.ord_to_term(facet_ord as u64, &mut facet);
+            // TODO handle errors.
-            // TODO
+            if facet_dict.ord_to_term(facet_ord as u64, &mut facet).is_ok() {
-            facet_counts.insert(Facet::from_encoded(facet).unwrap(), count);
+                if let Ok(facet) = Facet::from_encoded(facet) {
+                    facet_counts.insert(facet, count);
+                }
+            }
         }
         FacetCounts { facet_counts }
     }
@@ -398,6 +401,8 @@ impl<'a> Iterator for FacetChildIterator<'a> {
 }
 
 impl FacetCounts {
+    /// Returns an iterator over all of the facet count pairs inside this result.
+    /// See the documentation for `FacetCollector` for a usage example.
     pub fn get<T>(&self, facet_from: T) -> FacetChildIterator<'_>
     where
         Facet: From<T>,
@@ -417,6 +422,8 @@ impl FacetCounts {
         FacetChildIterator { underlying }
     }
 
+    /// Returns a vector of top `k` facets with their counts, sorted highest-to-lowest by counts.
+    /// See the documentation for `FacetCollector` for a usage example.
     pub fn top_k<T>(&self, facet: T, k: usize) -> Vec<(&Facet, u64)>
     where
         Facet: From<T>,
@@ -454,9 +461,11 @@ impl FacetCounts {
 #[cfg(test)]
 mod tests {
     use super::{FacetCollector, FacetCounts};
+    use crate::collector::Count;
     use crate::core::Index;
-    use crate::query::AllQuery;
+    use crate::query::{AllQuery, QueryParser, TermQuery};
-    use crate::schema::{Document, Facet, Field, Schema};
+    use crate::schema::{Document, Facet, Field, IndexRecordOption, Schema, INDEXED};
+    use crate::Term;
     use rand::distributions::Uniform;
     use rand::prelude::SliceRandom;
     use rand::{thread_rng, Rng};
@@ -465,11 +474,11 @@ mod tests {
     #[test]
     fn test_facet_collector_drilldown() {
         let mut schema_builder = Schema::builder();
-        let facet_field = schema_builder.add_facet_field("facet");
+        let facet_field = schema_builder.add_facet_field("facet", INDEXED);
         let schema = schema_builder.build();
         let index = Index::create_in_ram(schema);
 
-        let mut index_writer = index.writer_with_num_threads(1, 3_000_000).unwrap();
+        let mut index_writer = index.writer_for_tests().unwrap();
         let num_facets: usize = 3 * 4 * 5;
         let facets: Vec<Facet> = (0..num_facets)
             .map(|mut n| {
@@ -517,7 +526,7 @@ mod tests {
     #[should_panic(expected = "Tried to add a facet which is a descendant of \
                                an already added facet.")]
     fn test_misused_facet_collector() {
-        let mut facet_collector = FacetCollector::for_field(Field(0));
+        let mut facet_collector = FacetCollector::for_field(Field::from_field_id(0));
         facet_collector.add_facet(Facet::from("/country"));
         facet_collector.add_facet(Facet::from("/country/europe"));
     }
@@ -525,15 +534,15 @@ mod tests {
     #[test]
     fn test_doc_unsorted_multifacet() {
         let mut schema_builder = Schema::builder();
-        let facet_field = schema_builder.add_facet_field("facets");
+        let facet_field = schema_builder.add_facet_field("facets", INDEXED);
         let schema = schema_builder.build();
         let index = Index::create_in_ram(schema);
-        let mut index_writer = index.writer_with_num_threads(1, 3_000_000).unwrap();
+        let mut index_writer = index.writer_for_tests().unwrap();
         index_writer.add_document(doc!(
-            facet_field => Facet::from_text(&"/subjects/A/a"),
+            facet_field => Facet::from_text(&"/subjects/A/a").unwrap(),
-            facet_field => Facet::from_text(&"/subjects/B/a"),
+            facet_field => Facet::from_text(&"/subjects/B/a").unwrap(),
-            facet_field => Facet::from_text(&"/subjects/A/b"),
+            facet_field => Facet::from_text(&"/subjects/A/b").unwrap(),
-            facet_field => Facet::from_text(&"/subjects/B/b"),
+            facet_field => Facet::from_text(&"/subjects/B/b").unwrap(),
         ));
         index_writer.commit().unwrap();
         let reader = index.reader().unwrap();
@@ -546,9 +555,59 @@ mod tests {
         assert_eq!(facets[0].1, 1);
     }
 
+    #[test]
+    fn test_doc_search_by_facet() -> crate::Result<()> {
+        let mut schema_builder = Schema::builder();
+        let facet_field = schema_builder.add_facet_field("facet", INDEXED);
+        let schema = schema_builder.build();
+        let index = Index::create_in_ram(schema);
+        let mut index_writer = index.writer_for_tests()?;
+        index_writer.add_document(doc!(
+            facet_field => Facet::from_text(&"/A/A").unwrap(),
+        ));
+        index_writer.add_document(doc!(
+            facet_field => Facet::from_text(&"/A/B").unwrap(),
+        ));
+        index_writer.add_document(doc!(
+            facet_field => Facet::from_text(&"/A/C/A").unwrap(),
+        ));
+        index_writer.add_document(doc!(
+            facet_field => Facet::from_text(&"/D/C/A").unwrap(),
+        ));
+        index_writer.commit()?;
+        let reader = index.reader()?;
+        let searcher = reader.searcher();
+        assert_eq!(searcher.num_docs(), 4);
+
+        let count_facet = |facet_str: &str| {
+            let term = Term::from_facet(facet_field, &Facet::from_text(facet_str).unwrap());
+            searcher
+                .search(&TermQuery::new(term, IndexRecordOption::Basic), &Count)
+                .unwrap()
+        };
+
+        assert_eq!(count_facet("/"), 4);
+        assert_eq!(count_facet("/A"), 3);
+        assert_eq!(count_facet("/A/B"), 1);
+        assert_eq!(count_facet("/A/C"), 1);
+        assert_eq!(count_facet("/A/C/A"), 1);
+        assert_eq!(count_facet("/C/A"), 0);
+
+        let query_parser = QueryParser::for_index(&index, vec![]);
+        {
+            let query = query_parser.parse_query("facet:/A/B")?;
+            assert_eq!(1, searcher.search(&query, &Count).unwrap());
+        }
+        {
+            let query = query_parser.parse_query("facet:/A")?;
+            assert_eq!(3, searcher.search(&query, &Count)?);
+        }
+        Ok(())
+    }
+
     #[test]
     fn test_non_used_facet_collector() {
-        let mut facet_collector = FacetCollector::for_field(Field(0));
+        let mut facet_collector = FacetCollector::for_field(Field::from_field_id(0));
         facet_collector.add_facet(Facet::from("/country"));
         facet_collector.add_facet(Facet::from("/countryeurope"));
     }
@@ -556,7 +615,7 @@ mod tests {
     #[test]
     fn test_facet_collector_topk() {
         let mut schema_builder = Schema::builder();
-        let facet_field = schema_builder.add_facet_field("facet");
+        let facet_field = schema_builder.add_facet_field("facet", INDEXED);
         let schema = schema_builder.build();
         let index = Index::create_in_ram(schema);
 
@@ -578,7 +637,7 @@ mod tests {
             .collect();
         docs[..].shuffle(&mut thread_rng());
 
-        let mut index_writer = index.writer_with_num_threads(1, 3_000_000).unwrap();
+        let mut index_writer = index.writer_for_tests().unwrap();
         for doc in docs {
             index_writer.add_document(doc);
         }
@@ -602,23 +661,57 @@ mod tests {
         }
     }
 
+    #[test]
+    fn test_facet_collector_topk_tie_break() -> crate::Result<()> {
+        let mut schema_builder = Schema::builder();
+        let facet_field = schema_builder.add_facet_field("facet", INDEXED);
+        let schema = schema_builder.build();
+        let index = Index::create_in_ram(schema);
+
+        let docs: Vec<Document> = vec![("b", 2), ("a", 2), ("c", 4)]
+            .into_iter()
+            .flat_map(|(c, count)| {
+                let facet = Facet::from(&format!("/facet/{}", c));
+                let doc = doc!(facet_field => facet);
+                iter::repeat(doc).take(count)
+            })
+            .collect();
+
+        let mut index_writer = index.writer_for_tests()?;
+        for doc in docs {
+            index_writer.add_document(doc);
+        }
+        index_writer.commit()?;
+
+        let searcher = index.reader()?.searcher();
+        let mut facet_collector = FacetCollector::for_field(facet_field);
+        facet_collector.add_facet("/facet");
+        let counts: FacetCounts = searcher.search(&AllQuery, &facet_collector)?;
+
+        let facets: Vec<(&Facet, u64)> = counts.top_k("/facet", 2);
+        assert_eq!(
+            facets,
+            vec![(&Facet::from("/facet/c"), 4), (&Facet::from("/facet/a"), 2)]
+        );
+        Ok(())
+    }
 }
 
 #[cfg(all(test, feature = "unstable"))]
 mod bench {
 
-    use collector::FacetCollector;
+    use crate::collector::FacetCollector;
-    use query::AllQuery;
+    use crate::query::AllQuery;
-    use rand::{thread_rng, Rng};
+    use crate::schema::{Facet, Schema, INDEXED};
-    use schema::Facet;
+    use crate::Index;
-    use schema::Schema;
+    use rand::seq::SliceRandom;
+    use rand::thread_rng;
     use test::Bencher;
-    use Index;
 
     #[bench]
     fn bench_facet_collector(b: &mut Bencher) {
         let mut schema_builder = Schema::builder();
-        let facet_field = schema_builder.add_facet_field("facet");
+        let facet_field = schema_builder.add_facet_field("facet", INDEXED);
         let schema = schema_builder.build();
         let index = Index::create_in_ram(schema);
 
@@ -630,16 +723,16 @@ mod bench {
             }
         }
         // 40425 docs
-        thread_rng().shuffle(&mut docs[..]);
+        docs[..].shuffle(&mut thread_rng());
 
-        let mut index_writer = index.writer_with_num_threads(1, 3_000_000).unwrap();
+        let mut index_writer = index.writer_for_tests().unwrap();
         for doc in docs {
             index_writer.add_document(doc);
         }
         index_writer.commit().unwrap();
         let reader = index.reader().unwrap();
         b.iter(|| {
-            let searcher = index.searcher();
+            let searcher = reader.searcher();
             let facet_collector = FacetCollector::for_field(facet_field);
             searcher.search(&AllQuery, &facet_collector).unwrap();
         });

src/collector/filter_collector_wrapper.rs

@@ -0,0 +1,183 @@
+// # Custom collector example
+//
+// This example shows how you can implement your own
+// collector. As an example, we will compute a collector
+// that computes the standard deviation of a given fast field.
+//
+// Of course, you can have a look at the tantivy's built-in collectors
+// such as the `CountCollector` for more examples.
+
+// ---
+// Importing tantivy...
+use std::marker::PhantomData;
+
+use crate::collector::{Collector, SegmentCollector};
+use crate::fastfield::{DynamicFastFieldReader, FastFieldReader, FastValue};
+use crate::schema::Field;
+use crate::{Score, SegmentReader, TantivyError};
+
+/// The `FilterCollector` collector filters docs using a u64 fast field value and a predicate.
+/// Only the documents for which the predicate returned "true" will be passed on to the next collector.
+///
+/// ```rust
+/// use tantivy::collector::{TopDocs, FilterCollector};
+/// use tantivy::query::QueryParser;
+/// use tantivy::schema::{Schema, TEXT, INDEXED, FAST};
+/// use tantivy::{doc, DocAddress, Index};
+///
+/// let mut schema_builder = Schema::builder();
+/// let title = schema_builder.add_text_field("title", TEXT);
+/// let price = schema_builder.add_u64_field("price", INDEXED | FAST);
+/// let schema = schema_builder.build();
+/// let index = Index::create_in_ram(schema);
+///
+/// let mut index_writer = index.writer_with_num_threads(1, 10_000_000).unwrap();
+/// index_writer.add_document(doc!(title => "The Name of the Wind", price => 30_200u64));
+/// index_writer.add_document(doc!(title => "The Diary of Muadib", price => 29_240u64));
+/// index_writer.add_document(doc!(title => "A Dairy Cow", price => 21_240u64));
+/// index_writer.add_document(doc!(title => "The Diary of a Young Girl", price => 20_120u64));
+/// assert!(index_writer.commit().is_ok());
+///
+/// let reader = index.reader().unwrap();
+/// let searcher = reader.searcher();
+///
+/// let query_parser = QueryParser::for_index(&index, vec![title]);
+/// let query = query_parser.parse_query("diary").unwrap();
+/// let no_filter_collector = FilterCollector::new(price, &|value: u64| value > 20_120u64, TopDocs::with_limit(2));
+/// let top_docs = searcher.search(&query, &no_filter_collector).unwrap();
+///
+/// assert_eq!(top_docs.len(), 1);
+/// assert_eq!(top_docs[0].1, DocAddress::new(0, 1));
+///
+/// let filter_all_collector: FilterCollector<_, _, u64> = FilterCollector::new(price, &|value| value < 5u64, TopDocs::with_limit(2));
+/// let filtered_top_docs = searcher.search(&query, &filter_all_collector).unwrap();
+///
+/// assert_eq!(filtered_top_docs.len(), 0);
+/// ```
+pub struct FilterCollector<TCollector, TPredicate, TPredicateValue: FastValue>
+where
+    TPredicate: 'static + Clone,
+{
+    field: Field,
+    collector: TCollector,
+    predicate: TPredicate,
+    t_predicate_value: PhantomData<TPredicateValue>,
+}
+
+impl<TCollector, TPredicate, TPredicateValue: FastValue>
+    FilterCollector<TCollector, TPredicate, TPredicateValue>
+where
+    TCollector: Collector + Send + Sync,
+    TPredicate: Fn(TPredicateValue) -> bool + Send + Sync + Clone,
+{
+    /// Create a new FilterCollector.
+    pub fn new(
+        field: Field,
+        predicate: TPredicate,
+        collector: TCollector,
+    ) -> FilterCollector<TCollector, TPredicate, TPredicateValue> {
+        FilterCollector {
+            field,
+            predicate,
+            collector,
+            t_predicate_value: PhantomData,
+        }
+    }
+}
+
+impl<TCollector, TPredicate, TPredicateValue: FastValue> Collector
+    for FilterCollector<TCollector, TPredicate, TPredicateValue>
+where
+    TCollector: Collector + Send + Sync,
+    TPredicate: 'static + Fn(TPredicateValue) -> bool + Send + Sync + Clone,
+    TPredicateValue: FastValue,
+{
+    // That's the type of our result.
+    // Our standard deviation will be a float.
+    type Fruit = TCollector::Fruit;
+
+    type Child = FilterSegmentCollector<TCollector::Child, TPredicate, TPredicateValue>;
+
+    fn for_segment(
+        &self,
+        segment_local_id: u32,
+        segment_reader: &SegmentReader,
+    ) -> crate::Result<FilterSegmentCollector<TCollector::Child, TPredicate, TPredicateValue>> {
+        let schema = segment_reader.schema();
+        let field_entry = schema.get_field_entry(self.field);
+        if !field_entry.is_fast() {
+            return Err(TantivyError::SchemaError(format!(
+                "Field {:?} is not a fast field.",
+                field_entry.name()
+            )));
+        }
+        let requested_type = TPredicateValue::to_type();
+        let field_schema_type = field_entry.field_type().value_type();
+        if requested_type != field_schema_type {
+            return Err(TantivyError::SchemaError(format!(
+                "Field {:?} is of type {:?}!={:?}",
+                field_entry.name(),
+                requested_type,
+                field_schema_type
+            )));
+        }
+
+        let fast_field_reader = segment_reader
+            .fast_fields()
+            .typed_fast_field_reader(self.field)?;
+
+        let segment_collector = self
+            .collector
+            .for_segment(segment_local_id, segment_reader)?;
+
+        Ok(FilterSegmentCollector {
+            fast_field_reader,
+            segment_collector,
+            predicate: self.predicate.clone(),
+            t_predicate_value: PhantomData,
+        })
+    }
+
+    fn requires_scoring(&self) -> bool {
+        self.collector.requires_scoring()
+    }
+
+    fn merge_fruits(
+        &self,
+        segment_fruits: Vec<<TCollector::Child as SegmentCollector>::Fruit>,
+    ) -> crate::Result<TCollector::Fruit> {
+        self.collector.merge_fruits(segment_fruits)
+    }
+}
+
+pub struct FilterSegmentCollector<TSegmentCollector, TPredicate, TPredicateValue>
+where
+    TPredicate: 'static,
+    TPredicateValue: FastValue,
+{
+    fast_field_reader: DynamicFastFieldReader<TPredicateValue>,
+    segment_collector: TSegmentCollector,
+    predicate: TPredicate,
+    t_predicate_value: PhantomData<TPredicateValue>,
+}
+
+impl<TSegmentCollector, TPredicate, TPredicateValue> SegmentCollector
+    for FilterSegmentCollector<TSegmentCollector, TPredicate, TPredicateValue>
+where
+    TSegmentCollector: SegmentCollector,
+    TPredicate: 'static + Fn(TPredicateValue) -> bool + Send + Sync,
+    TPredicateValue: FastValue,
+{
+    type Fruit = TSegmentCollector::Fruit;
+
+    fn collect(&mut self, doc: u32, score: Score) {
+        let value = self.fast_field_reader.get(doc);
+        if (self.predicate)(value) {
+            self.segment_collector.collect(doc, score)
+        }
+    }
+
+    fn harvest(self) -> <TSegmentCollector as SegmentCollector>::Fruit {
+        self.segment_collector.harvest()
+    }
+}

src/collector/histogram_collector.rs

@@ -0,0 +1,291 @@
+use crate::collector::{Collector, SegmentCollector};
+use crate::fastfield::{DynamicFastFieldReader, FastFieldReader, FastValue};
+use crate::schema::{Field, Type};
+use crate::{DocId, Score};
+use fastdivide::DividerU64;
+
+/// Histogram builds an histogram of the values of a fastfield for the
+/// collected DocSet.
+///
+/// At construction, it is given parameters that define a partition of an interval
+/// [min_val, max_val) into N buckets with the same width.
+/// The ith bucket is then defined by `[min_val + i * bucket_width, min_val + (i+1) * bucket_width)`
+///
+/// An histogram is then defined as a `Vec<u64>` of length `num_buckets`, that contains a count of
+/// documents for each value bucket.
+///
+/// See also [`HistogramCollector::new()`].
+///
+/// # Warning
+///
+/// f64 field. are not supported.
+#[derive(Clone)]
+pub struct HistogramCollector {
+    min_value: u64,
+    num_buckets: usize,
+    divider: DividerU64,
+    field: Field,
+}
+
+impl HistogramCollector {
+    /// Builds a new HistogramCollector.
+    ///
+    /// The scale/range of the histogram is not dynamic. It is required to
+    /// define it by supplying following parameter:
+    ///  - `min_value`: the minimum value that can be recorded in the histogram.
+    ///  - `bucket_width`: the length of the interval that is associated to each buckets.
+    ///  - `num_buckets`: The overall number of buckets.
+    ///
+    /// Together, this parameters define a partition of `[min_value, min_value + num_buckets * bucket_width)`
+    /// into `num_buckets` intervals of width bucket that we call `bucket`.
+    ///
+    /// # Disclaimer
+    /// This function panics if the field given is of type f64.
+    pub fn new<TFastValue: FastValue>(
+        field: Field,
+        min_value: TFastValue,
+        bucket_width: u64,
+        num_buckets: usize,
+    ) -> HistogramCollector {
+        let fast_type = TFastValue::to_type();
+        assert!(fast_type == Type::U64 || fast_type == Type::I64 || fast_type == Type::Date);
+        HistogramCollector {
+            min_value: min_value.to_u64(),
+            num_buckets,
+            field,
+            divider: DividerU64::divide_by(bucket_width),
+        }
+    }
+}
+
+struct HistogramComputer {
+    counts: Vec<u64>,
+    min_value: u64,
+    divider: DividerU64,
+}
+
+impl HistogramComputer {
+    #[inline]
+    pub(crate) fn add_value(&mut self, value: u64) {
+        if value < self.min_value {
+            return;
+        }
+        let delta = value - self.min_value;
+        let delta_u64 = delta.to_u64();
+        let bucket_id: usize = self.divider.divide(delta_u64) as usize;
+        if bucket_id < self.counts.len() {
+            self.counts[bucket_id] += 1;
+        }
+    }
+
+    fn harvest(self) -> Vec<u64> {
+        self.counts
+    }
+}
+pub struct SegmentHistogramCollector {
+    histogram_computer: HistogramComputer,
+    ff_reader: DynamicFastFieldReader<u64>,
+}
+
+impl SegmentCollector for SegmentHistogramCollector {
+    type Fruit = Vec<u64>;
+
+    fn collect(&mut self, doc: DocId, _score: Score) {
+        let value = self.ff_reader.get(doc);
+        self.histogram_computer.add_value(value);
+    }
+
+    fn harvest(self) -> Self::Fruit {
+        self.histogram_computer.harvest()
+    }
+}
+
+impl Collector for HistogramCollector {
+    type Fruit = Vec<u64>;
+    type Child = SegmentHistogramCollector;
+
+    fn for_segment(
+        &self,
+        _segment_local_id: crate::SegmentOrdinal,
+        segment: &crate::SegmentReader,
+    ) -> crate::Result<Self::Child> {
+        let ff_reader = segment.fast_fields().u64_lenient(self.field)?;
+        Ok(SegmentHistogramCollector {
+            histogram_computer: HistogramComputer {
+                counts: vec![0; self.num_buckets],
+                min_value: self.min_value,
+                divider: self.divider,
+            },
+            ff_reader,
+        })
+    }
+
+    fn requires_scoring(&self) -> bool {
+        false
+    }
+
+    fn merge_fruits(&self, child_histograms: Vec<Vec<u64>>) -> crate::Result<Vec<u64>> {
+        Ok(add_vecs(child_histograms, self.num_buckets))
+    }
+}
+
+pub fn add_arrays_into(acc: &mut [u64], add: &[u64]) {
+    assert_eq!(acc.len(), add.len());
+    for (dest_bucket, bucket_count) in acc.iter_mut().zip(add) {
+        *dest_bucket += bucket_count;
+    }
+}
+
+fn add_vecs(mut vals_list: Vec<Vec<u64>>, len: usize) -> Vec<u64> {
+    let mut acc = vals_list.pop().unwrap_or_else(|| vec![0u64; len]);
+    assert_eq!(acc.len(), len);
+    for vals in vals_list {
+        add_arrays_into(&mut acc, &vals);
+    }
+    acc
+}
+
+#[cfg(test)]
+mod tests {
+    use super::{add_vecs, HistogramCollector, HistogramComputer};
+    use crate::chrono::{TimeZone, Utc};
+    use crate::schema::{Schema, FAST};
+    use crate::{doc, query, Index};
+    use fastdivide::DividerU64;
+    use query::AllQuery;
+
+    #[test]
+    fn test_add_histograms_simple() {
+        assert_eq!(
+            add_vecs(vec![vec![1, 0, 3], vec![11, 2, 3], vec![0, 0, 1]], 3),
+            vec![12, 2, 7]
+        )
+    }
+
+    #[test]
+    fn test_add_histograms_empty() {
+        assert_eq!(add_vecs(vec![], 3), vec![0, 0, 0])
+    }
+
+    #[test]
+    fn test_histogram_builder_simple() {
+        // [1..3)
+        // [3..5)
+        // ..
+        // [9..11)
+        let mut histogram_computer = HistogramComputer {
+            counts: vec![0; 5],
+            min_value: 1,
+            divider: DividerU64::divide_by(2),
+        };
+        histogram_computer.add_value(1);
+        histogram_computer.add_value(7);
+        assert_eq!(histogram_computer.harvest(), vec![1, 0, 0, 1, 0]);
+    }
+
+    #[test]
+    fn test_histogram_too_low_is_ignored() {
+        let mut histogram_computer = HistogramComputer {
+            counts: vec![0; 5],
+            min_value: 2,
+            divider: DividerU64::divide_by(2),
+        };
+        histogram_computer.add_value(0);
+        assert_eq!(histogram_computer.harvest(), vec![0, 0, 0, 0, 0]);
+    }
+
+    #[test]
+    fn test_histogram_too_high_is_ignored() {
+        let mut histogram_computer = HistogramComputer {
+            counts: vec![0u64; 5],
+            min_value: 0,
+            divider: DividerU64::divide_by(2),
+        };
+        histogram_computer.add_value(10);
+        assert_eq!(histogram_computer.harvest(), vec![0, 0, 0, 0, 0]);
+    }
+    #[test]
+    fn test_no_segments() -> crate::Result<()> {
+        let mut schema_builder = Schema::builder();
+        let val_field = schema_builder.add_u64_field("val_field", FAST);
+        let schema = schema_builder.build();
+        let index = Index::create_in_ram(schema);
+        let reader = index.reader()?;
+        let searcher = reader.searcher();
+        let all_query = AllQuery;
+        let histogram_collector = HistogramCollector::new(val_field, 0u64, 2, 5);
+        let histogram = searcher.search(&all_query, &histogram_collector)?;
+        assert_eq!(histogram, vec![0; 5]);
+        Ok(())
+    }
+
+    #[test]
+    fn test_histogram_i64() -> crate::Result<()> {
+        let mut schema_builder = Schema::builder();
+        let val_field = schema_builder.add_i64_field("val_field", FAST);
+        let schema = schema_builder.build();
+        let index = Index::create_in_ram(schema);
+        let mut writer = index.writer_with_num_threads(1, 4_000_000)?;
+        writer.add_document(doc!(val_field=>12i64));
+        writer.add_document(doc!(val_field=>-30i64));
+        writer.add_document(doc!(val_field=>-12i64));
+        writer.add_document(doc!(val_field=>-10i64));
+        writer.commit()?;
+        let reader = index.reader()?;
+        let searcher = reader.searcher();
+        let all_query = AllQuery;
+        let histogram_collector = HistogramCollector::new(val_field, -20i64, 10u64, 4);
+        let histogram = searcher.search(&all_query, &histogram_collector)?;
+        assert_eq!(histogram, vec![1, 1, 0, 1]);
+        Ok(())
+    }
+
+    #[test]
+    fn test_histogram_merge() -> crate::Result<()> {
+        let mut schema_builder = Schema::builder();
+        let val_field = schema_builder.add_i64_field("val_field", FAST);
+        let schema = schema_builder.build();
+        let index = Index::create_in_ram(schema);
+        let mut writer = index.writer_with_num_threads(1, 4_000_000)?;
+        writer.add_document(doc!(val_field=>12i64));
+        writer.commit()?;
+        writer.add_document(doc!(val_field=>-30i64));
+        writer.commit()?;
+        writer.add_document(doc!(val_field=>-12i64));
+        writer.commit()?;
+        writer.add_document(doc!(val_field=>-10i64));
+        writer.commit()?;
+        let reader = index.reader()?;
+        let searcher = reader.searcher();
+        let all_query = AllQuery;
+        let histogram_collector = HistogramCollector::new(val_field, -20i64, 10u64, 4);
+        let histogram = searcher.search(&all_query, &histogram_collector)?;
+        assert_eq!(histogram, vec![1, 1, 0, 1]);
+        Ok(())
+    }
+
+    #[test]
+    fn test_histogram_dates() -> crate::Result<()> {
+        let mut schema_builder = Schema::builder();
+        let date_field = schema_builder.add_date_field("date_field", FAST);
+        let schema = schema_builder.build();
+        let index = Index::create_in_ram(schema);
+        let mut writer = index.writer_with_num_threads(1, 4_000_000)?;
+        writer.add_document(doc!(date_field=>Utc.ymd(1982, 9, 17).and_hms(0, 0,0)));
+        writer.add_document(doc!(date_field=>Utc.ymd(1986, 3, 9).and_hms(0, 0, 0)));
+        writer.add_document(doc!(date_field=>Utc.ymd(1983, 9, 27).and_hms(0, 0, 0)));
+        writer.commit()?;
+        let reader = index.reader()?;
+        let searcher = reader.searcher();
+        let all_query = AllQuery;
+        let week_histogram_collector = HistogramCollector::new(
+            date_field,
+            Utc.ymd(1980, 1, 1).and_hms(0, 0, 0),
+            3600 * 24 * 365, // it is just for a unit test... sorry leap years.
+            10,
+        );
+        let week_histogram = searcher.search(&all_query, &week_histogram_collector)?;
+        assert_eq!(week_histogram, vec![0, 0, 1, 1, 0, 0, 1, 0, 0, 0]);
+        Ok(())
+    }
+}

src/collector/int_facet_collector.rs

@@ -1,127 +0,0 @@
-use std::cmp::Eq;
-use std::collections::HashMap;
-use std::hash::Hash;
-
-use collector::Collector;
-use fastfield::FastFieldReader;
-use schema::Field;
-
-use DocId;
-use Result;
-use Score;
-use SegmentReader;
-use SegmentLocalId;
-
-
-/// Facet collector  for i64/u64 fast field
-pub struct IntFacetCollector<T>
-where
-    T: FastFieldReader,
-    T::ValueType: Eq + Hash,
-{
-    counters: HashMap<T::ValueType, u64>,
-    field: Field,
-    ff_reader: Option<T>,
-}
-
-
-impl<T> IntFacetCollector<T>
-where
-    T: FastFieldReader,
-    T::ValueType: Eq + Hash,
-{
-    /// Creates a new facet collector for aggregating a given field.
-    pub fn new(field: Field) -> IntFacetCollector<T> {
-        IntFacetCollector {
-            counters: HashMap::new(),
-            field: field,
-            ff_reader: None,
-        }
-    }
-}
-
-
-impl<T> Collector for IntFacetCollector<T>
-where
-    T: FastFieldReader,
-    T::ValueType: Eq + Hash,
-{
-    fn set_segment(&mut self, _: SegmentLocalId, reader: &SegmentReader) -> Result<()> {
-        self.ff_reader = Some(reader.get_fast_field_reader(self.field)?);
-        Ok(())
-    }
-
-    fn collect(&mut self, doc: DocId, _: Score) {
-        let val = self.ff_reader
-            .as_ref()
-            .expect(
-                "collect() was called before set_segment. \
-                This should never happen.",
-            )
-            .get(doc);
-        *(self.counters.entry(val).or_insert(0)) += 1;
-    }
-}
-
-
-
-#[cfg(test)]
-mod tests {
-
-    use collector::{chain, IntFacetCollector};
-    use query::QueryParser;
-    use fastfield::{I64FastFieldReader, U64FastFieldReader};
-    use schema::{self, FAST, STRING};
-    use Index;
-
-    #[test]
-    // create 10 documents, set num field value to 0 or 1 for even/odd ones
-    // make sure we have facet counters correctly filled
-    fn test_facet_collector_results() {
-
-        let mut schema_builder = schema::Schema::builder();
-        let num_field_i64 = schema_builder.add_i64_field("num_i64", FAST);
-        let num_field_u64 = schema_builder.add_u64_field("num_u64", FAST);
-        let num_field_f64 = schema_builder.add_f64_field("num_f64", FAST);
-        let text_field = schema_builder.add_text_field("text", STRING);
-        let schema = schema_builder.build();
-
-        let index = Index::create_in_ram(schema.clone());
-
-        {
-            let mut index_writer = index.writer_with_num_threads(1, 3_000_000).unwrap();
-            {
-                for i in 0u64..10u64 {
-                    index_writer.add_document(doc!(
-                        num_field_i64 => ((i as i64) % 3i64) as i64,
-                        num_field_u64 => (i % 2u64) as u64,
-                        num_field_f64 => (i % 4u64) as f64,
-                        text_field => "text"
-                    ));
-                }
-            }
-            assert_eq!(index_writer.commit().unwrap(), 10u64);
-        }
-
-        let searcher = index.reader().searcher();
-        let mut ffvf_i64: IntFacetCollector<I64FastFieldReader> = IntFacetCollector::new(num_field_i64);
-        let mut ffvf_u64: IntFacetCollector<U64FastFieldReader> = IntFacetCollector::new(num_field_u64);
-        let mut ffvf_f64: IntFacetCollector<F64FastFieldReader> = IntFacetCollector::new(num_field_f64);
-
-        {
-            // perform the query
-            let mut facet_collectors = chain().push(&mut ffvf_i64).push(&mut ffvf_u64).push(&mut ffvf_f64);
-            let mut query_parser = QueryParser::for_index(index, vec![text_field]);
-            let query = query_parser.parse_query("text:text").unwrap();
-            query.search(&searcher, &mut facet_collectors).unwrap();
-        }
-
-        assert_eq!(ffvf_u64.counters[&0], 5);
-        assert_eq!(ffvf_u64.counters[&1], 5);
-        assert_eq!(ffvf_i64.counters[&0], 4);
-        assert_eq!(ffvf_i64.counters[&1], 3);
-        assert_eq!(ffvf_f64.counters[&0.0], 3);
-        assert_eq!(ffvf_f64.counters[&2.0], 2);
-
-    }
-}

src/collector/mod.rs

@@ -35,7 +35,6 @@ The resulting `Fruit` will then be a typed tuple with each collector's original
 in their respective position.
 
 ```rust
-# extern crate tantivy;
 # use tantivy::schema::*;
 # use tantivy::*;
 # use tantivy::query::*;
@@ -86,15 +85,17 @@ See the `custom_collector` example.
 */
 
 use crate::DocId;
-use crate::Result;
 use crate::Score;
-use crate::SegmentLocalId;
+use crate::SegmentOrdinal;
 use crate::SegmentReader;
 use downcast_rs::impl_downcast;
 
 mod count_collector;
 pub use self::count_collector::Count;
 
+mod histogram_collector;
+pub use histogram_collector::HistogramCollector;
+
 mod multi_collector;
 pub use self::multi_collector::MultiCollector;
 
@@ -111,6 +112,14 @@ pub use self::tweak_score_top_collector::{ScoreSegmentTweaker, ScoreTweaker};
 
 mod facet_collector;
 pub use self::facet_collector::FacetCollector;
+pub use self::facet_collector::FacetCounts;
+use crate::query::Weight;
+
+mod docset_collector;
+pub use self::docset_collector::DocSetCollector;
+
+mod filter_collector_wrapper;
+pub use self::filter_collector_wrapper::FilterCollector;
 
 /// `Fruit` is the type for the result of our collection.
 /// e.g. `usize` for the `Count` collector.
@@ -134,28 +143,109 @@ impl<T> Fruit for T where T: Send + downcast_rs::Downcast {}
 /// The collection logic itself is in the `SegmentCollector`.
 ///
 /// Segments are not guaranteed to be visited in any specific order.
-pub trait Collector: Sync {
+pub trait Collector: Sync + Send {
     /// `Fruit` is the type for the result of our collection.
     /// e.g. `usize` for the `Count` collector.
     type Fruit: Fruit;
 
     /// Type of the `SegmentCollector` associated to this collector.
-    type Child: SegmentCollector<Fruit = Self::Fruit>;
+    type Child: SegmentCollector;
 
     /// `set_segment` is called before beginning to enumerate
     /// on this segment.
     fn for_segment(
         &self,
-        segment_local_id: SegmentLocalId,
+        segment_local_id: SegmentOrdinal,
         segment: &SegmentReader,
-    ) -> Result<Self::Child>;
+    ) -> crate::Result<Self::Child>;
 
     /// Returns true iff the collector requires to compute scores for documents.
     fn requires_scoring(&self) -> bool;
 
     /// Combines the fruit associated to the collection of each segments
     /// into one fruit.
-    fn merge_fruits(&self, segment_fruits: Vec<Self::Fruit>) -> Result<Self::Fruit>;
+    fn merge_fruits(
+        &self,
+        segment_fruits: Vec<<Self::Child as SegmentCollector>::Fruit>,
+    ) -> crate::Result<Self::Fruit>;
+
+    /// Created a segment collector and
+    fn collect_segment(
+        &self,
+        weight: &dyn Weight,
+        segment_ord: u32,
+        reader: &SegmentReader,
+    ) -> crate::Result<<Self::Child as SegmentCollector>::Fruit> {
+        let mut segment_collector = self.for_segment(segment_ord as u32, reader)?;
+
+        if let Some(delete_bitset) = reader.delete_bitset() {
+            weight.for_each(reader, &mut |doc, score| {
+                if delete_bitset.is_alive(doc) {
+                    segment_collector.collect(doc, score);
+                }
+            })?;
+        } else {
+            weight.for_each(reader, &mut |doc, score| {
+                segment_collector.collect(doc, score);
+            })?;
+        }
+        Ok(segment_collector.harvest())
+    }
+}
+
+impl<TSegmentCollector: SegmentCollector> SegmentCollector for Option<TSegmentCollector> {
+    type Fruit = Option<TSegmentCollector::Fruit>;
+
+    fn collect(&mut self, doc: DocId, score: Score) {
+        if let Some(segment_collector) = self {
+            segment_collector.collect(doc, score);
+        }
+    }
+
+    fn harvest(self) -> Self::Fruit {
+        self.map(|segment_collector| segment_collector.harvest())
+    }
+}
+
+impl<TCollector: Collector> Collector for Option<TCollector> {
+    type Fruit = Option<TCollector::Fruit>;
+
+    type Child = Option<<TCollector as Collector>::Child>;
+
+    fn for_segment(
+        &self,
+        segment_local_id: SegmentOrdinal,
+        segment: &SegmentReader,
+    ) -> crate::Result<Self::Child> {
+        Ok(if let Some(inner) = self {
+            let inner_segment_collector = inner.for_segment(segment_local_id, segment)?;
+            Some(inner_segment_collector)
+        } else {
+            None
+        })
+    }
+
+    fn requires_scoring(&self) -> bool {
+        self.as_ref()
+            .map(|inner| inner.requires_scoring())
+            .unwrap_or(false)
+    }
+
+    fn merge_fruits(
+        &self,
+        segment_fruits: Vec<<Self::Child as SegmentCollector>::Fruit>,
+    ) -> crate::Result<Self::Fruit> {
+        if let Some(inner) = self.as_ref() {
+            let inner_segment_fruits: Vec<_> = segment_fruits
+                .into_iter()
+                .flat_map(|fruit_opt| fruit_opt.into_iter())
+                .collect();
+            let fruit = inner.merge_fruits(inner_segment_fruits)?;
+            Ok(Some(fruit))
+        } else {
+            Ok(None)
+        }
+    }
 }
 
 /// The `SegmentCollector` is the trait in charge of defining the
@@ -186,7 +276,11 @@ where
     type Fruit = (Left::Fruit, Right::Fruit);
     type Child = (Left::Child, Right::Child);
 
-    fn for_segment(&self, segment_local_id: u32, segment: &SegmentReader) -> Result<Self::Child> {
+    fn for_segment(
+        &self,
+        segment_local_id: u32,
+        segment: &SegmentReader,
+    ) -> crate::Result<Self::Child> {
         let left = self.0.for_segment(segment_local_id, segment)?;
         let right = self.1.for_segment(segment_local_id, segment)?;
         Ok((left, right))
@@ -198,11 +292,11 @@ where
 
     fn merge_fruits(
         &self,
-        children: Vec<(Left::Fruit, Right::Fruit)>,
+        segment_fruits: Vec<<Self::Child as SegmentCollector>::Fruit>,
-    ) -> Result<(Left::Fruit, Right::Fruit)> {
+    ) -> crate::Result<(Left::Fruit, Right::Fruit)> {
         let mut left_fruits = vec![];
         let mut right_fruits = vec![];
-        for (left_fruit, right_fruit) in children {
+        for (left_fruit, right_fruit) in segment_fruits {
             left_fruits.push(left_fruit);
             right_fruits.push(right_fruit);
         }
@@ -241,7 +335,11 @@ where
     type Fruit = (One::Fruit, Two::Fruit, Three::Fruit);
     type Child = (One::Child, Two::Child, Three::Child);
 
-    fn for_segment(&self, segment_local_id: u32, segment: &SegmentReader) -> Result<Self::Child> {
+    fn for_segment(
+        &self,
+        segment_local_id: u32,
+        segment: &SegmentReader,
+    ) -> crate::Result<Self::Child> {
         let one = self.0.for_segment(segment_local_id, segment)?;
         let two = self.1.for_segment(segment_local_id, segment)?;
         let three = self.2.for_segment(segment_local_id, segment)?;
@@ -252,7 +350,10 @@ where
         self.0.requires_scoring() || self.1.requires_scoring() || self.2.requires_scoring()
     }
 
-    fn merge_fruits(&self, children: Vec<Self::Fruit>) -> Result<Self::Fruit> {
+    fn merge_fruits(
+        &self,
+        children: Vec<<Self::Child as SegmentCollector>::Fruit>,
+    ) -> crate::Result<Self::Fruit> {
         let mut one_fruits = vec![];
         let mut two_fruits = vec![];
         let mut three_fruits = vec![];
@@ -300,7 +401,11 @@ where
     type Fruit = (One::Fruit, Two::Fruit, Three::Fruit, Four::Fruit);
     type Child = (One::Child, Two::Child, Three::Child, Four::Child);
 
-    fn for_segment(&self, segment_local_id: u32, segment: &SegmentReader) -> Result<Self::Child> {
+    fn for_segment(
+        &self,
+        segment_local_id: u32,
+        segment: &SegmentReader,
+    ) -> crate::Result<Self::Child> {
         let one = self.0.for_segment(segment_local_id, segment)?;
         let two = self.1.for_segment(segment_local_id, segment)?;
         let three = self.2.for_segment(segment_local_id, segment)?;
@@ -315,7 +420,10 @@ where
             || self.3.requires_scoring()
     }
 
-    fn merge_fruits(&self, children: Vec<Self::Fruit>) -> Result<Self::Fruit> {
+    fn merge_fruits(
+        &self,
+        children: Vec<<Self::Child as SegmentCollector>::Fruit>,
+    ) -> crate::Result<Self::Fruit> {
         let mut one_fruits = vec![];
         let mut two_fruits = vec![];
         let mut three_fruits = vec![];

src/collector/multi_collector.rs

@@ -2,9 +2,8 @@ use super::Collector;
 use super::SegmentCollector;
 use crate::collector::Fruit;
 use crate::DocId;
-use crate::Result;
 use crate::Score;
-use crate::SegmentLocalId;
+use crate::SegmentOrdinal;
 use crate::SegmentReader;
 use crate::TantivyError;
 use std::marker::PhantomData;
@@ -24,7 +23,7 @@ impl<TCollector: Collector> Collector for CollectorWrapper<TCollector> {
         &self,
         segment_local_id: u32,
         reader: &SegmentReader,
-    ) -> Result<Box<dyn BoxableSegmentCollector>> {
+    ) -> crate::Result<Box<dyn BoxableSegmentCollector>> {
         let child = self.0.for_segment(segment_local_id, reader)?;
         Ok(Box::new(SegmentCollectorWrapper(child)))
     }
@@ -33,18 +32,21 @@ impl<TCollector: Collector> Collector for CollectorWrapper<TCollector> {
         self.0.requires_scoring()
     }
 
-    fn merge_fruits(&self, children: Vec<<Self as Collector>::Fruit>) -> Result<Box<dyn Fruit>> {
+    fn merge_fruits(
-        let typed_fruit: Vec<TCollector::Fruit> = children
+        &self,
+        children: Vec<<Self::Child as SegmentCollector>::Fruit>,
+    ) -> crate::Result<Box<dyn Fruit>> {
+        let typed_fruit: Vec<<TCollector::Child as SegmentCollector>::Fruit> = children
             .into_iter()
             .map(|untyped_fruit| {
                 untyped_fruit
-                    .downcast::<TCollector::Fruit>()
+                    .downcast::<<TCollector::Child as SegmentCollector>::Fruit>()
                     .map(|boxed_but_typed| *boxed_but_typed)
                     .map_err(|_| {
                         TantivyError::InvalidArgument("Failed to cast child fruit.".to_string())
                     })
             })
-            .collect::<Result<_>>()?;
+            .collect::<crate::Result<_>>()?;
         let merged_fruit = self.0.merge_fruits(typed_fruit)?;
         Ok(Box::new(merged_fruit))
     }
@@ -53,7 +55,7 @@ impl<TCollector: Collector> Collector for CollectorWrapper<TCollector> {
 impl SegmentCollector for Box<dyn BoxableSegmentCollector> {
     type Fruit = Box<dyn Fruit>;
 
-    fn collect(&mut self, doc: u32, score: f32) {
+    fn collect(&mut self, doc: u32, score: Score) {
         self.as_mut().collect(doc, score);
     }
 
@@ -63,7 +65,7 @@ impl SegmentCollector for Box<dyn BoxableSegmentCollector> {
 }
 
 pub trait BoxableSegmentCollector {
-    fn collect(&mut self, doc: u32, score: f32);
+    fn collect(&mut self, doc: u32, score: Score);
     fn harvest_from_box(self: Box<Self>) -> Box<dyn Fruit>;
 }
 
@@ -72,7 +74,7 @@ pub struct SegmentCollectorWrapper<TSegmentCollector: SegmentCollector>(TSegment
 impl<TSegmentCollector: SegmentCollector> BoxableSegmentCollector
     for SegmentCollectorWrapper<TSegmentCollector>
 {
-    fn collect(&mut self, doc: u32, score: f32) {
+    fn collect(&mut self, doc: u32, score: Score) {
         self.0.collect(doc, score);
     }
 
@@ -105,54 +107,38 @@ impl<TFruit: Fruit> FruitHandle<TFruit> {
 /// [Combining several collectors section of the collector documentation](./index.html#combining-several-collectors).
 ///
 /// ```rust
-/// #[macro_use]
-/// extern crate tantivy;
-/// use tantivy::schema::{Schema, TEXT};
-/// use tantivy::{Index, Result};
 /// use tantivy::collector::{Count, TopDocs, MultiCollector};
 /// use tantivy::query::QueryParser;
+/// use tantivy::schema::{Schema, TEXT};
+/// use tantivy::{doc, Index};
 ///
-/// # fn main() { example().unwrap(); }
+/// let mut schema_builder = Schema::builder();
-/// fn example() -> Result<()> {
+/// let title = schema_builder.add_text_field("title", TEXT);
-///     let mut schema_builder = Schema::builder();
+/// let schema = schema_builder.build();
-///     let title = schema_builder.add_text_field("title", TEXT);
+/// let index = Index::create_in_ram(schema);
-///     let schema = schema_builder.build();
-///     let index = Index::create_in_ram(schema);
-///     {
-///         let mut index_writer = index.writer(3_000_000)?;
-///         index_writer.add_document(doc!(
-///             title => "The Name of the Wind",
-///         ));
-///         index_writer.add_document(doc!(
-///             title => "The Diary of Muadib",
-///         ));
-///         index_writer.add_document(doc!(
-///             title => "A Dairy Cow",
-///         ));
-///         index_writer.add_document(doc!(
-///             title => "The Diary of a Young Girl",
-///         ));
-///         index_writer.commit().unwrap();
-///     }
 ///
-///     let reader = index.reader()?;
+/// let mut index_writer = index.writer(3_000_000).unwrap();
-///     let searcher = reader.searcher();
+/// index_writer.add_document(doc!(title => "The Name of the Wind"));
+/// index_writer.add_document(doc!(title => "The Diary of Muadib"));
+/// index_writer.add_document(doc!(title => "A Dairy Cow"));
+/// index_writer.add_document(doc!(title => "The Diary of a Young Girl"));
+/// assert!(index_writer.commit().is_ok());
 ///
-///     let mut collectors = MultiCollector::new();
+/// let reader = index.reader().unwrap();
-///     let top_docs_handle = collectors.add_collector(TopDocs::with_limit(2));
+/// let searcher = reader.searcher();
-///     let count_handle = collectors.add_collector(Count);
-///     let query_parser = QueryParser::for_index(&index, vec![title]);
-///     let query = query_parser.parse_query("diary")?;
-///     let mut multi_fruit = searcher.search(&query, &collectors)?;
 ///
-///     let count = count_handle.extract(&mut multi_fruit);
+/// let mut collectors = MultiCollector::new();
-///     let top_docs = top_docs_handle.extract(&mut multi_fruit);
+/// let top_docs_handle = collectors.add_collector(TopDocs::with_limit(2));
+/// let count_handle = collectors.add_collector(Count);
+/// let query_parser = QueryParser::for_index(&index, vec![title]);
+/// let query = query_parser.parse_query("diary").unwrap();
+/// let mut multi_fruit = searcher.search(&query, &collectors).unwrap();
 ///
-///     # assert_eq!(count, 2);
+/// let count = count_handle.extract(&mut multi_fruit);
-///     # assert_eq!(top_docs.len(), 2);
+/// let top_docs = top_docs_handle.extract(&mut multi_fruit);
 ///
-///     Ok(())
+/// assert_eq!(count, 2);
-/// }
+/// assert_eq!(top_docs.len(), 2);
 /// ```
 #[allow(clippy::type_complexity)]
 #[derive(Default)]
@@ -189,14 +175,14 @@ impl<'a> Collector for MultiCollector<'a> {
 
     fn for_segment(
         &self,
-        segment_local_id: SegmentLocalId,
+        segment_local_id: SegmentOrdinal,
         segment: &SegmentReader,
-    ) -> Result<MultiCollectorChild> {
+    ) -> crate::Result<MultiCollectorChild> {
         let children = self
             .collector_wrappers
             .iter()
             .map(|collector_wrapper| collector_wrapper.for_segment(segment_local_id, segment))
-            .collect::<Result<Vec<_>>>()?;
+            .collect::<crate::Result<Vec<_>>>()?;
         Ok(MultiCollectorChild { children })
     }
 
@@ -207,7 +193,7 @@ impl<'a> Collector for MultiCollector<'a> {
             .any(Collector::requires_scoring)
     }
 
-    fn merge_fruits(&self, segments_multifruits: Vec<MultiFruit>) -> Result<MultiFruit> {
+    fn merge_fruits(&self, segments_multifruits: Vec<MultiFruit>) -> crate::Result<MultiFruit> {
         let mut segment_fruits_list: Vec<Vec<Box<dyn Fruit>>> = (0..self.collector_wrappers.len())
             .map(|_| Vec::with_capacity(segments_multifruits.len()))
             .collect::<Vec<_>>();
@@ -225,7 +211,7 @@ impl<'a> Collector for MultiCollector<'a> {
             .map(|(child_collector, segment_fruits)| {
                 Ok(Some(child_collector.merge_fruits(segment_fruits)?))
             })
-            .collect::<Result<_>>()?;
+            .collect::<crate::Result<_>>()?;
         Ok(MultiFruit { sub_fruits })
     }
 }
@@ -273,7 +259,7 @@ mod tests {
 
         let index = Index::create_in_ram(schema);
         {
-            let mut index_writer = index.writer_with_num_threads(1, 3_000_000).unwrap();
+            let mut index_writer = index.writer_for_tests().unwrap();
             index_writer.add_document(doc!(text=>"abc"));
             index_writer.add_document(doc!(text=>"abc abc abc"));
             index_writer.add_document(doc!(text=>"abc abc"));

src/collector/tests.rs

@@ -1,12 +1,20 @@
 use super::*;
 use crate::core::SegmentReader;
 use crate::fastfield::BytesFastFieldReader;
+use crate::fastfield::DynamicFastFieldReader;
 use crate::fastfield::FastFieldReader;
 use crate::schema::Field;
-use crate::DocAddress;
 use crate::DocId;
 use crate::Score;
-use crate::SegmentLocalId;
+use crate::SegmentOrdinal;
+use crate::{DocAddress, Document, Searcher};
+
+use crate::collector::{Count, FilterCollector, TopDocs};
+use crate::query::{AllQuery, QueryParser};
+use crate::schema::{Schema, FAST, TEXT};
+use crate::DateTime;
+use crate::{doc, Index};
+use std::str::FromStr;
 
 pub const TEST_COLLECTOR_WITH_SCORE: TestCollector = TestCollector {
     compute_score: true,
@@ -16,6 +24,54 @@ pub const TEST_COLLECTOR_WITHOUT_SCORE: TestCollector = TestCollector {
     compute_score: true,
 };
 
+#[test]
+pub fn test_filter_collector() {
+    let mut schema_builder = Schema::builder();
+    let title = schema_builder.add_text_field("title", TEXT);
+    let price = schema_builder.add_u64_field("price", FAST);
+    let date = schema_builder.add_date_field("date", FAST);
+    let schema = schema_builder.build();
+    let index = Index::create_in_ram(schema);
+
+    let mut index_writer = index.writer_with_num_threads(1, 10_000_000).unwrap();
+    index_writer.add_document(doc!(title => "The Name of the Wind", price => 30_200u64, date => DateTime::from_str("1898-04-09T00:00:00+00:00").unwrap()));
+    index_writer.add_document(doc!(title => "The Diary of Muadib", price => 29_240u64, date => DateTime::from_str("2020-04-09T00:00:00+00:00").unwrap()));
+    index_writer.add_document(doc!(title => "The Diary of Anne Frank", price => 18_240u64, date => DateTime::from_str("2019-04-20T00:00:00+00:00").unwrap()));
+    index_writer.add_document(doc!(title => "A Dairy Cow", price => 21_240u64, date => DateTime::from_str("2019-04-09T00:00:00+00:00").unwrap()));
+    index_writer.add_document(doc!(title => "The Diary of a Young Girl", price => 20_120u64, date => DateTime::from_str("2018-04-09T00:00:00+00:00").unwrap()));
+    assert!(index_writer.commit().is_ok());
+
+    let reader = index.reader().unwrap();
+    let searcher = reader.searcher();
+
+    let query_parser = QueryParser::for_index(&index, vec![title]);
+    let query = query_parser.parse_query("diary").unwrap();
+    let filter_some_collector = FilterCollector::new(
+        price,
+        &|value: u64| value > 20_120u64,
+        TopDocs::with_limit(2),
+    );
+    let top_docs = searcher.search(&query, &filter_some_collector).unwrap();
+
+    assert_eq!(top_docs.len(), 1);
+    assert_eq!(top_docs[0].1, DocAddress::new(0, 1));
+
+    let filter_all_collector: FilterCollector<_, _, u64> =
+        FilterCollector::new(price, &|value| value < 5u64, TopDocs::with_limit(2));
+    let filtered_top_docs = searcher.search(&query, &filter_all_collector).unwrap();
+
+    assert_eq!(filtered_top_docs.len(), 0);
+
+    fn date_filter(value: DateTime) -> bool {
+        (value - DateTime::from_str("2019-04-09T00:00:00+00:00").unwrap()).num_weeks() > 0
+    }
+
+    let filter_dates_collector = FilterCollector::new(date, &date_filter, TopDocs::with_limit(5));
+    let filtered_date_docs = searcher.search(&query, &filter_dates_collector).unwrap();
+
+    assert_eq!(filtered_date_docs.len(), 2);
+}
+
 /// Stores all of the doc ids.
 /// This collector is only used for tests.
 /// It is unusable in pr
@@ -27,7 +83,7 @@ pub struct TestCollector {
 }
 
 pub struct TestSegmentCollector {
-    segment_id: SegmentLocalId,
+    segment_id: SegmentOrdinal,
     fruit: TestFruit,
 }
 
@@ -53,9 +109,9 @@ impl Collector for TestCollector {
 
     fn for_segment(
         &self,
-        segment_id: SegmentLocalId,
+        segment_id: SegmentOrdinal,
         _reader: &SegmentReader,
-    ) -> Result<TestSegmentCollector> {
+    ) -> crate::Result<TestSegmentCollector> {
         Ok(TestSegmentCollector {
             segment_id,
             fruit: TestFruit::default(),
@@ -66,12 +122,12 @@ impl Collector for TestCollector {
         self.compute_score
     }
 
-    fn merge_fruits(&self, mut children: Vec<TestFruit>) -> Result<TestFruit> {
+    fn merge_fruits(&self, mut children: Vec<TestFruit>) -> crate::Result<TestFruit> {
         children.sort_by_key(|fruit| {
             if fruit.docs().is_empty() {
                 0
             } else {
-                fruit.docs()[0].segment_ord()
+                fruit.docs()[0].segment_ord
             }
         });
         let mut docs = vec![];
@@ -88,7 +144,7 @@ impl SegmentCollector for TestSegmentCollector {
     type Fruit = TestFruit;
 
     fn collect(&mut self, doc: DocId, score: Score) {
-        self.fruit.docs.push(DocAddress(self.segment_id, doc));
+        self.fruit.docs.push(DocAddress::new(self.segment_id, doc));
         self.fruit.scores.push(score);
     }
 
@@ -107,7 +163,7 @@ pub struct FastFieldTestCollector {
 
 pub struct FastFieldSegmentCollector {
     vals: Vec<u64>,
-    reader: FastFieldReader<u64>,
+    reader: DynamicFastFieldReader<u64>,
 }
 
 impl FastFieldTestCollector {
@@ -122,9 +178,9 @@ impl Collector for FastFieldTestCollector {
 
     fn for_segment(
         &self,
-        _: SegmentLocalId,
+        _: SegmentOrdinal,
         segment_reader: &SegmentReader,
-    ) -> Result<FastFieldSegmentCollector> {
+    ) -> crate::Result<FastFieldSegmentCollector> {
         let reader = segment_reader
             .fast_fields()
             .u64(self.field)
@@ -139,7 +195,7 @@ impl Collector for FastFieldTestCollector {
         false
     }
 
-    fn merge_fruits(&self, children: Vec<Vec<u64>>) -> Result<Vec<u64>> {
+    fn merge_fruits(&self, children: Vec<Vec<u64>>) -> crate::Result<Vec<u64>> {
         Ok(children.into_iter().flat_map(|v| v.into_iter()).collect())
     }
 }
@@ -184,13 +240,11 @@ impl Collector for BytesFastFieldTestCollector {
         &self,
         _segment_local_id: u32,
         segment_reader: &SegmentReader,
-    ) -> Result<BytesFastFieldSegmentCollector> {
+    ) -> crate::Result<BytesFastFieldSegmentCollector> {
+        let reader = segment_reader.fast_fields().bytes(self.field)?;
         Ok(BytesFastFieldSegmentCollector {
             vals: Vec::new(),
-            reader: segment_reader
+            reader,
-                .fast_fields()
-                .bytes(self.field)
-                .expect("Field is not a bytes fast field."),
         })
     }
 
@@ -198,7 +252,7 @@ impl Collector for BytesFastFieldTestCollector {
         false
     }
 
-    fn merge_fruits(&self, children: Vec<Vec<u8>>) -> Result<Vec<u8>> {
+    fn merge_fruits(&self, children: Vec<Vec<u8>>) -> crate::Result<Vec<u8>> {
         Ok(children.into_iter().flat_map(|c| c.into_iter()).collect())
     }
 }
@@ -206,7 +260,7 @@ impl Collector for BytesFastFieldTestCollector {
 impl SegmentCollector for BytesFastFieldSegmentCollector {
     type Fruit = Vec<u8>;
 
-    fn collect(&mut self, doc: u32, _score: f32) {
+    fn collect(&mut self, doc: u32, _score: Score) {
         let data = self.reader.get_bytes(doc);
         self.vals.extend(data);
     }
@@ -215,3 +269,30 @@ impl SegmentCollector for BytesFastFieldSegmentCollector {
         self.vals
     }
 }
+
+fn make_test_searcher() -> crate::Result<crate::LeasedItem<Searcher>> {
+    let schema = Schema::builder().build();
+    let index = Index::create_in_ram(schema);
+    let mut index_writer = index.writer_for_tests()?;
+    index_writer.add_document(Document::default());
+    index_writer.add_document(Document::default());
+    index_writer.commit()?;
+    Ok(index.reader()?.searcher())
+}
+
+#[test]
+fn test_option_collector_some() -> crate::Result<()> {
+    let searcher = make_test_searcher()?;
+    let counts = searcher.search(&AllQuery, &Some(Count))?;
+    assert_eq!(counts, Some(2));
+    Ok(())
+}
+
+#[test]
+fn test_option_collector_none() -> crate::Result<()> {
+    let searcher = make_test_searcher()?;
+    let none_collector: Option<Count> = None;
+    let counts = searcher.search(&AllQuery, &none_collector)?;
+    assert_eq!(counts, None);
+    Ok(())
+}

src/collector/top_collector.rs

@@ -1,52 +1,63 @@
 use crate::DocAddress;
 use crate::DocId;
-use crate::Result;
+use crate::SegmentOrdinal;
-use crate::SegmentLocalId;
 use crate::SegmentReader;
-use serde::export::PhantomData;
 use std::cmp::Ordering;
 use std::collections::BinaryHeap;
+use std::marker::PhantomData;
 
 /// Contains a feature (field, score, etc.) of a document along with the document address.
 ///
 /// It has a custom implementation of `PartialOrd` that reverses the order. This is because the
 /// default Rust heap is a max heap, whereas a min heap is needed.
 ///
+/// Additionally, it guarantees stable sorting: in case of a tie on the feature, the document
+/// address is used.
+///
 /// WARNING: equality is not what you would expect here.
 /// Two elements are equal if their feature is equal, and regardless of whether `doc`
 /// is equal. This should be perfectly fine for this usage, but let's make sure this
 /// struct is never public.
-struct ComparableDoc<T, D> {
+pub(crate) struct ComparableDoc<T, D> {
-    feature: T,
+    pub feature: T,
-    doc: D,
+    pub doc: D,
 }
 
-impl<T: PartialOrd, D> PartialOrd for ComparableDoc<T, D> {
+impl<T: PartialOrd, D: PartialOrd> PartialOrd for ComparableDoc<T, D> {
     fn partial_cmp(&self, other: &Self) -> Option<Ordering> {
         Some(self.cmp(other))
     }
 }
 
-impl<T: PartialOrd, D> Ord for ComparableDoc<T, D> {
+impl<T: PartialOrd, D: PartialOrd> Ord for ComparableDoc<T, D> {
     #[inline]
     fn cmp(&self, other: &Self) -> Ordering {
-        other
+        // Reversed to make BinaryHeap work as a min-heap
+        let by_feature = other
             .feature
             .partial_cmp(&self.feature)
-            .unwrap_or_else(|| Ordering::Equal)
+            .unwrap_or(Ordering::Equal);
+
+        let lazy_by_doc_address = || self.doc.partial_cmp(&other.doc).unwrap_or(Ordering::Equal);
+
+        // In case of a tie on the feature, we sort by ascending
+        // `DocAddress` in order to ensure a stable sorting of the
+        // documents.
+        by_feature.then_with(lazy_by_doc_address)
     }
 }
 
-impl<T: PartialOrd, D> PartialEq for ComparableDoc<T, D> {
+impl<T: PartialOrd, D: PartialOrd> PartialEq for ComparableDoc<T, D> {
     fn eq(&self, other: &Self) -> bool {
         self.cmp(other) == Ordering::Equal
     }
 }
 
-impl<T: PartialOrd, D> Eq for ComparableDoc<T, D> {}
+impl<T: PartialOrd, D: PartialOrd> Eq for ComparableDoc<T, D> {}
 
 pub(crate) struct TopCollector<T> {
-    limit: usize,
+    pub limit: usize,
+    pub offset: usize,
     _marker: PhantomData<T>,
 }
 
@@ -62,27 +73,33 @@ where
         if limit < 1 {
             panic!("Limit must be strictly greater than 0.");
         }
-        TopCollector {
+        Self {
             limit,
+            offset: 0,
             _marker: PhantomData,
         }
     }
 
-    pub fn limit(&self) -> usize {
+    /// Skip the first "offset" documents when collecting.
-        self.limit
+    ///
+    /// This is equivalent to `OFFSET` in MySQL or PostgreSQL and `start` in
+    /// Lucene's TopDocsCollector.
+    pub fn and_offset(mut self, offset: usize) -> TopCollector<T> {
+        self.offset = offset;
+        self
     }
 
     pub fn merge_fruits(
         &self,
         children: Vec<Vec<(T, DocAddress)>>,
-    ) -> Result<Vec<(T, DocAddress)>> {
+    ) -> crate::Result<Vec<(T, DocAddress)>> {
         if self.limit == 0 {
             return Ok(Vec::new());
         }
         let mut top_collector = BinaryHeap::new();
         for child_fruit in children {
             for (feature, doc) in child_fruit {
-                if top_collector.len() < self.limit {
+                if top_collector.len() < (self.limit + self.offset) {
                     top_collector.push(ComparableDoc { feature, doc });
                 } else if let Some(mut head) = top_collector.peek_mut() {
                     if head.feature < feature {
@@ -94,16 +111,30 @@ where
         Ok(top_collector
             .into_sorted_vec()
             .into_iter()
+            .skip(self.offset)
             .map(|cdoc| (cdoc.feature, cdoc.doc))
             .collect())
     }
 
     pub(crate) fn for_segment<F: PartialOrd>(
         &self,
-        segment_id: SegmentLocalId,
+        segment_id: SegmentOrdinal,
         _: &SegmentReader,
-    ) -> Result<TopSegmentCollector<F>> {
+    ) -> TopSegmentCollector<F> {
-        Ok(TopSegmentCollector::new(segment_id, self.limit))
+        TopSegmentCollector::new(segment_id, self.limit + self.offset)
+    }
+
+    /// Create a new TopCollector with the same limit and offset.
+    ///
+    /// Ideally we would use Into but the blanket implementation seems to cause the Scorer traits
+    /// to fail.
+    #[doc(hidden)]
+    pub(crate) fn into_tscore<TScore: PartialOrd + Clone>(self) -> TopCollector<TScore> {
+        TopCollector {
+            limit: self.limit,
+            offset: self.offset,
+            _marker: PhantomData,
+        }
     }
 }
 
@@ -116,29 +147,32 @@ where
 pub(crate) struct TopSegmentCollector<T> {
     limit: usize,
     heap: BinaryHeap<ComparableDoc<T, DocId>>,
-    segment_id: u32,
+    segment_ord: u32,
 }
 
 impl<T: PartialOrd> TopSegmentCollector<T> {
-    fn new(segment_id: SegmentLocalId, limit: usize) -> TopSegmentCollector<T> {
+    fn new(segment_ord: SegmentOrdinal, limit: usize) -> TopSegmentCollector<T> {
         TopSegmentCollector {
             limit,
             heap: BinaryHeap::with_capacity(limit),
-            segment_id,
+            segment_ord,
         }
     }
 }
 
 impl<T: PartialOrd + Clone> TopSegmentCollector<T> {
     pub fn harvest(self) -> Vec<(T, DocAddress)> {
-        let segment_id = self.segment_id;
+        let segment_ord = self.segment_ord;
         self.heap
             .into_sorted_vec()
             .into_iter()
             .map(|comparable_doc| {
                 (
                     comparable_doc.feature,
-                    DocAddress(segment_id, comparable_doc.doc),
+                    DocAddress {
+                        segment_ord,
+                        doc_id: comparable_doc.doc,
+                    },
                 )
             })
             .collect()
@@ -146,7 +180,7 @@ impl<T: PartialOrd + Clone> TopSegmentCollector<T> {
 
     /// Return true iff at least K documents have gone through
     /// the collector.
-    #[inline(always)]
+    #[inline]
     pub(crate) fn at_capacity(&self) -> bool {
         self.heap.len() >= self.limit
     }
@@ -155,7 +189,7 @@ impl<T: PartialOrd + Clone> TopSegmentCollector<T> {
     ///
     /// It collects documents until it has reached the max capacity. Once it reaches capacity, it
     /// will compare the lowest scoring item with the given one and keep whichever is greater.
-    #[inline(always)]
+    #[inline]
     pub fn collect(&mut self, doc: DocId, feature: T) {
         if self.at_capacity() {
             // It's ok to unwrap as long as a limit of 0 is forbidden.
@@ -177,7 +211,7 @@ impl<T: PartialOrd + Clone> TopSegmentCollector<T> {
 
 #[cfg(test)]
 mod tests {
-    use super::TopSegmentCollector;
+    use super::{TopCollector, TopSegmentCollector};
     use crate::DocAddress;
 
     #[test]
@@ -189,9 +223,9 @@ mod tests {
         assert_eq!(
             top_collector.harvest(),
             vec![
-                (0.8, DocAddress(0, 1)),
+                (0.8, DocAddress::new(0, 1)),
-                (0.3, DocAddress(0, 5)),
+                (0.3, DocAddress::new(0, 5)),
-                (0.2, DocAddress(0, 3))
+                (0.2, DocAddress::new(0, 3))
             ]
         );
     }
@@ -207,11 +241,149 @@ mod tests {
         assert_eq!(
             top_collector.harvest(),
             vec![
-                (0.9, DocAddress(0, 7)),
+                (0.9, DocAddress::new(0, 7)),
-                (0.8, DocAddress(0, 1)),
+                (0.8, DocAddress::new(0, 1)),
-                (0.3, DocAddress(0, 5)),
+                (0.3, DocAddress::new(0, 5)),
-                (0.2, DocAddress(0, 3))
+                (0.2, DocAddress::new(0, 3))
             ]
         );
     }
+
+    #[test]
+    fn test_top_segment_collector_stable_ordering_for_equal_feature() {
+        // given that the documents are collected in ascending doc id order,
+        // when harvesting we have to guarantee stable sorting in case of a tie
+        // on the score
+        let doc_ids_collection = [4, 5, 6];
+        let score = 3.14;
+
+        let mut top_collector_limit_2 = TopSegmentCollector::new(0, 2);
+        for id in &doc_ids_collection {
+            top_collector_limit_2.collect(*id, score);
+        }
+
+        let mut top_collector_limit_3 = TopSegmentCollector::new(0, 3);
+        for id in &doc_ids_collection {
+            top_collector_limit_3.collect(*id, score);
+        }
+
+        assert_eq!(
+            top_collector_limit_2.harvest(),
+            top_collector_limit_3.harvest()[..2].to_vec(),
+        );
+    }
+
+    #[test]
+    fn test_top_collector_with_limit_and_offset() {
+        let collector = TopCollector::with_limit(2).and_offset(1);
+
+        let results = collector
+            .merge_fruits(vec![vec![
+                (0.9, DocAddress::new(0, 1)),
+                (0.8, DocAddress::new(0, 2)),
+                (0.7, DocAddress::new(0, 3)),
+                (0.6, DocAddress::new(0, 4)),
+                (0.5, DocAddress::new(0, 5)),
+            ]])
+            .unwrap();
+
+        assert_eq!(
+            results,
+            vec![(0.8, DocAddress::new(0, 2)), (0.7, DocAddress::new(0, 3)),]
+        );
+    }
+
+    #[test]
+    fn test_top_collector_with_limit_larger_than_set_and_offset() {
+        let collector = TopCollector::with_limit(2).and_offset(1);
+
+        let results = collector
+            .merge_fruits(vec![vec![
+                (0.9, DocAddress::new(0, 1)),
+                (0.8, DocAddress::new(0, 2)),
+            ]])
+            .unwrap();
+
+        assert_eq!(results, vec![(0.8, DocAddress::new(0, 2)),]);
+    }
+
+    #[test]
+    fn test_top_collector_with_limit_and_offset_larger_than_set() {
+        let collector = TopCollector::with_limit(2).and_offset(20);
+
+        let results = collector
+            .merge_fruits(vec![vec![
+                (0.9, DocAddress::new(0, 1)),
+                (0.8, DocAddress::new(0, 2)),
+            ]])
+            .unwrap();
+
+        assert_eq!(results, vec![]);
+    }
+}
+
+#[cfg(all(test, feature = "unstable"))]
+mod bench {
+    use super::TopSegmentCollector;
+    use test::Bencher;
+
+    #[bench]
+    fn bench_top_segment_collector_collect_not_at_capacity(b: &mut Bencher) {
+        let mut top_collector = TopSegmentCollector::new(0, 400);
+
+        b.iter(|| {
+            for i in 0..100 {
+                top_collector.collect(i, 0.8);
+            }
+        });
+    }
+
+    #[bench]
+    fn bench_top_segment_collector_collect_at_capacity(b: &mut Bencher) {
+        let mut top_collector = TopSegmentCollector::new(0, 100);
+
+        for i in 0..100 {
+            top_collector.collect(i, 0.8);
+        }
+
+        b.iter(|| {
+            for i in 0..100 {
+                top_collector.collect(i, 0.8);
+            }
+        });
+    }
+
+    #[bench]
+    fn bench_top_segment_collector_collect_and_harvest_many_ties(b: &mut Bencher) {
+        b.iter(|| {
+            let mut top_collector = TopSegmentCollector::new(0, 100);
+
+            for i in 0..100 {
+                top_collector.collect(i, 0.8);
+            }
+
+            // it would be nice to be able to do the setup N times but still
+            // measure only harvest(). We can't since harvest() consumes
+            // the top_collector.
+            top_collector.harvest()
+        });
+    }
+
+    #[bench]
+    fn bench_top_segment_collector_collect_and_harvest_no_tie(b: &mut Bencher) {
+        b.iter(|| {
+            let mut top_collector = TopSegmentCollector::new(0, 100);
+            let mut score = 1.0;
+
+            for i in 0..100 {
+                score += 1.0;
+                top_collector.collect(i, score);
+            }
+
+            // it would be nice to be able to do the setup N times but still
+            // measure only harvest(). We can't since harvest() consumes
+            // the top_collector.
+            top_collector.harvest()
+        });
+    }
 }

src/collector/tweak_score_top_collector.rs

@@ -14,11 +14,11 @@ where
 {
     pub fn new(
         score_tweaker: TScoreTweaker,
-        limit: usize,
+        collector: TopCollector<TScore>,
     ) -> TweakedScoreTopCollector<TScoreTweaker, TScore> {
         TweakedScoreTopCollector {
             score_tweaker,
-            collector: TopCollector::with_limit(limit),
+            collector,
         }
     }
 }
@@ -29,7 +29,7 @@ where
 /// It is the segment local version of the [`ScoreTweaker`](./trait.ScoreTweaker.html).
 pub trait ScoreSegmentTweaker<TScore>: 'static {
     /// Tweak the given `score` for the document `doc`.
-    fn score(&self, doc: DocId, score: Score) -> TScore;
+    fn score(&mut self, doc: DocId, score: Score) -> TScore;
 }
 
 /// `ScoreTweaker` makes it possible to tweak the score
@@ -49,7 +49,7 @@ pub trait ScoreTweaker<TScore>: Sync {
 
 impl<TScoreTweaker, TScore> Collector for TweakedScoreTopCollector<TScoreTweaker, TScore>
 where
-    TScoreTweaker: ScoreTweaker<TScore>,
+    TScoreTweaker: ScoreTweaker<TScore> + Send + Sync,
     TScore: 'static + PartialOrd + Clone + Send + Sync,
 {
     type Fruit = Vec<(TScore, DocAddress)>;
@@ -62,9 +62,7 @@ where
         segment_reader: &SegmentReader,
     ) -> Result<Self::Child> {
         let segment_scorer = self.score_tweaker.segment_tweaker(segment_reader)?;
-        let segment_collector = self
+        let segment_collector = self.collector.for_segment(segment_local_id, segment_reader);
-            .collector
-            .for_segment(segment_local_id, segment_reader)?;
         Ok(TopTweakedScoreSegmentCollector {
             segment_collector,
             segment_scorer,
@@ -121,9 +119,9 @@ where
 
 impl<F, TScore> ScoreSegmentTweaker<TScore> for F
 where
-    F: 'static + Sync + Send + Fn(DocId, Score) -> TScore,
+    F: 'static + FnMut(DocId, Score) -> TScore,
 {
-    fn score(&self, doc: DocId, score: Score) -> TScore {
+    fn score(&mut self, doc: DocId, score: Score) -> TScore {
         (self)(doc, score)
     }
 }

src/common/bitset.rs

@@ -33,6 +33,10 @@ impl TinySet {
         TinySet(0u64)
     }
 
+    pub fn clear(&mut self) {
+        self.0 = 0u64;
+    }
+
     /// Returns the complement of the set in `[0, 64[`.
     fn complement(self) -> TinySet {
         TinySet(!self.0)
@@ -43,6 +47,11 @@ impl TinySet {
         !self.intersect(TinySet::singleton(el)).is_empty()
     }
 
+    /// Returns the number of elements in the TinySet.
+    pub fn len(self) -> u32 {
+        self.0.count_ones()
+    }
+
     /// Returns the intersection of `self` and `other`
     pub fn intersect(self, other: TinySet) -> TinySet {
         TinySet(self.0 & other.0)
@@ -50,19 +59,19 @@ impl TinySet {
 
     /// Creates a new `TinySet` containing only one element
     /// within `[0; 64[`
-    #[inline(always)]
+    #[inline]
     pub fn singleton(el: u32) -> TinySet {
         TinySet(1u64 << u64::from(el))
     }
 
     /// Insert a new element within [0..64[
-    #[inline(always)]
+    #[inline]
     pub fn insert(self, el: u32) -> TinySet {
         self.union(TinySet::singleton(el))
     }
 
     /// Insert a new element within [0..64[
-    #[inline(always)]
+    #[inline]
     pub fn insert_mut(&mut self, el: u32) -> bool {
         let old = *self;
         *self = old.insert(el);
@@ -70,20 +79,20 @@ impl TinySet {
     }
 
     /// Returns the union of two tinysets
-    #[inline(always)]
+    #[inline]
     pub fn union(self, other: TinySet) -> TinySet {
         TinySet(self.0 | other.0)
     }
 
     /// Returns true iff the `TinySet` is empty.
-    #[inline(always)]
+    #[inline]
     pub fn is_empty(self) -> bool {
         self.0 == 0u64
     }
 
     /// Returns the lowest element in the `TinySet`
     /// and removes it.
-    #[inline(always)]
+    #[inline]
     pub fn pop_lowest(&mut self) -> Option<u32> {
         if self.is_empty() {
             None
@@ -109,22 +118,12 @@ impl TinySet {
     pub fn range_greater_or_equal(from_included: u32) -> TinySet {
         TinySet::range_lower(from_included).complement()
     }
-
-    pub fn clear(&mut self) {
-        self.0 = 0u64;
-    }
-
-    pub fn len(self) -> u32 {
-        self.0.count_ones()
-    }
 }
 
 #[derive(Clone)]
 pub struct BitSet {
     tinysets: Box<[TinySet]>,
-    len: usize, //< Technically it should be u32, but we
+    len: usize,
-    // count multiple inserts.
-    // `usize` guards us from overflow.
     max_value: u32,
 }
 
@@ -204,7 +203,7 @@ mod tests {
 
     use super::BitSet;
     use super::TinySet;
-    use crate::docset::DocSet;
+    use crate::docset::{DocSet, TERMINATED};
     use crate::query::BitSetDocSet;
     use crate::tests;
     use crate::tests::generate_nonunique_unsorted;
@@ -278,11 +277,13 @@ mod tests {
         }
         assert_eq!(btreeset.len(), bitset.len());
         let mut bitset_docset = BitSetDocSet::from(bitset);
+        let mut remaining = true;
         for el in btreeset.into_iter() {
-            bitset_docset.advance();
+            assert!(remaining);
             assert_eq!(bitset_docset.doc(), el);
+            remaining = bitset_docset.advance() != TERMINATED;
         }
-        assert!(!bitset_docset.advance());
+        assert!(!remaining);
     }
 
     #[test]

src/common/composite_file.rs

@@ -1,14 +1,17 @@
 use crate::common::BinarySerializable;
 use crate::common::CountingWriter;
 use crate::common::VInt;
-use crate::directory::ReadOnlySource;
+use crate::directory::FileSlice;
-use crate::directory::WritePtr;
+use crate::directory::{TerminatingWrite, WritePtr};
 use crate::schema::Field;
 use crate::space_usage::FieldUsage;
 use crate::space_usage::PerFieldSpaceUsage;
 use std::collections::HashMap;
-use std::io::Write;
+use std::io::{self, Read, Write};
-use std::io::{self, Read};
+use std::iter::ExactSizeIterator;
+use std::ops::Range;
+
+use super::HasLen;
 
 #[derive(Eq, PartialEq, Hash, Copy, Ord, PartialOrd, Clone, Debug)]
 pub struct FileAddr {
@@ -42,7 +45,7 @@ pub struct CompositeWrite<W = WritePtr> {
     offsets: HashMap<FileAddr, u64>,
 }
 
-impl<W: Write> CompositeWrite<W> {
+impl<W: TerminatingWrite + Write> CompositeWrite<W> {
     /// Crate a new API writer that writes a composite file
     /// in a given write.
     pub fn wrap(w: W) -> CompositeWrite<W> {
@@ -91,8 +94,7 @@ impl<W: Write> CompositeWrite<W> {
 
         let footer_len = (self.write.written_bytes() - footer_offset) as u32;
         footer_len.serialize(&mut self.write)?;
-        self.write.flush()?;
+        self.write.terminate()
-        Ok(())
     }
 }
 
@@ -104,25 +106,26 @@ impl<W: Write> CompositeWrite<W> {
 /// for each field.
 #[derive(Clone)]
 pub struct CompositeFile {
-    data: ReadOnlySource,
+    data: FileSlice,
-    offsets_index: HashMap<FileAddr, (usize, usize)>,
+    offsets_index: HashMap<FileAddr, Range<usize>>,
 }
 
 impl CompositeFile {
     /// Opens a composite file stored in a given
-    /// `ReadOnlySource`.
+    /// `FileSlice`.
-    pub fn open(data: &ReadOnlySource) -> io::Result<CompositeFile> {
+    pub fn open(data: &FileSlice) -> io::Result<CompositeFile> {
         let end = data.len();
-        let footer_len_data = data.slice_from(end - 4);
+        let footer_len_data = data.slice_from(end - 4).read_bytes()?;
         let footer_len = u32::deserialize(&mut footer_len_data.as_slice())? as usize;
         let footer_start = end - 4 - footer_len;
-        let footer_data = data.slice(footer_start, footer_start + footer_len);
+        let footer_data = data
+            .slice(footer_start..footer_start + footer_len)
+            .read_bytes()?;
         let mut footer_buffer = footer_data.as_slice();
         let num_fields = VInt::deserialize(&mut footer_buffer)?.0 as usize;
 
         let mut file_addrs = vec![];
         let mut offsets = vec![];
-
         let mut field_index = HashMap::new();
 
         let mut offset = 0;
@@ -137,7 +140,7 @@ impl CompositeFile {
             let file_addr = file_addrs[i];
             let start_offset = offsets[i];
             let end_offset = offsets[i + 1];
-            field_index.insert(file_addr, (start_offset, end_offset));
+            field_index.insert(file_addr, start_offset..end_offset);
         }
 
         Ok(CompositeFile {
@@ -151,31 +154,31 @@ impl CompositeFile {
     pub fn empty() -> CompositeFile {
         CompositeFile {
             offsets_index: HashMap::new(),
-            data: ReadOnlySource::empty(),
+            data: FileSlice::empty(),
         }
     }
 
-    /// Returns the `ReadOnlySource` associated
+    /// Returns the `FileSlice` associated
     /// to a given `Field` and stored in a `CompositeFile`.
-    pub fn open_read(&self, field: Field) -> Option<ReadOnlySource> {
+    pub fn open_read(&self, field: Field) -> Option<FileSlice> {
         self.open_read_with_idx(field, 0)
     }
 
-    /// Returns the `ReadOnlySource` associated
+    /// Returns the `FileSlice` associated
     /// to a given `Field` and stored in a `CompositeFile`.
-    pub fn open_read_with_idx(&self, field: Field, idx: usize) -> Option<ReadOnlySource> {
+    pub fn open_read_with_idx(&self, field: Field, idx: usize) -> Option<FileSlice> {
         self.offsets_index
             .get(&FileAddr { field, idx })
-            .map(|&(from, to)| self.data.slice(from, to))
+            .map(|byte_range| self.data.slice(byte_range.clone()))
     }
 
     pub fn space_usage(&self) -> PerFieldSpaceUsage {
         let mut fields = HashMap::new();
-        for (&field_addr, &(start, end)) in self.offsets_index.iter() {
+        for (&field_addr, byte_range) in &self.offsets_index {
             fields
                 .entry(field_addr.field)
                 .or_insert_with(|| FieldUsage::empty(field_addr.field))
-                .add_field_idx(field_addr.idx, end - start);
+                .add_field_idx(field_addr.idx, byte_range.len());
         }
         PerFieldSpaceUsage::new(fields)
     }
@@ -187,49 +190,50 @@ mod test {
     use super::{CompositeFile, CompositeWrite};
     use crate::common::BinarySerializable;
     use crate::common::VInt;
-    use crate::directory::{Directory, RAMDirectory};
+    use crate::directory::{Directory, RamDirectory};
     use crate::schema::Field;
     use std::io::Write;
     use std::path::Path;
 
     #[test]
-    fn test_composite_file() {
+    fn test_composite_file() -> crate::Result<()> {
         let path = Path::new("test_path");
-        let mut directory = RAMDirectory::create();
+        let directory = RamDirectory::create();
         {
             let w = directory.open_write(path).unwrap();
             let mut composite_write = CompositeWrite::wrap(w);
-            {
+            let mut write_0 = composite_write.for_field(Field::from_field_id(0u32));
-                let mut write_0 = composite_write.for_field(Field(0u32));
+            VInt(32431123u64).serialize(&mut write_0)?;
-                VInt(32431123u64).serialize(&mut write_0).unwrap();
+            write_0.flush()?;
-                write_0.flush().unwrap();
+            let mut write_4 = composite_write.for_field(Field::from_field_id(4u32));
-            }
+            VInt(2).serialize(&mut write_4)?;
-
+            write_4.flush()?;
-            {
+            composite_write.close()?;
-                let mut write_4 = composite_write.for_field(Field(4u32));
-                VInt(2).serialize(&mut write_4).unwrap();
-                write_4.flush().unwrap();
-            }
-            composite_write.close().unwrap();
         }
         {
-            let r = directory.open_read(path).unwrap();
+            let r = directory.open_read(path)?;
-            let composite_file = CompositeFile::open(&r).unwrap();
+            let composite_file = CompositeFile::open(&r)?;
             {
-                let file0 = composite_file.open_read(Field(0u32)).unwrap();
+                let file0 = composite_file
+                    .open_read(Field::from_field_id(0u32))
+                    .unwrap()
+                    .read_bytes()?;
                 let mut file0_buf = file0.as_slice();
-                let payload_0 = VInt::deserialize(&mut file0_buf).unwrap().0;
+                let payload_0 = VInt::deserialize(&mut file0_buf)?.0;
                 assert_eq!(file0_buf.len(), 0);
                 assert_eq!(payload_0, 32431123u64);
             }
             {
-                let file4 = composite_file.open_read(Field(4u32)).unwrap();
+                let file4 = composite_file
+                    .open_read(Field::from_field_id(4u32))
+                    .unwrap()
+                    .read_bytes()?;
                 let mut file4_buf = file4.as_slice();
-                let payload_4 = VInt::deserialize(&mut file4_buf).unwrap().0;
+                let payload_4 = VInt::deserialize(&mut file4_buf)?.0;
                 assert_eq!(file4_buf.len(), 0);
                 assert_eq!(payload_4, 2u64);
             }
         }
+        Ok(())
     }
-
 }

src/common/counting_writer.rs

@@ -1,3 +1,5 @@
+use crate::directory::AntiCallToken;
+use crate::directory::TerminatingWrite;
 use std::io;
 use std::io::Write;
 
@@ -18,9 +20,10 @@ impl<W: Write> CountingWriter<W> {
         self.written_bytes
     }
 
-    pub fn finish(mut self) -> io::Result<(W, u64)> {
+    /// Returns the underlying write object.
-        self.flush()?;
+    /// Note that this method does not trigger any flushing.
-        Ok((self.underlying, self.written_bytes))
+    pub fn finish(self) -> W {
+        self.underlying
     }
 }
 
@@ -42,6 +45,12 @@ impl<W: Write> Write for CountingWriter<W> {
     }
 }
 
+impl<W: TerminatingWrite> TerminatingWrite for CountingWriter<W> {
+    fn terminate_ref(&mut self, token: AntiCallToken) -> io::Result<()> {
+        self.underlying.terminate_ref(token)
+    }
+}
+
 #[cfg(test)]
 mod test {
 
@@ -54,8 +63,9 @@ mod test {
         let mut counting_writer = CountingWriter::wrap(buffer);
         let bytes = (0u8..10u8).collect::<Vec<u8>>();
         counting_writer.write_all(&bytes).unwrap();
-        let (w, len): (Vec<u8>, u64) = counting_writer.finish().unwrap();
+        let len = counting_writer.written_bytes();
+        let buffer_restituted: Vec<u8> = counting_writer.finish();
         assert_eq!(len, 10u64);
-        assert_eq!(w.len(), 10);
+        assert_eq!(buffer_restituted.len(), 10);
     }
 }

src/common/mod.rs

@@ -1,4 +1,3 @@
-pub mod bitpacker;
 mod bitset;
 mod composite_file;
 mod counting_writer;
@@ -9,8 +8,10 @@ pub use self::bitset::BitSet;
 pub(crate) use self::bitset::TinySet;
 pub(crate) use self::composite_file::{CompositeFile, CompositeWrite};
 pub use self::counting_writer::CountingWriter;
-pub use self::serialize::{BinarySerializable, FixedSize};
+pub use self::serialize::{BinarySerializable, DeserializeFrom, FixedSize};
-pub use self::vint::{read_u32_vint, serialize_vint_u32, write_u32_vint, VInt};
+pub use self::vint::{
+    read_u32_vint, read_u32_vint_no_advance, serialize_vint_u32, write_u32_vint, VInt,
+};
 pub use byteorder::LittleEndian as Endianness;
 
 /// Segment's max doc must be `< MAX_DOC_LIMIT`.
@@ -18,43 +19,6 @@ pub use byteorder::LittleEndian as Endianness;
 /// We do not allow segments with more than
 pub const MAX_DOC_LIMIT: u32 = 1 << 31;
 
-/// Computes the number of bits that will be used for bitpacking.
-///
-/// In general the target is the minimum number of bits
-/// required to express the amplitude given in argument.
-///
-/// e.g. If the amplitude is 10, we can store all ints on simply 4bits.
-///
-/// The logic is slightly more convoluted here as for optimization
-/// reasons, we want to ensure that a value spawns over at most 8 bytes
-/// of aligns bytes.
-///
-/// Spanning over 9 bytes is possible for instance, if we do
-/// bitpacking with an amplitude of 63 bits.
-/// In this case, the second int will start on bit
-/// 63 (which belongs to byte 7) and ends at byte 15;
-/// Hence 9 bytes (from byte 7 to byte 15 included).
-///
-/// To avoid this, we force the number of bits to 64bits
-/// when the result is greater than `64-8 = 56 bits`.
-///
-/// Note that this only affects rare use cases spawning over
-/// a very large range of values. Even in this case, it results
-/// in an extra cost of at most 12% compared to the optimal
-/// number of bits.
-pub(crate) fn compute_num_bits(n: u64) -> u8 {
-    let amplitude = (64u32 - n.leading_zeros()) as u8;
-    if amplitude <= 64 - 8 {
-        amplitude
-    } else {
-        64
-    }
-}
-
-pub(crate) fn is_power_of_2(n: usize) -> bool {
-    (n > 0) && (n & (n - 1) == 0)
-}
-
 /// Has length trait
 pub trait HasLen {
     /// Return length
@@ -88,13 +52,13 @@ const HIGHEST_BIT: u64 = 1 << 63;
 ///
 /// # See also
 /// The [reverse mapping is `u64_to_i64`](./fn.u64_to_i64.html).
-#[inline(always)]
+#[inline]
 pub fn i64_to_u64(val: i64) -> u64 {
     (val as u64) ^ HIGHEST_BIT
 }
 
 /// Reverse the mapping given by [`i64_to_u64`](./fn.i64_to_u64.html).
-#[inline(always)]
+#[inline]
 pub fn u64_to_i64(val: u64) -> i64 {
     (val ^ HIGHEST_BIT) as i64
 }
@@ -104,14 +68,19 @@ pub fn u64_to_i64(val: u64) -> i64 {
 /// For simplicity, tantivy internally handles `f64` as `u64`.
 /// The mapping is defined by this function.
 ///
-/// Maps `f64` to `u64` so that lexical order is preserved.
+/// Maps `f64` to `u64` in a monotonic manner, so that bytes lexical order is preserved.
 ///
 /// This is more suited than simply casting (`val as u64`)
 /// which would truncate the result
 ///
+/// # Reference
+///
+/// Daniel Lemire's [blog post](https://lemire.me/blog/2020/12/14/converting-floating-point-numbers-to-integers-while-preserving-order/)
+/// explains the mapping in a clear manner.
+///
 /// # See also
 /// The [reverse mapping is `u64_to_f64`](./fn.u64_to_f64.html).
-#[inline(always)]
+#[inline]
 pub fn f64_to_u64(val: f64) -> u64 {
     let bits = val.to_bits();
     if val.is_sign_positive() {
@@ -122,23 +91,24 @@ pub fn f64_to_u64(val: f64) -> u64 {
 }
 
 /// Reverse the mapping given by [`i64_to_u64`](./fn.i64_to_u64.html).
-#[inline(always)]
+#[inline]
 pub fn u64_to_f64(val: u64) -> f64 {
-    f64::from_bits(
+    f64::from_bits(if val & HIGHEST_BIT != 0 {
-        if val & HIGHEST_BIT != 0 {
+        val ^ HIGHEST_BIT
-            val ^ HIGHEST_BIT
+    } else {
-        } else {
+        !val
-            !val
+    })
-        }
-    )
 }
 
 #[cfg(test)]
 pub(crate) mod test {
 
     pub use super::serialize::test::fixed_size_test;
-    use super::{compute_num_bits, i64_to_u64, u64_to_i64, f64_to_u64, u64_to_f64};
+    use super::{f64_to_u64, i64_to_u64, u64_to_f64, u64_to_i64};
+    use proptest::prelude::*;
     use std::f64;
+    use tantivy_bitpacker::compute_num_bits;
+    pub use tantivy_bitpacker::minmax;
 
     fn test_i64_converter_helper(val: i64) {
         assert_eq!(u64_to_i64(i64_to_u64(val)), val);
@@ -148,6 +118,15 @@ pub(crate) mod test {
         assert_eq!(u64_to_f64(f64_to_u64(val)), val);
     }
 
+    proptest! {
+        #[test]
+        fn test_f64_converter_monotonicity_proptest((left, right) in (proptest::num::f64::NORMAL, proptest::num::f64::NORMAL)) {
+            let left_u64 = f64_to_u64(left);
+            let right_u64 = f64_to_u64(right);
+            assert_eq!(left_u64 < right_u64,  left < right);
+        }
+    }
+
     #[test]
     fn test_i64_converter() {
         assert_eq!(i64_to_u64(i64::min_value()), u64::min_value());
@@ -172,7 +151,8 @@ pub(crate) mod test {
 
     #[test]
     fn test_f64_order() {
-        assert!(!(f64_to_u64(f64::NEG_INFINITY)..f64_to_u64(f64::INFINITY)).contains(&f64_to_u64(f64::NAN))); //nan is not a number
+        assert!(!(f64_to_u64(f64::NEG_INFINITY)..f64_to_u64(f64::INFINITY))
+            .contains(&f64_to_u64(f64::NAN))); //nan is not a number
         assert!(f64_to_u64(1.5) > f64_to_u64(1.0)); //same exponent, different mantissa
         assert!(f64_to_u64(2.0) > f64_to_u64(1.0)); //same mantissa, different exponent
         assert!(f64_to_u64(2.0) > f64_to_u64(1.5)); //different exponent and mantissa
@@ -200,4 +180,21 @@ pub(crate) mod test {
         assert!(((super::MAX_DOC_LIMIT - 1) as i32) >= 0);
         assert!((super::MAX_DOC_LIMIT as i32) < 0);
     }
+
+    #[test]
+    fn test_minmax_empty() {
+        let vals: Vec<u32> = vec![];
+        assert_eq!(minmax(vals.into_iter()), None);
+    }
+
+    #[test]
+    fn test_minmax_one() {
+        assert_eq!(minmax(vec![1].into_iter()), Some((1, 1)));
+    }
+
+    #[test]
+    fn test_minmax_two() {
+        assert_eq!(minmax(vec![1, 2].into_iter()), Some((1, 2)));
+        assert_eq!(minmax(vec![2, 1].into_iter()), Some((1, 2)));
+    }
 }

src/common/serialize.rs

@@ -14,6 +14,20 @@ pub trait BinarySerializable: fmt::Debug + Sized {
     fn deserialize<R: Read>(reader: &mut R) -> io::Result<Self>;
 }
 
+pub trait DeserializeFrom<T: BinarySerializable> {
+    fn deserialize(&mut self) -> io::Result<T>;
+}
+
+/// Implement deserialize from &[u8] for all types which implement BinarySerializable.
+///
+/// TryFrom would actually be preferrable, but not possible because of the orphan
+/// rules (not completely sure if this could be resolved)
+impl<T: BinarySerializable> DeserializeFrom<T> for &[u8] {
+    fn deserialize(&mut self) -> io::Result<T> {
+        T::deserialize(self)
+    }
+}
+
 /// `FixedSize` marks a `BinarySerializable` as
 /// always serializing to the same size.
 pub trait FixedSize: BinarySerializable {
@@ -61,6 +75,11 @@ impl<Left: BinarySerializable, Right: BinarySerializable> BinarySerializable for
         Ok((Left::deserialize(reader)?, Right::deserialize(reader)?))
     }
 }
+impl<Left: BinarySerializable + FixedSize, Right: BinarySerializable + FixedSize> FixedSize
+    for (Left, Right)
+{
+    const SIZE_IN_BYTES: usize = Left::SIZE_IN_BYTES + Right::SIZE_IN_BYTES;
+}
 
 impl BinarySerializable for u32 {
     fn serialize<W: Write>(&self, writer: &mut W) -> io::Result<()> {
@@ -89,6 +108,19 @@ impl FixedSize for u64 {
     const SIZE_IN_BYTES: usize = 8;
 }
 
+impl BinarySerializable for f32 {
+    fn serialize<W: Write>(&self, writer: &mut W) -> io::Result<()> {
+        writer.write_f32::<Endianness>(*self)
+    }
+    fn deserialize<R: Read>(reader: &mut R) -> io::Result<Self> {
+        reader.read_f32::<Endianness>()
+    }
+}
+
+impl FixedSize for f32 {
+    const SIZE_IN_BYTES: usize = 4;
+}
+
 impl BinarySerializable for i64 {
     fn serialize<W: Write>(&self, writer: &mut W) -> io::Result<()> {
         writer.write_i64::<Endianness>(*self)
@@ -199,10 +231,7 @@ pub mod test {
     fn test_serialize_string() {
         assert_eq!(serialize_test(String::from("")), 1);
         assert_eq!(serialize_test(String::from("ぽよぽよ")), 1 + 3 * 4);
-        assert_eq!(
+        assert_eq!(serialize_test(String::from("富士さん見える。")), 1 + 3 * 8);
-            serialize_test(String::from("富士さん見える。")),
-            1 + 3 * 8
-        );
     }
 
     #[test]

src/common/vint.rs

@@ -5,12 +5,12 @@ use std::io::Read;
 use std::io::Write;
 
 ///   Wrapper over a `u64` that serializes as a variable int.
-#[derive(Debug, Eq, PartialEq)]
+#[derive(Clone, Copy, Debug, Eq, PartialEq)]
 pub struct VInt(pub u64);
 
 const STOP_BIT: u8 = 128;
 
-pub fn serialize_vint_u32(val: u32) -> (u64, usize) {
+pub fn serialize_vint_u32(val: u32, buf: &mut [u8; 8]) -> &[u8] {
     const START_2: u64 = 1 << 7;
     const START_3: u64 = 1 << 14;
     const START_4: u64 = 1 << 21;
@@ -29,7 +29,7 @@ pub fn serialize_vint_u32(val: u32) -> (u64, usize) {
 
     let val = u64::from(val);
     const STOP_BIT: u64 = 128u64;
-    match val {
+    let (res, num_bytes) = match val {
         0..=STOP_1 => (val | STOP_BIT, 1),
         START_2..=STOP_2 => (
             (val & MASK_1) | ((val & MASK_2) << 1) | (STOP_BIT << (8)),
@@ -56,7 +56,9 @@ pub fn serialize_vint_u32(val: u32) -> (u64, usize) {
                 | (STOP_BIT << (8 * 4)),
             5,
         ),
-    }
+    };
+    LittleEndian::write_u64(&mut buf[..], res);
+    &buf[0..num_bytes]
 }
 
 /// Returns the number of bytes covered by a
@@ -85,23 +87,26 @@ fn vint_len(data: &[u8]) -> usize {
 /// If the buffer does not start by a valid
 /// vint payload
 pub fn read_u32_vint(data: &mut &[u8]) -> u32 {
-    let vlen = vint_len(*data);
+    let (result, vlen) = read_u32_vint_no_advance(*data);
+    *data = &data[vlen..];
+    result
+}
+
+pub fn read_u32_vint_no_advance(data: &[u8]) -> (u32, usize) {
+    let vlen = vint_len(data);
     let mut result = 0u32;
     let mut shift = 0u64;
     for &b in &data[..vlen] {
         result |= u32::from(b & 127u8) << shift;
         shift += 7;
     }
-    *data = &data[vlen..];
+    (result, vlen)
-    result
 }
-
 /// Write a `u32` as a vint payload.
 pub fn write_u32_vint<W: io::Write>(val: u32, writer: &mut W) -> io::Result<()> {
-    let (val, num_bytes) = serialize_vint_u32(val);
+    let mut buf = [0u8; 8];
-    let mut buffer = [0u8; 8];
+    let data = serialize_vint_u32(val, &mut buf);
-    LittleEndian::write_u64(&mut buffer, val);
+    writer.write_all(&data)
-    writer.write_all(&buffer[..num_bytes])
 }
 
 impl VInt {
@@ -172,7 +177,6 @@ mod tests {
     use super::serialize_vint_u32;
     use super::VInt;
     use crate::common::BinarySerializable;
-    use byteorder::{ByteOrder, LittleEndian};
 
     fn aux_test_vint(val: u64) {
         let mut v = [14u8; 10];
@@ -208,12 +212,10 @@ mod tests {
 
     fn aux_test_serialize_vint_u32(val: u32) {
         let mut buffer = [0u8; 10];
-        let mut buffer2 = [0u8; 10];
+        let mut buffer2 = [0u8; 8];
         let len_vint = VInt(val as u64).serialize_into(&mut buffer);
-        let (vint, len) = serialize_vint_u32(val);
+        let res2 = serialize_vint_u32(val, &mut buffer2);
-        assert_eq!(len, len_vint, "len wrong for val {}", val);
+        assert_eq!(&buffer[..len_vint], res2, "array wrong for {}", val);
-        LittleEndian::write_u64(&mut buffer2, vint);
-        assert_eq!(&buffer[..len], &buffer2[..len], "array wrong for {}", val);
     }
 
     #[test]

src/core/executor.rs

@@ -1,6 +1,5 @@
-use crate::Result;
 use crossbeam::channel;
-use scoped_pool::{Pool, ThreadConfig};
+use rayon::{ThreadPool, ThreadPoolBuilder};
 
 /// Search executor whether search request are single thread or multithread.
 ///
@@ -10,8 +9,10 @@ use scoped_pool::{Pool, ThreadConfig};
 /// API of a dependency, knowing it might conflict with a different version
 /// used by the client. Second, we may stop using rayon in the future.
 pub enum Executor {
+    /// Single thread variant of an Executor
     SingleThread,
-    ThreadPool(Pool),
+    /// Thread pool variant of an Executor
+    ThreadPool(ThreadPool),
 }
 
 impl Executor {
@@ -20,37 +21,39 @@ impl Executor {
         Executor::SingleThread
     }
 
-    // Creates an Executor that dispatches the tasks in a thread pool.
+    /// Creates an Executor that dispatches the tasks in a thread pool.
-    pub fn multi_thread(num_threads: usize, prefix: &'static str) -> Executor {
+    pub fn multi_thread(num_threads: usize, prefix: &'static str) -> crate::Result<Executor> {
-        let thread_config = ThreadConfig::new().prefix(prefix);
+        let pool = ThreadPoolBuilder::new()
-        let pool = Pool::with_thread_config(num_threads, thread_config);
+            .num_threads(num_threads)
-        Executor::ThreadPool(pool)
+            .thread_name(move |num| format!("{}{}", prefix, num))
+            .build()?;
+        Ok(Executor::ThreadPool(pool))
     }
 
-    // Perform a map in the thread pool.
+    /// Perform a map in the thread pool.
-    //
+    ///
-    // Regardless of the executor (`SingleThread` or `ThreadPool`), panics in the task
+    /// Regardless of the executor (`SingleThread` or `ThreadPool`), panics in the task
-    // will propagate to the caller.
+    /// will propagate to the caller.
     pub fn map<
         A: Send,
         R: Send,
         AIterator: Iterator<Item = A>,
-        F: Sized + Sync + Fn(A) -> Result<R>,
+        F: Sized + Sync + Fn(A) -> crate::Result<R>,
     >(
         &self,
         f: F,
         args: AIterator,
-    ) -> Result<Vec<R>> {
+    ) -> crate::Result<Vec<R>> {
         match self {
-            Executor::SingleThread => args.map(f).collect::<Result<_>>(),
+            Executor::SingleThread => args.map(f).collect::<crate::Result<_>>(),
             Executor::ThreadPool(pool) => {
                 let args_with_indices: Vec<(usize, A)> = args.enumerate().collect();
                 let num_fruits = args_with_indices.len();
                 let fruit_receiver = {
                     let (fruit_sender, fruit_receiver) = channel::unbounded();
-                    pool.scoped(|scope| {
+                    pool.scope(|scope| {
                         for arg_with_idx in args_with_indices {
-                            scope.execute(|| {
+                            scope.spawn(|_| {
                                 let (idx, arg) = arg_with_idx;
                                 let fruit = f(arg);
                                 if let Err(err) = fruit_sender.send((idx, fruit)) {
@@ -103,6 +106,7 @@ mod tests {
     #[should_panic] //< unfortunately the panic message is not propagated
     fn test_panic_propagates_multi_thread() {
         let _result: Vec<usize> = Executor::multi_thread(1, "search-test")
+            .unwrap()
             .map(
                 |_| {
                     panic!("panic should propagate");
@@ -126,6 +130,7 @@ mod tests {
     #[test]
     fn test_map_multithread() {
         let result: Vec<usize> = Executor::multi_thread(3, "search-test")
+            .unwrap()
             .map(|i| Ok(i * 2), 0..10)
             .unwrap();
         assert_eq!(result.len(), 10);

src/core/index.rs

@@ -1,62 +1,204 @@
-use super::segment::create_segment;
+use super::{segment::Segment, IndexSettings};
-use super::segment::Segment;
 use crate::core::Executor;
 use crate::core::IndexMeta;
 use crate::core::SegmentId;
 use crate::core::SegmentMeta;
 use crate::core::SegmentMetaInventory;
 use crate::core::META_FILEPATH;
+use crate::directory::error::OpenReadError;
 use crate::directory::ManagedDirectory;
 #[cfg(feature = "mmap")]
 use crate::directory::MmapDirectory;
 use crate::directory::INDEX_WRITER_LOCK;
-use crate::directory::{Directory, RAMDirectory};
+use crate::directory::{Directory, RamDirectory};
 use crate::error::DataCorruption;
 use crate::error::TantivyError;
-use crate::indexer::index_writer::HEAP_SIZE_MIN;
+use crate::indexer::index_writer::{HEAP_SIZE_MIN, MAX_NUM_THREAD};
 use crate::indexer::segment_updater::save_new_metas;
 use crate::reader::IndexReader;
 use crate::reader::IndexReaderBuilder;
 use crate::schema::Field;
 use crate::schema::FieldType;
 use crate::schema::Schema;
-use crate::tokenizer::BoxedTokenizer;
+use crate::tokenizer::{TextAnalyzer, TokenizerManager};
-use crate::tokenizer::TokenizerManager;
 use crate::IndexWriter;
-use crate::Result;
+use std::collections::HashSet;
-use num_cpus;
-use std::borrow::BorrowMut;
 use std::fmt;
+
 #[cfg(feature = "mmap")]
 use std::path::Path;
+use std::path::PathBuf;
 use std::sync::Arc;
 
-fn load_metas(directory: &dyn Directory, inventory: &SegmentMetaInventory) -> Result<IndexMeta> {
+fn load_metas(
+    directory: &dyn Directory,
+    inventory: &SegmentMetaInventory,
+) -> crate::Result<IndexMeta> {
     let meta_data = directory.atomic_read(&META_FILEPATH)?;
-    let meta_string = String::from_utf8_lossy(&meta_data);
+    let meta_string = String::from_utf8(meta_data).map_err(|_utf8_err| {
+        error!("Meta data is not valid utf8.");
+        DataCorruption::new(
+            META_FILEPATH.to_path_buf(),
+            "Meta file does not contain valid utf8 file.".to_string(),
+        )
+    })?;
     IndexMeta::deserialize(&meta_string, &inventory)
         .map_err(|e| {
             DataCorruption::new(
                 META_FILEPATH.to_path_buf(),
-                format!("Meta file cannot be deserialized. {:?}.", e),
+                format!(
+                    "Meta file cannot be deserialized. {:?}. Content: {:?}",
+                    e, meta_string
+                ),
             )
         })
         .map_err(From::from)
 }
 
+/// IndexBuilder can be used to create an index.
+///
+/// Use in conjunction with `SchemaBuilder`. Global index settings
+/// can be configured with `IndexSettings`
+///
+/// # Examples
+///
+/// ```
+/// use tantivy::schema::*;
+/// use tantivy::{Index, IndexSettings, IndexSortByField, Order};
+///
+/// let mut schema_builder = Schema::builder();
+/// let id_field = schema_builder.add_text_field("id", STRING);
+/// let title_field = schema_builder.add_text_field("title", TEXT);
+/// let body_field = schema_builder.add_text_field("body", TEXT);
+/// let number_field = schema_builder.add_u64_field(
+///     "number",
+///     IntOptions::default().set_fast(Cardinality::SingleValue),
+/// );
+///
+/// let schema = schema_builder.build();
+/// let settings = IndexSettings{sort_by_field: Some(IndexSortByField{field:"number".to_string(), order:Order::Asc}), ..Default::default()};
+/// let index = Index::builder().schema(schema).settings(settings).create_in_ram();
+///
+/// ```
+pub struct IndexBuilder {
+    schema: Option<Schema>,
+    index_settings: IndexSettings,
+}
+impl Default for IndexBuilder {
+    fn default() -> Self {
+        IndexBuilder::new()
+    }
+}
+impl IndexBuilder {
+    /// Creates a new `IndexBuilder`
+    pub fn new() -> Self {
+        Self {
+            schema: None,
+            index_settings: IndexSettings::default(),
+        }
+    }
+    /// Set the settings
+    pub fn settings(mut self, settings: IndexSettings) -> Self {
+        self.index_settings = settings;
+        self
+    }
+    /// Set the schema
+    pub fn schema(mut self, schema: Schema) -> Self {
+        self.schema = Some(schema);
+        self
+    }
+    /// Creates a new index using the `RAMDirectory`.
+    ///
+    /// The index will be allocated in anonymous memory.
+    /// This should only be used for unit tests.
+    pub fn create_in_ram(self) -> Result<Index, TantivyError> {
+        let ram_directory = RamDirectory::create();
+        Ok(self
+            .create(ram_directory)
+            .expect("Creating a RAMDirectory should never fail"))
+    }
+    /// Creates a new index in a given filepath.
+    /// The index will use the `MMapDirectory`.
+    ///
+    /// If a previous index was in this directory, then its meta file will be destroyed.
+    #[cfg(feature = "mmap")]
+    pub fn create_in_dir<P: AsRef<Path>>(self, directory_path: P) -> crate::Result<Index> {
+        let mmap_directory = MmapDirectory::open(directory_path)?;
+        if Index::exists(&mmap_directory)? {
+            return Err(TantivyError::IndexAlreadyExists);
+        }
+        self.create(mmap_directory)
+    }
+    /// Creates a new index in a temp directory.
+    ///
+    /// The index will use the `MMapDirectory` in a newly created directory.
+    /// The temp directory will be destroyed automatically when the `Index` object
+    /// is destroyed.
+    ///
+    /// The temp directory is only used for testing the `MmapDirectory`.
+    /// For other unit tests, prefer the `RAMDirectory`, see: `create_in_ram`.
+    #[cfg(feature = "mmap")]
+    pub fn create_from_tempdir(self) -> crate::Result<Index> {
+        let mmap_directory = MmapDirectory::create_from_tempdir()?;
+        self.create(mmap_directory)
+    }
+    fn get_expect_schema(&self) -> crate::Result<Schema> {
+        self.schema
+            .as_ref()
+            .cloned()
+            .ok_or(TantivyError::IndexBuilderMissingArgument("schema"))
+    }
+    /// Opens or creates a new index in the provided directory
+    pub fn open_or_create<Dir: Directory>(self, dir: Dir) -> crate::Result<Index> {
+        if !Index::exists(&dir)? {
+            return self.create(dir);
+        }
+        let index = Index::open(dir)?;
+        if index.schema() == self.get_expect_schema()? {
+            Ok(index)
+        } else {
+            Err(TantivyError::SchemaError(
+                "An index exists but the schema does not match.".to_string(),
+            ))
+        }
+    }
+    /// Creates a new index given an implementation of the trait `Directory`.
+    ///
+    /// If a directory previously existed, it will be erased.
+    fn create<Dir: Directory>(self, dir: Dir) -> crate::Result<Index> {
+        let directory = ManagedDirectory::wrap(dir)?;
+        save_new_metas(
+            self.get_expect_schema()?,
+            self.index_settings.clone(),
+            &directory,
+        )?;
+        let mut metas = IndexMeta::with_schema(self.get_expect_schema()?);
+        metas.index_settings = self.index_settings;
+        let index = Index::open_from_metas(directory, &metas, SegmentMetaInventory::default());
+        Ok(index)
+    }
+}
+
 /// Search Index
 #[derive(Clone)]
 pub struct Index {
     directory: ManagedDirectory,
     schema: Schema,
+    settings: IndexSettings,
     executor: Arc<Executor>,
     tokenizers: TokenizerManager,
     inventory: SegmentMetaInventory,
 }
 
 impl Index {
-    /// Examines the director to see if it contains an index
+    /// Creates a new builder.
-    pub fn exists<Dir: Directory>(dir: &Dir) -> bool {
+    pub fn builder() -> IndexBuilder {
+        IndexBuilder::new()
+    }
+    /// Examines the directory to see if it contains an index.
+    ///
+    /// Effectively, it only checks for the presence of the `meta.json` file.
+    pub fn exists<Dir: Directory>(dir: &Dir) -> Result<bool, OpenReadError> {
         dir.exists(&META_FILEPATH)
     }
 
@@ -72,24 +214,24 @@ impl Index {
 
     /// Replace the default single thread search executor pool
     /// by a thread pool with a given number of threads.
-    pub fn set_multithread_executor(&mut self, num_threads: usize) {
+    pub fn set_multithread_executor(&mut self, num_threads: usize) -> crate::Result<()> {
-        self.executor = Arc::new(Executor::multi_thread(num_threads, "thrd-tantivy-search-"));
+        self.executor = Arc::new(Executor::multi_thread(num_threads, "thrd-tantivy-search-")?);
+        Ok(())
     }
 
     /// Replace the default single thread search executor pool
     /// by a thread pool with a given number of threads.
-    pub fn set_default_multithread_executor(&mut self) {
+    pub fn set_default_multithread_executor(&mut self) -> crate::Result<()> {
         let default_num_threads = num_cpus::get();
-        self.set_multithread_executor(default_num_threads);
+        self.set_multithread_executor(default_num_threads)
     }
 
-    /// Creates a new index using the `RAMDirectory`.
+    /// Creates a new index using the `RamDirectory`.
     ///
     /// The index will be allocated in anonymous memory.
     /// This should only be used for unit tests.
     pub fn create_in_ram(schema: Schema) -> Index {
-        let ram_directory = RAMDirectory::create();
+        IndexBuilder::new().schema(schema).create_in_ram().unwrap()
-        Index::create(ram_directory, schema).expect("Creating a RAMDirectory should never fail")
     }
 
     /// Creates a new index in a given filepath.
@@ -97,29 +239,18 @@ impl Index {
     ///
     /// If a previous index was in this directory, then its meta file will be destroyed.
     #[cfg(feature = "mmap")]
-    pub fn create_in_dir<P: AsRef<Path>>(directory_path: P, schema: Schema) -> Result<Index> {
+    pub fn create_in_dir<P: AsRef<Path>>(
-        let mmap_directory = MmapDirectory::open(directory_path)?;
+        directory_path: P,
-        if Index::exists(&mmap_directory) {
+        schema: Schema,
-            return Err(TantivyError::IndexAlreadyExists);
+    ) -> crate::Result<Index> {
-        }
+        IndexBuilder::new()
-
+            .schema(schema)
-        Index::create(mmap_directory, schema)
+            .create_in_dir(directory_path)
     }
 
     /// Opens or creates a new index in the provided directory
-    pub fn open_or_create<Dir: Directory>(dir: Dir, schema: Schema) -> Result<Index> {
+    pub fn open_or_create<Dir: Directory>(dir: Dir, schema: Schema) -> crate::Result<Index> {
-        if Index::exists(&dir) {
+        IndexBuilder::new().schema(schema).open_or_create(dir)
-            let index = Index::open(dir)?;
-            if index.schema() == schema {
-                Ok(index)
-            } else {
-                Err(TantivyError::SchemaError(
-                    "An index exists but the schema does not match.".to_string(),
-                ))
-            }
-        } else {
-            Index::create(dir, schema)
-        }
     }
 
     /// Creates a new index in a temp directory.
@@ -129,42 +260,40 @@ impl Index {
     /// is destroyed.
     ///
     /// The temp directory is only used for testing the `MmapDirectory`.
-    /// For other unit tests, prefer the `RAMDirectory`, see: `create_in_ram`.
+    /// For other unit tests, prefer the `RamDirectory`, see: `create_in_ram`.
     #[cfg(feature = "mmap")]
-    pub fn create_from_tempdir(schema: Schema) -> Result<Index> {
+    pub fn create_from_tempdir(schema: Schema) -> crate::Result<Index> {
-        let mmap_directory = MmapDirectory::create_from_tempdir()?;
+        IndexBuilder::new().schema(schema).create_from_tempdir()
-        Index::create(mmap_directory, schema)
     }
 
-    /// Creates a new index given an implementation of the trait `Directory`
+    /// Creates a new index given an implementation of the trait `Directory`.
-    pub fn create<Dir: Directory>(dir: Dir, schema: Schema) -> Result<Index> {
-        let directory = ManagedDirectory::wrap(dir)?;
-        Index::from_directory(directory, schema)
-    }
-
-    /// Create a new index from a directory.
     ///
-    /// This will overwrite existing meta.json
+    /// If a directory previously existed, it will be erased.
-    fn from_directory(mut directory: ManagedDirectory, schema: Schema) -> Result<Index> {
+    pub fn create<Dir: Directory>(
-        save_new_metas(schema.clone(), directory.borrow_mut())?;
+        dir: Dir,
-        let metas = IndexMeta::with_schema(schema);
+        schema: Schema,
-        Index::create_from_metas(directory, &metas, SegmentMetaInventory::default())
+        settings: IndexSettings,
+    ) -> crate::Result<Index> {
+        let mut builder = IndexBuilder::new().schema(schema);
+        builder = builder.settings(settings);
+        builder.create(dir)
     }
 
     /// Creates a new index given a directory and an `IndexMeta`.
-    fn create_from_metas(
+    fn open_from_metas(
         directory: ManagedDirectory,
         metas: &IndexMeta,
         inventory: SegmentMetaInventory,
-    ) -> Result<Index> {
+    ) -> Index {
         let schema = metas.schema.clone();
-        Ok(Index {
+        Index {
+            settings: metas.index_settings.clone(),
             directory,
             schema,
             tokenizers: TokenizerManager::default(),
             executor: Arc::new(Executor::single_thread()),
             inventory,
-        })
+        }
     }
 
     /// Accessor for the tokenizer manager.
@@ -173,11 +302,11 @@ impl Index {
     }
 
     /// Helper to access the tokenizer associated to a specific field.
-    pub fn tokenizer_for_field(&self, field: Field) -> Result<Box<dyn BoxedTokenizer>> {
+    pub fn tokenizer_for_field(&self, field: Field) -> crate::Result<TextAnalyzer> {
         let field_entry = self.schema.get_field_entry(field);
         let field_type = field_entry.field_type();
         let tokenizer_manager: &TokenizerManager = self.tokenizers();
-        let tokenizer_name_opt: Option<Box<dyn BoxedTokenizer>> = match field_type {
+        let tokenizer_name_opt: Option<TextAnalyzer> = match field_type {
             FieldType::Str(text_options) => text_options
                 .get_indexing_options()
                 .map(|text_indexing_options| text_indexing_options.tokenizer().to_string())
@@ -196,7 +325,7 @@ impl Index {
     /// Create a default `IndexReader` for the given index.
     ///
     /// See [`Index.reader_builder()`](#method.reader_builder).
-    pub fn reader(&self) -> Result<IndexReader> {
+    pub fn reader(&self) -> crate::Result<IndexReader> {
         self.reader_builder().try_into()
     }
 
@@ -211,25 +340,40 @@ impl Index {
 
     /// Opens a new directory from an index path.
     #[cfg(feature = "mmap")]
-    pub fn open_in_dir<P: AsRef<Path>>(directory_path: P) -> Result<Index> {
+    pub fn open_in_dir<P: AsRef<Path>>(directory_path: P) -> crate::Result<Index> {
         let mmap_directory = MmapDirectory::open(directory_path)?;
         Index::open(mmap_directory)
     }
 
-    pub(crate) fn inventory(&self) -> &SegmentMetaInventory {
+    /// Returns the list of the segment metas tracked by the index.
-        &self.inventory
+    ///
+    /// Such segments can of course be part of the index,
+    /// but also they could be segments being currently built or in the middle of a merge
+    /// operation.
+    pub(crate) fn list_all_segment_metas(&self) -> Vec<SegmentMeta> {
+        self.inventory.all()
+    }
+
+    /// Creates a new segment_meta (Advanced user only).
+    ///
+    /// As long as the `SegmentMeta` lives, the files associated with the
+    /// `SegmentMeta` are guaranteed to not be garbage collected, regardless of
+    /// whether the segment is recorded as part of the index or not.
+    pub fn new_segment_meta(&self, segment_id: SegmentId, max_doc: u32) -> SegmentMeta {
+        self.inventory.new_segment_meta(segment_id, max_doc)
     }
 
     /// Open the index using the provided directory
-    pub fn open<D: Directory>(directory: D) -> Result<Index> {
+    pub fn open<D: Directory>(directory: D) -> crate::Result<Index> {
         let directory = ManagedDirectory::wrap(directory)?;
         let inventory = SegmentMetaInventory::default();
         let metas = load_metas(&directory, &inventory)?;
-        Index::create_from_metas(directory, &metas, inventory)
+        let index = Index::open_from_metas(directory, &metas, inventory);
+        Ok(index)
     }
 
     /// Reads the index meta file from the directory.
-    pub fn load_metas(&self) -> Result<IndexMeta> {
+    pub fn load_metas(&self) -> crate::Result<IndexMeta> {
         load_metas(self.directory(), &self.inventory)
     }
 
@@ -249,7 +393,7 @@ impl Index {
     /// Each thread will receive a budget of  `overall_heap_size_in_bytes / num_threads`.
     ///
     /// # Errors
-    /// If the lockfile already exists, returns `Error::DirectoryLockBusy` or an `Error::IOError`.
+    /// If the lockfile already exists, returns `Error::DirectoryLockBusy` or an `Error::IoError`.
     ///
     /// # Panics
     /// If the heap size per thread is too small, panics.
@@ -257,7 +401,7 @@ impl Index {
         &self,
         num_threads: usize,
         overall_heap_size_in_bytes: usize,
-    ) -> Result<IndexWriter> {
+    ) -> crate::Result<IndexWriter> {
         let directory_lock = self
             .directory
             .acquire_lock(&INDEX_WRITER_LOCK)
@@ -265,7 +409,7 @@ impl Index {
                 TantivyError::LockFailure(
                     err,
                     Some(
-                        "Failed to acquire index lock. If you are using\
+                        "Failed to acquire index lock. If you are using \
                          a regular directory, this means there is already an \
                          `IndexWriter` working on this `Directory`, in this process \
                          or in a different process."
@@ -282,9 +426,19 @@ impl Index {
         )
     }
 
+    /// Helper to create an index writer for tests.
+    ///
+    /// That index writer only simply has a single thread and a heap of 10 MB.
+    /// Using a single thread gives us a deterministic allocation of DocId.
+    #[cfg(test)]
+    pub fn writer_for_tests(&self) -> crate::Result<IndexWriter> {
+        self.writer_with_num_threads(1, 10_000_000)
+    }
+
     /// Creates a multithreaded writer
     ///
-    /// Tantivy will automatically define the number of threads to use.
+    /// Tantivy will automatically define the number of threads to use, but
+    /// no more than [`MAX_NUM_THREAD`] threads.
     /// `overall_heap_size_in_bytes` is the total target memory usage that will be split
     /// between a given number of threads.
     ///
@@ -292,8 +446,8 @@ impl Index {
     /// If the lockfile already exists, returns `Error::FileAlreadyExists`.
     /// # Panics
     /// If the heap size per thread is too small, panics.
-    pub fn writer(&self, overall_heap_size_in_bytes: usize) -> Result<IndexWriter> {
+    pub fn writer(&self, overall_heap_size_in_bytes: usize) -> crate::Result<IndexWriter> {
-        let mut num_threads = num_cpus::get();
+        let mut num_threads = std::cmp::min(num_cpus::get(), MAX_NUM_THREAD);
         let heap_size_in_bytes_per_thread = overall_heap_size_in_bytes / num_threads;
         if heap_size_in_bytes_per_thread < HEAP_SIZE_MIN {
             num_threads = (overall_heap_size_in_bytes / HEAP_SIZE_MIN).max(1);
@@ -301,6 +455,18 @@ impl Index {
         self.writer_with_num_threads(num_threads, overall_heap_size_in_bytes)
     }
 
+    /// Accessor to the index settings
+    ///
+    pub fn settings(&self) -> &IndexSettings {
+        &self.settings
+    }
+
+    /// Accessor to the index settings
+    ///
+    pub fn settings_mut(&mut self) -> &mut IndexSettings {
+        &mut self.settings
+    }
+
     /// Accessor to the index schema
     ///
     /// The schema is actually cloned.
@@ -309,7 +475,7 @@ impl Index {
     }
 
     /// Returns the list of segments that are searchable
-    pub fn searchable_segments(&self) -> Result<Vec<Segment>> {
+    pub fn searchable_segments(&self) -> crate::Result<Vec<Segment>> {
         Ok(self
             .searchable_segment_metas()?
             .into_iter()
@@ -319,7 +485,7 @@ impl Index {
 
     #[doc(hidden)]
     pub fn segment(&self, segment_meta: SegmentMeta) -> Segment {
-        create_segment(self.clone(), segment_meta)
+        Segment::for_index(self.clone(), segment_meta)
     }
 
     /// Creates a new segment.
@@ -342,18 +508,23 @@ impl Index {
 
     /// Reads the meta.json and returns the list of
     /// `SegmentMeta` from the last commit.
-    pub fn searchable_segment_metas(&self) -> Result<Vec<SegmentMeta>> {
+    pub fn searchable_segment_metas(&self) -> crate::Result<Vec<SegmentMeta>> {
         Ok(self.load_metas()?.segments)
     }
 
     /// Returns the list of segment ids that are searchable.
-    pub fn searchable_segment_ids(&self) -> Result<Vec<SegmentId>> {
+    pub fn searchable_segment_ids(&self) -> crate::Result<Vec<SegmentId>> {
         Ok(self
             .searchable_segment_metas()?
             .iter()
             .map(SegmentMeta::id)
             .collect())
     }
+
+    /// Returns the set of corrupted files
+    pub fn validate_checksum(&self) -> crate::Result<HashSet<PathBuf>> {
+        self.directory.list_damaged().map_err(Into::into)
+    }
 }
 
 impl fmt::Debug for Index {
@@ -364,15 +535,15 @@ impl fmt::Debug for Index {
 
 #[cfg(test)]
 mod tests {
-    use crate::directory::RAMDirectory;
     use crate::schema::Field;
     use crate::schema::{Schema, INDEXED, TEXT};
-    use crate::Index;
     use crate::IndexReader;
-    use crate::IndexWriter;
     use crate::ReloadPolicy;
-    use std::thread;
+    use crate::{
-    use std::time::Duration;
+        directory::{RamDirectory, WatchCallback},
+        IndexSettings,
+    };
+    use crate::{Directory, Index};
 
     #[test]
     fn test_indexer_for_field() {
@@ -390,41 +561,66 @@ mod tests {
 
     #[test]
     fn test_index_exists() {
-        let directory = RAMDirectory::create();
+        let directory = RamDirectory::create();
-        assert!(!Index::exists(&directory));
+        assert!(!Index::exists(&directory).unwrap());
-        assert!(Index::create(directory.clone(), throw_away_schema()).is_ok());
+        assert!(Index::create(
-        assert!(Index::exists(&directory));
+            directory.clone(),
+            throw_away_schema(),
+            IndexSettings::default()
+        )
+        .is_ok());
+        assert!(Index::exists(&directory).unwrap());
     }
 
     #[test]
     fn open_or_create_should_create() {
-        let directory = RAMDirectory::create();
+        let directory = RamDirectory::create();
-        assert!(!Index::exists(&directory));
+        assert!(!Index::exists(&directory).unwrap());
         assert!(Index::open_or_create(directory.clone(), throw_away_schema()).is_ok());
-        assert!(Index::exists(&directory));
+        assert!(Index::exists(&directory).unwrap());
     }
 
     #[test]
     fn open_or_create_should_open() {
-        let directory = RAMDirectory::create();
+        let directory = RamDirectory::create();
-        assert!(Index::create(directory.clone(), throw_away_schema()).is_ok());
+        assert!(Index::create(
-        assert!(Index::exists(&directory));
+            directory.clone(),
+            throw_away_schema(),
+            IndexSettings::default()
+        )
+        .is_ok());
+        assert!(Index::exists(&directory).unwrap());
         assert!(Index::open_or_create(directory, throw_away_schema()).is_ok());
     }
 
     #[test]
     fn create_should_wipeoff_existing() {
-        let directory = RAMDirectory::create();
+        let directory = RamDirectory::create();
-        assert!(Index::create(directory.clone(), throw_away_schema()).is_ok());
+        assert!(Index::create(
-        assert!(Index::exists(&directory));
+            directory.clone(),
-        assert!(Index::create(directory.clone(), Schema::builder().build()).is_ok());
+            throw_away_schema(),
+            IndexSettings::default()
+        )
+        .is_ok());
+        assert!(Index::exists(&directory).unwrap());
+        assert!(Index::create(
+            directory.clone(),
+            Schema::builder().build(),
+            IndexSettings::default()
+        )
+        .is_ok());
     }
 
     #[test]
     fn open_or_create_exists_but_schema_does_not_match() {
-        let directory = RAMDirectory::create();
+        let directory = RamDirectory::create();
-        assert!(Index::create(directory.clone(), throw_away_schema()).is_ok());
+        assert!(Index::create(
-        assert!(Index::exists(&directory));
+            directory.clone(),
+            throw_away_schema(),
+            IndexSettings::default()
+        )
+        .is_ok());
+        assert!(Index::exists(&directory).unwrap());
         assert!(Index::open_or_create(directory.clone(), throw_away_schema()).is_ok());
         let err = Index::open_or_create(directory, Schema::builder().build());
         assert_eq!(
@@ -450,61 +646,63 @@ mod tests {
             .try_into()
             .unwrap();
         assert_eq!(reader.searcher().num_docs(), 0);
-        let mut writer = index.writer_with_num_threads(1, 3_000_000).unwrap();
+        test_index_on_commit_reload_policy_aux(field, &index, &reader);
-        test_index_on_commit_reload_policy_aux(field, &mut writer, &reader);
     }
 
     #[cfg(feature = "mmap")]
     mod mmap_specific {
 
         use super::*;
+        use crate::Directory;
         use std::path::PathBuf;
-        use tempdir::TempDir;
+        use tempfile::TempDir;
 
         #[test]
         fn test_index_on_commit_reload_policy_mmap() {
             let schema = throw_away_schema();
             let field = schema.get_field("num_likes").unwrap();
-            let tempdir = TempDir::new("index").unwrap();
+            let tempdir = TempDir::new().unwrap();
             let tempdir_path = PathBuf::from(tempdir.path());
             let index = Index::create_in_dir(&tempdir_path, schema).unwrap();
-            let mut writer = index.writer_with_num_threads(1, 3_000_000).unwrap();
-            writer.commit().unwrap();
             let reader = index
                 .reader_builder()
                 .reload_policy(ReloadPolicy::OnCommit)
                 .try_into()
                 .unwrap();
             assert_eq!(reader.searcher().num_docs(), 0);
-            test_index_on_commit_reload_policy_aux(field, &mut writer, &reader);
+            test_index_on_commit_reload_policy_aux(field, &index, &reader);
         }
 
         #[test]
-        fn test_index_manual_policy_mmap() {
+        fn test_index_manual_policy_mmap() -> crate::Result<()> {
             let schema = throw_away_schema();
             let field = schema.get_field("num_likes").unwrap();
-            let index = Index::create_from_tempdir(schema).unwrap();
+            let mut index = Index::create_from_tempdir(schema)?;
-            let mut writer = index.writer_with_num_threads(1, 3_000_000).unwrap();
+            let mut writer = index.writer_for_tests()?;
-            writer.commit().unwrap();
+            writer.commit()?;
             let reader = index
                 .reader_builder()
                 .reload_policy(ReloadPolicy::Manual)
-                .try_into()
+                .try_into()?;
-                .unwrap();
             assert_eq!(reader.searcher().num_docs(), 0);
             writer.add_document(doc!(field=>1u64));
-            writer.commit().unwrap();
+            let (sender, receiver) = crossbeam::channel::unbounded();
-            thread::sleep(Duration::from_millis(500));
+            let _handle = index.directory_mut().watch(WatchCallback::new(move || {
+                let _ = sender.send(());
+            }));
+            writer.commit()?;
+            assert!(receiver.recv().is_ok());
             assert_eq!(reader.searcher().num_docs(), 0);
-            reader.reload().unwrap();
+            reader.reload()?;
             assert_eq!(reader.searcher().num_docs(), 1);
+            Ok(())
         }
 
         #[test]
         fn test_index_on_commit_reload_policy_different_directories() {
             let schema = throw_away_schema();
             let field = schema.get_field("num_likes").unwrap();
-            let tempdir = TempDir::new("index").unwrap();
+            let tempdir = TempDir::new().unwrap();
             let tempdir_path = PathBuf::from(tempdir.path());
             let write_index = Index::create_in_dir(&tempdir_path, schema).unwrap();
             let read_index = Index::open_in_dir(&tempdir_path).unwrap();
@@ -514,39 +712,38 @@ mod tests {
                 .try_into()
                 .unwrap();
             assert_eq!(reader.searcher().num_docs(), 0);
-            let mut writer = write_index.writer_with_num_threads(1, 3_000_000).unwrap();
+            test_index_on_commit_reload_policy_aux(field, &write_index, &reader);
-            test_index_on_commit_reload_policy_aux(field, &mut writer, &reader);
         }
     }
-
+    fn test_index_on_commit_reload_policy_aux(field: Field, index: &Index, reader: &IndexReader) {
-    fn test_index_on_commit_reload_policy_aux(
+        let mut reader_index = reader.index();
-        field: Field,
+        let (sender, receiver) = crossbeam::channel::unbounded();
-        writer: &mut IndexWriter,
+        let _watch_handle = reader_index
-        reader: &IndexReader,
+            .directory_mut()
-    ) {
+            .watch(WatchCallback::new(move || {
+                let _ = sender.send(());
+            }));
+        let mut writer = index.writer_for_tests().unwrap();
         assert_eq!(reader.searcher().num_docs(), 0);
         writer.add_document(doc!(field=>1u64));
         writer.commit().unwrap();
-        let mut count = 0;
+        // We need a loop here because it is possible for notify to send more than
-        for _ in 0..100 {
+        // one modify event. It was observed on CI on MacOS.
-            count = reader.searcher().num_docs();
+        loop {
-            if count > 0 {
+            assert!(receiver.recv().is_ok());
+            if reader.searcher().num_docs() == 1 {
                 break;
             }
-            thread::sleep(Duration::from_millis(100));
         }
-        assert_eq!(count, 1);
         writer.add_document(doc!(field=>2u64));
         writer.commit().unwrap();
-        let mut count = 0;
+        // ... Same as above
-        for _ in 0..10 {
+        loop {
-            count = reader.searcher().num_docs();
+            assert!(receiver.recv().is_ok());
-            if count > 1 {
+            if reader.searcher().num_docs() == 2 {
                 break;
             }
-            thread::sleep(Duration::from_millis(100));
         }
-        assert_eq!(count, 2);
     }
 
     // This test will not pass on windows, because windows
@@ -554,18 +751,22 @@ mod tests {
     #[cfg(not(target_os = "windows"))]
     #[test]
     fn garbage_collect_works_as_intended() {
-        let directory = RAMDirectory::create();
+        let directory = RamDirectory::create();
         let schema = throw_away_schema();
         let field = schema.get_field("num_likes").unwrap();
-        let index = Index::create(directory.clone(), schema).unwrap();
+        let index = Index::create(directory.clone(), schema, IndexSettings::default()).unwrap();
 
         let mut writer = index.writer_with_num_threads(8, 24_000_000).unwrap();
         for i in 0u64..8_000u64 {
             writer.add_document(doc!(field => i));
         }
+        let (sender, receiver) = crossbeam::channel::unbounded();
+        let _handle = directory.watch(WatchCallback::new(move || {
+            let _ = sender.send(());
+        }));
         writer.commit().unwrap();
         let mem_right_after_commit = directory.total_mem_usage();
-        thread::sleep(Duration::from_millis(1_000));
+        assert!(receiver.recv().is_ok());
         let reader = index
             .reader_builder()
             .reload_policy(ReloadPolicy::Manual)
@@ -579,7 +780,11 @@ mod tests {
         reader.reload().unwrap();
         let searcher = reader.searcher();
         assert_eq!(searcher.num_docs(), 8_000);
-        assert!(mem_right_after_merge_finished < mem_right_after_commit);
+        assert!(
+            mem_right_after_merge_finished < mem_right_after_commit,
+            "(mem after merge){} is expected < (mem before merge){}",
+            mem_right_after_merge_finished,
+            mem_right_after_commit
+        );
     }
-
 }

src/core/index_meta.rs

@@ -1,13 +1,12 @@
 use super::SegmentComponent;
-use crate::core::SegmentId;
 use crate::schema::Schema;
 use crate::Opstamp;
+use crate::{core::SegmentId, store::Compressor};
 use census::{Inventory, TrackedObject};
-use serde;
+use serde::{Deserialize, Serialize};
-use serde_json;
-use std::collections::HashSet;
-use std::fmt;
 use std::path::PathBuf;
+use std::{collections::HashSet, sync::atomic::AtomicBool};
+use std::{fmt, sync::Arc};
 
 #[derive(Clone, Debug, Serialize, Deserialize)]
 struct DeleteMeta {
@@ -30,11 +29,11 @@ impl SegmentMetaInventory {
             .collect::<Vec<_>>()
     }
 
-    #[doc(hidden)]
     pub fn new_segment_meta(&self, segment_id: SegmentId, max_doc: u32) -> SegmentMeta {
         let inner = InnerSegmentMeta {
             segment_id,
             max_doc,
+            include_temp_doc_store: Arc::new(AtomicBool::new(true)),
             deletes: None,
         };
         SegmentMeta::from(self.inventory.track(inner))
@@ -82,6 +81,15 @@ impl SegmentMeta {
         self.tracked.segment_id
     }
 
+    /// Removes the Component::TempStore from the alive list and
+    /// therefore marks the temp docstore file to be deleted by
+    /// the garbage collection.
+    pub fn untrack_temp_docstore(&self) {
+        self.tracked
+            .include_temp_doc_store
+            .store(false, std::sync::atomic::Ordering::Relaxed);
+    }
+
     /// Returns the number of deleted documents.
     pub fn num_deleted_docs(&self) -> u32 {
         self.tracked
@@ -98,9 +106,20 @@ impl SegmentMeta {
     /// is by removing all files that have been created by tantivy
     /// and are not used by any segment anymore.
     pub fn list_files(&self) -> HashSet<PathBuf> {
-        SegmentComponent::iterator()
+        if self
-            .map(|component| self.relative_path(*component))
+            .tracked
-            .collect::<HashSet<PathBuf>>()
+            .include_temp_doc_store
+            .load(std::sync::atomic::Ordering::Relaxed)
+        {
+            SegmentComponent::iterator()
+                .map(|component| self.relative_path(*component))
+                .collect::<HashSet<PathBuf>>()
+        } else {
+            SegmentComponent::iterator()
+                .filter(|comp| *comp != &SegmentComponent::TempStore)
+                .map(|component| self.relative_path(*component))
+                .collect::<HashSet<PathBuf>>()
+        }
     }
 
     /// Returns the relative path of a component of our segment.
@@ -110,14 +129,14 @@ impl SegmentMeta {
     pub fn relative_path(&self, component: SegmentComponent) -> PathBuf {
         let mut path = self.id().uuid_string();
         path.push_str(&*match component {
-            SegmentComponent::POSTINGS => ".idx".to_string(),
+            SegmentComponent::Postings => ".idx".to_string(),
-            SegmentComponent::POSITIONS => ".pos".to_string(),
+            SegmentComponent::Positions => ".pos".to_string(),
-            SegmentComponent::POSITIONSSKIP => ".posidx".to_string(),
+            SegmentComponent::Terms => ".term".to_string(),
-            SegmentComponent::TERMS => ".term".to_string(),
+            SegmentComponent::Store => ".store".to_string(),
-            SegmentComponent::STORE => ".store".to_string(),
+            SegmentComponent::TempStore => ".store.temp".to_string(),
-            SegmentComponent::FASTFIELDS => ".fast".to_string(),
+            SegmentComponent::FastFields => ".fast".to_string(),
-            SegmentComponent::FIELDNORMS => ".fieldnorm".to_string(),
+            SegmentComponent::FieldNorms => ".fieldnorm".to_string(),
-            SegmentComponent::DELETE => format!(".{}.del", self.delete_opstamp().unwrap_or(0)),
+            SegmentComponent::Delete => format!(".{}.del", self.delete_opstamp().unwrap_or(0)),
         });
         PathBuf::from(path)
     }
@@ -151,6 +170,22 @@ impl SegmentMeta {
         self.num_deleted_docs() > 0
     }
 
+    /// Updates the max_doc value from the `SegmentMeta`.
+    ///
+    /// This method is only used when updating `max_doc` from 0
+    /// as we finalize a fresh new segment.
+    pub(crate) fn with_max_doc(self, max_doc: u32) -> SegmentMeta {
+        assert_eq!(self.tracked.max_doc, 0);
+        assert!(self.tracked.deletes.is_none());
+        let tracked = self.tracked.map(move |inner_meta| InnerSegmentMeta {
+            segment_id: inner_meta.segment_id,
+            max_doc,
+            deletes: None,
+            include_temp_doc_store: Arc::new(AtomicBool::new(true)),
+        });
+        SegmentMeta { tracked }
+    }
+
     #[doc(hidden)]
     pub fn with_delete_meta(self, num_deleted_docs: u32, opstamp: Opstamp) -> SegmentMeta {
         let delete_meta = DeleteMeta {
@@ -160,6 +195,7 @@ impl SegmentMeta {
         let tracked = self.tracked.map(move |inner_meta| InnerSegmentMeta {
             segment_id: inner_meta.segment_id,
             max_doc: inner_meta.max_doc,
+            include_temp_doc_store: Arc::new(AtomicBool::new(true)),
             deletes: Some(delete_meta),
         });
         SegmentMeta { tracked }
@@ -171,6 +207,14 @@ struct InnerSegmentMeta {
     segment_id: SegmentId,
     max_doc: u32,
     deletes: Option<DeleteMeta>,
+    /// If you want to avoid the SegmentComponent::TempStore file to be covered by
+    /// garbage collection and deleted, set this to true. This is used during merge.
+    #[serde(skip)]
+    #[serde(default = "default_temp_store")]
+    pub(crate) include_temp_doc_store: Arc<AtomicBool>,
+}
+fn default_temp_store() -> Arc<AtomicBool> {
+    Arc::new(AtomicBool::new(false))
 }
 
 impl InnerSegmentMeta {
@@ -181,6 +225,51 @@ impl InnerSegmentMeta {
     }
 }
 
+/// Search Index Settings.
+///
+/// Contains settings which are applied on the whole
+/// index, like presort documents.
+#[derive(Clone, Default, Serialize, Deserialize, Eq, PartialEq)]
+pub struct IndexSettings {
+    /// Sorts the documents by information
+    /// provided in `IndexSortByField`
+    #[serde(skip_serializing_if = "Option::is_none")]
+    pub sort_by_field: Option<IndexSortByField>,
+    /// The `Compressor` used to compress the doc store.
+    #[serde(default)]
+    pub docstore_compression: Compressor,
+}
+/// Settings to presort the documents in an index
+///
+/// Presorting documents can greatly performance
+/// in some scenarios, by applying top n
+/// optimizations.
+#[derive(Clone, Serialize, Deserialize, Eq, PartialEq)]
+pub struct IndexSortByField {
+    /// The field to sort the documents by
+    pub field: String,
+    /// The order to sort the documents by
+    pub order: Order,
+}
+/// The order to sort by
+#[derive(Clone, Serialize, Deserialize, Eq, PartialEq)]
+pub enum Order {
+    /// Ascending Order
+    Asc,
+    /// Descending Order
+    Desc,
+}
+impl Order {
+    /// return if the Order is ascending
+    pub fn is_asc(&self) -> bool {
+        self == &Order::Asc
+    }
+    /// return if the Order is descending
+    pub fn is_desc(&self) -> bool {
+        self == &Order::Desc
+    }
+}
+
 /// Meta information about the `Index`.
 ///
 /// This object is serialized on disk in the `meta.json` file.
@@ -191,6 +280,9 @@ impl InnerSegmentMeta {
 ///
 #[derive(Clone, Serialize)]
 pub struct IndexMeta {
+    /// `IndexSettings` to configure index options.
+    #[serde(default)]
+    pub index_settings: IndexSettings,
     /// List of `SegmentMeta` informations associated to each finalized segment of the index.
     pub segments: Vec<SegmentMeta>,
     /// Index `Schema`
@@ -200,7 +292,7 @@ pub struct IndexMeta {
     #[serde(skip_serializing_if = "Option::is_none")]
     /// Payload associated to the last commit.
     ///
-    /// Upon commit, clients can optionally add a small `Striing` payload to their commit
+    /// Upon commit, clients can optionally add a small `String` payload to their commit
     /// to help identify this commit.
     /// This payload is entirely unused by tantivy.
     pub payload: Option<String>,
@@ -209,6 +301,8 @@ pub struct IndexMeta {
 #[derive(Deserialize)]
 struct UntrackedIndexMeta {
     pub segments: Vec<InnerSegmentMeta>,
+    #[serde(default)]
+    pub index_settings: IndexSettings,
     pub schema: Schema,
     pub opstamp: Opstamp,
     #[serde(skip_serializing_if = "Option::is_none")]
@@ -218,6 +312,7 @@ struct UntrackedIndexMeta {
 impl UntrackedIndexMeta {
     pub fn track(self, inventory: &SegmentMetaInventory) -> IndexMeta {
         IndexMeta {
+            index_settings: self.index_settings,
             segments: self
                 .segments
                 .into_iter()
@@ -238,6 +333,7 @@ impl IndexMeta {
     /// Opstamp will the value `0u64`.
     pub fn with_schema(schema: Schema) -> IndexMeta {
         IndexMeta {
+            index_settings: IndexSettings::default(),
             segments: vec![],
             schema,
             opstamp: 0u64,
@@ -269,7 +365,10 @@ impl fmt::Debug for IndexMeta {
 mod tests {
 
     use super::IndexMeta;
-    use crate::schema::{Schema, TEXT};
+    use crate::{
+        schema::{Schema, TEXT},
+        IndexSettings, IndexSortByField, Order,
+    };
     use serde_json;
 
     #[test]
@@ -280,12 +379,22 @@ mod tests {
             schema_builder.build()
         };
         let index_metas = IndexMeta {
+            index_settings: IndexSettings {
+                sort_by_field: Some(IndexSortByField {
+                    field: "text".to_string(),
+                    order: Order::Asc,
+                }),
+                ..Default::default()
+            },
             segments: Vec::new(),
             schema,
             opstamp: 0u64,
             payload: None,
         };
         let json = serde_json::ser::to_string(&index_metas).expect("serialization failed");
-        assert_eq!(json, r#"{"segments":[],"schema":[{"name":"text","type":"text","options":{"indexing":{"record":"position","tokenizer":"default"},"stored":false}}],"opstamp":0}"#);
+        assert_eq!(
+            json,
+            r#"{"index_settings":{"sort_by_field":{"field":"text","order":"Asc"},"docstore_compression":"lz4"},"segments":[],"schema":[{"name":"text","type":"text","options":{"indexing":{"record":"position","tokenizer":"default"},"stored":false}}],"opstamp":0}"#
+        );
     }
 }

src/core/inverted_index_reader.rs

@@ -1,13 +1,13 @@
+use std::io;
+
 use crate::common::BinarySerializable;
-use crate::directory::ReadOnlySource;
+use crate::directory::FileSlice;
 use crate::positions::PositionReader;
 use crate::postings::TermInfo;
 use crate::postings::{BlockSegmentPostings, SegmentPostings};
-use crate::schema::FieldType;
 use crate::schema::IndexRecordOption;
 use crate::schema::Term;
 use crate::termdict::TermDictionary;
-use owned_read::OwnedRead;
 
 /// The inverted index reader is in charge of accessing
 /// the inverted index associated to a specific field.
@@ -16,7 +16,7 @@ use owned_read::OwnedRead;
 ///
 /// It is safe to delete the segment associated to
 /// an `InvertedIndexReader`. As long as it is open,
-/// the `ReadOnlySource` it is relying on should
+/// the `FileSlice` it is relying on should
 /// stay available.
 ///
 ///
@@ -24,9 +24,8 @@ use owned_read::OwnedRead;
 /// the `SegmentReader`'s [`.inverted_index(...)`] method
 pub struct InvertedIndexReader {
     termdict: TermDictionary,
-    postings_source: ReadOnlySource,
+    postings_file_slice: FileSlice,
-    positions_source: ReadOnlySource,
+    positions_file_slice: FileSlice,
-    positions_idx_source: ReadOnlySource,
     record_option: IndexRecordOption,
     total_num_tokens: u64,
 }
@@ -35,42 +34,35 @@ impl InvertedIndexReader {
     #[cfg_attr(feature = "cargo-clippy", allow(clippy::needless_pass_by_value))] // for symmetry
     pub(crate) fn new(
         termdict: TermDictionary,
-        postings_source: ReadOnlySource,
+        postings_file_slice: FileSlice,
-        positions_source: ReadOnlySource,
+        positions_file_slice: FileSlice,
-        positions_idx_source: ReadOnlySource,
         record_option: IndexRecordOption,
-    ) -> InvertedIndexReader {
+    ) -> io::Result<InvertedIndexReader> {
-        let total_num_tokens_data = postings_source.slice(0, 8);
+        let (total_num_tokens_slice, postings_body) = postings_file_slice.split(8);
-        let mut total_num_tokens_cursor = total_num_tokens_data.as_slice();
+        let total_num_tokens = u64::deserialize(&mut total_num_tokens_slice.read_bytes()?)?;
-        let total_num_tokens = u64::deserialize(&mut total_num_tokens_cursor).unwrap_or(0u64);
+        Ok(InvertedIndexReader {
-        InvertedIndexReader {
             termdict,
-            postings_source: postings_source.slice_from(8),
+            postings_file_slice: postings_body,
-            positions_source,
+            positions_file_slice,
-            positions_idx_source,
             record_option,
             total_num_tokens,
-        }
+        })
     }
 
     /// Creates an empty `InvertedIndexReader` object, which
     /// contains no terms at all.
-    pub fn empty(field_type: &FieldType) -> InvertedIndexReader {
+    pub fn empty(record_option: IndexRecordOption) -> InvertedIndexReader {
-        let record_option = field_type
-            .get_index_record_option()
-            .unwrap_or(IndexRecordOption::Basic);
         InvertedIndexReader {
-            termdict: TermDictionary::empty(&field_type),
+            termdict: TermDictionary::empty(),
-            postings_source: ReadOnlySource::empty(),
+            postings_file_slice: FileSlice::empty(),
-            positions_source: ReadOnlySource::empty(),
+            positions_file_slice: FileSlice::empty(),
-            positions_idx_source: ReadOnlySource::empty(),
             record_option,
             total_num_tokens: 0u64,
         }
     }
 
     /// Returns the term info associated with the term.
-    pub fn get_term_info(&self, term: &Term) -> Option<TermInfo> {
+    pub fn get_term_info(&self, term: &Term) -> io::Result<Option<TermInfo>> {
         self.termdict.get(term.value_bytes())
     }
 
@@ -93,12 +85,12 @@ impl InvertedIndexReader {
         &self,
         term_info: &TermInfo,
         block_postings: &mut BlockSegmentPostings,
-    ) {
+    ) -> io::Result<()> {
-        let offset = term_info.postings_offset as usize;
+        let postings_slice = self
-        let end_source = self.postings_source.len();
+            .postings_file_slice
-        let postings_slice = self.postings_source.slice(offset, end_source);
+            .slice(term_info.postings_range.clone());
-        let postings_reader = OwnedRead::new(postings_slice);
+        block_postings.reset(term_info.doc_freq, postings_slice.read_bytes()?);
-        block_postings.reset(term_info.doc_freq, postings_reader);
+        Ok(())
     }
 
     /// Returns a block postings given a `Term`.
@@ -109,9 +101,10 @@ impl InvertedIndexReader {
         &self,
         term: &Term,
         option: IndexRecordOption,
-    ) -> Option<BlockSegmentPostings> {
+    ) -> io::Result<Option<BlockSegmentPostings>> {
-        self.get_term_info(term)
+        self.get_term_info(term)?
             .map(move |term_info| self.read_block_postings_from_terminfo(&term_info, option))
+            .transpose()
     }
 
     /// Returns a block postings given a `term_info`.
@@ -122,12 +115,13 @@ impl InvertedIndexReader {
         &self,
         term_info: &TermInfo,
         requested_option: IndexRecordOption,
-    ) -> BlockSegmentPostings {
+    ) -> io::Result<BlockSegmentPostings> {
-        let offset = term_info.postings_offset as usize;
+        let postings_data = self
-        let postings_data = self.postings_source.slice_from(offset);
+            .postings_file_slice
-        BlockSegmentPostings::from_data(
+            .slice(term_info.postings_range.clone());
+        BlockSegmentPostings::open(
             term_info.doc_freq,
-            OwnedRead::new(postings_data),
+            postings_data,
             self.record_option,
             requested_option,
         )
@@ -141,20 +135,23 @@ impl InvertedIndexReader {
         &self,
         term_info: &TermInfo,
         option: IndexRecordOption,
-    ) -> SegmentPostings {
+    ) -> io::Result<SegmentPostings> {
-        let block_postings = self.read_block_postings_from_terminfo(term_info, option);
+        let block_postings = self.read_block_postings_from_terminfo(term_info, option)?;
-        let position_stream = {
+        let position_reader = {
             if option.has_positions() {
-                let position_reader = self.positions_source.clone();
+                let positions_data = self
-                let skip_reader = self.positions_idx_source.clone();
+                    .positions_file_slice
-                let position_reader =
+                    .read_bytes_slice(term_info.positions_range.clone())?;
-                    PositionReader::new(position_reader, skip_reader, term_info.positions_idx);
+                let position_reader = PositionReader::open(positions_data)?;
                 Some(position_reader)
             } else {
                 None
             }
         };
-        SegmentPostings::from_block_postings(block_postings, position_stream)
+        Ok(SegmentPostings::from_block_postings(
+            block_postings,
+            position_reader,
+        ))
     }
 
     /// Returns the total number of tokens recorded for all documents
@@ -173,24 +170,31 @@ impl InvertedIndexReader {
     /// For instance, requesting `IndexRecordOption::Freq` for a
     /// `TextIndexingOptions` that does not index position will return a `SegmentPostings`
     /// with `DocId`s and frequencies.
-    pub fn read_postings(&self, term: &Term, option: IndexRecordOption) -> Option<SegmentPostings> {
+    pub fn read_postings(
-        self.get_term_info(term)
+        &self,
+        term: &Term,
+        option: IndexRecordOption,
+    ) -> io::Result<Option<SegmentPostings>> {
+        self.get_term_info(term)?
             .map(move |term_info| self.read_postings_from_terminfo(&term_info, option))
+            .transpose()
     }
 
     pub(crate) fn read_postings_no_deletes(
         &self,
         term: &Term,
         option: IndexRecordOption,
-    ) -> Option<SegmentPostings> {
+    ) -> io::Result<Option<SegmentPostings>> {
-        self.get_term_info(term)
+        self.get_term_info(term)?
             .map(|term_info| self.read_postings_from_terminfo(&term_info, option))
+            .transpose()
     }
 
     /// Returns the number of documents containing the term.
-    pub fn doc_freq(&self, term: &Term) -> u32 {
+    pub fn doc_freq(&self, term: &Term) -> io::Result<u32> {
-        self.get_term_info(term)
+        Ok(self
+            .get_term_info(term)?
             .map(|term_info| term_info.doc_freq)
-            .unwrap_or(0u32)
+            .unwrap_or(0u32))
     }
 }

src/core/mod.rs

@@ -9,12 +9,13 @@ mod segment_id;
 mod segment_reader;
 
 pub use self::executor::Executor;
-pub use self::index::Index;
+pub use self::index::{Index, IndexBuilder};
-pub use self::index_meta::{IndexMeta, SegmentMeta, SegmentMetaInventory};
+pub use self::index_meta::{
+    IndexMeta, IndexSettings, IndexSortByField, Order, SegmentMeta, SegmentMetaInventory,
+};
 pub use self::inverted_index_reader::InvertedIndexReader;
 pub use self::searcher::Searcher;
 pub use self::segment::Segment;
-pub use self::segment::SerializableSegment;
 pub use self::segment_component::SegmentComponent;
 pub use self::segment_id::SegmentId;
 pub use self::segment_reader::SegmentReader;

src/core/searcher.rs

@@ -1,42 +1,17 @@
 use crate::collector::Collector;
-use crate::collector::SegmentCollector;
 use crate::core::Executor;
-use crate::core::InvertedIndexReader;
+
 use crate::core::SegmentReader;
 use crate::query::Query;
-use crate::query::Scorer;
-use crate::query::Weight;
 use crate::schema::Document;
 use crate::schema::Schema;
-use crate::schema::{Field, Term};
+use crate::schema::Term;
 use crate::space_usage::SearcherSpaceUsage;
 use crate::store::StoreReader;
-use crate::termdict::TermMerger;
 use crate::DocAddress;
 use crate::Index;
-use crate::Result;
-use std::fmt;
-use std::sync::Arc;
 
-fn collect_segment<C: Collector>(
+use std::{fmt, io};
-    collector: &C,
-    weight: &dyn Weight,
-    segment_ord: u32,
-    segment_reader: &SegmentReader,
-) -> Result<C::Fruit> {
-    let mut scorer = weight.scorer(segment_reader)?;
-    let mut segment_collector = collector.for_segment(segment_ord as u32, segment_reader)?;
-    if let Some(delete_bitset) = segment_reader.delete_bitset() {
-        scorer.for_each(&mut |doc, score| {
-            if delete_bitset.is_alive(doc) {
-                segment_collector.collect(doc, score);
-            }
-        });
-    } else {
-        scorer.for_each(&mut |doc, score| segment_collector.collect(doc, score));
-    }
-    Ok(segment_collector.harvest())
-}
 
 /// Holds a list of `SegmentReader`s ready for search.
 ///
@@ -56,17 +31,17 @@ impl Searcher {
         schema: Schema,
         index: Index,
         segment_readers: Vec<SegmentReader>,
-    ) -> Searcher {
+    ) -> io::Result<Searcher> {
-        let store_readers = segment_readers
+        let store_readers: Vec<StoreReader> = segment_readers
             .iter()
             .map(SegmentReader::get_store_reader)
-            .collect();
+            .collect::<io::Result<Vec<_>>>()?;
-        Searcher {
+        Ok(Searcher {
             schema,
             index,
             segment_readers,
             store_readers,
-        }
+        })
     }
 
     /// Returns the `Index` associated to the `Searcher`
@@ -78,10 +53,9 @@ impl Searcher {
     ///
     /// The searcher uses the segment ordinal to route the
     /// the request to the right `Segment`.
-    pub fn doc(&self, doc_address: DocAddress) -> Result<Document> {
+    pub fn doc(&self, doc_address: DocAddress) -> crate::Result<Document> {
-        let DocAddress(segment_local_id, doc_id) = doc_address;
+        let store_reader = &self.store_readers[doc_address.segment_ord as usize];
-        let store_reader = &self.store_readers[segment_local_id as usize];
+        store_reader.get(doc_address.doc_id)
-        store_reader.get(doc_id)
     }
 
     /// Access the schema associated to the index of this searcher.
@@ -99,13 +73,14 @@ impl Searcher {
 
     /// Return the overall number of documents containing
     /// the given term.
-    pub fn doc_freq(&self, term: &Term) -> u64 {
+    pub fn doc_freq(&self, term: &Term) -> crate::Result<u64> {
-        self.segment_readers
+        let mut total_doc_freq = 0;
-            .iter()
+        for segment_reader in &self.segment_readers {
-            .map(|segment_reader| {
+            let inverted_index = segment_reader.inverted_index(term.field())?;
-                u64::from(segment_reader.inverted_index(term.field()).doc_freq(term))
+            let doc_freq = inverted_index.doc_freq(term)?;
-            })
+            total_doc_freq += u64::from(doc_freq);
-            .sum::<u64>()
+        }
+        Ok(total_doc_freq)
     }
 
     /// Return the list of segment readers
@@ -132,7 +107,11 @@ impl Searcher {
     ///
     ///  Finally, the Collector merges each of the child collectors into itself for result usability
     ///  by the caller.
-    pub fn search<C: Collector>(&self, query: &dyn Query, collector: &C) -> Result<C::Fruit> {
+    pub fn search<C: Collector>(
+        &self,
+        query: &dyn Query,
+        collector: &C,
+    ) -> crate::Result<C::Fruit> {
         let executor = self.index.search_executor();
         self.search_with_executor(query, collector, executor)
     }
@@ -154,62 +133,26 @@ impl Searcher {
         query: &dyn Query,
         collector: &C,
         executor: &Executor,
-    ) -> Result<C::Fruit> {
+    ) -> crate::Result<C::Fruit> {
         let scoring_enabled = collector.requires_scoring();
         let weight = query.weight(self, scoring_enabled)?;
         let segment_readers = self.segment_readers();
         let fruits = executor.map(
             |(segment_ord, segment_reader)| {
-                collect_segment(
+                collector.collect_segment(weight.as_ref(), segment_ord as u32, segment_reader)
-                    collector,
-                    weight.as_ref(),
-                    segment_ord as u32,
-                    segment_reader,
-                )
             },
             segment_readers.iter().enumerate(),
         )?;
         collector.merge_fruits(fruits)
     }
 
-    /// Return the field searcher associated to a `Field`.
-    pub fn field(&self, field: Field) -> FieldSearcher {
-        let inv_index_readers = self
-            .segment_readers
-            .iter()
-            .map(|segment_reader| segment_reader.inverted_index(field))
-            .collect::<Vec<_>>();
-        FieldSearcher::new(inv_index_readers)
-    }
-
     /// Summarize total space usage of this searcher.
-    pub fn space_usage(&self) -> SearcherSpaceUsage {
+    pub fn space_usage(&self) -> io::Result<SearcherSpaceUsage> {
         let mut space_usage = SearcherSpaceUsage::new();
-        for segment_reader in self.segment_readers.iter() {
+        for segment_reader in &self.segment_readers {
-            space_usage.add_segment(segment_reader.space_usage());
+            space_usage.add_segment(segment_reader.space_usage()?);
         }
-        space_usage
+        Ok(space_usage)
-    }
-}
-
-pub struct FieldSearcher {
-    inv_index_readers: Vec<Arc<InvertedIndexReader>>,
-}
-
-impl FieldSearcher {
-    fn new(inv_index_readers: Vec<Arc<InvertedIndexReader>>) -> FieldSearcher {
-        FieldSearcher { inv_index_readers }
-    }
-
-    /// Returns a Stream over all of the sorted unique terms of
-    /// for the given field.
-    pub fn terms(&self) -> TermMerger<'_> {
-        let term_streamers: Vec<_> = self
-            .inv_index_readers
-            .iter()
-            .map(|inverted_index| inverted_index.terms().stream())
-            .collect();
-        TermMerger::new(term_streamers)
     }
 }
 

src/core/segment.rs

@@ -4,14 +4,11 @@ use crate::core::SegmentId;
 use crate::core::SegmentMeta;
 use crate::directory::error::{OpenReadError, OpenWriteError};
 use crate::directory::Directory;
-use crate::directory::{ReadOnlySource, WritePtr};
+use crate::directory::{FileSlice, WritePtr};
-use crate::indexer::segment_serializer::SegmentSerializer;
 use crate::schema::Schema;
 use crate::Opstamp;
-use crate::Result;
 use std::fmt;
 use std::path::PathBuf;
-use std::result;
 
 /// A segment is a piece of the index.
 #[derive(Clone)]
@@ -26,15 +23,12 @@ impl fmt::Debug for Segment {
     }
 }
 
-/// Creates a new segment given an `Index` and a `SegmentId`
-///
-/// The function is here to make it private outside `tantivy`.
-/// #[doc(hidden)]
-pub fn create_segment(index: Index, meta: SegmentMeta) -> Segment {
-    Segment { index, meta }
-}
-
 impl Segment {
+    /// Creates a new segment given an `Index` and a `SegmentId`
+    pub(crate) fn for_index(index: Index, meta: SegmentMeta) -> Segment {
+        Segment { index, meta }
+    }
+
     /// Returns the index the segment belongs to.
     pub fn index(&self) -> &Index {
         &self.index
@@ -50,6 +44,17 @@ impl Segment {
         &self.meta
     }
 
+    /// Updates the max_doc value from the `SegmentMeta`.
+    ///
+    /// This method is only used when updating `max_doc` from 0
+    /// as we finalize a fresh new segment.
+    pub(crate) fn with_max_doc(self, max_doc: u32) -> Segment {
+        Segment {
+            index: self.index,
+            meta: self.meta.with_max_doc(max_doc),
+        }
+    }
+
     #[doc(hidden)]
     pub fn with_delete_meta(self, num_deleted_docs: u32, opstamp: Opstamp) -> Segment {
         Segment {
@@ -72,31 +77,15 @@ impl Segment {
     }
 
     /// Open one of the component file for a *regular* read.
-    pub fn open_read(
+    pub fn open_read(&self, component: SegmentComponent) -> Result<FileSlice, OpenReadError> {
-        &self,
-        component: SegmentComponent,
-    ) -> result::Result<ReadOnlySource, OpenReadError> {
         let path = self.relative_path(component);
-        let source = self.index.directory().open_read(&path)?;
+        self.index.directory().open_read(&path)
-        Ok(source)
     }
 
     /// Open one of the component file for *regular* write.
-    pub fn open_write(
+    pub fn open_write(&mut self, component: SegmentComponent) -> Result<WritePtr, OpenWriteError> {
-        &mut self,
-        component: SegmentComponent,
-    ) -> result::Result<WritePtr, OpenWriteError> {
         let path = self.relative_path(component);
         let write = self.index.directory_mut().open_write(&path)?;
         Ok(write)
     }
 }
-
-pub trait SerializableSegment {
-    /// Writes a view of a segment by pushing information
-    /// to the `SegmentSerializer`.
-    ///
-    /// # Returns
-    /// The number of documents in the segment.
-    fn write(&self, serializer: SegmentSerializer) -> Result<u32>;
-}

src/core/segment_component.rs

@@ -4,42 +4,42 @@ use std::slice;
 /// Each component is stored in its own file,
 /// using the pattern `segment_uuid`.`component_extension`,
 /// except the delete component that takes an `segment_uuid`.`delete_opstamp`.`component_extension`
-#[derive(Copy, Clone)]
+#[derive(Copy, Clone, Eq, PartialEq)]
 pub enum SegmentComponent {
     /// Postings (or inverted list). Sorted lists of document ids, associated to terms
-    POSTINGS,
+    Postings,
     /// Positions of terms in each document.
-    POSITIONS,
+    Positions,
-    /// Index to seek within the position file
-    POSITIONSSKIP,
     /// Column-oriented random-access storage of fields.
-    FASTFIELDS,
+    FastFields,
     /// Stores the sum  of the length (in terms) of each field for each document.
     /// Field norms are stored as a special u64 fast field.
-    FIELDNORMS,
+    FieldNorms,
     /// Dictionary associating `Term`s to `TermInfo`s which is
     /// simply an address into the `postings` file and the `positions` file.
-    TERMS,
+    Terms,
-    /// Row-oriented, LZ4-compressed storage of the documents.
+    /// Row-oriented, compressed storage of the documents.
     /// Accessing a document from the store is relatively slow, as it
     /// requires to decompress the entire block it belongs to.
-    STORE,
+    Store,
+    /// Temporary storage of the documents, before streamed to `Store`.
+    TempStore,
     /// Bitset describing which document of the segment is deleted.
-    DELETE,
+    Delete,
 }
 
 impl SegmentComponent {
     /// Iterates through the components.
     pub fn iterator() -> slice::Iter<'static, SegmentComponent> {
         static SEGMENT_COMPONENTS: [SegmentComponent; 8] = [
-            SegmentComponent::POSTINGS,
+            SegmentComponent::Postings,
-            SegmentComponent::POSITIONS,
+            SegmentComponent::Positions,
-            SegmentComponent::POSITIONSSKIP,
+            SegmentComponent::FastFields,
-            SegmentComponent::FASTFIELDS,
+            SegmentComponent::FieldNorms,
-            SegmentComponent::FIELDNORMS,
+            SegmentComponent::Terms,
-            SegmentComponent::TERMS,
+            SegmentComponent::Store,
-            SegmentComponent::STORE,
+            SegmentComponent::TempStore,
-            SegmentComponent::DELETE,
+            SegmentComponent::Delete,
         ];
         SEGMENT_COMPONENTS.iter()
     }

src/core/segment_id.rs

@@ -4,6 +4,9 @@ use uuid::Uuid;
 
 #[cfg(test)]
 use once_cell::sync::Lazy;
+use serde::{Deserialize, Serialize};
+use std::error::Error;
+use std::str::FromStr;
 #[cfg(test)]
 use std::sync::atomic;
 
@@ -52,15 +55,51 @@ impl SegmentId {
     /// and the rest is random.
     ///
     /// Picking the first 8 chars is ok to identify
-    /// segments in a display message.
+    /// segments in a display message (e.g. a5c4dfcb).
     pub fn short_uuid_string(&self) -> String {
         (&self.0.to_simple_ref().to_string()[..8]).to_string()
     }
 
     /// Returns a segment uuid string.
+    ///
+    /// It consists in 32 lowercase hexadecimal chars
+    /// (e.g. a5c4dfcbdfe645089129e308e26d5523)
     pub fn uuid_string(&self) -> String {
         self.0.to_simple_ref().to_string()
     }
+
+    /// Build a `SegmentId` string from the full uuid string.
+    ///
+    /// E.g. "a5c4dfcbdfe645089129e308e26d5523"
+    pub fn from_uuid_string(uuid_string: &str) -> Result<SegmentId, SegmentIdParseError> {
+        FromStr::from_str(uuid_string)
+    }
+}
+
+/// Error type used when parsing a `SegmentId` from a string fails.
+pub struct SegmentIdParseError(uuid::Error);
+
+impl Error for SegmentIdParseError {}
+
+impl fmt::Debug for SegmentIdParseError {
+    fn fmt(&self, f: &mut fmt::Formatter) -> fmt::Result {
+        self.0.fmt(f)
+    }
+}
+
+impl fmt::Display for SegmentIdParseError {
+    fn fmt(&self, f: &mut fmt::Formatter) -> fmt::Result {
+        self.0.fmt(f)
+    }
+}
+
+impl FromStr for SegmentId {
+    type Err = SegmentIdParseError;
+
+    fn from_str(uuid_string: &str) -> Result<Self, SegmentIdParseError> {
+        let uuid = Uuid::parse_str(uuid_string).map_err(SegmentIdParseError)?;
+        Ok(SegmentId(uuid))
+    }
 }
 
 impl fmt::Debug for SegmentId {
@@ -80,3 +119,18 @@ impl Ord for SegmentId {
         self.0.as_bytes().cmp(other.0.as_bytes())
     }
 }
+
+#[cfg(test)]
+mod tests {
+    use super::SegmentId;
+
+    #[test]
+    fn test_to_uuid_string() {
+        let full_uuid = "a5c4dfcbdfe645089129e308e26d5523";
+        let segment_id = SegmentId::from_uuid_string(full_uuid).unwrap();
+        assert_eq!(segment_id.uuid_string(), full_uuid);
+        assert_eq!(segment_id.short_uuid_string(), "a5c4dfcb");
+        // one extra char
+        assert!(SegmentId::from_uuid_string("a5c4dfcbdfe645089129e308e26d5523b").is_err());
+    }
+}

src/core/segment_reader.rs

@@ -1,27 +1,26 @@
-use crate::common::CompositeFile;
 use crate::common::HasLen;
 use crate::core::InvertedIndexReader;
 use crate::core::Segment;
 use crate::core::SegmentComponent;
 use crate::core::SegmentId;
-use crate::directory::ReadOnlySource;
+use crate::directory::FileSlice;
 use crate::fastfield::DeleteBitSet;
 use crate::fastfield::FacetReader;
 use crate::fastfield::FastFieldReaders;
-use crate::fieldnorm::FieldNormReader;
+use crate::fieldnorm::{FieldNormReader, FieldNormReaders};
-use crate::schema::Field;
 use crate::schema::FieldType;
 use crate::schema::Schema;
+use crate::schema::{Field, IndexRecordOption};
 use crate::space_usage::SegmentSpaceUsage;
 use crate::store::StoreReader;
 use crate::termdict::TermDictionary;
 use crate::DocId;
-use crate::Result;
+use crate::{common::CompositeFile, error::DataCorruption};
 use fail::fail_point;
-use std::collections::HashMap;
 use std::fmt;
 use std::sync::Arc;
 use std::sync::RwLock;
+use std::{collections::HashMap, io};
 
 /// Entry point to access all of the datastructures of the `Segment`
 ///
@@ -47,11 +46,10 @@ pub struct SegmentReader {
     termdict_composite: CompositeFile,
     postings_composite: CompositeFile,
     positions_composite: CompositeFile,
-    positions_idx_composite: CompositeFile,
     fast_fields_readers: Arc<FastFieldReaders>,
-    fieldnorms_composite: CompositeFile,
+    fieldnorm_readers: FieldNormReaders,
 
-    store_source: ReadOnlySource,
+    store_file: FileSlice,
     delete_bitset_opt: Option<DeleteBitSet>,
     schema: Schema,
 }
@@ -107,16 +105,24 @@ impl SegmentReader {
     }
 
     /// Accessor to the `FacetReader` associated to a given `Field`.
-    pub fn facet_reader(&self, field: Field) -> Option<FacetReader> {
+    pub fn facet_reader(&self, field: Field) -> crate::Result<FacetReader> {
         let field_entry = self.schema.get_field_entry(field);
-        if field_entry.field_type() != &FieldType::HierarchicalFacet {
+
-            return None;
+        match field_entry.field_type() {
+            FieldType::HierarchicalFacet(_) => {
+                let term_ords_reader = self.fast_fields().u64s(field)?;
+                let termdict = self
+                    .termdict_composite
+                    .open_read(field)
+                    .map(TermDictionary::open)
+                    .unwrap_or_else(|| Ok(TermDictionary::empty()))?;
+                Ok(FacetReader::new(term_ords_reader, termdict))
+            }
+            _ => Err(crate::TantivyError::InvalidArgument(format!(
+                "Field {:?} is not a facet field.",
+                field_entry.name()
+            ))),
         }
-        let term_ords_reader = self.fast_fields().u64s(field)?;
-        let termdict_source = self.termdict_composite.open_read(field)?;
-        let termdict = TermDictionary::from_source(&termdict_source);
-        let facet_reader = FacetReader::new(term_ords_reader, termdict);
-        Some(facet_reader)
     }
 
     /// Accessor to the segment's `Field norms`'s reader.
@@ -126,47 +132,37 @@ impl SegmentReader {
     ///
     /// They are simply stored as a fast field, serialized in
     /// the `.fieldnorm` file of the segment.
-    pub fn get_fieldnorms_reader(&self, field: Field) -> FieldNormReader {
+    pub fn get_fieldnorms_reader(&self, field: Field) -> crate::Result<FieldNormReader> {
-        if let Some(fieldnorm_source) = self.fieldnorms_composite.open_read(field) {
+        self.fieldnorm_readers.get_field(field)?.ok_or_else(|| {
-            FieldNormReader::open(fieldnorm_source)
-        } else {
             let field_name = self.schema.get_field_name(field);
             let err_msg = format!(
-                "Field norm not found for field {:?}. Was it market as indexed during indexing.",
+                "Field norm not found for field {:?}. Was it marked as indexed during indexing?",
                 field_name
             );
-            panic!(err_msg);
+            crate::TantivyError::SchemaError(err_msg)
-        }
+        })
     }
 
     /// Accessor to the segment's `StoreReader`.
-    pub fn get_store_reader(&self) -> StoreReader {
+    pub fn get_store_reader(&self) -> io::Result<StoreReader> {
-        StoreReader::from_source(self.store_source.clone())
+        StoreReader::open(self.store_file.clone())
     }
 
     /// Open a new segment for reading.
-    pub fn open(segment: &Segment) -> Result<SegmentReader> {
+    pub fn open(segment: &Segment) -> crate::Result<SegmentReader> {
-        let termdict_source = segment.open_read(SegmentComponent::TERMS)?;
+        let termdict_file = segment.open_read(SegmentComponent::Terms)?;
-        let termdict_composite = CompositeFile::open(&termdict_source)?;
+        let termdict_composite = CompositeFile::open(&termdict_file)?;
 
-        let store_source = segment.open_read(SegmentComponent::STORE)?;
+        let store_file = segment.open_read(SegmentComponent::Store)?;
 
         fail_point!("SegmentReader::open#middle");
 
-        let postings_source = segment.open_read(SegmentComponent::POSTINGS)?;
+        let postings_file = segment.open_read(SegmentComponent::Postings)?;
-        let postings_composite = CompositeFile::open(&postings_source)?;
+        let postings_composite = CompositeFile::open(&postings_file)?;
 
         let positions_composite = {
-            if let Ok(source) = segment.open_read(SegmentComponent::POSITIONS) {
+            if let Ok(positions_file) = segment.open_read(SegmentComponent::Positions) {
-                CompositeFile::open(&source)?
+                CompositeFile::open(&positions_file)?
-            } else {
-                CompositeFile::empty()
-            }
-        };
-
-        let positions_idx_composite = {
-            if let Ok(source) = segment.open_read(SegmentComponent::POSITIONSSKIP) {
-                CompositeFile::open(&source)?
             } else {
                 CompositeFile::empty()
             }
@@ -174,34 +170,34 @@ impl SegmentReader {
 
         let schema = segment.schema();
 
-        let fast_fields_data = segment.open_read(SegmentComponent::FASTFIELDS)?;
+        let fast_fields_data = segment.open_read(SegmentComponent::FastFields)?;
         let fast_fields_composite = CompositeFile::open(&fast_fields_data)?;
         let fast_field_readers =
-            Arc::new(FastFieldReaders::load_all(&schema, &fast_fields_composite)?);
+            Arc::new(FastFieldReaders::new(schema.clone(), fast_fields_composite));
 
-        let fieldnorms_data = segment.open_read(SegmentComponent::FIELDNORMS)?;
+        let fieldnorm_data = segment.open_read(SegmentComponent::FieldNorms)?;
-        let fieldnorms_composite = CompositeFile::open(&fieldnorms_data)?;
+        let fieldnorm_readers = FieldNormReaders::open(fieldnorm_data)?;
 
         let delete_bitset_opt = if segment.meta().has_deletes() {
-            let delete_data = segment.open_read(SegmentComponent::DELETE)?;
+            let delete_data = segment.open_read(SegmentComponent::Delete)?;
-            Some(DeleteBitSet::open(delete_data))
+            let delete_bitset = DeleteBitSet::open(delete_data)?;
+            Some(delete_bitset)
         } else {
             None
         };
 
         Ok(SegmentReader {
-            inv_idx_reader_cache: Arc::new(RwLock::new(HashMap::new())),
+            inv_idx_reader_cache: Default::default(),
             max_doc: segment.meta().max_doc(),
             num_docs: segment.meta().num_docs(),
             termdict_composite,
             postings_composite,
             fast_fields_readers: fast_field_readers,
-            fieldnorms_composite,
+            fieldnorm_readers,
             segment_id: segment.id(),
-            store_source,
+            store_file,
             delete_bitset_opt,
             positions_composite,
-            positions_idx_composite,
             schema,
         })
     }
@@ -213,58 +209,61 @@ impl SegmentReader {
     /// The field reader is in charge of iterating through the
     /// term dictionary associated to a specific field,
     /// and opening the posting list associated to any term.
-    pub fn inverted_index(&self, field: Field) -> Arc<InvertedIndexReader> {
+    ///
+    /// If the field is marked as index, a warn is logged and an empty `InvertedIndexReader`
+    /// is returned.
+    /// Similarly if the field is marked as indexed but no term has been indexed for the given
+    /// index. an empty `InvertedIndexReader` is returned (but no warning is logged).
+    pub fn inverted_index(&self, field: Field) -> crate::Result<Arc<InvertedIndexReader>> {
         if let Some(inv_idx_reader) = self
             .inv_idx_reader_cache
             .read()
             .expect("Lock poisoned. This should never happen")
             .get(&field)
         {
-            return Arc::clone(inv_idx_reader);
+            return Ok(Arc::clone(inv_idx_reader));
         }
         let field_entry = self.schema.get_field_entry(field);
         let field_type = field_entry.field_type();
         let record_option_opt = field_type.get_index_record_option();
 
         if record_option_opt.is_none() {
-            panic!("Field {:?} does not seem indexed.", field_entry.name());
+            warn!("Field {:?} does not seem indexed.", field_entry.name());
         }
 
-        let record_option = record_option_opt.unwrap();
+        let postings_file_opt = self.postings_composite.open_read(field);
 
-        let postings_source_opt = self.postings_composite.open_read(field);
+        if postings_file_opt.is_none() || record_option_opt.is_none() {
-
-        if postings_source_opt.is_none() {
             // no documents in the segment contained this field.
             // As a result, no data is associated to the inverted index.
             //
             // Returns an empty inverted index.
-            return Arc::new(InvertedIndexReader::empty(field_type));
+            let record_option = record_option_opt.unwrap_or(IndexRecordOption::Basic);
+            return Ok(Arc::new(InvertedIndexReader::empty(record_option)));
         }
 
-        let postings_source = postings_source_opt.unwrap();
+        let record_option = record_option_opt.unwrap();
+        let postings_file = postings_file_opt.unwrap();
 
-        let termdict_source = self.termdict_composite.open_read(field).expect(
+        let termdict_file: FileSlice = self.termdict_composite.open_read(field)
-            "Failed to open field term dictionary in composite file. Is the field indexed?",
+            .ok_or_else(||
-        );
+               DataCorruption::comment_only(format!("Failed to open field {:?}'s term dictionary in the composite file. Has the schema been modified?", field_entry.name()))
+            )?;
 
-        let positions_source = self
+        let positions_file = self
             .positions_composite
             .open_read(field)
-            .expect("Index corrupted. Failed to open field positions in composite file.");
+            .ok_or_else(|| {
-
+                let error_msg = format!("Failed to open field {:?}'s positions in the composite file. Has the schema been modified?", field_entry.name());
-        let positions_idx_source = self
+               DataCorruption::comment_only(error_msg)
-            .positions_idx_composite
+            })?;
-            .open_read(field)
-            .expect("Index corrupted. Failed to open field positions in composite file.");
 
         let inv_idx_reader = Arc::new(InvertedIndexReader::new(
-            TermDictionary::from_source(&termdict_source),
+            TermDictionary::open(termdict_file)?,
-            postings_source,
+            postings_file,
-            positions_source,
+            positions_file,
-            positions_idx_source,
             record_option,
-        ));
+        )?);
 
         // by releasing the lock in between, we may end up opening the inverting index
         // twice, but this is fine.
@@ -273,7 +272,7 @@ impl SegmentReader {
             .expect("Field reader cache lock poisoned. This should never happen.")
             .insert(field, Arc::clone(&inv_idx_reader));
 
-        inv_idx_reader
+        Ok(inv_idx_reader)
     }
 
     /// Returns the segment id
@@ -296,26 +295,25 @@ impl SegmentReader {
     }
 
     /// Returns an iterator that will iterate over the alive document ids
-    pub fn doc_ids_alive(&self) -> SegmentReaderAliveDocsIterator<'_> {
+    pub fn doc_ids_alive(&self) -> impl Iterator<Item = DocId> + '_ {
-        SegmentReaderAliveDocsIterator::new(&self)
+        (0u32..self.max_doc).filter(move |doc| !self.is_deleted(*doc))
     }
 
     /// Summarize total space usage of this segment.
-    pub fn space_usage(&self) -> SegmentSpaceUsage {
+    pub fn space_usage(&self) -> io::Result<SegmentSpaceUsage> {
-        SegmentSpaceUsage::new(
+        Ok(SegmentSpaceUsage::new(
             self.num_docs(),
             self.termdict_composite.space_usage(),
             self.postings_composite.space_usage(),
             self.positions_composite.space_usage(),
-            self.positions_idx_composite.space_usage(),
             self.fast_fields_readers.space_usage(),
-            self.fieldnorms_composite.space_usage(),
+            self.fieldnorm_readers.space_usage(),
-            self.get_store_reader().space_usage(),
+            self.get_store_reader()?.space_usage(),
             self.delete_bitset_opt
                 .as_ref()
                 .map(DeleteBitSet::space_usage)
                 .unwrap_or(0),
-        )
+        ))
     }
 }
 
@@ -325,52 +323,6 @@ impl fmt::Debug for SegmentReader {
     }
 }
 
-/// Implements the iterator trait to allow easy iteration
-/// over non-deleted ("alive") DocIds in a SegmentReader
-pub struct SegmentReaderAliveDocsIterator<'a> {
-    reader: &'a SegmentReader,
-    max_doc: DocId,
-    current: DocId,
-}
-
-impl<'a> SegmentReaderAliveDocsIterator<'a> {
-    pub fn new(reader: &'a SegmentReader) -> SegmentReaderAliveDocsIterator<'a> {
-        SegmentReaderAliveDocsIterator {
-            reader,
-            max_doc: reader.max_doc(),
-            current: 0,
-        }
-    }
-}
-
-impl<'a> Iterator for SegmentReaderAliveDocsIterator<'a> {
-    type Item = DocId;
-
-    fn next(&mut self) -> Option<Self::Item> {
-        // TODO: Use TinySet (like in BitSetDocSet) to speed this process up
-        if self.current >= self.max_doc {
-            return None;
-        }
-
-        // find the next alive doc id
-        while self.reader.is_deleted(self.current) {
-            self.current += 1;
-
-            if self.current >= self.max_doc {
-                return None;
-            }
-        }
-
-        // capture the current alive DocId
-        let result = Some(self.current);
-
-        // move down the chain
-        self.current += 1;
-
-        result
-    }
-}
-
 #[cfg(test)]
 mod test {
     use crate::core::Index;
@@ -378,7 +330,7 @@ mod test {
     use crate::DocId;
 
     #[test]
-    fn test_alive_docs_iterator() {
+    fn test_alive_docs_iterator() -> crate::Result<()> {
         let mut schema_builder = Schema::builder();
         schema_builder.add_text_field("name", TEXT | STORED);
         let schema = schema_builder.build();
@@ -386,26 +338,26 @@ mod test {
         let name = schema.get_field("name").unwrap();
 
         {
-            let mut index_writer = index.writer_with_num_threads(1, 3_000_000).unwrap();
+            let mut index_writer = index.writer_for_tests()?;
             index_writer.add_document(doc!(name => "tantivy"));
             index_writer.add_document(doc!(name => "horse"));
             index_writer.add_document(doc!(name => "jockey"));
             index_writer.add_document(doc!(name => "cap"));
-
             // we should now have one segment with two docs
-            index_writer.commit().unwrap();
+            index_writer.commit()?;
         }
 
         {
-            let mut index_writer2 = index.writer(50_000_000).unwrap();
+            let mut index_writer2 = index.writer(50_000_000)?;
             index_writer2.delete_term(Term::from_field_text(name, "horse"));
             index_writer2.delete_term(Term::from_field_text(name, "cap"));
 
             // ok, now we should have a deleted doc
-            index_writer2.commit().unwrap();
+            index_writer2.commit()?;
         }
-        let searcher = index.reader().unwrap().searcher();
+        let searcher = index.reader()?.searcher();
         let docs: Vec<DocId> = searcher.segment_reader(0).doc_ids_alive().collect();
         assert_eq!(vec![0u32, 2u32], docs);
+        Ok(())
     }
 }

src/directory/directory.rs

@@ -1,9 +1,9 @@
 use crate::directory::directory_lock::Lock;
 use crate::directory::error::LockError;
 use crate::directory::error::{DeleteError, OpenReadError, OpenWriteError};
-use crate::directory::WatchCallback;
 use crate::directory::WatchHandle;
-use crate::directory::{ReadOnlySource, WritePtr};
+use crate::directory::{FileHandle, WatchCallback};
+use crate::directory::{FileSlice, WritePtr};
 use std::fmt;
 use std::io;
 use std::io::Write;
@@ -11,7 +11,6 @@ use std::marker::Send;
 use std::marker::Sync;
 use std::path::Path;
 use std::path::PathBuf;
-use std::result;
 use std::thread;
 use std::time::Duration;
 
@@ -48,14 +47,14 @@ impl RetryPolicy {
 ///
 /// It is transparently associated to a lock file, that gets deleted
 /// on `Drop.` The lock is released automatically on `Drop`.
-pub struct DirectoryLock(Box<dyn Drop + Send + Sync + 'static>);
+pub struct DirectoryLock(Box<dyn Send + Sync + 'static>);
 
 struct DirectoryLockGuard {
     directory: Box<dyn Directory>,
     path: PathBuf,
 }
 
-impl<T: Drop + Send + Sync + 'static> From<Box<T>> for DirectoryLock {
+impl<T: Send + Sync + 'static> From<Box<T>> for DirectoryLock {
     fn from(underlying: Box<T>) -> Self {
         DirectoryLock(underlying)
     }
@@ -71,7 +70,7 @@ impl Drop for DirectoryLockGuard {
 
 enum TryAcquireLockError {
     FileExists,
-    IOError(io::Error),
+    IoError(io::Error),
 }
 
 fn try_acquire_lock(
@@ -80,9 +79,9 @@ fn try_acquire_lock(
 ) -> Result<DirectoryLock, TryAcquireLockError> {
     let mut write = directory.open_write(filepath).map_err(|e| match e {
         OpenWriteError::FileAlreadyExists(_) => TryAcquireLockError::FileExists,
-        OpenWriteError::IOError(io_error) => TryAcquireLockError::IOError(io_error.into()),
+        OpenWriteError::IoError { io_error, .. } => TryAcquireLockError::IoError(io_error),
     })?;
-    write.flush().map_err(TryAcquireLockError::IOError)?;
+    write.flush().map_err(TryAcquireLockError::IoError)?;
     Ok(DirectoryLock::from(Box::new(DirectoryLockGuard {
         directory: directory.box_clone(),
         path: filepath.to_owned(),
@@ -107,37 +106,45 @@ fn retry_policy(is_blocking: bool) -> RetryPolicy {
 ///
 /// - The [`MMapDirectory`](struct.MmapDirectory.html), this
 /// should be your default choice.
-/// - The [`RAMDirectory`](struct.RAMDirectory.html), which
+/// - The [`RamDirectory`](struct.RamDirectory.html), which
 /// should be used mostly for tests.
-///
 pub trait Directory: DirectoryClone + fmt::Debug + Send + Sync + 'static {
-    /// Opens a virtual file for read.
+    /// Opens a file and returns a boxed `FileHandle`.
     ///
+    /// Users of `Directory` should typically call `Directory::open_read(...)`,
+    /// while `Directory` implementor should implement `get_file_handle()`.
+    fn get_file_handle(&self, path: &Path) -> Result<Box<dyn FileHandle>, OpenReadError>;
+
     /// Once a virtual file is open, its data may not
     /// change.
     ///
     /// Specifically, subsequent writes or flushes should
-    /// have no effect on the returned `ReadOnlySource` object.
+    /// have no effect on the returned `FileSlice` object.
-    fn open_read(&self, path: &Path) -> result::Result<ReadOnlySource, OpenReadError>;
+    ///
+    /// You should only use this to read files create with [Directory::open_write].
+    fn open_read(&self, path: &Path) -> Result<FileSlice, OpenReadError> {
+        let file_handle = self.get_file_handle(path)?;
+        Ok(FileSlice::new(file_handle))
+    }
 
     /// Removes a file
     ///
     /// Removing a file will not affect an eventual
-    /// existing ReadOnlySource pointing to it.
+    /// existing FileSlice pointing to it.
     ///
     /// Removing a nonexistent file, yields a
     /// `DeleteError::DoesNotExist`.
-    fn delete(&self, path: &Path) -> result::Result<(), DeleteError>;
+    fn delete(&self, path: &Path) -> Result<(), DeleteError>;
 
     /// Returns true iff the file exists
-    fn exists(&self, path: &Path) -> bool;
+    fn exists(&self, path: &Path) -> Result<bool, OpenReadError>;
 
     /// Opens a writer for the *virtual file* associated with
     /// a Path.
     ///
     /// Right after this call, the file should be created
     /// and any subsequent call to `open_read` for the
-    /// same path should return a `ReadOnlySource`.
+    /// same path should return a `FileSlice`.
     ///
     /// Write operations may be aggressively buffered.
     /// The client of this trait is responsible for calling flush
@@ -147,16 +154,18 @@ pub trait Directory: DirectoryClone + fmt::Debug + Send + Sync + 'static {
     /// Flush operation should also be persistent.
     ///
     /// The user shall not rely on `Drop` triggering `flush`.
-    /// Note that `RAMDirectory` will panic! if `flush`
+    /// Note that `RamDirectory` will panic! if `flush`
     /// was not called.
     ///
     /// The file may not previously exist.
-    fn open_write(&mut self, path: &Path) -> Result<WritePtr, OpenWriteError>;
+    fn open_write(&self, path: &Path) -> Result<WritePtr, OpenWriteError>;
 
     /// Reads the full content file that has been written using
     /// atomic_write.
     ///
     /// This should only be used for small files.
+    ///
+    /// You should only use this to read files create with [Directory::atomic_write].
     fn atomic_read(&self, path: &Path) -> Result<Vec<u8>, OpenReadError>;
 
     /// Atomically replace the content of a file with data.
@@ -165,7 +174,7 @@ pub trait Directory: DirectoryClone + fmt::Debug + Send + Sync + 'static {
     /// a partially written file.
     ///
     /// The file may or may not previously exist.
-    fn atomic_write(&mut self, path: &Path, data: &[u8]) -> io::Result<()>;
+    fn atomic_write(&self, path: &Path, data: &[u8]) -> io::Result<()>;
 
     /// Acquire a lock in the given directory.
     ///
@@ -183,8 +192,8 @@ pub trait Directory: DirectoryClone + fmt::Debug + Send + Sync + 'static {
                         return Err(LockError::LockBusy);
                     }
                 }
-                Err(TryAcquireLockError::IOError(io_error)) => {
+                Err(TryAcquireLockError::IoError(io_error)) => {
-                    return Err(LockError::IOError(io_error));
+                    return Err(LockError::IoError(io_error));
                 }
             }
         }
@@ -193,7 +202,7 @@ pub trait Directory: DirectoryClone + fmt::Debug + Send + Sync + 'static {
     /// Registers a callback that will be called whenever a change on the `meta.json`
     /// using the `atomic_write` API is detected.
     ///
-    /// The behavior when using `.watch()` on a file using `.open_write(...)` is, on the other
+    /// The behavior when using `.watch()` on a file using [Directory::open_write] is, on the other
     /// hand, undefined.
     ///
     /// The file will be watched for the lifetime of the returned `WatchHandle`. The caller is

src/directory/error.rs

@@ -1,243 +1,173 @@
-use std::error::Error as StdError;
+use crate::Version;
 use std::fmt;
 use std::io;
 use std::path::PathBuf;
 
 /// Error while trying to acquire a directory lock.
-#[derive(Debug, Fail)]
+#[derive(Debug, Error)]
 pub enum LockError {
     /// Failed to acquired a lock as it is already held by another
     /// client.
     /// - In the context of a blocking lock, this means the lock was not released within some `timeout` period.
     /// - In the context of a non-blocking lock, this means the lock was busy at the moment of the call.
-    #[fail(
+    #[error("Could not acquire lock as it is already held, possibly by a different process.")]
-        display = "Could not acquire lock as it is already held, possibly by a different process."
-    )]
     LockBusy,
-    /// Trying to acquire a lock failed with an `IOError`
+    /// Trying to acquire a lock failed with an `IoError`
-    #[fail(display = "Failed to acquire the lock due to an io:Error.")]
+    #[error("Failed to acquire the lock due to an io:Error.")]
-    IOError(io::Error),
-}
-
-/// General IO error with an optional path to the offending file.
-#[derive(Debug)]
-pub struct IOError {
-    path: Option<PathBuf>,
-    err: io::Error,
-}
-
-impl Into<io::Error> for IOError {
-    fn into(self) -> io::Error {
-        self.err
-    }
-}
-
-impl fmt::Display for IOError {
-    fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
-        match self.path {
-            Some(ref path) => write!(f, "io error occurred on path '{:?}': '{}'", path, self.err),
-            None => write!(f, "io error occurred: '{}'", self.err),
-        }
-    }
-}
-
-impl StdError for IOError {
-    fn description(&self) -> &str {
-        "io error occurred"
-    }
-
-    fn cause(&self) -> Option<&dyn StdError> {
-        Some(&self.err)
-    }
-}
-
-impl IOError {
-    pub(crate) fn with_path(path: PathBuf, err: io::Error) -> Self {
-        IOError {
-            path: Some(path),
-            err,
-        }
-    }
-}
-
-impl From<io::Error> for IOError {
-    fn from(err: io::Error) -> IOError {
-        IOError { path: None, err }
-    }
-}
-
-/// Error that may occur when opening a directory
-#[derive(Debug)]
-pub enum OpenDirectoryError {
-    /// The underlying directory does not exists.
-    DoesNotExist(PathBuf),
-    /// The path exists but is not a directory.
-    NotADirectory(PathBuf),
-    /// IoError
     IoError(io::Error),
 }
 
-impl From<io::Error> for OpenDirectoryError {
+/// Error that may occur when opening a directory
-    fn from(io_err: io::Error) -> Self {
+#[derive(Debug, Error)]
-        OpenDirectoryError::IoError(io_err)
+pub enum OpenDirectoryError {
-    }
+    /// The underlying directory does not exists.
-}
+    #[error("Directory does not exist: '{0}'.")]
-
+    DoesNotExist(PathBuf),
-impl fmt::Display for OpenDirectoryError {
+    /// The path exists but is not a directory.
-    fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
+    #[error("Path exists but is not a directory: '{0}'.")]
-        match *self {
+    NotADirectory(PathBuf),
-            OpenDirectoryError::DoesNotExist(ref path) => {
+    /// Failed to create a temp directory.
-                write!(f, "the underlying directory '{:?}' does not exist", path)
+    #[error("Failed to create a temporary directory: '{0}'.")]
-            }
+    FailedToCreateTempDir(io::Error),
-            OpenDirectoryError::NotADirectory(ref path) => {
+    /// IoError
-                write!(f, "the path '{:?}' exists but is not a directory", path)
+    #[error("IoError '{io_error:?}' while create directory in: '{directory_path:?}'.")]
-            }
+    IoError {
-            OpenDirectoryError::IoError(ref err) => write!(
+        /// underlying io Error.
-                f,
+        io_error: io::Error,
-                "IOError while trying to open/create the directory. {:?}",
+        /// directory we tried to open.
-                err
+        directory_path: PathBuf,
-            ),
+    },
-        }
-    }
-}
-
-impl StdError for OpenDirectoryError {
-    fn description(&self) -> &str {
-        "error occurred while opening a directory"
-    }
-
-    fn cause(&self) -> Option<&dyn StdError> {
-        None
-    }
 }
 
 /// Error that may occur when starting to write in a file
-#[derive(Debug)]
+#[derive(Debug, Error)]
 pub enum OpenWriteError {
     /// Our directory is WORM, writing an existing file is forbidden.
     /// Checkout the `Directory` documentation.
+    #[error("File already exists: '{0}'")]
     FileAlreadyExists(PathBuf),
     /// Any kind of IO error that happens when
     /// writing in the underlying IO device.
-    IOError(IOError),
+    #[error("IoError '{io_error:?}' while opening file for write: '{filepath}'.")]
+    IoError {
+        /// The underlying `io::Error`.
+        io_error: io::Error,
+        /// File path of the file that tantivy failed to open for write.
+        filepath: PathBuf,
+    },
 }
 
-impl From<IOError> for OpenWriteError {
+impl OpenWriteError {
-    fn from(err: IOError) -> OpenWriteError {
+    /// Wraps an io error.
-        OpenWriteError::IOError(err)
+    pub fn wrap_io_error(io_error: io::Error, filepath: PathBuf) -> Self {
+        Self::IoError { io_error, filepath }
     }
 }
+/// Type of index incompatibility between the library and the index found on disk
+/// Used to catch and provide a hint to solve this incompatibility issue
+pub enum Incompatibility {
+    /// This library cannot decompress the index found on disk
+    CompressionMismatch {
+        /// Compression algorithm used by the current version of tantivy
+        library_compression_format: String,
+        /// Compression algorithm that was used to serialise the index
+        index_compression_format: String,
+    },
+    /// The index format found on disk isn't supported by this version of the library
+    IndexMismatch {
+        /// Version used by the library
+        library_version: Version,
+        /// Version the index was built with
+        index_version: Version,
+    },
+}
 
-impl fmt::Display for OpenWriteError {
+impl fmt::Debug for Incompatibility {
-    fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
+    fn fmt(&self, f: &mut fmt::Formatter<'_>) -> Result<(), fmt::Error> {
-        match *self {
+        match self {
-            OpenWriteError::FileAlreadyExists(ref path) => {
+            Incompatibility::CompressionMismatch {
-                write!(f, "the file '{:?}' already exists", path)
+                library_compression_format,
+                index_compression_format,
+            } => {
+                let err = format!(
+                    "Library was compiled with {:?} compression, index was compressed with {:?}",
+                    library_compression_format, index_compression_format
+                );
+                let advice = format!(
+                    "Change the feature flag to {:?} and rebuild the library",
+                    index_compression_format
+                );
+                write!(f, "{}. {}", err, advice)?;
+            }
+            Incompatibility::IndexMismatch {
+                library_version,
+                index_version,
+            } => {
+                let err = format!(
+                    "Library version: {}, index version: {}",
+                    library_version.index_format_version, index_version.index_format_version
+                );
+                // TODO make a more useful error message
+                // include the version range that supports this index_format_version
+                let advice = format!(
+                    "Change tantivy to a version compatible with index format {} (e.g. {}.{}.x) \
+                     and rebuild your project.",
+                    index_version.index_format_version, index_version.major, index_version.minor
+                );
+                write!(f, "{}. {}", err, advice)?;
             }
-            OpenWriteError::IOError(ref err) => write!(
-                f,
-                "an io error occurred while opening a file for writing: '{}'",
-                err
-            ),
         }
-    }
-}
 
-impl StdError for OpenWriteError {
+        Ok(())
-    fn description(&self) -> &str {
-        "error occurred while opening a file for writing"
-    }
-
-    fn cause(&self) -> Option<&dyn StdError> {
-        match *self {
-            OpenWriteError::FileAlreadyExists(_) => None,
-            OpenWriteError::IOError(ref err) => Some(err),
-        }
     }
 }
 
 /// Error that may occur when accessing a file read
-#[derive(Debug)]
+#[derive(Debug, Error)]
 pub enum OpenReadError {
     /// The file does not exists.
+    #[error("Files does not exists: {0:?}")]
     FileDoesNotExist(PathBuf),
-    /// Any kind of IO error that happens when
+    /// Any kind of io::Error.
-    /// interacting with the underlying IO device.
+    #[error(
-    IOError(IOError),
+        "IoError: '{io_error:?}' happened while opening the following file for Read: {filepath}."
+    )]
+    IoError {
+        /// The underlying `io::Error`.
+        io_error: io::Error,
+        /// File path of the file that tantivy failed to open for read.
+        filepath: PathBuf,
+    },
+    /// This library does not support the index version found in file footer.
+    #[error("Index version unsupported: {0:?}")]
+    IncompatibleIndex(Incompatibility),
 }
 
-impl From<IOError> for OpenReadError {
+impl OpenReadError {
-    fn from(err: IOError) -> OpenReadError {
+    /// Wraps an io error.
-        OpenReadError::IOError(err)
+    pub fn wrap_io_error(io_error: io::Error, filepath: PathBuf) -> Self {
+        Self::IoError { io_error, filepath }
     }
 }
-
-impl fmt::Display for OpenReadError {
-    fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
-        match *self {
-            OpenReadError::FileDoesNotExist(ref path) => {
-                write!(f, "the file '{:?}' does not exist", path)
-            }
-            OpenReadError::IOError(ref err) => write!(
-                f,
-                "an io error occurred while opening a file for reading: '{}'",
-                err
-            ),
-        }
-    }
-}
-
-impl StdError for OpenReadError {
-    fn description(&self) -> &str {
-        "error occurred while opening a file for reading"
-    }
-
-    fn cause(&self) -> Option<&dyn StdError> {
-        match *self {
-            OpenReadError::FileDoesNotExist(_) => None,
-            OpenReadError::IOError(ref err) => Some(err),
-        }
-    }
-}
-
 /// Error that may occur when trying to delete a file
-#[derive(Debug)]
+#[derive(Debug, Error)]
 pub enum DeleteError {
     /// The file does not exists.
+    #[error("File does not exists: '{0}'.")]
     FileDoesNotExist(PathBuf),
     /// Any kind of IO error that happens when
     /// interacting with the underlying IO device.
-    IOError(IOError),
+    #[error("The following IO error happened while deleting file '{filepath}': '{io_error:?}'.")]
+    IoError {
+        /// The underlying `io::Error`.
+        io_error: io::Error,
+        /// File path of the file that tantivy failed to delete.
+        filepath: PathBuf,
+    },
 }
 
-impl From<IOError> for DeleteError {
+impl From<Incompatibility> for OpenReadError {
-    fn from(err: IOError) -> DeleteError {
+    fn from(incompatibility: Incompatibility) -> Self {
-        DeleteError::IOError(err)
+        OpenReadError::IncompatibleIndex(incompatibility)
-    }
-}
-
-impl fmt::Display for DeleteError {
-    fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
-        match *self {
-            DeleteError::FileDoesNotExist(ref path) => {
-                write!(f, "the file '{:?}' does not exist", path)
-            }
-            DeleteError::IOError(ref err) => {
-                write!(f, "an io error occurred while deleting a file: '{}'", err)
-            }
-        }
-    }
-}
-
-impl StdError for DeleteError {
-    fn description(&self) -> &str {
-        "error occurred while deleting a file"
-    }
-
-    fn cause(&self) -> Option<&dyn StdError> {
-        match *self {
-            DeleteError::FileDoesNotExist(_) => None,
-            DeleteError::IOError(ref err) => Some(err),
-        }
     }
 }

src/directory/file_slice.rs

@@ -0,0 +1,253 @@
+use stable_deref_trait::StableDeref;
+
+use crate::common::HasLen;
+use crate::directory::OwnedBytes;
+use std::fmt;
+use std::ops::Range;
+use std::sync::{Arc, Weak};
+use std::{io, ops::Deref};
+
+pub type ArcBytes = Arc<dyn Deref<Target = [u8]> + Send + Sync + 'static>;
+pub type WeakArcBytes = Weak<dyn Deref<Target = [u8]> + Send + Sync + 'static>;
+
+/// Objects that represents files sections in tantivy.
+///
+/// By contract, whatever happens to the directory file, as long as a FileHandle
+/// is alive, the data associated with it cannot be altered or destroyed.
+///
+/// The underlying behavior is therefore specific to the `Directory` that created it.
+/// Despite its name, a `FileSlice` may or may not directly map to an actual file
+/// on the filesystem.
+pub trait FileHandle: 'static + Send + Sync + HasLen + fmt::Debug {
+    /// Reads a slice of bytes.
+    ///
+    /// This method may panic if the range requested is invalid.
+    fn read_bytes(&self, range: Range<usize>) -> io::Result<OwnedBytes>;
+}
+
+impl FileHandle for &'static [u8] {
+    fn read_bytes(&self, range: Range<usize>) -> io::Result<OwnedBytes> {
+        let bytes = &self[range];
+        Ok(OwnedBytes::new(bytes))
+    }
+}
+
+impl<T: Deref<Target = [u8]>> HasLen for T {
+    fn len(&self) -> usize {
+        self.deref().len()
+    }
+}
+
+impl<B> From<B> for FileSlice
+where
+    B: StableDeref + Deref<Target = [u8]> + 'static + Send + Sync,
+{
+    fn from(bytes: B) -> FileSlice {
+        FileSlice::new(Box::new(OwnedBytes::new(bytes)))
+    }
+}
+
+/// Logical slice of read only file in tantivy.
+///
+/// It can be cloned and sliced cheaply.
+///
+#[derive(Clone)]
+pub struct FileSlice {
+    data: Arc<dyn FileHandle>,
+    range: Range<usize>,
+}
+
+impl fmt::Debug for FileSlice {
+    fn fmt(&self, f: &mut fmt::Formatter) -> fmt::Result {
+        write!(f, "FileSlice({:?}, {:?})", &self.data, self.range)
+    }
+}
+
+impl FileSlice {
+    /// Wraps a FileHandle.
+    pub fn new(file_handle: Box<dyn FileHandle>) -> Self {
+        let num_bytes = file_handle.len();
+        FileSlice::new_with_num_bytes(file_handle, num_bytes)
+    }
+
+    /// Wraps a FileHandle.
+    #[doc(hidden)]
+    pub fn new_with_num_bytes(file_handle: Box<dyn FileHandle>, num_bytes: usize) -> Self {
+        FileSlice {
+            data: Arc::from(file_handle),
+            range: 0..num_bytes,
+        }
+    }
+
+    /// Creates a fileslice that is just a view over a slice of the data.
+    ///
+    /// # Panics
+    ///
+    /// Panics if `byte_range.end` exceeds the filesize.
+    pub fn slice(&self, byte_range: Range<usize>) -> FileSlice {
+        assert!(byte_range.end <= self.len());
+        FileSlice {
+            data: self.data.clone(),
+            range: self.range.start + byte_range.start..self.range.start + byte_range.end,
+        }
+    }
+
+    /// Creates an empty FileSlice
+    pub fn empty() -> FileSlice {
+        const EMPTY_SLICE: &[u8] = &[];
+        FileSlice::from(EMPTY_SLICE)
+    }
+
+    /// Returns a `OwnedBytes` with all of the data in the `FileSlice`.
+    ///
+    /// The behavior is strongly dependant on the implementation of the underlying
+    /// `Directory` and the `FileSliceTrait` it creates.
+    /// In particular, it is  up to the `Directory` implementation
+    /// to handle caching if needed.
+    pub fn read_bytes(&self) -> io::Result<OwnedBytes> {
+        self.data.read_bytes(self.range.clone())
+    }
+
+    /// Reads a specific slice of data.
+    ///
+    /// This is equivalent to running `file_slice.slice(from, to).read_bytes()`.
+    pub fn read_bytes_slice(&self, range: Range<usize>) -> io::Result<OwnedBytes> {
+        assert!(
+            range.end <= self.len(),
+            "end of requested range exceeds the fileslice length ({} > {})",
+            range.end,
+            self.len()
+        );
+        self.data
+            .read_bytes(self.range.start + range.start..self.range.start + range.end)
+    }
+
+    /// Splits the FileSlice at the given offset and return two file slices.
+    /// `file_slice[..split_offset]` and `file_slice[split_offset..]`.
+    ///
+    /// This operation is cheap and must not copy any underlying data.
+    pub fn split(self, left_len: usize) -> (FileSlice, FileSlice) {
+        let left = self.slice_to(left_len);
+        let right = self.slice_from(left_len);
+        (left, right)
+    }
+
+    /// Splits the file slice at the given offset and return two file slices.
+    /// `file_slice[..split_offset]` and `file_slice[split_offset..]`.
+    pub fn split_from_end(self, right_len: usize) -> (FileSlice, FileSlice) {
+        let left_len = self.len() - right_len;
+        self.split(left_len)
+    }
+
+    /// Like `.slice(...)` but enforcing only the `from`
+    /// boundary.
+    ///
+    /// Equivalent to `.slice(from_offset, self.len())`
+    pub fn slice_from(&self, from_offset: usize) -> FileSlice {
+        self.slice(from_offset..self.len())
+    }
+
+    /// Returns a slice from the end.
+    ///
+    /// Equivalent to `.slice(self.len() - from_offset, self.len())`
+    pub fn slice_from_end(&self, from_offset: usize) -> FileSlice {
+        self.slice(self.len() - from_offset..self.len())
+    }
+
+    /// Like `.slice(...)` but enforcing only the `to`
+    /// boundary.
+    ///
+    /// Equivalent to `.slice(0, to_offset)`
+    pub fn slice_to(&self, to_offset: usize) -> FileSlice {
+        self.slice(0..to_offset)
+    }
+}
+
+impl FileHandle for FileSlice {
+    fn read_bytes(&self, range: Range<usize>) -> io::Result<OwnedBytes> {
+        self.read_bytes_slice(range)
+    }
+}
+
+impl HasLen for FileSlice {
+    fn len(&self) -> usize {
+        self.range.len()
+    }
+}
+
+#[cfg(test)]
+mod tests {
+    use super::{FileHandle, FileSlice};
+    use crate::common::HasLen;
+    use std::io;
+
+    #[test]
+    fn test_file_slice() -> io::Result<()> {
+        let file_slice = FileSlice::new(Box::new(b"abcdef".as_ref()));
+        assert_eq!(file_slice.len(), 6);
+        assert_eq!(file_slice.slice_from(2).read_bytes()?.as_slice(), b"cdef");
+        assert_eq!(file_slice.slice_to(2).read_bytes()?.as_slice(), b"ab");
+        assert_eq!(
+            file_slice
+                .slice_from(1)
+                .slice_to(2)
+                .read_bytes()?
+                .as_slice(),
+            b"bc"
+        );
+        {
+            let (left, right) = file_slice.clone().split(0);
+            assert_eq!(left.read_bytes()?.as_slice(), b"");
+            assert_eq!(right.read_bytes()?.as_slice(), b"abcdef");
+        }
+        {
+            let (left, right) = file_slice.clone().split(2);
+            assert_eq!(left.read_bytes()?.as_slice(), b"ab");
+            assert_eq!(right.read_bytes()?.as_slice(), b"cdef");
+        }
+        {
+            let (left, right) = file_slice.clone().split_from_end(0);
+            assert_eq!(left.read_bytes()?.as_slice(), b"abcdef");
+            assert_eq!(right.read_bytes()?.as_slice(), b"");
+        }
+        {
+            let (left, right) = file_slice.clone().split_from_end(2);
+            assert_eq!(left.read_bytes()?.as_slice(), b"abcd");
+            assert_eq!(right.read_bytes()?.as_slice(), b"ef");
+        }
+        Ok(())
+    }
+
+    #[test]
+    fn test_file_slice_trait_slice_len() {
+        let blop: &'static [u8] = b"abc";
+        let owned_bytes: Box<dyn FileHandle> = Box::new(blop);
+        assert_eq!(owned_bytes.len(), 3);
+    }
+
+    #[test]
+    fn test_slice_simple_read() -> io::Result<()> {
+        let slice = FileSlice::new(Box::new(&b"abcdef"[..]));
+        assert_eq!(slice.len(), 6);
+        assert_eq!(slice.read_bytes()?.as_ref(), b"abcdef");
+        assert_eq!(slice.slice(1..4).read_bytes()?.as_ref(), b"bcd");
+        Ok(())
+    }
+
+    #[test]
+    fn test_slice_read_slice() -> io::Result<()> {
+        let slice_deref = FileSlice::new(Box::new(&b"abcdef"[..]));
+        assert_eq!(slice_deref.read_bytes_slice(1..4)?.as_ref(), b"bcd");
+        Ok(())
+    }
+
+    #[test]
+    #[should_panic(expected = "end of requested range exceeds the fileslice length (10 > 6)")]
+    fn test_slice_read_slice_invalid_range_exceeds() {
+        let slice_deref = FileSlice::new(Box::new(&b"abcdef"[..]));
+        assert_eq!(
+            slice_deref.read_bytes_slice(0..10).unwrap().as_ref(),
+            b"bcd"
+        );
+    }
+}

src/directory/file_watcher.rs

@@ -0,0 +1,182 @@
+use crate::directory::{WatchCallback, WatchCallbackList, WatchHandle};
+use crc32fast::Hasher;
+use std::fs;
+use std::io;
+use std::io::BufRead;
+use std::path::Path;
+use std::sync::atomic::{AtomicUsize, Ordering};
+use std::sync::Arc;
+use std::thread;
+use std::time::Duration;
+
+pub const POLLING_INTERVAL: Duration = Duration::from_millis(if cfg!(test) { 1 } else { 500 });
+
+// Watches a file and executes registered callbacks when the file is modified.
+pub struct FileWatcher {
+    path: Arc<Path>,
+    callbacks: Arc<WatchCallbackList>,
+    state: Arc<AtomicUsize>, // 0: new, 1: runnable, 2: terminated
+}
+
+impl FileWatcher {
+    pub fn new(path: &Path) -> FileWatcher {
+        FileWatcher {
+            path: Arc::from(path),
+            callbacks: Default::default(),
+            state: Default::default(),
+        }
+    }
+
+    pub fn spawn(&self) {
+        if self
+            .state
+            .compare_exchange(0, 1, Ordering::SeqCst, Ordering::SeqCst)
+            .is_err()
+        {
+            return;
+        }
+
+        let path = self.path.clone();
+        let callbacks = self.callbacks.clone();
+        let state = self.state.clone();
+
+        thread::Builder::new()
+            .name("thread-tantivy-meta-file-watcher".to_string())
+            .spawn(move || {
+                let mut current_checksum = None;
+
+                while state.load(Ordering::SeqCst) == 1 {
+                    if let Ok(checksum) = FileWatcher::compute_checksum(&path) {
+                        // `None.unwrap_or_else(|| !checksum) != checksum` evaluates to `true`
+                        if current_checksum.unwrap_or_else(|| !checksum) != checksum {
+                            info!("Meta file {:?} was modified", path);
+                            current_checksum = Some(checksum);
+                            futures::executor::block_on(callbacks.broadcast());
+                        }
+                    }
+
+                    thread::sleep(POLLING_INTERVAL);
+                }
+            })
+            .expect("Failed to spawn meta file watcher thread");
+    }
+
+    pub fn watch(&self, callback: WatchCallback) -> WatchHandle {
+        let handle = self.callbacks.subscribe(callback);
+        self.spawn();
+        handle
+    }
+
+    fn compute_checksum(path: &Path) -> Result<u32, io::Error> {
+        let reader = match fs::File::open(path) {
+            Ok(f) => io::BufReader::new(f),
+            Err(e) => {
+                warn!("Failed to open meta file {:?}: {:?}", path, e);
+                return Err(e);
+            }
+        };
+
+        let mut hasher = Hasher::new();
+
+        for line in reader.lines() {
+            hasher.update(line?.as_bytes())
+        }
+
+        Ok(hasher.finalize())
+    }
+}
+
+impl Drop for FileWatcher {
+    fn drop(&mut self) {
+        self.state.store(2, Ordering::SeqCst);
+    }
+}
+
+#[cfg(test)]
+mod tests {
+
+    use std::mem;
+
+    use crate::directory::mmap_directory::atomic_write;
+
+    use super::*;
+
+    #[test]
+    fn test_file_watcher_drop_watcher() -> crate::Result<()> {
+        let tmp_dir = tempfile::TempDir::new()?;
+        let tmp_file = tmp_dir.path().join("watched.txt");
+
+        let counter: Arc<AtomicUsize> = Default::default();
+        let (tx, rx) = crossbeam::channel::unbounded();
+        let timeout = Duration::from_millis(100);
+
+        let watcher = FileWatcher::new(&tmp_file);
+
+        let state = watcher.state.clone();
+        assert_eq!(state.load(Ordering::SeqCst), 0);
+
+        let counter_clone = counter.clone();
+
+        let _handle = watcher.watch(WatchCallback::new(move || {
+            let val = counter_clone.fetch_add(1, Ordering::SeqCst);
+            tx.send(val + 1).unwrap();
+        }));
+
+        assert_eq!(counter.load(Ordering::SeqCst), 0);
+        assert_eq!(state.load(Ordering::SeqCst), 1);
+
+        atomic_write(&tmp_file, b"foo")?;
+        assert_eq!(rx.recv_timeout(timeout), Ok(1));
+
+        atomic_write(&tmp_file, b"foo")?;
+        assert!(rx.recv_timeout(timeout).is_err());
+
+        atomic_write(&tmp_file, b"bar")?;
+        assert_eq!(rx.recv_timeout(timeout), Ok(2));
+
+        mem::drop(watcher);
+
+        atomic_write(&tmp_file, b"qux")?;
+        thread::sleep(Duration::from_millis(10));
+        assert_eq!(counter.load(Ordering::SeqCst), 2);
+        assert_eq!(state.load(Ordering::SeqCst), 2);
+
+        Ok(())
+    }
+
+    #[test]
+    fn test_file_watcher_drop_handle() -> crate::Result<()> {
+        let tmp_dir = tempfile::TempDir::new()?;
+        let tmp_file = tmp_dir.path().join("watched.txt");
+
+        let counter: Arc<AtomicUsize> = Default::default();
+        let (tx, rx) = crossbeam::channel::unbounded();
+        let timeout = Duration::from_millis(100);
+
+        let watcher = FileWatcher::new(&tmp_file);
+
+        let state = watcher.state.clone();
+        assert_eq!(state.load(Ordering::SeqCst), 0);
+
+        let counter_clone = counter.clone();
+
+        let handle = watcher.watch(WatchCallback::new(move || {
+            let val = counter_clone.fetch_add(1, Ordering::SeqCst);
+            tx.send(val + 1).unwrap();
+        }));
+
+        assert_eq!(counter.load(Ordering::SeqCst), 0);
+        assert_eq!(state.load(Ordering::SeqCst), 1);
+
+        atomic_write(&tmp_file, b"foo")?;
+        assert_eq!(rx.recv_timeout(timeout), Ok(1));
+
+        mem::drop(handle);
+
+        atomic_write(&tmp_file, b"qux")?;
+        assert_eq!(counter.load(Ordering::SeqCst), 1);
+        assert_eq!(state.load(Ordering::SeqCst), 1);
+
+        Ok(())
+    }
+}

src/directory/footer.rs

@@ -0,0 +1,228 @@
+use crate::directory::error::Incompatibility;
+use crate::directory::FileSlice;
+use crate::{
+    common::{BinarySerializable, CountingWriter, DeserializeFrom, FixedSize, HasLen},
+    directory::{AntiCallToken, TerminatingWrite},
+    Version, INDEX_FORMAT_VERSION,
+};
+use crc32fast::Hasher;
+use serde::{Deserialize, Serialize};
+use std::io;
+use std::io::Write;
+
+const FOOTER_MAX_LEN: u32 = 50_000;
+
+/// The magic byte of the footer to identify corruption
+/// or an old version of the footer.
+const FOOTER_MAGIC_NUMBER: u32 = 1337;
+
+type CrcHashU32 = u32;
+
+/// A Footer is appended to every file
+#[derive(Debug, Clone, PartialEq, Serialize, Deserialize)]
+pub struct Footer {
+    pub version: Version,
+    pub crc: CrcHashU32,
+}
+
+impl Footer {
+    pub fn new(crc: CrcHashU32) -> Self {
+        let version = crate::VERSION.clone();
+        Footer { version, crc }
+    }
+
+    pub fn crc(&self) -> CrcHashU32 {
+        self.crc
+    }
+    pub fn append_footer<W: io::Write>(&self, mut write: &mut W) -> io::Result<()> {
+        let mut counting_write = CountingWriter::wrap(&mut write);
+        counting_write.write_all(serde_json::to_string(&self)?.as_ref())?;
+        let footer_payload_len = counting_write.written_bytes();
+        BinarySerializable::serialize(&(footer_payload_len as u32), write)?;
+        BinarySerializable::serialize(&(FOOTER_MAGIC_NUMBER as u32), write)?;
+        Ok(())
+    }
+
+    pub fn extract_footer(file: FileSlice) -> io::Result<(Footer, FileSlice)> {
+        if file.len() < 4 {
+            return Err(io::Error::new(
+                io::ErrorKind::UnexpectedEof,
+                format!(
+                    "File corrupted. The file is smaller than 4 bytes (len={}).",
+                    file.len()
+                ),
+            ));
+        }
+
+        let footer_metadata_len = <(u32, u32)>::SIZE_IN_BYTES;
+        let (footer_len, footer_magic_byte): (u32, u32) = file
+            .slice_from_end(footer_metadata_len)
+            .read_bytes()?
+            .as_ref()
+            .deserialize()?;
+
+        if footer_magic_byte != FOOTER_MAGIC_NUMBER {
+            return Err(io::Error::new(
+                io::ErrorKind::InvalidData,
+                    "Footer magic byte mismatch. File corrupted or index was created using old an tantivy version which is not supported anymore. Please use tantivy 0.15 or above to recreate the index.",
+            ));
+        }
+
+        if footer_len > FOOTER_MAX_LEN {
+            return Err(io::Error::new(
+                io::ErrorKind::InvalidData,
+                format!(
+                    "Footer seems invalid as it suggests a footer len of {}. File is corrupted, \
+            or the index was created with a different & old version of tantivy.",
+                    footer_len
+                ),
+            ));
+        }
+        let total_footer_size = footer_len as usize + footer_metadata_len;
+        if file.len() < total_footer_size {
+            return Err(io::Error::new(
+                io::ErrorKind::UnexpectedEof,
+                format!(
+                    "File corrupted. The file is smaller than it's footer bytes (len={}).",
+                    total_footer_size
+                ),
+            ));
+        }
+
+        let footer: Footer = serde_json::from_slice(&file.read_bytes_slice(
+            file.len() - total_footer_size..file.len() - footer_metadata_len as usize,
+        )?)?;
+
+        let body = file.slice_to(file.len() - total_footer_size);
+        Ok((footer, body))
+    }
+
+    /// Confirms that the index will be read correctly by this version of tantivy
+    /// Has to be called after `extract_footer` to make sure it's not accessing uninitialised memory
+    pub fn is_compatible(&self) -> Result<(), Incompatibility> {
+        let library_version = crate::version();
+        if self.version.index_format_version < 4
+            || self.version.index_format_version > INDEX_FORMAT_VERSION
+        {
+            return Err(Incompatibility::IndexMismatch {
+                library_version: library_version.clone(),
+                index_version: self.version.clone(),
+            });
+        }
+        Ok(())
+    }
+}
+
+pub(crate) struct FooterProxy<W: TerminatingWrite> {
+    /// always Some except after terminate call
+    hasher: Option<Hasher>,
+    /// always Some except after terminate call
+    writer: Option<W>,
+}
+
+impl<W: TerminatingWrite> FooterProxy<W> {
+    pub fn new(writer: W) -> Self {
+        FooterProxy {
+            hasher: Some(Hasher::new()),
+            writer: Some(writer),
+        }
+    }
+}
+
+impl<W: TerminatingWrite> Write for FooterProxy<W> {
+    fn write(&mut self, buf: &[u8]) -> io::Result<usize> {
+        let count = self.writer.as_mut().unwrap().write(buf)?;
+        self.hasher.as_mut().unwrap().update(&buf[..count]);
+        Ok(count)
+    }
+
+    fn flush(&mut self) -> io::Result<()> {
+        self.writer.as_mut().unwrap().flush()
+    }
+}
+
+impl<W: TerminatingWrite> TerminatingWrite for FooterProxy<W> {
+    fn terminate_ref(&mut self, _: AntiCallToken) -> io::Result<()> {
+        let crc32 = self.hasher.take().unwrap().finalize();
+        let footer = Footer::new(crc32);
+        let mut writer = self.writer.take().unwrap();
+        footer.append_footer(&mut writer)?;
+        writer.terminate()
+    }
+}
+
+#[cfg(test)]
+mod tests {
+
+    use crate::directory::footer::Footer;
+    use crate::directory::OwnedBytes;
+    use crate::{
+        common::BinarySerializable,
+        directory::{footer::FOOTER_MAGIC_NUMBER, FileSlice},
+    };
+    use std::io;
+
+    #[test]
+    fn test_deserialize_footer() {
+        let mut buf: Vec<u8> = vec![];
+        let footer = Footer::new(123);
+        footer.append_footer(&mut buf).unwrap();
+        let owned_bytes = OwnedBytes::new(buf);
+        let fileslice = FileSlice::new(Box::new(owned_bytes));
+        let (footer_deser, _body) = Footer::extract_footer(fileslice).unwrap();
+        assert_eq!(footer_deser.crc(), footer.crc());
+    }
+    #[test]
+    fn test_deserialize_footer_missing_magic_byte() {
+        let mut buf: Vec<u8> = vec![];
+        BinarySerializable::serialize(&0_u32, &mut buf).unwrap();
+        let wrong_magic_byte: u32 = 5555;
+        BinarySerializable::serialize(&wrong_magic_byte, &mut buf).unwrap();
+
+        let owned_bytes = OwnedBytes::new(buf);
+
+        let fileslice = FileSlice::new(Box::new(owned_bytes));
+        let err = Footer::extract_footer(fileslice).unwrap_err();
+        assert_eq!(
+            err.to_string(),
+            "Footer magic byte mismatch. File corrupted or index was created using old an tantivy version which \
+            is not supported anymore. Please use tantivy 0.15 or above to recreate the index."
+        );
+    }
+    #[test]
+    fn test_deserialize_footer_wrong_filesize() {
+        let mut buf: Vec<u8> = vec![];
+        BinarySerializable::serialize(&100_u32, &mut buf).unwrap();
+        BinarySerializable::serialize(&FOOTER_MAGIC_NUMBER, &mut buf).unwrap();
+
+        let owned_bytes = OwnedBytes::new(buf);
+
+        let fileslice = FileSlice::new(Box::new(owned_bytes));
+        let err = Footer::extract_footer(fileslice).unwrap_err();
+        assert_eq!(err.kind(), io::ErrorKind::UnexpectedEof);
+        assert_eq!(
+            err.to_string(),
+            "File corrupted. The file is smaller than it\'s footer bytes (len=108)."
+        );
+    }
+
+    #[test]
+    fn test_deserialize_too_large_footer() {
+        let mut buf: Vec<u8> = vec![];
+
+        let footer_length = super::FOOTER_MAX_LEN + 1;
+        BinarySerializable::serialize(&footer_length, &mut buf).unwrap();
+        BinarySerializable::serialize(&FOOTER_MAGIC_NUMBER, &mut buf).unwrap();
+
+        let owned_bytes = OwnedBytes::new(buf);
+
+        let fileslice = FileSlice::new(Box::new(owned_bytes));
+        let err = Footer::extract_footer(fileslice).unwrap_err();
+        assert_eq!(err.kind(), io::ErrorKind::InvalidData);
+        assert_eq!(
+            err.to_string(),
+            "Footer seems invalid as it suggests a footer len of 50001. File is corrupted, \
+    or the index was created with a different & old version of tantivy."
+        );
+    }
+}

src/directory/managed_directory.rs

@@ -1,14 +1,16 @@
-use crate::core::MANAGED_FILEPATH;
+use crate::core::{MANAGED_FILEPATH, META_FILEPATH};
-use crate::directory::error::{DeleteError, IOError, LockError, OpenReadError, OpenWriteError};
+use crate::directory::error::{DeleteError, LockError, OpenReadError, OpenWriteError};
-use crate::directory::DirectoryLock;
+use crate::directory::footer::{Footer, FooterProxy};
+use crate::directory::GarbageCollectionResult;
 use crate::directory::Lock;
 use crate::directory::META_LOCK;
-use crate::directory::{ReadOnlySource, WritePtr};
+use crate::directory::{DirectoryLock, FileHandle};
+use crate::directory::{FileSlice, WritePtr};
 use crate::directory::{WatchCallback, WatchHandle};
 use crate::error::DataCorruption;
 use crate::Directory;
-use crate::Result;
+
-use serde_json;
+use crc32fast::Hasher;
 use std::collections::HashSet;
 use std::io;
 use std::io::Write;
@@ -51,7 +53,7 @@ struct MetaInformation {
 /// Saves the file containing the list of existing files
 /// that were created by tantivy.
 fn save_managed_paths(
-    directory: &mut dyn Directory,
+    directory: &dyn Directory,
     wlock: &RwLockWriteGuard<'_, MetaInformation>,
 ) -> io::Result<()> {
     let mut w = serde_json::to_vec(&wlock.managed_paths)?;
@@ -62,7 +64,7 @@ fn save_managed_paths(
 
 impl ManagedDirectory {
     /// Wraps a directory as managed directory.
-    pub fn wrap<Dir: Directory>(directory: Dir) -> Result<ManagedDirectory> {
+    pub fn wrap<Dir: Directory>(directory: Dir) -> crate::Result<ManagedDirectory> {
         match directory.atomic_read(&MANAGED_FILEPATH) {
             Ok(data) => {
                 let managed_files_json = String::from_utf8_lossy(&data);
@@ -84,7 +86,12 @@ impl ManagedDirectory {
                 directory: Box::new(directory),
                 meta_informations: Arc::default(),
             }),
-            Err(OpenReadError::IOError(e)) => Err(From::from(e)),
+            io_err @ Err(OpenReadError::IoError { .. }) => Err(io_err.err().unwrap().into()),
+            Err(OpenReadError::IncompatibleIndex(incompatibility)) => {
+                // For the moment, this should never happen  `meta.json`
+                // do not have any footer and cannot detect incompatibility.
+                Err(crate::TantivyError::IncompatibleIndex(incompatibility))
+            }
         }
     }
 
@@ -102,7 +109,10 @@ impl ManagedDirectory {
     /// If a file cannot be deleted (for permission reasons for instance)
     /// an error is simply logged, and the file remains in the list of managed
     /// files.
-    pub fn garbage_collect<L: FnOnce() -> HashSet<PathBuf>>(&mut self, get_living_files: L) {
+    pub fn garbage_collect<L: FnOnce() -> HashSet<PathBuf>>(
+        &mut self,
+        get_living_files: L,
+    ) -> crate::Result<GarbageCollectionResult> {
         info!("Garbage collect");
         let mut files_to_delete = vec![];
 
@@ -128,19 +138,25 @@ impl ManagedDirectory {
             // 2) writer change meta.json (for instance after a merge or a commit)
             // 3) gc kicks in.
             // 4) gc removes a file that was useful for process B, before process B opened it.
-            if let Ok(_meta_lock) = self.acquire_lock(&META_LOCK) {
+            match self.acquire_lock(&META_LOCK) {
-                let living_files = get_living_files();
+                Ok(_meta_lock) => {
-                for managed_path in &meta_informations_rlock.managed_paths {
+                    let living_files = get_living_files();
-                    if !living_files.contains(managed_path) {
+                    for managed_path in &meta_informations_rlock.managed_paths {
-                        files_to_delete.push(managed_path.clone());
+                        if !living_files.contains(managed_path) {
+                            files_to_delete.push(managed_path.clone());
+                        }
                     }
                 }
-            } else {
+                Err(err) => {
-                error!("Failed to acquire lock for GC");
+                    error!("Failed to acquire lock for GC");
+                    return Err(crate::TantivyError::from(err));
+                }
             }
         }
 
+        let mut failed_to_delete_files = vec![];
         let mut deleted_files = vec![];
+
         for file_to_delete in files_to_delete {
             match self.delete(&file_to_delete) {
                 Ok(_) => {
@@ -150,9 +166,10 @@ impl ManagedDirectory {
                 Err(file_error) => {
                     match file_error {
                         DeleteError::FileDoesNotExist(_) => {
-                            deleted_files.push(file_to_delete);
+                            deleted_files.push(file_to_delete.clone());
                         }
-                        DeleteError::IOError(_) => {
+                        DeleteError::IoError { .. } => {
+                            failed_to_delete_files.push(file_to_delete.clone());
                             if !cfg!(target_os = "windows") {
                                 // On windows, delete is expected to fail if the file
                                 // is mmapped.
@@ -175,10 +192,13 @@ impl ManagedDirectory {
             for delete_file in &deleted_files {
                 managed_paths_write.remove(delete_file);
             }
-            if save_managed_paths(self.directory.as_mut(), &meta_informations_wlock).is_err() {
+            save_managed_paths(self.directory.as_mut(), &meta_informations_wlock)?;
-                error!("Failed to save the list of managed files.");
-            }
         }
+
+        Ok(GarbageCollectionResult {
+            deleted_files,
+            failed_to_delete_files,
+        })
     }
 
     /// Registers a file as managed
@@ -192,7 +212,7 @@ impl ManagedDirectory {
     /// File starting by "." are reserved to locks.
     /// They are not managed and cannot be subjected
     /// to garbage collection.
-    fn register_file_as_managed(&mut self, filepath: &Path) -> io::Result<()> {
+    fn register_file_as_managed(&self, filepath: &Path) -> io::Result<()> {
         // Files starting by "." (e.g. lock files) are not managed.
         if !is_managed(filepath) {
             return Ok(());
@@ -203,24 +223,79 @@ impl ManagedDirectory {
             .expect("Managed file lock poisoned");
         let has_changed = meta_wlock.managed_paths.insert(filepath.to_owned());
         if has_changed {
-            save_managed_paths(self.directory.as_mut(), &meta_wlock)?;
+            save_managed_paths(self.directory.as_ref(), &meta_wlock)?;
         }
         Ok(())
     }
+
+    /// Verify checksum of a managed file
+    pub fn validate_checksum(&self, path: &Path) -> result::Result<bool, OpenReadError> {
+        let reader = self.directory.open_read(path)?;
+        let (footer, data) =
+            Footer::extract_footer(reader).map_err(|io_error| OpenReadError::IoError {
+                io_error,
+                filepath: path.to_path_buf(),
+            })?;
+        let bytes = data
+            .read_bytes()
+            .map_err(|io_error| OpenReadError::IoError {
+                filepath: path.to_path_buf(),
+                io_error,
+            })?;
+        let mut hasher = Hasher::new();
+        hasher.update(bytes.as_slice());
+        let crc = hasher.finalize();
+        Ok(footer.crc() == crc)
+    }
+
+    /// List files for which checksum does not match content
+    pub fn list_damaged(&self) -> result::Result<HashSet<PathBuf>, OpenReadError> {
+        let mut managed_paths = self
+            .meta_informations
+            .read()
+            .expect("Managed directory rlock poisoned in list damaged.")
+            .managed_paths
+            .clone();
+
+        managed_paths.remove(*META_FILEPATH);
+
+        let mut damaged_files = HashSet::new();
+        for path in managed_paths {
+            if !self.validate_checksum(&path)? {
+                damaged_files.insert(path);
+            }
+        }
+        Ok(damaged_files)
+    }
 }
 
 impl Directory for ManagedDirectory {
-    fn open_read(&self, path: &Path) -> result::Result<ReadOnlySource, OpenReadError> {
+    fn get_file_handle(&self, path: &Path) -> Result<Box<dyn FileHandle>, OpenReadError> {
-        self.directory.open_read(path)
+        let file_slice = self.open_read(path)?;
+        Ok(Box::new(file_slice))
     }
 
-    fn open_write(&mut self, path: &Path) -> result::Result<WritePtr, OpenWriteError> {
+    fn open_read(&self, path: &Path) -> result::Result<FileSlice, OpenReadError> {
+        let file_slice = self.directory.open_read(path)?;
+        let (footer, reader) = Footer::extract_footer(file_slice)
+            .map_err(|io_error| OpenReadError::wrap_io_error(io_error, path.to_path_buf()))?;
+        footer.is_compatible()?;
+        Ok(reader)
+    }
+
+    fn open_write(&self, path: &Path) -> result::Result<WritePtr, OpenWriteError> {
         self.register_file_as_managed(path)
-            .map_err(|e| IOError::with_path(path.to_owned(), e))?;
+            .map_err(|io_error| OpenWriteError::wrap_io_error(io_error, path.to_path_buf()))?;
-        self.directory.open_write(path)
+        Ok(io::BufWriter::new(Box::new(FooterProxy::new(
+            self.directory
+                .open_write(path)?
+                .into_inner()
+                .map_err(|_| ())
+                .expect("buffer should be empty"),
+        ))))
     }
 
-    fn atomic_write(&mut self, path: &Path, data: &[u8]) -> io::Result<()> {
+    fn atomic_write(&self, path: &Path, data: &[u8]) -> io::Result<()> {
         self.register_file_as_managed(path)?;
         self.directory.atomic_write(path, data)
     }
@@ -233,7 +308,7 @@ impl Directory for ManagedDirectory {
         self.directory.delete(path)
     }
 
-    fn exists(&self, path: &Path) -> bool {
+    fn exists(&self, path: &Path) -> Result<bool, OpenReadError> {
         self.directory.exists(path)
     }
 
@@ -259,15 +334,16 @@ impl Clone for ManagedDirectory {
 #[cfg(test)]
 mod tests_mmap_specific {
 
-    use crate::directory::{Directory, ManagedDirectory, MmapDirectory};
+    use crate::directory::{Directory, ManagedDirectory, MmapDirectory, TerminatingWrite};
     use std::collections::HashSet;
+    use std::fs::OpenOptions;
     use std::io::Write;
     use std::path::{Path, PathBuf};
-    use tempdir::TempDir;
+    use tempfile::TempDir;
 
     #[test]
     fn test_managed_directory() {
-        let tempdir = TempDir::new("tantivy-test").unwrap();
+        let tempdir = TempDir::new().unwrap();
         let tempdir_path = PathBuf::from(tempdir.path());
 
         let test_path1: &'static Path = Path::new("some_path_for_test");
@@ -275,28 +351,27 @@ mod tests_mmap_specific {
         {
             let mmap_directory = MmapDirectory::open(&tempdir_path).unwrap();
             let mut managed_directory = ManagedDirectory::wrap(mmap_directory).unwrap();
-            let mut write_file = managed_directory.open_write(test_path1).unwrap();
+            let write_file = managed_directory.open_write(test_path1).unwrap();
-            write_file.flush().unwrap();
+            write_file.terminate().unwrap();
             managed_directory
                 .atomic_write(test_path2, &[0u8, 1u8])
                 .unwrap();
-            assert!(managed_directory.exists(test_path1));
+            assert!(managed_directory.exists(test_path1).unwrap());
-            assert!(managed_directory.exists(test_path2));
+            assert!(managed_directory.exists(test_path2).unwrap());
-            let living_files: HashSet<PathBuf> =
+            let living_files: HashSet<PathBuf> = [test_path1.to_owned()].iter().cloned().collect();
-                [test_path1.to_owned()].into_iter().cloned().collect();
+            assert!(managed_directory.garbage_collect(|| living_files).is_ok());
-            managed_directory.garbage_collect(|| living_files);
+            assert!(managed_directory.exists(test_path1).unwrap());
-            assert!(managed_directory.exists(test_path1));
+            assert!(!managed_directory.exists(test_path2).unwrap());
-            assert!(!managed_directory.exists(test_path2));
         }
         {
             let mmap_directory = MmapDirectory::open(&tempdir_path).unwrap();
             let mut managed_directory = ManagedDirectory::wrap(mmap_directory).unwrap();
-            assert!(managed_directory.exists(test_path1));
+            assert!(managed_directory.exists(test_path1).unwrap());
-            assert!(!managed_directory.exists(test_path2));
+            assert!(!managed_directory.exists(test_path2).unwrap());
             let living_files: HashSet<PathBuf> = HashSet::new();
-            managed_directory.garbage_collect(|| living_files);
+            assert!(managed_directory.garbage_collect(|| living_files).is_ok());
-            assert!(!managed_directory.exists(test_path1));
+            assert!(!managed_directory.exists(test_path1).unwrap());
-            assert!(!managed_directory.exists(test_path2));
+            assert!(!managed_directory.exists(test_path2).unwrap());
         }
     }
 
@@ -304,31 +379,67 @@ mod tests_mmap_specific {
     fn test_managed_directory_gc_while_mmapped() {
         let test_path1: &'static Path = Path::new("some_path_for_test");
 
-        let tempdir = TempDir::new("index").unwrap();
+        let tempdir = TempDir::new().unwrap();
         let tempdir_path = PathBuf::from(tempdir.path());
         let living_files = HashSet::new();
 
         let mmap_directory = MmapDirectory::open(&tempdir_path).unwrap();
         let mut managed_directory = ManagedDirectory::wrap(mmap_directory).unwrap();
-        managed_directory
+        let mut write = managed_directory.open_write(test_path1).unwrap();
-            .atomic_write(test_path1, &vec![0u8, 1u8])
+        write.write_all(&[0u8, 1u8]).unwrap();
-            .unwrap();
+        write.terminate().unwrap();
-        assert!(managed_directory.exists(test_path1));
+        assert!(managed_directory.exists(test_path1).unwrap());
 
         let _mmap_read = managed_directory.open_read(test_path1).unwrap();
-        managed_directory.garbage_collect(|| living_files.clone());
+        assert!(managed_directory
+            .garbage_collect(|| living_files.clone())
+            .is_ok());
         if cfg!(target_os = "windows") {
             // On Windows, gc should try and fail the file as it is mmapped.
-            assert!(managed_directory.exists(test_path1));
+            assert!(managed_directory.exists(test_path1).unwrap());
             // unmap should happen here.
             drop(_mmap_read);
             // The file should still be in the list of managed file and
             // eventually be deleted once mmap is released.
-            managed_directory.garbage_collect(|| living_files);
+            assert!(managed_directory.garbage_collect(|| living_files).is_ok());
-            assert!(!managed_directory.exists(test_path1));
+            assert!(!managed_directory.exists(test_path1).unwrap());
         } else {
-            assert!(!managed_directory.exists(test_path1));
+            assert!(!managed_directory.exists(test_path1).unwrap());
         }
     }
 
+    #[test]
+    fn test_checksum() -> crate::Result<()> {
+        let test_path1: &'static Path = Path::new("some_path_for_test");
+        let test_path2: &'static Path = Path::new("other_test_path");
+
+        let tempdir = TempDir::new().unwrap();
+        let tempdir_path = PathBuf::from(tempdir.path());
+
+        let mmap_directory = MmapDirectory::open(&tempdir_path)?;
+        let managed_directory = ManagedDirectory::wrap(mmap_directory)?;
+        let mut write = managed_directory.open_write(test_path1)?;
+        write.write_all(&[0u8, 1u8])?;
+        write.terminate()?;
+
+        let mut write = managed_directory.open_write(test_path2)?;
+        write.write_all(&[3u8, 4u8, 5u8])?;
+        write.terminate()?;
+
+        let read_file = managed_directory.open_read(test_path2)?.read_bytes()?;
+        assert_eq!(read_file.as_slice(), &[3u8, 4u8, 5u8]);
+        assert!(managed_directory.list_damaged().unwrap().is_empty());
+
+        let mut corrupted_path = tempdir_path.clone();
+        corrupted_path.push(test_path2);
+        let mut file = OpenOptions::new().write(true).open(&corrupted_path)?;
+        file.write_all(&[255u8])?;
+        file.flush()?;
+        drop(file);
+
+        let damaged = managed_directory.list_damaged()?;
+        assert_eq!(damaged.len(), 1);
+        assert!(damaged.contains(test_path2));
+        Ok(())
+    }
 }

src/directory/mmap_directory.rs

@@ -1,27 +1,19 @@
-use fs2;
-use notify;
-
-use self::fs2::FileExt;
-use self::notify::RawEvent;
-use self::notify::RecursiveMode;
-use self::notify::Watcher;
 use crate::core::META_FILEPATH;
 use crate::directory::error::LockError;
-use crate::directory::error::{
+use crate::directory::error::{DeleteError, OpenDirectoryError, OpenReadError, OpenWriteError};
-    DeleteError, IOError, OpenDirectoryError, OpenReadError, OpenWriteError,
+use crate::directory::file_watcher::FileWatcher;
-};
-use crate::directory::read_only_source::BoxedData;
 use crate::directory::Directory;
 use crate::directory::DirectoryLock;
 use crate::directory::Lock;
-use crate::directory::ReadOnlySource;
 use crate::directory::WatchCallback;
-use crate::directory::WatchCallbackList;
 use crate::directory::WatchHandle;
-use crate::directory::WritePtr;
+use crate::directory::{AntiCallToken, FileHandle, OwnedBytes};
-use atomicwrites;
+use crate::directory::{ArcBytes, WeakArcBytes};
+use crate::directory::{TerminatingWrite, WritePtr};
+use fs2::FileExt;
 use memmap::Mmap;
-use std::collections::HashMap;
+use serde::{Deserialize, Serialize};
+use stable_deref_trait::StableDeref;
 use std::convert::From;
 use std::fmt;
 use std::fs::OpenOptions;
@@ -30,13 +22,10 @@ use std::io::{self, Seek, SeekFrom};
 use std::io::{BufWriter, Read, Write};
 use std::path::{Path, PathBuf};
 use std::result;
-use std::sync::mpsc::{channel, Receiver, Sender};
 use std::sync::Arc;
-use std::sync::Mutex;
 use std::sync::RwLock;
-use std::sync::Weak;
+use std::{collections::HashMap, ops::Deref};
-use std::thread;
+use tempfile::TempDir;
-use tempdir::TempDir;
 
 /// Create a default io error given a string.
 pub(crate) fn make_io_err(msg: String) -> io::Error {
@@ -46,17 +35,17 @@ pub(crate) fn make_io_err(msg: String) -> io::Error {
 /// Returns None iff the file exists, can be read, but is empty (and hence
 /// cannot be mmapped)
 fn open_mmap(full_path: &Path) -> result::Result<Option<Mmap>, OpenReadError> {
-    let file = File::open(full_path).map_err(|e| {
+    let file = File::open(full_path).map_err(|io_err| {
-        if e.kind() == io::ErrorKind::NotFound {
+        if io_err.kind() == io::ErrorKind::NotFound {
-            OpenReadError::FileDoesNotExist(full_path.to_owned())
+            OpenReadError::FileDoesNotExist(full_path.to_path_buf())
         } else {
-            OpenReadError::IOError(IOError::with_path(full_path.to_owned(), e))
+            OpenReadError::wrap_io_error(io_err, full_path.to_path_buf())
         }
     })?;
 
     let meta_data = file
         .metadata()
-        .map_err(|e| IOError::with_path(full_path.to_owned(), e))?;
+        .map_err(|io_err| OpenReadError::wrap_io_error(io_err, full_path.to_owned()))?;
     if meta_data.len() == 0 {
         // if the file size is 0, it will not be possible
         // to mmap the file, so we return None
@@ -66,7 +55,7 @@ fn open_mmap(full_path: &Path) -> result::Result<Option<Mmap>, OpenReadError> {
     unsafe {
         memmap::Mmap::map(&file)
             .map(Some)
-            .map_err(|e| From::from(IOError::with_path(full_path.to_owned(), e)))
+            .map_err(|io_err| OpenReadError::wrap_io_error(io_err, full_path.to_path_buf()))
     }
 }
 
@@ -87,7 +76,7 @@ pub struct CacheInfo {
 
 struct MmapCache {
     counters: CacheCounters,
-    cache: HashMap<PathBuf, Weak<BoxedData>>,
+    cache: HashMap<PathBuf, WeakArcBytes>,
 }
 
 impl Default for MmapCache {
@@ -121,7 +110,7 @@ impl MmapCache {
     }
 
     // Returns None if the file exists but as a len of 0 (and hence is not mmappable).
-    fn get_mmap(&mut self, full_path: &Path) -> Result<Option<Arc<BoxedData>>, OpenReadError> {
+    fn get_mmap(&mut self, full_path: &Path) -> Result<Option<ArcBytes>, OpenReadError> {
         if let Some(mmap_weak) = self.cache.get(full_path) {
             if let Some(mmap_arc) = mmap_weak.upgrade() {
                 self.counters.hit += 1;
@@ -130,83 +119,13 @@ impl MmapCache {
         }
         self.cache.remove(full_path);
         self.counters.miss += 1;
-        Ok(if let Some(mmap) = open_mmap(full_path)? {
+        let mmap_opt = open_mmap(full_path)?;
-            let mmap_arc: Arc<BoxedData> = Arc::new(Box::new(mmap));
+        Ok(mmap_opt.map(|mmap| {
+            let mmap_arc: ArcBytes = Arc::new(mmap);
             let mmap_weak = Arc::downgrade(&mmap_arc);
             self.cache.insert(full_path.to_owned(), mmap_weak);
-            Some(mmap_arc)
+            mmap_arc
-        } else {
+        }))
-            None
-        })
-    }
-}
-
-struct InnerWatcherWrapper {
-    _watcher: Mutex<notify::RecommendedWatcher>,
-    watcher_router: WatchCallbackList,
-}
-
-impl InnerWatcherWrapper {
-    pub fn new(path: &Path) -> Result<(Self, Receiver<notify::RawEvent>), notify::Error> {
-        let (tx, watcher_recv): (Sender<RawEvent>, Receiver<RawEvent>) = channel();
-        // We need to initialize the
-        let mut watcher = notify::raw_watcher(tx)?;
-        watcher.watch(path, RecursiveMode::Recursive)?;
-        let inner = InnerWatcherWrapper {
-            _watcher: Mutex::new(watcher),
-            watcher_router: Default::default(),
-        };
-        Ok((inner, watcher_recv))
-    }
-}
-
-#[derive(Clone)]
-struct WatcherWrapper {
-    inner: Arc<InnerWatcherWrapper>,
-}
-
-impl WatcherWrapper {
-    pub fn new(path: &Path) -> Result<Self, OpenDirectoryError> {
-        let (inner, watcher_recv) = InnerWatcherWrapper::new(path).map_err(|err| match err {
-            notify::Error::PathNotFound => OpenDirectoryError::DoesNotExist(path.to_owned()),
-            _ => {
-                panic!("Unknown error while starting watching directory {:?}", path);
-            }
-        })?;
-        let watcher_wrapper = WatcherWrapper {
-            inner: Arc::new(inner),
-        };
-        let watcher_wrapper_clone = watcher_wrapper.clone();
-        thread::Builder::new()
-            .name("meta-file-watch-thread".to_string())
-            .spawn(move || {
-                loop {
-                    match watcher_recv.recv().map(|evt| evt.path) {
-                        Ok(Some(changed_path)) => {
-                            // ... Actually subject to false positive.
-                            // We might want to be more accurate than this at one point.
-                            if let Some(filename) = changed_path.file_name() {
-                                if filename == *META_FILEPATH {
-                                    watcher_wrapper_clone.inner.watcher_router.broadcast();
-                                }
-                            }
-                        }
-                        Ok(None) => {
-                            // not an event we are interested in.
-                        }
-                        Err(_e) => {
-                            // the watch send channel was dropped
-                            break;
-                        }
-                    }
-                }
-            })
-            .expect("Failed to spawn thread to watch meta.json");
-        Ok(watcher_wrapper)
-    }
-
-    pub fn watch(&mut self, watch_callback: WatchCallback) -> WatchHandle {
-        self.inner.watcher_router.subscribe(watch_callback)
     }
 }
 
@@ -231,44 +150,21 @@ struct MmapDirectoryInner {
     root_path: PathBuf,
     mmap_cache: RwLock<MmapCache>,
     _temp_directory: Option<TempDir>,
-    watcher: RwLock<Option<WatcherWrapper>>,
+    watcher: FileWatcher,
 }
 
 impl MmapDirectoryInner {
-    fn new(
+    fn new(root_path: PathBuf, temp_directory: Option<TempDir>) -> MmapDirectoryInner {
-        root_path: PathBuf,
+        MmapDirectoryInner {
-        temp_directory: Option<TempDir>,
-    ) -> Result<MmapDirectoryInner, OpenDirectoryError> {
-        let mmap_directory_inner = MmapDirectoryInner {
-            root_path,
             mmap_cache: Default::default(),
             _temp_directory: temp_directory,
-            watcher: RwLock::new(None),
+            watcher: FileWatcher::new(&root_path.join(*META_FILEPATH)),
-        };
+            root_path,
-        Ok(mmap_directory_inner)
+        }
     }
 
-    fn watch(&self, watch_callback: WatchCallback) -> crate::Result<WatchHandle> {
+    fn watch(&self, callback: WatchCallback) -> WatchHandle {
-        // a lot of juggling here, to ensure we don't do anything that panics
+        self.watcher.watch(callback)
-        // while the rwlock is held. That way we ensure that the rwlock cannot
-        // be poisoned.
-        //
-        // The downside is that we might create a watch wrapper that is not useful.
-        let need_initialization = self.watcher.read().unwrap().is_none();
-        if need_initialization {
-            let watch_wrapper = WatcherWrapper::new(&self.root_path)?;
-            let mut watch_wlock = self.watcher.write().unwrap();
-            // the watcher could have been initialized when we released the lock, and
-            // we do not want to lose the watched files that were set.
-            if watch_wlock.is_none() {
-                *watch_wlock = Some(watch_wrapper);
-            }
-        }
-        if let Some(watch_wrapper) = self.watcher.write().unwrap().as_mut() {
-            return Ok(watch_wrapper.watch(watch_callback));
-        } else {
-            unreachable!("At this point, watch wrapper is supposed to be initialized");
-        }
     }
 }
 
@@ -279,24 +175,23 @@ impl fmt::Debug for MmapDirectory {
 }
 
 impl MmapDirectory {
-    fn new(
+    fn new(root_path: PathBuf, temp_directory: Option<TempDir>) -> MmapDirectory {
-        root_path: PathBuf,
+        let inner = MmapDirectoryInner::new(root_path, temp_directory);
-        temp_directory: Option<TempDir>,
+        MmapDirectory {
-    ) -> Result<MmapDirectory, OpenDirectoryError> {
-        let inner = MmapDirectoryInner::new(root_path, temp_directory)?;
-        Ok(MmapDirectory {
             inner: Arc::new(inner),
-        })
+        }
     }
 
     /// Creates a new MmapDirectory in a temporary directory.
     ///
     /// This is mostly useful to test the MmapDirectory itself.
-    /// For your unit tests, prefer the RAMDirectory.
+    /// For your unit tests, prefer the RamDirectory.
     pub fn create_from_tempdir() -> Result<MmapDirectory, OpenDirectoryError> {
-        let tempdir = TempDir::new("index").map_err(OpenDirectoryError::IoError)?;
+        let tempdir = TempDir::new().map_err(OpenDirectoryError::FailedToCreateTempDir)?;
-        let tempdir_path = PathBuf::from(tempdir.path());
+        Ok(MmapDirectory::new(
-        MmapDirectory::new(tempdir_path, Some(tempdir))
+            tempdir.path().to_path_buf(),
+            Some(tempdir),
+        ))
     }
 
     /// Opens a MmapDirectory in a directory.
@@ -314,7 +209,7 @@ impl MmapDirectory {
                 directory_path,
             )))
         } else {
-            Ok(MmapDirectory::new(PathBuf::from(directory_path), None)?)
+            Ok(MmapDirectory::new(PathBuf::from(directory_path), None))
         }
     }
 
@@ -412,8 +307,44 @@ impl Seek for SafeFileWriter {
     }
 }
 
+impl TerminatingWrite for SafeFileWriter {
+    fn terminate_ref(&mut self, _: AntiCallToken) -> io::Result<()> {
+        self.flush()
+    }
+}
+
+#[derive(Clone)]
+struct MmapArc(Arc<dyn Deref<Target = [u8]> + Send + Sync>);
+
+impl Deref for MmapArc {
+    type Target = [u8];
+
+    fn deref(&self) -> &[u8] {
+        self.0.deref()
+    }
+}
+unsafe impl StableDeref for MmapArc {}
+
+/// Writes a file in an atomic manner.
+pub(crate) fn atomic_write(path: &Path, content: &[u8]) -> io::Result<()> {
+    // We create the temporary file in the same directory as the target file.
+    // Indeed the canonical temp directory and the target file might sit in different
+    // filesystem, in which case the atomic write may actually not work.
+    let parent_path = path.parent().ok_or_else(|| {
+        io::Error::new(
+            io::ErrorKind::InvalidInput,
+            "Path {:?} does not have parent directory.",
+        )
+    })?;
+    let mut tempfile = tempfile::Builder::new().tempfile_in(&parent_path)?;
+    tempfile.write_all(content)?;
+    tempfile.flush()?;
+    tempfile.into_temp_path().persist(path)?;
+    Ok(())
+}
+
 impl Directory for MmapDirectory {
-    fn open_read(&self, path: &Path) -> result::Result<ReadOnlySource, OpenReadError> {
+    fn get_file_handle(&self, path: &Path) -> result::Result<Box<dyn FileHandle>, OpenReadError> {
         debug!("Open Read {:?}", path);
         let full_path = self.resolve_path(path);
 
@@ -423,12 +354,19 @@ impl Directory for MmapDirectory {
                  on mmap cache while reading {:?}",
                 path
             );
-            IOError::with_path(path.to_owned(), make_io_err(msg))
+            let io_err = make_io_err(msg);
+            OpenReadError::wrap_io_error(io_err, path.to_path_buf())
         })?;
-        Ok(mmap_cache
+
+        let owned_bytes = mmap_cache
             .get_mmap(&full_path)?
-            .map(ReadOnlySource::from)
+            .map(|mmap_arc| {
-            .unwrap_or_else(ReadOnlySource::empty))
+                let mmap_arc_obj = MmapArc(mmap_arc);
+                OwnedBytes::new(mmap_arc_obj)
+            })
+            .unwrap_or_else(OwnedBytes::empty);
+
+        Ok(Box::new(owned_bytes))
     }
 
     /// Any entry associated to the path in the mmap will be
@@ -436,25 +374,29 @@ impl Directory for MmapDirectory {
     fn delete(&self, path: &Path) -> result::Result<(), DeleteError> {
         let full_path = self.resolve_path(path);
         match fs::remove_file(&full_path) {
-            Ok(_) => self
+            Ok(_) => self.sync_directory().map_err(|e| DeleteError::IoError {
-                .sync_directory()
+                io_error: e,
-                .map_err(|e| IOError::with_path(path.to_owned(), e).into()),
+                filepath: path.to_path_buf(),
+            }),
             Err(e) => {
                 if e.kind() == io::ErrorKind::NotFound {
                     Err(DeleteError::FileDoesNotExist(path.to_owned()))
                 } else {
-                    Err(IOError::with_path(path.to_owned(), e).into())
+                    Err(DeleteError::IoError {
+                        io_error: e,
+                        filepath: path.to_path_buf(),
+                    })
                 }
             }
         }
     }
 
-    fn exists(&self, path: &Path) -> bool {
+    fn exists(&self, path: &Path) -> Result<bool, OpenReadError> {
         let full_path = self.resolve_path(path);
-        full_path.exists()
+        Ok(full_path.exists())
     }
 
-    fn open_write(&mut self, path: &Path) -> Result<WritePtr, OpenWriteError> {
+    fn open_write(&self, path: &Path) -> Result<WritePtr, OpenWriteError> {
         debug!("Open Write {:?}", path);
         let full_path = self.resolve_path(path);
 
@@ -463,22 +405,22 @@ impl Directory for MmapDirectory {
             .create_new(true)
             .open(full_path);
 
-        let mut file = open_res.map_err(|err| {
+        let mut file = open_res.map_err(|io_err| {
-            if err.kind() == io::ErrorKind::AlreadyExists {
+            if io_err.kind() == io::ErrorKind::AlreadyExists {
-                OpenWriteError::FileAlreadyExists(path.to_owned())
+                OpenWriteError::FileAlreadyExists(path.to_path_buf())
             } else {
-                IOError::with_path(path.to_owned(), err).into()
+                OpenWriteError::wrap_io_error(io_err, path.to_path_buf())
             }
         })?;
 
         // making sure the file is created.
         file.flush()
-            .map_err(|e| IOError::with_path(path.to_owned(), e))?;
+            .map_err(|io_error| OpenWriteError::wrap_io_error(io_error, path.to_path_buf()))?;
 
         // Apparetntly, on some filesystem syncing the parent
         // directory is required.
         self.sync_directory()
-            .map_err(|e| IOError::with_path(path.to_owned(), e))?;
+            .map_err(|io_err| OpenWriteError::wrap_io_error(io_err, path.to_path_buf()))?;
 
         let writer = SafeFileWriter::new(file);
         Ok(BufWriter::new(Box::new(writer)))
@@ -489,26 +431,26 @@ impl Directory for MmapDirectory {
         let mut buffer = Vec::new();
         match File::open(&full_path) {
             Ok(mut file) => {
-                file.read_to_end(&mut buffer)
+                file.read_to_end(&mut buffer).map_err(|io_error| {
-                    .map_err(|e| IOError::with_path(path.to_owned(), e))?;
+                    OpenReadError::wrap_io_error(io_error, path.to_path_buf())
+                })?;
                 Ok(buffer)
             }
-            Err(e) => {
+            Err(io_error) => {
-                if e.kind() == io::ErrorKind::NotFound {
+                if io_error.kind() == io::ErrorKind::NotFound {
                     Err(OpenReadError::FileDoesNotExist(path.to_owned()))
                 } else {
-                    Err(IOError::with_path(path.to_owned(), e).into())
+                    Err(OpenReadError::wrap_io_error(io_error, path.to_path_buf()))
                 }
             }
         }
     }
 
-    fn atomic_write(&mut self, path: &Path, data: &[u8]) -> io::Result<()> {
+    fn atomic_write(&self, path: &Path, content: &[u8]) -> io::Result<()> {
         debug!("Atomic Write {:?}", path);
         let full_path = self.resolve_path(path);
-        let meta_file = atomicwrites::AtomicFile::new(full_path, atomicwrites::AllowOverwrite);
+        atomic_write(&full_path, content)?;
-        meta_file.write(|f| f.write_all(data))?;
+        self.sync_directory()
-        Ok(())
     }
 
     fn acquire_lock(&self, lock: &Lock) -> Result<DirectoryLock, LockError> {
@@ -518,9 +460,9 @@ impl Directory for MmapDirectory {
             .write(true)
             .create(true) //< if the file does not exist yet, create it.
             .open(&full_path)
-            .map_err(LockError::IOError)?;
+            .map_err(LockError::IoError)?;
         if lock.is_blocking {
-            file.lock_exclusive().map_err(LockError::IOError)?;
+            file.lock_exclusive().map_err(LockError::IoError)?;
         } else {
             file.try_lock_exclusive().map_err(|_| LockError::LockBusy)?
         }
@@ -532,7 +474,7 @@ impl Directory for MmapDirectory {
     }
 
     fn watch(&self, watch_callback: WatchCallback) -> crate::Result<WatchHandle> {
-        self.inner.watch(watch_callback)
+        Ok(self.inner.watch(watch_callback))
     }
 }
 
@@ -543,16 +485,16 @@ mod tests {
     // The following tests are specific to the MmapDirectory
 
     use super::*;
-    use crate::schema::{Schema, SchemaBuilder, TEXT};
     use crate::Index;
     use crate::ReloadPolicy;
-    use std::fs;
+    use crate::{common::HasLen, indexer::LogMergePolicy};
-    use std::sync::atomic::{AtomicUsize, Ordering};
+    use crate::{
-    use std::thread;
+        schema::{Schema, SchemaBuilder, TEXT},
-    use std::time::Duration;
+        IndexSettings,
+    };
 
     #[test]
-    fn test_open_non_existant_path() {
+    fn test_open_non_existent_path() {
         assert!(MmapDirectory::open(PathBuf::from("./nowhere")).is_err());
     }
 
@@ -562,7 +504,7 @@ mod tests {
         // cannot be mmapped.
         //
         // In that case the directory returns a SharedVecSlice.
-        let mut mmap_directory = MmapDirectory::create_from_tempdir().unwrap();
+        let mmap_directory = MmapDirectory::create_from_tempdir().unwrap();
         let path = PathBuf::from("test");
         {
             let mut w = mmap_directory.open_write(&path).unwrap();
@@ -578,7 +520,7 @@ mod tests {
 
         // here we test if the cache releases
         // mmaps correctly.
-        let mut mmap_directory = MmapDirectory::create_from_tempdir().unwrap();
+        let mmap_directory = MmapDirectory::create_from_tempdir().unwrap();
         let num_paths = 10;
         let paths: Vec<PathBuf> = (0..num_paths)
             .map(|i| PathBuf::from(&*format!("file_{}", i)))
@@ -638,56 +580,51 @@ mod tests {
         assert_eq!(mmap_directory.get_cache_info().mmapped.len(), 0);
     }
 
-    #[test]
-    fn test_watch_wrapper() {
-        let counter: Arc<AtomicUsize> = Default::default();
-        let counter_clone = counter.clone();
-        let tmp_dir: TempDir = tempdir::TempDir::new("test_watch_wrapper").unwrap();
-        let tmp_dirpath = tmp_dir.path().to_owned();
-        let mut watch_wrapper = WatcherWrapper::new(&tmp_dirpath).unwrap();
-        let tmp_file = tmp_dirpath.join("coucou");
-        let _handle = watch_wrapper.watch(Box::new(move || {
-            counter_clone.fetch_add(1, Ordering::SeqCst);
-        }));
-        assert_eq!(counter.load(Ordering::SeqCst), 0);
-        fs::write(&tmp_file, b"whateverwilldo").unwrap();
-        thread::sleep(Duration::new(0, 1_000u32));
-    }
-
     #[test]
     fn test_mmap_released() {
         let mmap_directory = MmapDirectory::create_from_tempdir().unwrap();
         let mut schema_builder: SchemaBuilder = Schema::builder();
         let text_field = schema_builder.add_text_field("text", TEXT);
         let schema = schema_builder.build();
+
         {
-            let index = Index::create(mmap_directory.clone(), schema).unwrap();
+            let index =
-            let mut index_writer = index.writer_with_num_threads(1, 3_000_000).unwrap();
+                Index::create(mmap_directory.clone(), schema, IndexSettings::default()).unwrap();
-            for _num_commits in 0..16 {
+
+            let mut index_writer = index.writer_for_tests().unwrap();
+            let mut log_merge_policy = LogMergePolicy::default();
+            log_merge_policy.set_min_num_segments(3);
+            index_writer.set_merge_policy(Box::new(log_merge_policy));
+            for _num_commits in 0..10 {
                 for _ in 0..10 {
                     index_writer.add_document(doc!(text_field=>"abc"));
                 }
                 index_writer.commit().unwrap();
             }
+
             let reader = index
                 .reader_builder()
                 .reload_policy(ReloadPolicy::Manual)
                 .try_into()
                 .unwrap();
-            for _ in 0..30 {
+
+            for _ in 0..4 {
                 index_writer.add_document(doc!(text_field=>"abc"));
                 index_writer.commit().unwrap();
                 reader.reload().unwrap();
             }
             index_writer.wait_merging_threads().unwrap();
+
             reader.reload().unwrap();
             let num_segments = reader.searcher().segment_readers().len();
-            assert_eq!(num_segments, 4);
+            assert!(num_segments <= 4);
+            let num_components_except_deletes_and_tempstore =
+                crate::core::SegmentComponent::iterator().len() - 2;
             assert_eq!(
-                num_segments * 7,
+                num_segments * num_components_except_deletes_and_tempstore,
                 mmap_directory.get_cache_info().mmapped.len()
             );
         }
-        assert_eq!(mmap_directory.get_cache_info().mmapped.len(), 0);
+        assert!(mmap_directory.get_cache_info().mmapped.is_empty());
     }
 }

src/directory/mod.rs

@@ -9,9 +9,12 @@ mod mmap_directory;
 
 mod directory;
 mod directory_lock;
+mod file_slice;
+mod file_watcher;
+mod footer;
 mod managed_directory;
+mod owned_bytes;
 mod ram_directory;
-mod read_only_source;
 mod watch_event_router;
 
 /// Errors specific to the directory module.
@@ -20,22 +23,79 @@ pub mod error;
 pub use self::directory::DirectoryLock;
 pub use self::directory::{Directory, DirectoryClone};
 pub use self::directory_lock::{Lock, INDEX_WRITER_LOCK, META_LOCK};
-pub use self::ram_directory::RAMDirectory;
+pub(crate) use self::file_slice::{ArcBytes, WeakArcBytes};
-pub use self::read_only_source::ReadOnlySource;
+pub use self::file_slice::{FileHandle, FileSlice};
-pub(crate) use self::watch_event_router::WatchCallbackList;
+pub use self::owned_bytes::OwnedBytes;
-pub use self::watch_event_router::{WatchCallback, WatchHandle};
+pub use self::ram_directory::RamDirectory;
-use std::io::{BufWriter, Write};
+pub use self::watch_event_router::{WatchCallback, WatchCallbackList, WatchHandle};
+use std::io::{self, BufWriter, Write};
+use std::path::PathBuf;
+
+/// Outcome of the Garbage collection
+pub struct GarbageCollectionResult {
+    /// List of files that were deleted in this cycle
+    pub deleted_files: Vec<PathBuf>,
+    /// List of files that were schedule to be deleted in this cycle,
+    /// but deletion did not work. This typically happens on windows,
+    /// as deleting a memory mapped file is forbidden.
+    ///
+    /// If a searcher is still held, a file cannot be deleted.
+    /// This is not considered a bug, the file will simply be deleted
+    /// in the next GC.
+    pub failed_to_delete_files: Vec<PathBuf>,
+}
 
 #[cfg(feature = "mmap")]
 pub use self::mmap_directory::MmapDirectory;
 
 pub use self::managed_directory::ManagedDirectory;
 
+/// Struct used to prevent from calling [`terminate_ref`](trait.TerminatingWrite#method.terminate_ref) directly
+///
+/// The point is that while the type is public, it cannot be built by anyone
+/// outside of this module.
+pub struct AntiCallToken(());
+
+/// Trait used to indicate when no more write need to be done on a writer
+pub trait TerminatingWrite: Write {
+    /// Indicate that the writer will no longer be used. Internally call terminate_ref.
+    fn terminate(mut self) -> io::Result<()>
+    where
+        Self: Sized,
+    {
+        self.terminate_ref(AntiCallToken(()))
+    }
+
+    /// You should implement this function to define custom behavior.
+    /// This function should flush any buffer it may hold.
+    fn terminate_ref(&mut self, _: AntiCallToken) -> io::Result<()>;
+}
+
+impl<W: TerminatingWrite + ?Sized> TerminatingWrite for Box<W> {
+    fn terminate_ref(&mut self, token: AntiCallToken) -> io::Result<()> {
+        self.as_mut().terminate_ref(token)
+    }
+}
+
+impl<W: TerminatingWrite> TerminatingWrite for BufWriter<W> {
+    fn terminate_ref(&mut self, a: AntiCallToken) -> io::Result<()> {
+        self.flush()?;
+        self.get_mut().terminate_ref(a)
+    }
+}
+
+#[cfg(test)]
+impl<'a> TerminatingWrite for &'a mut Vec<u8> {
+    fn terminate_ref(&mut self, _a: AntiCallToken) -> io::Result<()> {
+        self.flush()
+    }
+}
+
 /// Write object for Directory.
 ///
 /// `WritePtr` are required to implement both Write
 /// and Seek.
-pub type WritePtr = BufWriter<Box<dyn Write>>;
+pub type WritePtr = BufWriter<Box<dyn TerminatingWrite>>;
 
 #[cfg(test)]
 mod tests;

src/directory/owned_bytes.rs

@@ -0,0 +1,290 @@
+use crate::directory::FileHandle;
+use stable_deref_trait::StableDeref;
+use std::convert::TryInto;
+use std::mem;
+use std::ops::{Deref, Range};
+use std::sync::Arc;
+use std::{fmt, io};
+
+/// An OwnedBytes simply wraps an object that owns a slice of data and exposes
+/// this data as a static slice.
+///
+/// The backing object is required to be `StableDeref`.
+#[derive(Clone)]
+pub struct OwnedBytes {
+    data: &'static [u8],
+    box_stable_deref: Arc<dyn Deref<Target = [u8]> + Sync + Send>,
+}
+
+impl FileHandle for OwnedBytes {
+    fn read_bytes(&self, range: Range<usize>) -> io::Result<OwnedBytes> {
+        Ok(self.slice(range))
+    }
+}
+
+impl OwnedBytes {
+    /// Creates an empty `OwnedBytes`.
+    pub fn empty() -> OwnedBytes {
+        OwnedBytes::new(&[][..])
+    }
+
+    /// Creates an `OwnedBytes` intance given a `StableDeref` object.
+    pub fn new<T: StableDeref + Deref<Target = [u8]> + 'static + Send + Sync>(
+        data_holder: T,
+    ) -> OwnedBytes {
+        let box_stable_deref = Arc::new(data_holder);
+        let bytes: &[u8] = box_stable_deref.as_ref();
+        let data = unsafe { mem::transmute::<_, &'static [u8]>(bytes.deref()) };
+        OwnedBytes {
+            data,
+            box_stable_deref,
+        }
+    }
+
+    /// creates a fileslice that is just a view over a slice of the data.
+    pub fn slice(&self, range: Range<usize>) -> Self {
+        OwnedBytes {
+            data: &self.data[range],
+            box_stable_deref: self.box_stable_deref.clone(),
+        }
+    }
+
+    /// Returns the underlying slice of data.
+    /// `Deref` and `AsRef` are also available.
+    #[inline]
+    pub fn as_slice(&self) -> &[u8] {
+        self.data
+    }
+
+    /// Returns the len of the slice.
+    #[inline]
+    pub fn len(&self) -> usize {
+        self.data.len()
+    }
+
+    /// Splits the OwnedBytes into two OwnedBytes `(left, right)`.
+    ///
+    /// Left will hold `split_len` bytes.
+    ///
+    /// This operation is cheap and does not require to copy any memory.
+    /// On the other hand, both `left` and `right` retain a handle over
+    /// the entire slice of memory. In other words, the memory will only
+    /// be released when both left and right are dropped.
+    pub fn split(self, split_len: usize) -> (OwnedBytes, OwnedBytes) {
+        let right_box_stable_deref = self.box_stable_deref.clone();
+        let left = OwnedBytes {
+            data: &self.data[..split_len],
+            box_stable_deref: self.box_stable_deref,
+        };
+        let right = OwnedBytes {
+            data: &self.data[split_len..],
+            box_stable_deref: right_box_stable_deref,
+        };
+        (left, right)
+    }
+
+    /// Returns true iff this `OwnedBytes` is empty.
+    #[inline]
+    pub fn is_empty(&self) -> bool {
+        self.as_slice().is_empty()
+    }
+
+    /// Drops the left most `advance_len` bytes.
+    ///
+    /// See also [.clip(clip_len: usize))](#method.clip).
+    #[inline]
+    pub fn advance(&mut self, advance_len: usize) {
+        self.data = &self.data[advance_len..]
+    }
+
+    /// Reads an `u8` from the `OwnedBytes` and advance by one byte.
+    pub fn read_u8(&mut self) -> u8 {
+        assert!(!self.is_empty());
+
+        let byte = self.as_slice()[0];
+        self.advance(1);
+        byte
+    }
+
+    /// Reads an `u64` encoded as little-endian from the `OwnedBytes` and advance by 8 bytes.
+    pub fn read_u64(&mut self) -> u64 {
+        assert!(self.len() > 7);
+
+        let octlet: [u8; 8] = self.as_slice()[..8].try_into().unwrap();
+        self.advance(8);
+        u64::from_le_bytes(octlet)
+    }
+}
+
+impl fmt::Debug for OwnedBytes {
+    fn fmt(&self, f: &mut fmt::Formatter) -> fmt::Result {
+        // We truncate the bytes in order to make sure the debug string
+        // is not too long.
+        let bytes_truncated: &[u8] = if self.len() > 8 {
+            &self.as_slice()[..10]
+        } else {
+            self.as_slice()
+        };
+        write!(f, "OwnedBytes({:?}, len={})", bytes_truncated, self.len())
+    }
+}
+
+impl Deref for OwnedBytes {
+    type Target = [u8];
+
+    fn deref(&self) -> &Self::Target {
+        self.as_slice()
+    }
+}
+
+impl io::Read for OwnedBytes {
+    fn read(&mut self, buf: &mut [u8]) -> io::Result<usize> {
+        let read_len = {
+            let data = self.as_slice();
+            if data.len() >= buf.len() {
+                let buf_len = buf.len();
+                buf.copy_from_slice(&data[..buf_len]);
+                buf.len()
+            } else {
+                let data_len = data.len();
+                buf[..data_len].copy_from_slice(data);
+                data_len
+            }
+        };
+        self.advance(read_len);
+        Ok(read_len)
+    }
+    fn read_to_end(&mut self, buf: &mut Vec<u8>) -> io::Result<usize> {
+        let read_len = {
+            let data = self.as_slice();
+            buf.extend(data);
+            data.len()
+        };
+        self.advance(read_len);
+        Ok(read_len)
+    }
+    fn read_exact(&mut self, buf: &mut [u8]) -> io::Result<()> {
+        let read_len = self.read(buf)?;
+        if read_len != buf.len() {
+            return Err(io::Error::new(
+                io::ErrorKind::UnexpectedEof,
+                "failed to fill whole buffer",
+            ));
+        }
+        Ok(())
+    }
+}
+
+impl AsRef<[u8]> for OwnedBytes {
+    fn as_ref(&self) -> &[u8] {
+        self.as_slice()
+    }
+}
+
+#[cfg(test)]
+mod tests {
+    use std::io::{self, Read};
+
+    use super::OwnedBytes;
+
+    #[test]
+    fn test_owned_bytes_debug() {
+        let short_bytes = OwnedBytes::new(b"abcd".as_ref());
+        assert_eq!(
+            format!("{:?}", short_bytes),
+            "OwnedBytes([97, 98, 99, 100], len=4)"
+        );
+        let long_bytes = OwnedBytes::new(b"abcdefghijklmnopq".as_ref());
+        assert_eq!(
+            format!("{:?}", long_bytes),
+            "OwnedBytes([97, 98, 99, 100, 101, 102, 103, 104, 105, 106], len=17)"
+        );
+    }
+
+    #[test]
+    fn test_owned_bytes_read() -> io::Result<()> {
+        let mut bytes = OwnedBytes::new(b"abcdefghiklmnopqrstuvwxyz".as_ref());
+        {
+            let mut buf = [0u8; 5];
+            bytes.read_exact(&mut buf[..]).unwrap();
+            assert_eq!(&buf, b"abcde");
+            assert_eq!(bytes.as_slice(), b"fghiklmnopqrstuvwxyz")
+        }
+        {
+            let mut buf = [0u8; 2];
+            bytes.read_exact(&mut buf[..]).unwrap();
+            assert_eq!(&buf, b"fg");
+            assert_eq!(bytes.as_slice(), b"hiklmnopqrstuvwxyz")
+        }
+        Ok(())
+    }
+
+    #[test]
+    fn test_owned_bytes_read_right_at_the_end() -> io::Result<()> {
+        let mut bytes = OwnedBytes::new(b"abcde".as_ref());
+        let mut buf = [0u8; 5];
+        assert_eq!(bytes.read(&mut buf[..]).unwrap(), 5);
+        assert_eq!(&buf, b"abcde");
+        assert_eq!(bytes.as_slice(), b"");
+        assert_eq!(bytes.read(&mut buf[..]).unwrap(), 0);
+        assert_eq!(&buf, b"abcde");
+        Ok(())
+    }
+    #[test]
+    fn test_owned_bytes_read_incomplete() -> io::Result<()> {
+        let mut bytes = OwnedBytes::new(b"abcde".as_ref());
+        let mut buf = [0u8; 7];
+        assert_eq!(bytes.read(&mut buf[..]).unwrap(), 5);
+        assert_eq!(&buf[..5], b"abcde");
+        assert_eq!(bytes.read(&mut buf[..]).unwrap(), 0);
+        Ok(())
+    }
+
+    #[test]
+    fn test_owned_bytes_read_to_end() -> io::Result<()> {
+        let mut bytes = OwnedBytes::new(b"abcde".as_ref());
+        let mut buf = Vec::new();
+        bytes.read_to_end(&mut buf)?;
+        assert_eq!(buf.as_slice(), b"abcde".as_ref());
+        Ok(())
+    }
+
+    #[test]
+    fn test_owned_bytes_read_u8() -> io::Result<()> {
+        let mut bytes = OwnedBytes::new(b"\xFF".as_ref());
+        assert_eq!(bytes.read_u8(), 255);
+        assert_eq!(bytes.len(), 0);
+        Ok(())
+    }
+
+    #[test]
+    fn test_owned_bytes_read_u64() -> io::Result<()> {
+        let mut bytes = OwnedBytes::new(b"\0\xFF\xFF\xFF\xFF\xFF\xFF\xFF".as_ref());
+        assert_eq!(bytes.read_u64(), u64::MAX - 255);
+        assert_eq!(bytes.len(), 0);
+        Ok(())
+    }
+
+    #[test]
+    fn test_owned_bytes_split() {
+        let bytes = OwnedBytes::new(b"abcdefghi".as_ref());
+        let (left, right) = bytes.split(3);
+        assert_eq!(left.as_slice(), b"abc");
+        assert_eq!(right.as_slice(), b"defghi");
+    }
+
+    #[test]
+    fn test_owned_bytes_split_boundary() {
+        let bytes = OwnedBytes::new(b"abcdefghi".as_ref());
+        {
+            let (left, right) = bytes.clone().split(0);
+            assert_eq!(left.as_slice(), b"");
+            assert_eq!(right.as_slice(), b"abcdefghi");
+        }
+        {
+            let (left, right) = bytes.split(9);
+            assert_eq!(left.as_slice(), b"abcdefghi");
+            assert_eq!(right.as_slice(), b"");
+        }
+    }
+}

src/directory/ram_directory.rs

@@ -1,8 +1,9 @@
-use crate::core::META_FILEPATH;
 use crate::directory::error::{DeleteError, OpenReadError, OpenWriteError};
+use crate::directory::AntiCallToken;
 use crate::directory::WatchCallbackList;
-use crate::directory::WritePtr;
+use crate::directory::{Directory, FileSlice, WatchCallback, WatchHandle};
-use crate::directory::{Directory, ReadOnlySource, WatchCallback, WatchHandle};
+use crate::directory::{TerminatingWrite, WritePtr};
+use crate::{common::HasLen, core::META_FILEPATH};
 use fail::fail_point;
 use std::collections::HashMap;
 use std::fmt;
@@ -11,7 +12,9 @@ use std::path::{Path, PathBuf};
 use std::result;
 use std::sync::{Arc, RwLock};
 
-/// Writer associated with the `RAMDirectory`
+use super::FileHandle;
+
+/// Writer associated with the `RamDirectory`
 ///
 /// The Writer just writes a buffer.
 ///
@@ -23,13 +26,13 @@ use std::sync::{Arc, RwLock};
 ///
 struct VecWriter {
     path: PathBuf,
-    shared_directory: RAMDirectory,
+    shared_directory: RamDirectory,
     data: Cursor<Vec<u8>>,
     is_flushed: bool,
 }
 
 impl VecWriter {
-    fn new(path_buf: PathBuf, shared_directory: RAMDirectory) -> VecWriter {
+    fn new(path_buf: PathBuf, shared_directory: RamDirectory) -> VecWriter {
         VecWriter {
             path: path_buf,
             data: Cursor::new(Vec::new()),
@@ -71,19 +74,25 @@ impl Write for VecWriter {
     }
 }
 
+impl TerminatingWrite for VecWriter {
+    fn terminate_ref(&mut self, _: AntiCallToken) -> io::Result<()> {
+        self.flush()
+    }
+}
+
 #[derive(Default)]
 struct InnerDirectory {
-    fs: HashMap<PathBuf, ReadOnlySource>,
+    fs: HashMap<PathBuf, FileSlice>,
     watch_router: WatchCallbackList,
 }
 
 impl InnerDirectory {
     fn write(&mut self, path: PathBuf, data: &[u8]) -> bool {
-        let data = ReadOnlySource::new(Vec::from(data));
+        let data = FileSlice::from(data.to_vec());
         self.fs.insert(path, data).is_some()
     }
 
-    fn open_read(&self, path: &Path) -> Result<ReadOnlySource, OpenReadError> {
+    fn open_read(&self, path: &Path) -> Result<FileSlice, OpenReadError> {
         self.fs
             .get(path)
             .ok_or_else(|| OpenReadError::FileDoesNotExist(PathBuf::from(path)))
@@ -110,9 +119,9 @@ impl InnerDirectory {
     }
 }
 
-impl fmt::Debug for RAMDirectory {
+impl fmt::Debug for RamDirectory {
     fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
-        write!(f, "RAMDirectory")
+        write!(f, "RamDirectory")
     }
 }
 
@@ -122,42 +131,71 @@ impl fmt::Debug for RAMDirectory {
 /// Writes are only made visible upon flushing.
 ///
 #[derive(Clone, Default)]
-pub struct RAMDirectory {
+pub struct RamDirectory {
     fs: Arc<RwLock<InnerDirectory>>,
 }
 
-impl RAMDirectory {
+impl RamDirectory {
     /// Constructor
-    pub fn create() -> RAMDirectory {
+    pub fn create() -> RamDirectory {
         Self::default()
     }
 
     /// Returns the sum of the size of the different files
-    /// in the RAMDirectory.
+    /// in the RamDirectory.
     pub fn total_mem_usage(&self) -> usize {
         self.fs.read().unwrap().total_mem_usage()
     }
+
+    /// Write a copy of all of the files saved in the RamDirectory in the target `Directory`.
+    ///
+    /// Files are all written using the `Directory::write` meaning, even if they were
+    /// written using the `atomic_write` api.
+    ///
+    /// If an error is encounterred, files may be persisted partially.
+    pub fn persist(&self, dest: &dyn Directory) -> crate::Result<()> {
+        let wlock = self.fs.write().unwrap();
+        for (path, file) in wlock.fs.iter() {
+            let mut dest_wrt = dest.open_write(path)?;
+            dest_wrt.write_all(file.read_bytes()?.as_slice())?;
+            dest_wrt.terminate()?;
+        }
+        Ok(())
+    }
 }
 
-impl Directory for RAMDirectory {
+impl Directory for RamDirectory {
-    fn open_read(&self, path: &Path) -> result::Result<ReadOnlySource, OpenReadError> {
+    fn get_file_handle(&self, path: &Path) -> Result<Box<dyn FileHandle>, OpenReadError> {
+        let file_slice = self.open_read(path)?;
+        Ok(Box::new(file_slice))
+    }
+
+    fn open_read(&self, path: &Path) -> result::Result<FileSlice, OpenReadError> {
         self.fs.read().unwrap().open_read(path)
     }
 
     fn delete(&self, path: &Path) -> result::Result<(), DeleteError> {
-        fail_point!("RAMDirectory::delete", |_| {
+        fail_point!("RamDirectory::delete", |_| {
-            use crate::directory::error::IOError;
+            Err(DeleteError::IoError {
-            let io_error = IOError::from(io::Error::from(io::ErrorKind::Other));
+                io_error: io::Error::from(io::ErrorKind::Other),
-            Err(DeleteError::from(io_error))
+                filepath: path.to_path_buf(),
+            })
         });
         self.fs.write().unwrap().delete(path)
     }
 
-    fn exists(&self, path: &Path) -> bool {
+    fn exists(&self, path: &Path) -> Result<bool, OpenReadError> {
-        self.fs.read().unwrap().exists(path)
+        Ok(self
+            .fs
+            .read()
+            .map_err(|e| OpenReadError::IoError {
+                io_error: io::Error::new(io::ErrorKind::Other, e.to_string()),
+                filepath: path.to_path_buf(),
+            })?
+            .exists(path))
     }
 
-    fn open_write(&mut self, path: &Path) -> Result<WritePtr, OpenWriteError> {
+    fn open_write(&self, path: &Path) -> Result<WritePtr, OpenWriteError> {
         let mut fs = self.fs.write().unwrap();
         let path_buf = PathBuf::from(path);
         let vec_writer = VecWriter::new(path_buf.clone(), self.clone());
@@ -171,24 +209,27 @@ impl Directory for RAMDirectory {
     }
 
     fn atomic_read(&self, path: &Path) -> Result<Vec<u8>, OpenReadError> {
-        Ok(self.open_read(path)?.as_slice().to_owned())
+        let bytes =
+            self.open_read(path)?
+                .read_bytes()
+                .map_err(|io_error| OpenReadError::IoError {
+                    io_error,
+                    filepath: path.to_path_buf(),
+                })?;
+        Ok(bytes.as_slice().to_owned())
     }
 
-    fn atomic_write(&mut self, path: &Path, data: &[u8]) -> io::Result<()> {
+    fn atomic_write(&self, path: &Path, data: &[u8]) -> io::Result<()> {
-        fail_point!("RAMDirectory::atomic_write", |msg| Err(io::Error::new(
+        fail_point!("RamDirectory::atomic_write", |msg| Err(io::Error::new(
             io::ErrorKind::Other,
-            msg.unwrap_or("Undefined".to_string())
+            msg.unwrap_or_else(|| "Undefined".to_string())
         )));
         let path_buf = PathBuf::from(path);
 
-        // Reserve the path to prevent calls to .write() to succeed.
+        self.fs.write().unwrap().write(path_buf, data);
-        self.fs.write().unwrap().write(path_buf.clone(), &[]);
 
-        let mut vec_writer = VecWriter::new(path_buf.clone(), self.clone());
+        if path == *META_FILEPATH {
-        vec_writer.write_all(data)?;
+            let _ = self.fs.write().unwrap().watch_router.broadcast();
-        vec_writer.flush()?;
-        if path == Path::new(&*META_FILEPATH) {
-            self.fs.write().unwrap().watch_router.broadcast();
         }
         Ok(())
     }
@@ -197,3 +238,28 @@ impl Directory for RAMDirectory {
         Ok(self.fs.write().unwrap().watch(watch_callback))
     }
 }
+
+#[cfg(test)]
+mod tests {
+    use super::RamDirectory;
+    use crate::Directory;
+    use std::io::Write;
+    use std::path::Path;
+
+    #[test]
+    fn test_persist() {
+        let msg_atomic: &'static [u8] = b"atomic is the way";
+        let msg_seq: &'static [u8] = b"sequential is the way";
+        let path_atomic: &'static Path = Path::new("atomic");
+        let path_seq: &'static Path = Path::new("seq");
+        let directory = RamDirectory::create();
+        assert!(directory.atomic_write(path_atomic, msg_atomic).is_ok());
+        let mut wrt = directory.open_write(path_seq).unwrap();
+        assert!(wrt.write_all(msg_seq).is_ok());
+        assert!(wrt.flush().is_ok());
+        let directory_copy = RamDirectory::create();
+        assert!(directory.persist(&directory_copy).is_ok());
+        assert_eq!(directory_copy.atomic_read(path_atomic).unwrap(), msg_atomic);
+        assert_eq!(directory_copy.atomic_read(path_seq).unwrap(), msg_seq);
+    }
+}

src/directory/read_only_source.rs

@@ -1,131 +0,0 @@
-use crate::common::HasLen;
-use stable_deref_trait::{CloneStableDeref, StableDeref};
-use std::ops::Deref;
-use std::sync::Arc;
-
-pub type BoxedData = Box<dyn Deref<Target = [u8]> + Send + Sync + 'static>;
-
-/// Read object that represents files in tantivy.
-///
-/// These read objects are only in charge to deliver
-/// the data in the form of a constant read-only `&[u8]`.
-/// Whatever happens to the directory file, the data
-/// hold by this object should never be altered or destroyed.
-pub struct ReadOnlySource {
-    data: Arc<BoxedData>,
-    start: usize,
-    stop: usize,
-}
-
-unsafe impl StableDeref for ReadOnlySource {}
-unsafe impl CloneStableDeref for ReadOnlySource {}
-
-impl Deref for ReadOnlySource {
-    type Target = [u8];
-
-    fn deref(&self) -> &[u8] {
-        self.as_slice()
-    }
-}
-
-impl From<Arc<BoxedData>> for ReadOnlySource {
-    fn from(data: Arc<BoxedData>) -> Self {
-        let len = data.len();
-        ReadOnlySource {
-            data,
-            start: 0,
-            stop: len,
-        }
-    }
-}
-
-impl ReadOnlySource {
-    pub(crate) fn new<D>(data: D) -> ReadOnlySource
-    where
-        D: Deref<Target = [u8]> + Send + Sync + 'static,
-    {
-        let len = data.len();
-        ReadOnlySource {
-            data: Arc::new(Box::new(data)),
-            start: 0,
-            stop: len,
-        }
-    }
-
-    /// Creates an empty ReadOnlySource
-    pub fn empty() -> ReadOnlySource {
-        ReadOnlySource::new(&[][..])
-    }
-
-    /// Returns the data underlying the ReadOnlySource object.
-    pub fn as_slice(&self) -> &[u8] {
-        &self.data[self.start..self.stop]
-    }
-
-    /// Splits into 2 `ReadOnlySource`, at the offset given
-    /// as an argument.
-    pub fn split(self, addr: usize) -> (ReadOnlySource, ReadOnlySource) {
-        let left = self.slice(0, addr);
-        let right = self.slice_from(addr);
-        (left, right)
-    }
-
-    /// Creates a ReadOnlySource that is just a
-    /// view over a slice of the data.
-    ///
-    /// Keep in mind that any living slice extends
-    /// the lifetime of the original ReadOnlySource,
-    ///
-    /// For instance, if `ReadOnlySource` wraps 500MB
-    /// worth of data in anonymous memory, and only a
-    /// 1KB slice is remaining, the whole `500MBs`
-    /// are retained in memory.
-    pub fn slice(&self, start: usize, stop: usize) -> ReadOnlySource {
-        assert!(
-            start <= stop,
-            "Requested negative slice [{}..{}]",
-            start,
-            stop
-        );
-        assert!(stop <= self.len());
-        ReadOnlySource {
-            data: self.data.clone(),
-            start: self.start + start,
-            stop: self.start + stop,
-        }
-    }
-
-    /// Like `.slice(...)` but enforcing only the `from`
-    /// boundary.
-    ///
-    /// Equivalent to `.slice(from_offset, self.len())`
-    pub fn slice_from(&self, from_offset: usize) -> ReadOnlySource {
-        self.slice(from_offset, self.len())
-    }
-
-    /// Like `.slice(...)` but enforcing only the `to`
-    /// boundary.
-    ///
-    /// Equivalent to `.slice(0, to_offset)`
-    pub fn slice_to(&self, to_offset: usize) -> ReadOnlySource {
-        self.slice(0, to_offset)
-    }
-}
-
-impl HasLen for ReadOnlySource {
-    fn len(&self) -> usize {
-        self.stop - self.start
-    }
-}
-
-impl Clone for ReadOnlySource {
-    fn clone(&self) -> Self {
-        self.slice_from(0)
-    }
-}
-
-impl From<Vec<u8>> for ReadOnlySource {
-    fn from(data: Vec<u8>) -> ReadOnlySource {
-        ReadOnlySource::new(data)
-    }
-}

src/directory/tests.rs

@@ -1,93 +1,182 @@
 use super::*;
+use futures::channel::oneshot;
+use futures::executor::block_on;
 use std::io::Write;
 use std::mem;
 use std::path::{Path, PathBuf};
-use std::sync::atomic::AtomicUsize;
+use std::sync::atomic::Ordering::SeqCst;
-use std::sync::atomic::Ordering;
+use std::sync::atomic::{AtomicBool, AtomicUsize};
 use std::sync::Arc;
-use std::thread;
-use std::time;
 use std::time::Duration;
 
-#[test]
+#[cfg(feature = "mmap")]
-fn test_ram_directory() {
+mod mmap_directory_tests {
-    let mut ram_directory = RAMDirectory::create();
+    use crate::directory::MmapDirectory;
-    test_directory(&mut ram_directory);
+
+    type DirectoryImpl = MmapDirectory;
+
+    fn make_directory() -> DirectoryImpl {
+        MmapDirectory::create_from_tempdir().unwrap()
+    }
+
+    #[test]
+    fn test_simple() -> crate::Result<()> {
+        let directory = make_directory();
+        super::test_simple(&directory)
+    }
+
+    #[test]
+    fn test_write_create_the_file() {
+        let directory = make_directory();
+        super::test_write_create_the_file(&directory);
+    }
+
+    #[test]
+    fn test_rewrite_forbidden() -> crate::Result<()> {
+        let directory = make_directory();
+        super::test_rewrite_forbidden(&directory)?;
+        Ok(())
+    }
+
+    #[test]
+    fn test_directory_delete() -> crate::Result<()> {
+        let directory = make_directory();
+        super::test_directory_delete(&directory)?;
+        Ok(())
+    }
+
+    #[test]
+    fn test_lock_non_blocking() {
+        let directory = make_directory();
+        super::test_lock_non_blocking(&directory);
+    }
+
+    #[test]
+    fn test_lock_blocking() {
+        let directory = make_directory();
+        super::test_lock_blocking(&directory);
+    }
+
+    #[test]
+    fn test_watch() {
+        let directory = make_directory();
+        super::test_watch(&directory);
+    }
 }
 
-#[test]
+mod ram_directory_tests {
-#[cfg(feature = "mmap")]
+    use crate::directory::RamDirectory;
-fn test_mmap_directory() {
+
-    let mut mmap_directory = MmapDirectory::create_from_tempdir().unwrap();
+    type DirectoryImpl = RamDirectory;
-    test_directory(&mut mmap_directory);
+
+    fn make_directory() -> DirectoryImpl {
+        RamDirectory::default()
+    }
+
+    #[test]
+    fn test_simple() -> crate::Result<()> {
+        let directory = make_directory();
+        super::test_simple(&directory)
+    }
+
+    #[test]
+    fn test_write_create_the_file() {
+        let directory = make_directory();
+        super::test_write_create_the_file(&directory);
+    }
+
+    #[test]
+    fn test_rewrite_forbidden() -> crate::Result<()> {
+        let directory = make_directory();
+        super::test_rewrite_forbidden(&directory)?;
+        Ok(())
+    }
+
+    #[test]
+    fn test_directory_delete() -> crate::Result<()> {
+        let directory = make_directory();
+        super::test_directory_delete(&directory)?;
+        Ok(())
+    }
+
+    #[test]
+    fn test_lock_non_blocking() {
+        let directory = make_directory();
+        super::test_lock_non_blocking(&directory);
+    }
+
+    #[test]
+    fn test_lock_blocking() {
+        let directory = make_directory();
+        super::test_lock_blocking(&directory);
+    }
+
+    #[test]
+    fn test_watch() {
+        let directory = make_directory();
+        super::test_watch(&directory);
+    }
 }
 
 #[test]
 #[should_panic]
 fn ram_directory_panics_if_flush_forgotten() {
     let test_path: &'static Path = Path::new("some_path_for_test");
-    let mut ram_directory = RAMDirectory::create();
+    let ram_directory = RamDirectory::create();
     let mut write_file = ram_directory.open_write(test_path).unwrap();
     assert!(write_file.write_all(&[4]).is_ok());
 }
 
-fn test_simple(directory: &mut dyn Directory) {
+fn test_simple(directory: &dyn Directory) -> crate::Result<()> {
     let test_path: &'static Path = Path::new("some_path_for_test");
-    {
+    let mut write_file = directory.open_write(test_path)?;
-        let mut write_file = directory.open_write(test_path).unwrap();
+    assert!(directory.exists(test_path).unwrap());
-        assert!(directory.exists(test_path));
+    write_file.write_all(&[4])?;
-        write_file.write_all(&[4]).unwrap();
+    write_file.write_all(&[3])?;
-        write_file.write_all(&[3]).unwrap();
+    write_file.write_all(&[7, 3, 5])?;
-        write_file.write_all(&[7, 3, 5]).unwrap();
+    write_file.flush()?;
-        write_file.flush().unwrap();
+    let read_file = directory.open_read(test_path)?.read_bytes()?;
-    }
+    assert_eq!(read_file.as_slice(), &[4u8, 3u8, 7u8, 3u8, 5u8]);
-    {
+    mem::drop(read_file);
-        let read_file = directory.open_read(test_path).unwrap();
-        let data: &[u8] = &*read_file;
-        assert_eq!(data, &[4u8, 3u8, 7u8, 3u8, 5u8]);
-    }
     assert!(directory.delete(test_path).is_ok());
-    assert!(!directory.exists(test_path));
+    assert!(!directory.exists(test_path).unwrap());
+    Ok(())
 }
 
-fn test_rewrite_forbidden(directory: &mut dyn Directory) {
+fn test_rewrite_forbidden(directory: &dyn Directory) -> crate::Result<()> {
     let test_path: &'static Path = Path::new("some_path_for_test");
-    {
+    directory.open_write(test_path)?;
-        directory.open_write(test_path).unwrap();
+    assert!(directory.exists(test_path).unwrap());
-        assert!(directory.exists(test_path));
+    assert!(directory.open_write(test_path).is_err());
-    }
-    {
-        assert!(directory.open_write(test_path).is_err());
-    }
     assert!(directory.delete(test_path).is_ok());
+    Ok(())
 }
 
-fn test_write_create_the_file(directory: &mut dyn Directory) {
+fn test_write_create_the_file(directory: &dyn Directory) {
     let test_path: &'static Path = Path::new("some_path_for_test");
     {
         assert!(directory.open_read(test_path).is_err());
         let _w = directory.open_write(test_path).unwrap();
-        assert!(directory.exists(test_path));
+        assert!(directory.exists(test_path).unwrap());
         assert!(directory.open_read(test_path).is_ok());
         assert!(directory.delete(test_path).is_ok());
     }
 }
 
-fn test_directory_delete(directory: &mut dyn Directory) {
+fn test_directory_delete(directory: &dyn Directory) -> crate::Result<()> {
     let test_path: &'static Path = Path::new("some_path_for_test");
     assert!(directory.open_read(test_path).is_err());
-    let mut write_file = directory.open_write(&test_path).unwrap();
+    let mut write_file = directory.open_write(&test_path)?;
-    write_file.write_all(&[1, 2, 3, 4]).unwrap();
+    write_file.write_all(&[1, 2, 3, 4])?;
-    write_file.flush().unwrap();
+    write_file.flush()?;
     {
-        let read_handle = directory.open_read(&test_path).unwrap();
+        let read_handle = directory.open_read(&test_path)?.read_bytes()?;
-        assert_eq!(&*read_handle, &[1u8, 2u8, 3u8, 4u8]);
+        assert_eq!(read_handle.as_slice(), &[1u8, 2u8, 3u8, 4u8]);
         // Mapped files can't be deleted on Windows
         if !cfg!(windows) {
             assert!(directory.delete(&test_path).is_ok());
-            assert_eq!(&*read_handle, &[1u8, 2u8, 3u8, 4u8]);
+            assert_eq!(read_handle.as_slice(), &[1u8, 2u8, 3u8, 4u8]);
         }
-
         assert!(directory.delete(Path::new("SomeOtherPath")).is_err());
     }
 
@@ -97,53 +186,40 @@ fn test_directory_delete(directory: &mut dyn Directory) {
 
     assert!(directory.open_read(&test_path).is_err());
     assert!(directory.delete(&test_path).is_err());
+    Ok(())
 }
 
-fn test_directory(directory: &mut dyn Directory) {
+fn test_watch(directory: &dyn Directory) {
-    test_simple(directory);
-    test_rewrite_forbidden(directory);
-    test_write_create_the_file(directory);
-    test_directory_delete(directory);
-    test_lock_non_blocking(directory);
-    test_lock_blocking(directory);
-    test_watch(directory);
-}
-
-fn test_watch(directory: &mut dyn Directory) {
     let counter: Arc<AtomicUsize> = Default::default();
-    let counter_clone = counter.clone();
+    let (tx, rx) = crossbeam::channel::unbounded();
-    let watch_callback = Box::new(move || {
+    let timeout = Duration::from_millis(500);
-        counter_clone.fetch_add(1, Ordering::SeqCst);
+
-    });
+    let handle = directory
-    assert!(directory
+        .watch(WatchCallback::new(move || {
-        .atomic_write(Path::new("meta.json"), b"random_test_data")
+            let val = counter.fetch_add(1, SeqCst);
-        .is_ok());
+            tx.send(val + 1).unwrap();
-    thread::sleep(Duration::new(0, 10_000));
+        }))
-    assert_eq!(0, counter.load(Ordering::SeqCst));
+        .unwrap();
 
-    let watch_handle = directory.watch(watch_callback).unwrap();
-    for i in 0..10 {
-        assert_eq!(i, counter.load(Ordering::SeqCst));
-        assert!(directory
-            .atomic_write(Path::new("meta.json"), b"random_test_data_2")
-            .is_ok());
-        for _ in 0..100 {
-            if counter.load(Ordering::SeqCst) > i {
-                break;
-            }
-            thread::sleep(Duration::from_millis(10));
-        }
-        assert_eq!(i + 1, counter.load(Ordering::SeqCst));
-    }
-    mem::drop(watch_handle);
     assert!(directory
-        .atomic_write(Path::new("meta.json"), b"random_test_data")
+        .atomic_write(Path::new("meta.json"), b"foo")
         .is_ok());
-    thread::sleep(Duration::from_millis(200));
+    assert_eq!(rx.recv_timeout(timeout), Ok(1));
-    assert_eq!(10, counter.load(Ordering::SeqCst));
+
+    assert!(directory
+        .atomic_write(Path::new("meta.json"), b"bar")
+        .is_ok());
+    assert_eq!(rx.recv_timeout(timeout), Ok(2));
+
+    mem::drop(handle);
+
+    assert!(directory
+        .atomic_write(Path::new("meta.json"), b"qux")
+        .is_ok());
+    assert!(rx.recv_timeout(timeout).is_err());
 }
 
-fn test_lock_non_blocking(directory: &mut dyn Directory) {
+fn test_lock_non_blocking(directory: &dyn Directory) {
     {
         let lock_a_res = directory.acquire_lock(&Lock {
             filepath: PathBuf::from("a.lock"),
@@ -168,15 +244,19 @@ fn test_lock_non_blocking(directory: &mut dyn Directory) {
     assert!(lock_a_res.is_ok());
 }
 
-fn test_lock_blocking(directory: &mut dyn Directory) {
+fn test_lock_blocking(directory: &dyn Directory) {
     let lock_a_res = directory.acquire_lock(&Lock {
         filepath: PathBuf::from("a.lock"),
         is_blocking: true,
     });
     assert!(lock_a_res.is_ok());
+    let in_thread = Arc::new(AtomicBool::default());
+    let in_thread_clone = in_thread.clone();
+    let (sender, receiver) = oneshot::channel();
     std::thread::spawn(move || {
         //< lock_a_res is sent to the thread.
-        std::thread::sleep(time::Duration::from_millis(10));
+        in_thread_clone.store(true, SeqCst);
+        let _just_sync = block_on(receiver);
         // explicitely droping lock_a_res. It would have been sufficient to just force it
         // to be part of the move, but the intent seems clearer that way.
         drop(lock_a_res);
@@ -189,14 +269,18 @@ fn test_lock_blocking(directory: &mut dyn Directory) {
         });
         assert!(lock_a_res.is_err());
     }
-    {
+    let directory_clone = directory.box_clone();
-        // the blocking call should wait for at least 10ms.
+    let (sender2, receiver2) = oneshot::channel();
-        let start = time::Instant::now();
+    let join_handle = std::thread::spawn(move || {
-        let lock_a_res = directory.acquire_lock(&Lock {
+        assert!(sender2.send(()).is_ok());
+        let lock_a_res = directory_clone.acquire_lock(&Lock {
             filepath: PathBuf::from("a.lock"),
             is_blocking: true,
         });
+        assert!(in_thread.load(SeqCst));
         assert!(lock_a_res.is_ok());
-        assert!(start.elapsed().subsec_millis() >= 10);
+    });
-    }
+    assert!(block_on(receiver2).is_ok());
+    assert!(sender.send(()).is_ok());
+    assert!(join_handle.join().is_ok());
 }

src/directory/watch_event_router.rs

@@ -1,9 +1,23 @@
+use futures::channel::oneshot;
+use futures::{Future, TryFutureExt};
 use std::sync::Arc;
 use std::sync::RwLock;
 use std::sync::Weak;
 
-/// Type alias for callbacks registered when watching files of a `Directory`.
+/// Cloneable wrapper for callbacks registered when watching files of a `Directory`.
-pub type WatchCallback = Box<dyn Fn() -> () + Sync + Send>;
+#[derive(Clone)]
+pub struct WatchCallback(Arc<dyn Fn() + Sync + Send>);
+
+impl WatchCallback {
+    /// Wraps a `Fn()` to create a WatchCallback.
+    pub fn new<F: Fn() + Sync + Send + 'static>(op: F) -> Self {
+        WatchCallback(Arc::new(op))
+    }
+
+    fn call(&self) {
+        self.0()
+    }
+}
 
 /// Helper struct to implement the watch method in `Directory` implementations.
 ///
@@ -22,22 +36,36 @@ pub struct WatchCallbackList {
 #[derive(Clone)]
 pub struct WatchHandle(Arc<WatchCallback>);
 
+impl WatchHandle {
+    /// Create a WatchHandle handle.
+    pub fn new(watch_callback: Arc<WatchCallback>) -> WatchHandle {
+        WatchHandle(watch_callback)
+    }
+
+    /// Returns an empty watch handle.
+    ///
+    /// This function is only useful when implementing a readonly directory.
+    pub fn empty() -> WatchHandle {
+        WatchHandle::new(Arc::new(WatchCallback::new(|| {})))
+    }
+}
+
 impl WatchCallbackList {
-    /// Suscribes a new callback and returns a handle that controls the lifetime of the callback.
+    /// Subscribes a new callback and returns a handle that controls the lifetime of the callback.
     pub fn subscribe(&self, watch_callback: WatchCallback) -> WatchHandle {
         let watch_callback_arc = Arc::new(watch_callback);
         let watch_callback_weak = Arc::downgrade(&watch_callback_arc);
         self.router.write().unwrap().push(watch_callback_weak);
-        WatchHandle(watch_callback_arc)
+        WatchHandle::new(watch_callback_arc)
     }
 
-    fn list_callback(&self) -> Vec<Arc<WatchCallback>> {
+    fn list_callback(&self) -> Vec<WatchCallback> {
-        let mut callbacks = vec![];
+        let mut callbacks: Vec<WatchCallback> = vec![];
         let mut router_wlock = self.router.write().unwrap();
         let mut i = 0;
         while i < router_wlock.len() {
             if let Some(watch) = router_wlock[i].upgrade() {
-                callbacks.push(watch);
+                callbacks.push(watch.as_ref().clone());
                 i += 1;
             } else {
                 router_wlock.swap_remove(i);
@@ -47,14 +75,21 @@ impl WatchCallbackList {
     }
 
     /// Triggers all callbacks
-    pub fn broadcast(&self) {
+    pub fn broadcast(&self) -> impl Future<Output = ()> {
         let callbacks = self.list_callback();
+        let (sender, receiver) = oneshot::channel();
+        let result = receiver.unwrap_or_else(|_| ());
+        if callbacks.is_empty() {
+            let _ = sender.send(());
+            return result;
+        }
         let spawn_res = std::thread::Builder::new()
             .name("watch-callbacks".to_string())
             .spawn(move || {
                 for callback in callbacks {
-                    callback();
+                    callback.call();
                 }
+                let _ = sender.send(());
             });
         if let Err(err) = spawn_res {
             error!(
@@ -62,44 +97,42 @@ impl WatchCallbackList {
                 err
             );
         }
+        result
     }
 }
 
 #[cfg(test)]
 mod tests {
-    use crate::directory::WatchCallbackList;
+    use crate::directory::{WatchCallback, WatchCallbackList};
+    use futures::executor::block_on;
     use std::mem;
     use std::sync::atomic::{AtomicUsize, Ordering};
     use std::sync::Arc;
-    use std::thread;
-    use std::time::Duration;
-
-    const WAIT_TIME: u64 = 20;
 
     #[test]
     fn test_watch_event_router_simple() {
         let watch_event_router = WatchCallbackList::default();
         let counter: Arc<AtomicUsize> = Default::default();
         let counter_clone = counter.clone();
-        let inc_callback = Box::new(move || {
+        let inc_callback = WatchCallback::new(move || {
             counter_clone.fetch_add(1, Ordering::SeqCst);
         });
-        watch_event_router.broadcast();
+        block_on(watch_event_router.broadcast());
         assert_eq!(0, counter.load(Ordering::SeqCst));
         let handle_a = watch_event_router.subscribe(inc_callback);
-        thread::sleep(Duration::from_millis(WAIT_TIME));
         assert_eq!(0, counter.load(Ordering::SeqCst));
-        watch_event_router.broadcast();
+        block_on(watch_event_router.broadcast());
-        thread::sleep(Duration::from_millis(WAIT_TIME));
         assert_eq!(1, counter.load(Ordering::SeqCst));
-        watch_event_router.broadcast();
+        block_on(async {
-        watch_event_router.broadcast();
+            (
-        watch_event_router.broadcast();
+                watch_event_router.broadcast().await,
-        thread::sleep(Duration::from_millis(WAIT_TIME));
+                watch_event_router.broadcast().await,
+                watch_event_router.broadcast().await,
+            )
+        });
         assert_eq!(4, counter.load(Ordering::SeqCst));
         mem::drop(handle_a);
-        watch_event_router.broadcast();
+        block_on(watch_event_router.broadcast());
-        thread::sleep(Duration::from_millis(WAIT_TIME));
         assert_eq!(4, counter.load(Ordering::SeqCst));
     }
 
@@ -109,26 +142,26 @@ mod tests {
         let counter: Arc<AtomicUsize> = Default::default();
         let inc_callback = |inc: usize| {
             let counter_clone = counter.clone();
-            Box::new(move || {
+            WatchCallback::new(move || {
                 counter_clone.fetch_add(inc, Ordering::SeqCst);
             })
         };
         let handle_a = watch_event_router.subscribe(inc_callback(1));
         let handle_a2 = watch_event_router.subscribe(inc_callback(10));
-        thread::sleep(Duration::from_millis(WAIT_TIME));
         assert_eq!(0, counter.load(Ordering::SeqCst));
-        watch_event_router.broadcast();
+        block_on(async {
-        watch_event_router.broadcast();
+            futures::join!(
-        thread::sleep(Duration::from_millis(WAIT_TIME));
+                watch_event_router.broadcast(),
+                watch_event_router.broadcast()
+            )
+        });
         assert_eq!(22, counter.load(Ordering::SeqCst));
         mem::drop(handle_a);
-        watch_event_router.broadcast();
+        block_on(watch_event_router.broadcast());
-        thread::sleep(Duration::from_millis(WAIT_TIME));
         assert_eq!(32, counter.load(Ordering::SeqCst));
         mem::drop(handle_a2);
-        watch_event_router.broadcast();
+        block_on(watch_event_router.broadcast());
-        watch_event_router.broadcast();
+        block_on(watch_event_router.broadcast());
-        thread::sleep(Duration::from_millis(WAIT_TIME));
         assert_eq!(32, counter.load(Ordering::SeqCst));
     }
 
@@ -137,20 +170,20 @@ mod tests {
         let watch_event_router = WatchCallbackList::default();
         let counter: Arc<AtomicUsize> = Default::default();
         let counter_clone = counter.clone();
-        let inc_callback = Box::new(move || {
+        let inc_callback = WatchCallback::new(move || {
             counter_clone.fetch_add(1, Ordering::SeqCst);
         });
         let handle_a = watch_event_router.subscribe(inc_callback);
         assert_eq!(0, counter.load(Ordering::SeqCst));
-        watch_event_router.broadcast();
+        block_on(async {
-        watch_event_router.broadcast();
+            let future1 = watch_event_router.broadcast();
-        thread::sleep(Duration::from_millis(WAIT_TIME));
+            let future2 = watch_event_router.broadcast();
+            futures::join!(future1, future2)
+        });
         assert_eq!(2, counter.load(Ordering::SeqCst));
-        thread::sleep(Duration::from_millis(WAIT_TIME));
         mem::drop(handle_a);
-        watch_event_router.broadcast();
+        let _ = watch_event_router.broadcast();
-        thread::sleep(Duration::from_millis(WAIT_TIME));
+        block_on(watch_event_router.broadcast());
         assert_eq!(2, counter.load(Ordering::SeqCst));
     }
-
 }

src/docset.rs

@@ -1,58 +1,48 @@
-use crate::common::BitSet;
 use crate::fastfield::DeleteBitSet;
 use crate::DocId;
 use std::borrow::Borrow;
 use std::borrow::BorrowMut;
-use std::cmp::Ordering;
 
-/// Expresses the outcome of a call to `DocSet`'s `.skip_next(...)`.
+/// Sentinel value returned when a DocSet has been entirely consumed.
-#[derive(PartialEq, Eq, Debug)]
+///
-pub enum SkipResult {
+/// This is not u32::MAX as one would have expected, due to the lack of SSE2 instructions
-    /// target was in the docset
+/// to compare [u32; 4].
-    Reached,
+pub const TERMINATED: DocId = std::i32::MAX as u32;
-    /// target was not in the docset, skipping stopped as a greater element was found
-    OverStep,
-    /// the docset was entirely consumed without finding the target, nor any
-    /// element greater than the target.
-    End,
-}
 
 /// Represents an iterable set of sorted doc ids.
-pub trait DocSet {
+pub trait DocSet: Send {
     /// Goes to the next element.
-    /// `.advance(...)` needs to be called a first time to point to the correct
+    ///
-    /// element.
+    /// The DocId of the next element is returned.
-    fn advance(&mut self) -> bool;
+    /// In other words we should always have :
+    /// ```ignore
+    /// let doc = docset.advance();
+    /// assert_eq!(doc, docset.doc());
+    /// ```
+    ///
+    /// If we reached the end of the DocSet, TERMINATED should be returned.
+    ///
+    /// Calling `.advance()` on a terminated DocSet should be supported, and TERMINATED should
+    /// be returned.
+    /// TODO Test existing docsets.
+    fn advance(&mut self) -> DocId;
 
-    /// After skipping, position the iterator in such a way that `.doc()`
+    /// Advances the DocSet forward until reaching the target, or going to the
-    /// will return a value greater than or equal to target.
+    /// lowest DocId greater than the target.
     ///
-    /// SkipResult expresses whether the `target value` was reached, overstepped,
+    /// If the end of the DocSet is reached, TERMINATED is returned.
-    /// or if the `DocSet` was entirely consumed without finding any value
-    /// greater or equal to the `target`.
     ///
-    /// WARNING: Calling skip always advances the docset.
+    /// Calling `.seek(target)` on a terminated DocSet is legal. Implementation
-    /// More specifically, if the docset is already positionned on the target
+    /// of DocSet should support it.
-    /// skipping will advance to the next position and return SkipResult::Overstep.
     ///
-    /// If `.skip_next()` oversteps, then the docset must be positionned correctly
+    /// Calling `seek(TERMINATED)` is also legal and is the normal way to consume a DocSet.
-    /// on an existing document. In other words, `.doc()` should return the first document
+    fn seek(&mut self, target: DocId) -> DocId {
-    /// greater than `DocId`.
+        let mut doc = self.doc();
-    fn skip_next(&mut self, target: DocId) -> SkipResult {
+        debug_assert!(doc <= target);
-        if !self.advance() {
+        while doc < target {
-            return SkipResult::End;
+            doc = self.advance();
-        }
-        loop {
-            match self.doc().cmp(&target) {
-                Ordering::Less => {
-                    if !self.advance() {
-                        return SkipResult::End;
-                    }
-                }
-                Ordering::Equal => return SkipResult::Reached,
-                Ordering::Greater => return SkipResult::OverStep,
-            }
         }
+        doc
     }
 
     /// Fills a given mutable buffer with the next doc ids from the
@@ -71,38 +61,38 @@ pub trait DocSet {
     /// use case where batching. The normal way to
     /// go through the `DocId`'s is to call `.advance()`.
     fn fill_buffer(&mut self, buffer: &mut [DocId]) -> usize {
+        if self.doc() == TERMINATED {
+            return 0;
+        }
         for (i, buffer_val) in buffer.iter_mut().enumerate() {
-            if self.advance() {
+            *buffer_val = self.doc();
-                *buffer_val = self.doc();
+            if self.advance() == TERMINATED {
-            } else {
+                return i + 1;
-                return i;
             }
         }
         buffer.len()
     }
 
     /// Returns the current document
+    /// Right after creating a new DocSet, the docset points to the first document.
+    ///
+    /// If the DocSet is empty, .doc() should return `TERMINATED`.
     fn doc(&self) -> DocId;
 
     /// Returns a best-effort hint of the
     /// length of the docset.
     fn size_hint(&self) -> u32;
 
-    /// Appends all docs to a `bitset`.
-    fn append_to_bitset(&mut self, bitset: &mut BitSet) {
-        while self.advance() {
-            bitset.insert(self.doc());
-        }
-    }
-
     /// Returns the number documents matching.
     /// Calling this method consumes the `DocSet`.
     fn count(&mut self, delete_bitset: &DeleteBitSet) -> u32 {
         let mut count = 0u32;
-        while self.advance() {
+        let mut doc = self.doc();
-            if !delete_bitset.is_deleted(self.doc()) {
+        while doc != TERMINATED {
+            if !delete_bitset.is_deleted(doc) {
                 count += 1u32;
             }
+            doc = self.advance();
         }
         count
     }
@@ -114,22 +104,50 @@ pub trait DocSet {
     /// given by `count()`.
     fn count_including_deleted(&mut self) -> u32 {
         let mut count = 0u32;
-        while self.advance() {
+        let mut doc = self.doc();
+        while doc != TERMINATED {
             count += 1u32;
+            doc = self.advance();
         }
         count
     }
 }
 
+impl<'a> DocSet for &'a mut dyn DocSet {
+    fn advance(&mut self) -> u32 {
+        (**self).advance()
+    }
+
+    fn seek(&mut self, target: DocId) -> DocId {
+        (**self).seek(target)
+    }
+
+    fn doc(&self) -> u32 {
+        (**self).doc()
+    }
+
+    fn size_hint(&self) -> u32 {
+        (**self).size_hint()
+    }
+
+    fn count(&mut self, delete_bitset: &DeleteBitSet) -> u32 {
+        (**self).count(delete_bitset)
+    }
+
+    fn count_including_deleted(&mut self) -> u32 {
+        (**self).count_including_deleted()
+    }
+}
+
 impl<TDocSet: DocSet + ?Sized> DocSet for Box<TDocSet> {
-    fn advance(&mut self) -> bool {
+    fn advance(&mut self) -> DocId {
         let unboxed: &mut TDocSet = self.borrow_mut();
         unboxed.advance()
     }
 
-    fn skip_next(&mut self, target: DocId) -> SkipResult {
+    fn seek(&mut self, target: DocId) -> DocId {
         let unboxed: &mut TDocSet = self.borrow_mut();
-        unboxed.skip_next(target)
+        unboxed.seek(target)
     }
 
     fn doc(&self) -> DocId {
@@ -151,9 +169,4 @@ impl<TDocSet: DocSet + ?Sized> DocSet for Box<TDocSet> {
         let unboxed: &mut TDocSet = self.borrow_mut();
         unboxed.count_including_deleted()
     }
-
-    fn append_to_bitset(&mut self, bitset: &mut BitSet) {
-        let unboxed: &mut TDocSet = self.borrow_mut();
-        unboxed.append_to_bitset(bitset);
-    }
 }

src/error.rs

@@ -2,22 +2,27 @@
 
 use std::io;
 
-use crate::directory::error::LockError;
+use crate::directory::error::{Incompatibility, LockError};
-use crate::directory::error::{IOError, OpenDirectoryError, OpenReadError, OpenWriteError};
 use crate::fastfield::FastFieldNotAvailableError;
 use crate::query;
-use crate::schema;
+use crate::{
-use serde_json;
+    directory::error::{OpenDirectoryError, OpenReadError, OpenWriteError},
+    schema,
+};
 use std::fmt;
 use std::path::PathBuf;
 use std::sync::PoisonError;
 
+/// Represents a `DataCorruption` error.
+///
+/// When facing data corruption, tantivy  actually panic or return this error.
 pub struct DataCorruption {
     filepath: Option<PathBuf>,
     comment: String,
 }
 
 impl DataCorruption {
+    /// Creates a `DataCorruption` Error.
     pub fn new(filepath: PathBuf, comment: String) -> DataCorruption {
         DataCorruption {
             filepath: Some(filepath),
@@ -25,10 +30,11 @@ impl DataCorruption {
         }
     }
 
-    pub fn comment_only(comment: String) -> DataCorruption {
+    /// Creates a `DataCorruption` Error, when the filepath is irrelevant.
+    pub fn comment_only<TStr: ToString>(comment: TStr) -> DataCorruption {
         DataCorruption {
             filepath: None,
-            comment,
+            comment: comment.to_string(),
         }
     }
 }
@@ -44,42 +50,51 @@ impl fmt::Debug for DataCorruption {
     }
 }
 
-/// The library's failure based error enum
+/// The library's error enum
-#[derive(Debug, Fail)]
+#[derive(Debug, Error)]
 pub enum TantivyError {
-    /// Path does not exist.
+    /// Failed to open the directory.
-    #[fail(display = "Path does not exist: '{:?}'", _0)]
+    #[error("Failed to open the directory: '{0:?}'")]
-    PathDoesNotExist(PathBuf),
+    OpenDirectoryError(#[from] OpenDirectoryError),
-    /// File already exists, this is a problem when we try to write into a new file.
+    /// Failed to open a file for read.
-    #[fail(display = "File already exists: '{:?}'", _0)]
+    #[error("Failed to open file for read: '{0:?}'")]
-    FileAlreadyExists(PathBuf),
+    OpenReadError(#[from] OpenReadError),
+    /// Failed to open a file for write.
+    #[error("Failed to open file for write: '{0:?}'")]
+    OpenWriteError(#[from] OpenWriteError),
     /// Index already exists in this directory
-    #[fail(display = "Index already exists")]
+    #[error("Index already exists")]
     IndexAlreadyExists,
     /// Failed to acquire file lock
-    #[fail(display = "Failed to acquire Lockfile: {:?}. {:?}", _0, _1)]
+    #[error("Failed to acquire Lockfile: {0:?}. {1:?}")]
     LockFailure(LockError, Option<String>),
     /// IO Error.
-    #[fail(display = "An IO error occurred: '{}'", _0)]
+    #[error("An IO error occurred: '{0}'")]
-    IOError(#[cause] IOError),
+    IoError(#[from] io::Error),
     /// Data corruption.
-    #[fail(display = "{:?}", _0)]
+    #[error("Data corrupted: '{0:?}'")]
     DataCorruption(DataCorruption),
     /// A thread holding the locked panicked and poisoned the lock.
-    #[fail(display = "A thread holding the locked panicked and poisoned the lock")]
+    #[error("A thread holding the locked panicked and poisoned the lock")]
     Poisoned,
     /// Invalid argument was passed by the user.
-    #[fail(display = "An invalid argument was passed: '{}'", _0)]
+    #[error("An invalid argument was passed: '{0}'")]
     InvalidArgument(String),
     /// An Error happened in one of the thread.
-    #[fail(display = "An error occurred in a thread: '{}'", _0)]
+    #[error("An error occurred in a thread: '{0}'")]
     ErrorInThread(String),
+    /// An Error appeared related to opening or creating a index.
+    #[error("Missing required index builder argument when open/create index: '{0}'")]
+    IndexBuilderMissingArgument(&'static str),
     /// An Error appeared related to the schema.
-    #[fail(display = "Schema error: '{}'", _0)]
+    #[error("Schema error: '{0}'")]
     SchemaError(String),
     /// System error. (e.g.: We failed spawning a new thread)
-    #[fail(display = "System error.'{}'", _0)]
+    #[error("System error.'{0}'")]
     SystemError(String),
+    /// Index incompatible with current version of tantivy
+    #[error("{0:?}")]
+    IncompatibleIndex(Incompatibility),
 }
 
 impl From<DataCorruption> for TantivyError {
@@ -87,31 +102,17 @@ impl From<DataCorruption> for TantivyError {
         TantivyError::DataCorruption(data_corruption)
     }
 }
-
 impl From<FastFieldNotAvailableError> for TantivyError {
     fn from(fastfield_error: FastFieldNotAvailableError) -> TantivyError {
         TantivyError::SchemaError(format!("{}", fastfield_error))
     }
 }
-
 impl From<LockError> for TantivyError {
     fn from(lock_error: LockError) -> TantivyError {
         TantivyError::LockFailure(lock_error, None)
     }
 }
 
-impl From<IOError> for TantivyError {
-    fn from(io_error: IOError) -> TantivyError {
-        TantivyError::IOError(io_error)
-    }
-}
-
-impl From<io::Error> for TantivyError {
-    fn from(io_error: io::Error) -> TantivyError {
-        TantivyError::IOError(io_error.into())
-    }
-}
-
 impl From<query::QueryParserError> for TantivyError {
     fn from(parsing_error: query::QueryParserError) -> TantivyError {
         TantivyError::InvalidArgument(format!("Query is invalid. {:?}", parsing_error))
@@ -124,12 +125,9 @@ impl<Guard> From<PoisonError<Guard>> for TantivyError {
     }
 }
 
-impl From<OpenReadError> for TantivyError {
+impl From<chrono::ParseError> for TantivyError {
-    fn from(error: OpenReadError) -> TantivyError {
+    fn from(err: chrono::ParseError) -> TantivyError {
-        match error {
+        TantivyError::InvalidArgument(err.to_string())
-            OpenReadError::FileDoesNotExist(filepath) => TantivyError::PathDoesNotExist(filepath),
-            OpenReadError::IOError(io_error) => TantivyError::IOError(io_error),
-        }
     }
 }
 
@@ -139,34 +137,14 @@ impl From<schema::DocParsingError> for TantivyError {
     }
 }
 
-impl From<OpenWriteError> for TantivyError {
-    fn from(error: OpenWriteError) -> TantivyError {
-        match error {
-            OpenWriteError::FileAlreadyExists(filepath) => {
-                TantivyError::FileAlreadyExists(filepath)
-            }
-            OpenWriteError::IOError(io_error) => TantivyError::IOError(io_error),
-        }
-    }
-}
-
-impl From<OpenDirectoryError> for TantivyError {
-    fn from(error: OpenDirectoryError) -> TantivyError {
-        match error {
-            OpenDirectoryError::DoesNotExist(directory_path) => {
-                TantivyError::PathDoesNotExist(directory_path)
-            }
-            OpenDirectoryError::NotADirectory(directory_path) => {
-                TantivyError::InvalidArgument(format!("{:?} is not a directory", directory_path))
-            }
-            OpenDirectoryError::IoError(err) => TantivyError::IOError(IOError::from(err)),
-        }
-    }
-}
-
 impl From<serde_json::Error> for TantivyError {
     fn from(error: serde_json::Error) -> TantivyError {
-        let io_err = io::Error::from(error);
+        TantivyError::IoError(error.into())
-        TantivyError::IOError(io_err.into())
+    }
+}
+
+impl From<rayon::ThreadPoolBuildError> for TantivyError {
+    fn from(error: rayon::ThreadPoolBuildError) -> TantivyError {
+        TantivyError::SystemError(error.to_string())
     }
 }

src/fastfield/bytes/mod.rs

@@ -6,31 +6,114 @@ pub use self::writer::BytesFastFieldWriter;
 
 #[cfg(test)]
 mod tests {
-    use crate::schema::Schema;
+    use crate::schema::{BytesOptions, IndexRecordOption, Schema, Value};
-    use crate::Index;
+    use crate::{query::TermQuery, schema::FAST, schema::INDEXED, schema::STORED};
+    use crate::{DocAddress, DocSet, Index, Searcher, Term};
+    use std::ops::Deref;
 
     #[test]
-    fn test_bytes() {
+    fn test_bytes() -> crate::Result<()> {
         let mut schema_builder = Schema::builder();
-        let field = schema_builder.add_bytes_field("bytesfield");
+        let bytes_field = schema_builder.add_bytes_field("bytesfield", FAST);
         let schema = schema_builder.build();
         let index = Index::create_in_ram(schema);
-        let mut index_writer = index.writer_with_num_threads(1, 3_000_000).unwrap();
+        let mut index_writer = index.writer_for_tests()?;
-        index_writer.add_document(doc!(field=>vec![0u8, 1, 2, 3]));
+        index_writer.add_document(doc!(bytes_field=>vec![0u8, 1, 2, 3]));
-        index_writer.add_document(doc!(field=>vec![]));
+        index_writer.add_document(doc!(bytes_field=>vec![]));
-        index_writer.add_document(doc!(field=>vec![255u8]));
+        index_writer.add_document(doc!(bytes_field=>vec![255u8]));
-        index_writer.add_document(doc!(field=>vec![1u8, 3, 5, 7, 9]));
+        index_writer.add_document(doc!(bytes_field=>vec![1u8, 3, 5, 7, 9]));
-        index_writer.add_document(doc!(field=>vec![0u8; 1000]));
+        index_writer.add_document(doc!(bytes_field=>vec![0u8; 1000]));
-        assert!(index_writer.commit().is_ok());
+        index_writer.commit()?;
-        let searcher = index.reader().unwrap().searcher();
+        let searcher = index.reader()?.searcher();
         let segment_reader = searcher.segment_reader(0);
-        let bytes_reader = segment_reader.fast_fields().bytes(field).unwrap();
+        let bytes_reader = segment_reader.fast_fields().bytes(bytes_field).unwrap();
-
         assert_eq!(bytes_reader.get_bytes(0), &[0u8, 1, 2, 3]);
         assert!(bytes_reader.get_bytes(1).is_empty());
         assert_eq!(bytes_reader.get_bytes(2), &[255u8]);
         assert_eq!(bytes_reader.get_bytes(3), &[1u8, 3, 5, 7, 9]);
         let long = vec![0u8; 1000];
         assert_eq!(bytes_reader.get_bytes(4), long.as_slice());
+        Ok(())
+    }
+
+    fn create_index_for_test<T: Into<BytesOptions>>(
+        byte_options: T,
+    ) -> crate::Result<impl Deref<Target = Searcher>> {
+        let mut schema_builder = Schema::builder();
+        let field = schema_builder.add_bytes_field("string_bytes", byte_options.into());
+        let schema = schema_builder.build();
+        let index = Index::create_in_ram(schema);
+        let mut index_writer = index.writer_for_tests()?;
+        index_writer.add_document(doc!(
+                field => b"tantivy".as_ref(),
+                field => b"lucene".as_ref()
+        ));
+        index_writer.commit()?;
+        Ok(index.reader()?.searcher())
+    }
+
+    #[test]
+    fn test_stored_bytes() -> crate::Result<()> {
+        let searcher = create_index_for_test(STORED)?;
+        assert_eq!(searcher.num_docs(), 1);
+        let retrieved_doc = searcher.doc(DocAddress::new(0u32, 0u32))?;
+        let field = searcher.schema().get_field("string_bytes").unwrap();
+        let values: Vec<&Value> = retrieved_doc.get_all(field).collect();
+        assert_eq!(values.len(), 2);
+        let values_bytes: Vec<&[u8]> = values
+            .into_iter()
+            .flat_map(|value| value.bytes_value())
+            .collect();
+        assert_eq!(values_bytes, &[&b"tantivy"[..], &b"lucene"[..]]);
+        Ok(())
+    }
+
+    #[test]
+    fn test_non_stored_bytes() -> crate::Result<()> {
+        let searcher = create_index_for_test(INDEXED)?;
+        assert_eq!(searcher.num_docs(), 1);
+        let retrieved_doc = searcher.doc(DocAddress::new(0u32, 0u32))?;
+        let field = searcher.schema().get_field("string_bytes").unwrap();
+        assert!(retrieved_doc.get_first(field).is_none());
+        Ok(())
+    }
+
+    #[test]
+    fn test_index_bytes() -> crate::Result<()> {
+        let searcher = create_index_for_test(INDEXED)?;
+        assert_eq!(searcher.num_docs(), 1);
+        let field = searcher.schema().get_field("string_bytes").unwrap();
+        let term = Term::from_field_bytes(field, b"lucene".as_ref());
+        let term_query = TermQuery::new(term, IndexRecordOption::Basic);
+        let term_weight = term_query.specialized_weight(&searcher, true)?;
+        let term_scorer = term_weight.specialized_scorer(searcher.segment_reader(0), 1.0)?;
+        assert_eq!(term_scorer.doc(), 0u32);
+        Ok(())
+    }
+
+    #[test]
+    fn test_non_index_bytes() -> crate::Result<()> {
+        let searcher = create_index_for_test(STORED)?;
+        assert_eq!(searcher.num_docs(), 1);
+        let field = searcher.schema().get_field("string_bytes").unwrap();
+        let term = Term::from_field_bytes(field, b"lucene".as_ref());
+        let term_query = TermQuery::new(term, IndexRecordOption::Basic);
+        let term_weight_err = term_query.specialized_weight(&searcher, false);
+        assert!(matches!(
+            term_weight_err,
+            Err(crate::TantivyError::SchemaError(_))
+        ));
+        Ok(())
+    }
+
+    #[test]
+    fn test_fast_bytes_multivalue_value() -> crate::Result<()> {
+        let searcher = create_index_for_test(FAST)?;
+        assert_eq!(searcher.num_docs(), 1);
+        let fast_fields = searcher.segment_reader(0u32).fast_fields();
+        let field = searcher.schema().get_field("string_bytes").unwrap();
+        let fast_field_reader = fast_fields.bytes(field).unwrap();
+        assert_eq!(fast_field_reader.get_bytes(0u32), b"tantivy");
+        Ok(())
     }
 }

src/fastfield/bytes/reader.rs

@@ -1,7 +1,6 @@
-use owning_ref::OwningRef;
+use crate::directory::FileSlice;
-
+use crate::directory::OwnedBytes;
-use crate::directory::ReadOnlySource;
+use crate::fastfield::{BitpackedFastFieldReader, FastFieldReader, MultiValueLength};
-use crate::fastfield::FastFieldReader;
 use crate::DocId;
 
 /// Reader for byte array fast fields
@@ -16,17 +15,17 @@ use crate::DocId;
 /// and the start index for the next document, and keeping the bytes in between.
 #[derive(Clone)]
 pub struct BytesFastFieldReader {
-    idx_reader: FastFieldReader<u64>,
+    idx_reader: BitpackedFastFieldReader<u64>,
-    values: OwningRef<ReadOnlySource, [u8]>,
+    values: OwnedBytes,
 }
 
 impl BytesFastFieldReader {
     pub(crate) fn open(
-        idx_reader: FastFieldReader<u64>,
+        idx_reader: BitpackedFastFieldReader<u64>,
-        values_source: ReadOnlySource,
+        values_file: FileSlice,
-    ) -> BytesFastFieldReader {
+    ) -> crate::Result<BytesFastFieldReader> {
-        let values = OwningRef::new(values_source).map(|source| &source[..]);
+        let values = values_file.read_bytes()?;
-        BytesFastFieldReader { idx_reader, values }
+        Ok(BytesFastFieldReader { idx_reader, values })
     }
 
     fn range(&self, doc: DocId) -> (usize, usize) {
@@ -38,7 +37,13 @@ impl BytesFastFieldReader {
     /// Returns the bytes associated to the given `doc`
     pub fn get_bytes(&self, doc: DocId) -> &[u8] {
         let (start, stop) = self.range(doc);
-        &self.values[start..stop]
+        &self.values.as_slice()[start..stop]
+    }
+
+    /// Returns the length of the bytes associated to the given `doc`
+    pub fn num_bytes(&self, doc: DocId) -> usize {
+        let (start, stop) = self.range(doc);
+        stop - start
     }
 
     /// Returns the overall number of bytes in this bytes fast field.
@@ -46,3 +51,12 @@ impl BytesFastFieldReader {
         self.values.len()
     }
 }
+
+impl MultiValueLength for BytesFastFieldReader {
+    fn get_len(&self, doc_id: DocId) -> u64 {
+        self.num_bytes(doc_id) as u64
+    }
+    fn get_total_len(&self) -> u64 {
+        self.total_num_bytes() as u64
+    }
+}

src/fastfield/bytes/writer.rs

@@ -3,6 +3,9 @@ use std::io;
 use crate::fastfield::serializer::FastFieldSerializer;
 use crate::schema::{Document, Field, Value};
 use crate::DocId;
+use crate::{
+    fastfield::serializer::CompositeFastFieldSerializer, indexer::doc_id_mapping::DocIdMapping,
+};
 
 /// Writer for byte array (as in, any number of bytes per document) fast fields
 ///
@@ -35,6 +38,10 @@ impl BytesFastFieldWriter {
         }
     }
 
+    /// The memory used (inclusive childs)
+    pub fn mem_usage(&self) -> usize {
+        self.vals.capacity() + self.doc_index.capacity() * std::mem::size_of::<u64>()
+    }
     /// Access the field associated to the `BytesFastFieldWriter`
     pub fn field(&self) -> Field {
         self.field
@@ -49,16 +56,10 @@ impl BytesFastFieldWriter {
     /// matching field values present in the document.
     pub fn add_document(&mut self, doc: &Document) {
         self.next_doc();
-        for field_value in doc.field_values() {
+        for field_value in doc.get_all(self.field) {
-            if field_value.field() == self.field {
+            if let Value::Bytes(ref bytes) = field_value {
-                if let Value::Bytes(ref bytes) = *field_value.value() {
+                self.vals.extend_from_slice(bytes);
-                    self.vals.extend_from_slice(bytes);
+                return;
-                } else {
-                    panic!(
-                        "Bytes field contained non-Bytes Value!. Field {:?} = {:?}",
-                        self.field, field_value
-                    );
-                }
             }
         }
     }
@@ -74,21 +75,60 @@ impl BytesFastFieldWriter {
         doc
     }
 
+    /// Returns an iterator over values per doc_id in ascending doc_id order.
+    ///
+    /// Normally the order is simply iterating self.doc_id_index.
+    /// With doc_id_map it accounts for the new mapping, returning values in the order of the
+    /// new doc_ids.
+    fn get_ordered_values<'a: 'b, 'b>(
+        &'a self,
+        doc_id_map: Option<&'b DocIdMapping>,
+    ) -> impl Iterator<Item = &'b [u8]> {
+        let doc_id_iter: Box<dyn Iterator<Item = u32>> = if let Some(doc_id_map) = doc_id_map {
+            Box::new(doc_id_map.iter_old_doc_ids())
+        } else {
+            let max_doc = self.doc_index.len() as u32;
+            Box::new(0..max_doc)
+        };
+        doc_id_iter.map(move |doc_id| self.get_values_for_doc_id(doc_id))
+    }
+
+    /// returns all values for a doc_ids
+    fn get_values_for_doc_id(&self, doc_id: u32) -> &[u8] {
+        let start_pos = self.doc_index[doc_id as usize] as usize;
+        let end_pos = self
+            .doc_index
+            .get(doc_id as usize + 1)
+            .cloned()
+            .unwrap_or(self.vals.len() as u64) as usize; // special case, last doc_id has no offset information
+        &self.vals[start_pos..end_pos]
+    }
+
     /// Serializes the fast field values by pushing them to the `FastFieldSerializer`.
-    pub fn serialize(&self, serializer: &mut FastFieldSerializer) -> io::Result<()> {
+    pub fn serialize(
-        {
+        &self,
-            // writing the offset index
+        serializer: &mut CompositeFastFieldSerializer,
-            let mut doc_index_serializer =
+        doc_id_map: Option<&DocIdMapping>,
-                serializer.new_u64_fast_field_with_idx(self.field, 0, self.vals.len() as u64, 0)?;
+    ) -> io::Result<()> {
-            for &offset in &self.doc_index {
+        // writing the offset index
-                doc_index_serializer.add_val(offset)?;
+        let mut doc_index_serializer =
-            }
+            serializer.new_u64_fast_field_with_idx(self.field, 0, self.vals.len() as u64, 0)?;
-            doc_index_serializer.add_val(self.vals.len() as u64)?;
+        let mut offset = 0;
-            doc_index_serializer.close_field()?;
+        for vals in self.get_ordered_values(doc_id_map) {
+            doc_index_serializer.add_val(offset)?;
+            offset += vals.len() as u64;
         }
-        {
+        doc_index_serializer.add_val(self.vals.len() as u64)?;
-            // writing the values themselves
+        doc_index_serializer.close_field()?;
-            let mut value_serializer = serializer.new_bytes_fast_field_with_idx(self.field, 1)?;
+        // writing the values themselves
+        let mut value_serializer = serializer.new_bytes_fast_field_with_idx(self.field, 1);
+        // the else could be removed, but this is faster (difference not benchmarked)
+        if let Some(doc_id_map) = doc_id_map {
+            for vals in self.get_ordered_values(Some(doc_id_map)) {
+                // sort values in case of remapped doc_ids?
+                value_serializer.write_all(vals)?;
+            }
+        } else {
             value_serializer.write_all(&self.vals)?;
         }
         Ok(())

src/fastfield/delete.rs

@@ -1,17 +1,22 @@
-use crate::common::HasLen;
+use crate::common::{BitSet, HasLen};
-use crate::directory::ReadOnlySource;
+use crate::directory::FileSlice;
+use crate::directory::OwnedBytes;
 use crate::directory::WritePtr;
 use crate::space_usage::ByteCount;
 use crate::DocId;
-use bit_set::BitSet;
 use std::io;
 use std::io::Write;
 
 /// Write a delete `BitSet`
 ///
 /// where `delete_bitset` is the set of deleted `DocId`.
-pub fn write_delete_bitset(delete_bitset: &BitSet, writer: &mut WritePtr) -> io::Result<()> {
+/// Warning: this function does not call terminate. The caller is in charge of
-    let max_doc = delete_bitset.capacity();
+/// closing the writer properly.
+pub fn write_delete_bitset(
+    delete_bitset: &BitSet,
+    max_doc: u32,
+    writer: &mut WritePtr,
+) -> io::Result<()> {
     let mut byte = 0u8;
     let mut shift = 0u8;
     for doc in 0..max_doc {
@@ -29,28 +34,47 @@ pub fn write_delete_bitset(delete_bitset: &BitSet, writer: &mut WritePtr) -> io:
     if max_doc % 8 > 0 {
         writer.write_all(&[byte])?;
     }
-    writer.flush()
+    Ok(())
 }
 
 /// Set of deleted `DocId`s.
 #[derive(Clone)]
 pub struct DeleteBitSet {
-    data: ReadOnlySource,
+    data: OwnedBytes,
-    len: usize,
+    num_deleted: usize,
 }
 
 impl DeleteBitSet {
-    /// Opens a delete bitset given its data source.
+    #[cfg(test)]
-    pub fn open(data: ReadOnlySource) -> DeleteBitSet {
+    pub(crate) fn for_test(docs: &[DocId], max_doc: u32) -> DeleteBitSet {
-        let num_deleted: usize = data
+        use crate::directory::{Directory, RamDirectory, TerminatingWrite};
+        use std::path::Path;
+        assert!(docs.iter().all(|&doc| doc < max_doc));
+        let mut bitset = BitSet::with_max_value(max_doc);
+        for &doc in docs {
+            bitset.insert(doc);
+        }
+        let directory = RamDirectory::create();
+        let path = Path::new("dummydeletebitset");
+        let mut wrt = directory.open_write(path).unwrap();
+        write_delete_bitset(&bitset, max_doc, &mut wrt).unwrap();
+        wrt.terminate().unwrap();
+        let file = directory.open_read(path).unwrap();
+        Self::open(file).unwrap()
+    }
+
+    /// Opens a delete bitset given its file.
+    pub fn open(file: FileSlice) -> crate::Result<DeleteBitSet> {
+        let bytes = file.read_bytes()?;
+        let num_deleted: usize = bytes
             .as_slice()
             .iter()
             .map(|b| b.count_ones() as usize)
             .sum();
-        DeleteBitSet {
+        Ok(DeleteBitSet {
-            data,
+            data: bytes,
-            len: num_deleted,
+            num_deleted,
-        }
+        })
     }
 
     /// Returns true iff the document is still "alive". In other words, if it has not been deleted.
@@ -59,10 +83,10 @@ impl DeleteBitSet {
     }
 
     /// Returns true iff the document has been marked as deleted.
-    #[inline(always)]
+    #[inline]
     pub fn is_deleted(&self, doc: DocId) -> bool {
         let byte_offset = doc / 8u32;
-        let b: u8 = (*self.data)[byte_offset as usize];
+        let b: u8 = self.data.as_slice()[byte_offset as usize];
         let shift = (doc & 7u32) as u8;
         b & (1u8 << shift) != 0
     }
@@ -75,51 +99,41 @@ impl DeleteBitSet {
 
 impl HasLen for DeleteBitSet {
     fn len(&self) -> usize {
-        self.len
+        self.num_deleted
     }
 }
 
 #[cfg(test)]
 mod tests {
-    use super::*;
+    use super::DeleteBitSet;
-    use crate::directory::*;
+    use crate::common::HasLen;
-    use bit_set::BitSet;
-    use std::path::PathBuf;
 
-    fn test_delete_bitset_helper(bitset: &BitSet) {
+    #[test]
-        let test_path = PathBuf::from("test");
+    fn test_delete_bitset_empty() {
-        let mut directory = RAMDirectory::create();
+        let delete_bitset = DeleteBitSet::for_test(&[], 10);
-        {
+        for doc in 0..10 {
-            let mut writer = directory.open_write(&*test_path).unwrap();
+            assert_eq!(delete_bitset.is_deleted(doc), !delete_bitset.is_alive(doc));
-            write_delete_bitset(bitset, &mut writer).unwrap();
-        }
-        {
-            let source = directory.open_read(&test_path).unwrap();
-            let delete_bitset = DeleteBitSet::open(source);
-            let n = bitset.capacity();
-            for doc in 0..n {
-                assert_eq!(bitset.contains(doc), delete_bitset.is_deleted(doc as DocId));
-            }
-            assert_eq!(delete_bitset.len(), bitset.len());
         }
+        assert_eq!(delete_bitset.len(), 0);
     }
 
     #[test]
     fn test_delete_bitset() {
-        {
+        let delete_bitset = DeleteBitSet::for_test(&[1, 9], 10);
-            let mut bitset = BitSet::with_capacity(10);
+        assert!(delete_bitset.is_alive(0));
-            bitset.insert(1);
+        assert!(delete_bitset.is_deleted(1));
-            bitset.insert(9);
+        assert!(delete_bitset.is_alive(2));
-            test_delete_bitset_helper(&bitset);
+        assert!(delete_bitset.is_alive(3));
-        }
+        assert!(delete_bitset.is_alive(4));
-        {
+        assert!(delete_bitset.is_alive(5));
-            let mut bitset = BitSet::with_capacity(8);
+        assert!(delete_bitset.is_alive(6));
-            bitset.insert(1);
+        assert!(delete_bitset.is_alive(6));
-            bitset.insert(2);
+        assert!(delete_bitset.is_alive(7));
-            bitset.insert(3);
+        assert!(delete_bitset.is_alive(8));
-            bitset.insert(5);
+        assert!(delete_bitset.is_deleted(9));
-            bitset.insert(7);
+        for doc in 0..10 {
-            test_delete_bitset_helper(&bitset);
+            assert_eq!(delete_bitset.is_deleted(doc), !delete_bitset.is_alive(doc));
         }
+        assert_eq!(delete_bitset.len(), 2);
     }
 }

src/fastfield/error.rs

@@ -4,8 +4,8 @@ use std::result;
 /// `FastFieldNotAvailableError` is returned when the
 /// user requested for a fast field reader, and the field was not
 /// defined in the schema as a fast field.
-#[derive(Debug, Fail)]
+#[derive(Debug, Error)]
-#[fail(display = "Fast field not available: '{:?}'", field_name)]
+#[error("Fast field not available: '{field_name:?}'")]
 pub struct FastFieldNotAvailableError {
     field_name: String,
 }

src/fastfield/facet_reader.rs

@@ -1,4 +1,5 @@
-use super::MultiValueIntFastFieldReader;
+use super::MultiValuedFastFieldReader;
+use crate::error::DataCorruption;
 use crate::schema::Facet;
 use crate::termdict::TermDictionary;
 use crate::termdict::TermOrdinal;
@@ -19,7 +20,7 @@ use std::str;
 /// list of facets. This ordinal is segment local and
 /// only makes sense for a given segment.
 pub struct FacetReader {
-    term_ords: MultiValueIntFastFieldReader<u64>,
+    term_ords: MultiValuedFastFieldReader<u64>,
     term_dict: TermDictionary,
     buffer: Vec<u8>,
 }
@@ -28,12 +29,12 @@ impl FacetReader {
     /// Creates a new `FacetReader`.
     ///
     /// A facet reader just wraps :
-    /// - a `MultiValueIntFastFieldReader` that makes it possible to
+    /// - a `MultiValuedFastFieldReader` that makes it possible to
     /// access the list of facet ords for a given document.
     /// - a `TermDictionary` that helps associating a facet to
     /// an ordinal and vice versa.
     pub fn new(
-        term_ords: MultiValueIntFastFieldReader<u64>,
+        term_ords: MultiValuedFastFieldReader<u64>,
         term_dict: TermDictionary,
     ) -> FacetReader {
         FacetReader {
@@ -62,18 +63,166 @@ impl FacetReader {
         &mut self,
         facet_ord: TermOrdinal,
         output: &mut Facet,
-    ) -> Result<(), str::Utf8Error> {
+    ) -> crate::Result<()> {
         let found_term = self
             .term_dict
-            .ord_to_term(facet_ord as u64, &mut self.buffer);
+            .ord_to_term(facet_ord as u64, &mut self.buffer)?;
         assert!(found_term, "Term ordinal {} no found.", facet_ord);
-        let facet_str = str::from_utf8(&self.buffer[..])?;
+        let facet_str = str::from_utf8(&self.buffer[..])
+            .map_err(|utf8_err| DataCorruption::comment_only(utf8_err.to_string()))?;
         output.set_facet_str(facet_str);
         Ok(())
     }
 
     /// Return the list of facet ordinals associated to a document.
-    pub fn facet_ords(&mut self, doc: DocId, output: &mut Vec<u64>) {
+    pub fn facet_ords(&self, doc: DocId, output: &mut Vec<u64>) {
         self.term_ords.get_vals(doc, output);
     }
 }
+
+#[cfg(test)]
+mod tests {
+    use crate::Index;
+    use crate::{
+        schema::{Facet, FacetOptions, SchemaBuilder, Value, INDEXED, STORED},
+        DocAddress, Document,
+    };
+
+    #[test]
+    fn test_facet_only_indexed() -> crate::Result<()> {
+        let mut schema_builder = SchemaBuilder::default();
+        let facet_field = schema_builder.add_facet_field("facet", INDEXED);
+        let schema = schema_builder.build();
+        let index = Index::create_in_ram(schema);
+        let mut index_writer = index.writer_for_tests()?;
+        index_writer.add_document(doc!(facet_field=>Facet::from_text("/a/b").unwrap()));
+        index_writer.commit()?;
+        let searcher = index.reader()?.searcher();
+        let facet_reader = searcher
+            .segment_reader(0u32)
+            .facet_reader(facet_field)
+            .unwrap();
+        let mut facet_ords = Vec::new();
+        facet_reader.facet_ords(0u32, &mut facet_ords);
+        assert_eq!(&facet_ords, &[2u64]);
+        let doc = searcher.doc(DocAddress::new(0u32, 0u32))?;
+        let value = doc.get_first(facet_field).and_then(Value::path);
+        assert_eq!(value, None);
+        Ok(())
+    }
+
+    #[test]
+    fn test_facet_only_stored() -> crate::Result<()> {
+        let mut schema_builder = SchemaBuilder::default();
+        let facet_field = schema_builder.add_facet_field("facet", STORED);
+        let schema = schema_builder.build();
+        let index = Index::create_in_ram(schema);
+        let mut index_writer = index.writer_for_tests()?;
+        index_writer.add_document(doc!(facet_field=>Facet::from_text("/a/b").unwrap()));
+        index_writer.commit()?;
+        let searcher = index.reader()?.searcher();
+        let facet_reader = searcher
+            .segment_reader(0u32)
+            .facet_reader(facet_field)
+            .unwrap();
+        let mut facet_ords = Vec::new();
+        facet_reader.facet_ords(0u32, &mut facet_ords);
+        assert!(facet_ords.is_empty());
+        let doc = searcher.doc(DocAddress::new(0u32, 0u32))?;
+        let value = doc.get_first(facet_field).and_then(Value::path);
+        assert_eq!(value, Some("/a/b".to_string()));
+        Ok(())
+    }
+
+    #[test]
+    fn test_facet_stored_and_indexed() -> crate::Result<()> {
+        let mut schema_builder = SchemaBuilder::default();
+        let facet_field = schema_builder.add_facet_field("facet", STORED | INDEXED);
+        let schema = schema_builder.build();
+        let index = Index::create_in_ram(schema);
+        let mut index_writer = index.writer_for_tests()?;
+        index_writer.add_document(doc!(facet_field=>Facet::from_text("/a/b").unwrap()));
+        index_writer.commit()?;
+        let searcher = index.reader()?.searcher();
+        let facet_reader = searcher
+            .segment_reader(0u32)
+            .facet_reader(facet_field)
+            .unwrap();
+        let mut facet_ords = Vec::new();
+        facet_reader.facet_ords(0u32, &mut facet_ords);
+        assert_eq!(&facet_ords, &[2u64]);
+        let doc = searcher.doc(DocAddress::new(0u32, 0u32))?;
+        let value = doc.get_first(facet_field).and_then(Value::path);
+        assert_eq!(value, Some("/a/b".to_string()));
+        Ok(())
+    }
+
+    #[test]
+    fn test_facet_neither_stored_and_indexed() -> crate::Result<()> {
+        let mut schema_builder = SchemaBuilder::default();
+        let facet_field = schema_builder.add_facet_field("facet", FacetOptions::default());
+        let schema = schema_builder.build();
+        let index = Index::create_in_ram(schema);
+        let mut index_writer = index.writer_for_tests()?;
+        index_writer.add_document(doc!(facet_field=>Facet::from_text("/a/b").unwrap()));
+        index_writer.commit()?;
+        let searcher = index.reader()?.searcher();
+        let facet_reader = searcher
+            .segment_reader(0u32)
+            .facet_reader(facet_field)
+            .unwrap();
+        let mut facet_ords = Vec::new();
+        facet_reader.facet_ords(0u32, &mut facet_ords);
+        assert!(facet_ords.is_empty());
+        let doc = searcher.doc(DocAddress::new(0u32, 0u32))?;
+        let value = doc.get_first(facet_field).and_then(Value::path);
+        assert_eq!(value, None);
+        Ok(())
+    }
+
+    #[test]
+    fn test_facet_not_populated_for_all_docs() -> crate::Result<()> {
+        let mut schema_builder = SchemaBuilder::default();
+        let facet_field = schema_builder.add_facet_field("facet", INDEXED);
+        let schema = schema_builder.build();
+        let index = Index::create_in_ram(schema);
+        let mut index_writer = index.writer_for_tests()?;
+        index_writer.add_document(doc!(facet_field=>Facet::from_text("/a/b").unwrap()));
+        index_writer.add_document(Document::default());
+        index_writer.commit()?;
+        let searcher = index.reader()?.searcher();
+        let facet_reader = searcher
+            .segment_reader(0u32)
+            .facet_reader(facet_field)
+            .unwrap();
+        let mut facet_ords = Vec::new();
+        facet_reader.facet_ords(0u32, &mut facet_ords);
+        assert_eq!(&facet_ords, &[2u64]);
+        facet_reader.facet_ords(1u32, &mut facet_ords);
+        assert!(facet_ords.is_empty());
+        Ok(())
+    }
+
+    #[test]
+    fn test_facet_not_populated_for_any_docs() -> crate::Result<()> {
+        let mut schema_builder = SchemaBuilder::default();
+        let facet_field = schema_builder.add_facet_field("facet", INDEXED);
+        let schema = schema_builder.build();
+        let index = Index::create_in_ram(schema);
+        let mut index_writer = index.writer_for_tests()?;
+        index_writer.add_document(Document::default());
+        index_writer.add_document(Document::default());
+        index_writer.commit()?;
+        let searcher = index.reader()?.searcher();
+        let facet_reader = searcher
+            .segment_reader(0u32)
+            .facet_reader(facet_field)
+            .unwrap();
+        let mut facet_ords = Vec::new();
+        facet_reader.facet_ords(0u32, &mut facet_ords);
+        assert!(facet_ords.is_empty());
+        facet_reader.facet_ords(1u32, &mut facet_ords);
+        assert!(facet_ords.is_empty());
+        Ok(())
+    }
+}

src/fastfield/mod.rs

@@ -28,15 +28,22 @@ pub use self::delete::write_delete_bitset;
 pub use self::delete::DeleteBitSet;
 pub use self::error::{FastFieldNotAvailableError, Result};
 pub use self::facet_reader::FacetReader;
-pub use self::multivalued::{MultiValueIntFastFieldReader, MultiValueIntFastFieldWriter};
+pub use self::multivalued::{MultiValuedFastFieldReader, MultiValuedFastFieldWriter};
+pub use self::reader::BitpackedFastFieldReader;
+pub use self::reader::DynamicFastFieldReader;
 pub use self::reader::FastFieldReader;
 pub use self::readers::FastFieldReaders;
+pub use self::serializer::CompositeFastFieldSerializer;
 pub use self::serializer::FastFieldSerializer;
 pub use self::writer::{FastFieldsWriter, IntFastFieldWriter};
-use crate::common;
 use crate::schema::Cardinality;
 use crate::schema::FieldType;
 use crate::schema::Value;
+use crate::{
+    chrono::{NaiveDateTime, Utc},
+    schema::Type,
+};
+use crate::{common, DocId};
 
 mod bytes;
 mod delete;
@@ -48,8 +55,17 @@ mod readers;
 mod serializer;
 mod writer;
 
+/// Trait for `BytesFastFieldReader` and `MultiValuedFastFieldReader` to return the length of data
+/// for a doc_id
+pub trait MultiValueLength {
+    /// returns the num of values associated to a doc_id
+    fn get_len(&self, doc_id: DocId) -> u64;
+    /// returns the sum of num values for all doc_ids
+    fn get_total_len(&self) -> u64;
+}
+
 /// Trait for types that are allowed for fast fields: (u64, i64 and f64).
-pub trait FastValue: Default + Clone + Copy + Send + Sync + PartialOrd {
+pub trait FastValue: Clone + Copy + Send + Sync + PartialOrd + 'static {
     /// Converts a value from u64
     ///
     /// Internally all fast field values are encoded as u64.
@@ -69,6 +85,15 @@ pub trait FastValue: Default + Clone + Copy + Send + Sync + PartialOrd {
     /// Cast value to `u64`.
     /// The value is just reinterpreted in memory.
     fn as_u64(&self) -> u64;
+
+    /// Build a default value. This default value is never used, so the value does not
+    /// really matter.
+    fn make_zero() -> Self {
+        Self::from_u64(0i64.to_u64())
+    }
+
+    /// Returns the `schema::Type` for this FastValue.
+    fn to_type() -> Type;
 }
 
 impl FastValue for u64 {
@@ -83,7 +108,7 @@ impl FastValue for u64 {
     fn fast_field_cardinality(field_type: &FieldType) -> Option<Cardinality> {
         match *field_type {
             FieldType::U64(ref integer_options) => integer_options.get_fastfield_cardinality(),
-            FieldType::HierarchicalFacet => Some(Cardinality::MultiValues),
+            FieldType::HierarchicalFacet(_) => Some(Cardinality::MultiValues),
             _ => None,
         }
     }
@@ -91,6 +116,10 @@ impl FastValue for u64 {
     fn as_u64(&self) -> u64 {
         *self
     }
+
+    fn to_type() -> Type {
+        Type::U64
+    }
 }
 
 impl FastValue for i64 {
@@ -112,6 +141,10 @@ impl FastValue for i64 {
     fn as_u64(&self) -> u64 {
         *self as u64
     }
+
+    fn to_type() -> Type {
+        Type::I64
+    }
 }
 
 impl FastValue for f64 {
@@ -133,6 +166,36 @@ impl FastValue for f64 {
     fn as_u64(&self) -> u64 {
         self.to_bits()
     }
+
+    fn to_type() -> Type {
+        Type::F64
+    }
+}
+
+impl FastValue for crate::DateTime {
+    fn from_u64(timestamp_u64: u64) -> Self {
+        let timestamp_i64 = i64::from_u64(timestamp_u64);
+        crate::DateTime::from_utc(NaiveDateTime::from_timestamp(timestamp_i64, 0), Utc)
+    }
+
+    fn to_u64(&self) -> u64 {
+        self.timestamp().to_u64()
+    }
+
+    fn fast_field_cardinality(field_type: &FieldType) -> Option<Cardinality> {
+        match *field_type {
+            FieldType::Date(ref integer_options) => integer_options.get_fastfield_cardinality(),
+            _ => None,
+        }
+    }
+
+    fn as_u64(&self) -> u64 {
+        self.timestamp().as_u64()
+    }
+
+    fn to_type() -> Type {
+        Type::Date
+    }
 }
 
 fn value_to_u64(value: &Value) -> u64 {
@@ -140,6 +203,7 @@ fn value_to_u64(value: &Value) -> u64 {
         Value::U64(ref val) => *val,
         Value::I64(ref val) => common::i64_to_u64(*val),
         Value::F64(ref val) => common::f64_to_u64(*val),
+        Value::Date(ref datetime) => common::i64_to_u64(datetime.timestamp()),
         _ => panic!("Expected a u64/i64/f64 field, got {:?} ", value),
     }
 }
@@ -149,12 +213,15 @@ mod tests {
 
     use super::*;
     use crate::common::CompositeFile;
-    use crate::directory::{Directory, RAMDirectory, WritePtr};
+    use crate::directory::{Directory, RamDirectory, WritePtr};
-    use crate::fastfield::FastFieldReader;
+    use crate::fastfield::BitpackedFastFieldReader;
-    use crate::schema::Document;
+    use crate::merge_policy::NoMergePolicy;
     use crate::schema::Field;
     use crate::schema::Schema;
     use crate::schema::FAST;
+    use crate::schema::{Document, IntOptions};
+    use crate::{Index, SegmentId, SegmentReader};
+    use common::HasLen;
     use once_cell::sync::Lazy;
     use rand::prelude::SliceRandom;
     use rand::rngs::StdRng;
@@ -172,49 +239,52 @@ mod tests {
 
     #[test]
     pub fn test_fastfield() {
-        let test_fastfield = FastFieldReader::<u64>::from(vec![100, 200, 300]);
+        let test_fastfield = BitpackedFastFieldReader::<u64>::from(vec![100, 200, 300]);
         assert_eq!(test_fastfield.get(0), 100);
         assert_eq!(test_fastfield.get(1), 200);
         assert_eq!(test_fastfield.get(2), 300);
     }
 
     #[test]
-    fn test_intfastfield_small() {
+    pub fn test_fastfield_i64_u64() {
+        let datetime = crate::DateTime::from_utc(NaiveDateTime::from_timestamp(0i64, 0), Utc);
+        assert_eq!(i64::from_u64(datetime.to_u64()), 0i64);
+    }
+
+    #[test]
+    fn test_intfastfield_small() -> crate::Result<()> {
         let path = Path::new("test");
-        let mut directory: RAMDirectory = RAMDirectory::create();
+        let directory: RamDirectory = RamDirectory::create();
         {
             let write: WritePtr = directory.open_write(Path::new("test")).unwrap();
-            let mut serializer = FastFieldSerializer::from_write(write).unwrap();
+            let mut serializer = CompositeFastFieldSerializer::from_write(write).unwrap();
             let mut fast_field_writers = FastFieldsWriter::from_schema(&SCHEMA);
             fast_field_writers.add_document(&doc!(*FIELD=>13u64));
             fast_field_writers.add_document(&doc!(*FIELD=>14u64));
             fast_field_writers.add_document(&doc!(*FIELD=>2u64));
             fast_field_writers
-                .serialize(&mut serializer, &HashMap::new())
+                .serialize(&mut serializer, &HashMap::new(), None)
                 .unwrap();
             serializer.close().unwrap();
         }
-        let source = directory.open_read(&path).unwrap();
+        let file = directory.open_read(&path).unwrap();
-        {
+        assert_eq!(file.len(), 36 as usize);
-            assert_eq!(source.len(), 36 as usize);
+        let composite_file = CompositeFile::open(&file)?;
-        }
+        let file = composite_file.open_read(*FIELD).unwrap();
-        {
+        let fast_field_reader = BitpackedFastFieldReader::<u64>::open(file)?;
-            let composite_file = CompositeFile::open(&source).unwrap();
+        assert_eq!(fast_field_reader.get(0), 13u64);
-            let field_source = composite_file.open_read(*FIELD).unwrap();
+        assert_eq!(fast_field_reader.get(1), 14u64);
-            let fast_field_reader = FastFieldReader::<u64>::open(field_source);
+        assert_eq!(fast_field_reader.get(2), 2u64);
-            assert_eq!(fast_field_reader.get(0), 13u64);
+        Ok(())
-            assert_eq!(fast_field_reader.get(1), 14u64);
-            assert_eq!(fast_field_reader.get(2), 2u64);
-        }
     }
 
     #[test]
-    fn test_intfastfield_large() {
+    fn test_intfastfield_large() -> crate::Result<()> {
         let path = Path::new("test");
-        let mut directory: RAMDirectory = RAMDirectory::create();
+        let directory: RamDirectory = RamDirectory::create();
         {
-            let write: WritePtr = directory.open_write(Path::new("test")).unwrap();
+            let write: WritePtr = directory.open_write(Path::new("test"))?;
-            let mut serializer = FastFieldSerializer::from_write(write).unwrap();
+            let mut serializer = CompositeFastFieldSerializer::from_write(write)?;
             let mut fast_field_writers = FastFieldsWriter::from_schema(&SCHEMA);
             fast_field_writers.add_document(&doc!(*FIELD=>4u64));
             fast_field_writers.add_document(&doc!(*FIELD=>14_082_001u64));
@@ -225,19 +295,15 @@ mod tests {
             fast_field_writers.add_document(&doc!(*FIELD=>1_002u64));
             fast_field_writers.add_document(&doc!(*FIELD=>1_501u64));
             fast_field_writers.add_document(&doc!(*FIELD=>215u64));
-            fast_field_writers
+            fast_field_writers.serialize(&mut serializer, &HashMap::new(), None)?;
-                .serialize(&mut serializer, &HashMap::new())
+            serializer.close()?;
-                .unwrap();
-            serializer.close().unwrap();
         }
-        let source = directory.open_read(&path).unwrap();
+        let file = directory.open_read(&path)?;
+        assert_eq!(file.len(), 61 as usize);
         {
-            assert_eq!(source.len(), 61 as usize);
+            let fast_fields_composite = CompositeFile::open(&file)?;
-        }
-        {
-            let fast_fields_composite = CompositeFile::open(&source).unwrap();
             let data = fast_fields_composite.open_read(*FIELD).unwrap();
-            let fast_field_reader = FastFieldReader::<u64>::open(data);
+            let fast_field_reader = BitpackedFastFieldReader::<u64>::open(data)?;
             assert_eq!(fast_field_reader.get(0), 4u64);
             assert_eq!(fast_field_reader.get(1), 14_082_001u64);
             assert_eq!(fast_field_reader.get(2), 3_052u64);
@@ -248,47 +314,47 @@ mod tests {
             assert_eq!(fast_field_reader.get(7), 1_501u64);
             assert_eq!(fast_field_reader.get(8), 215u64);
         }
+        Ok(())
     }
 
     #[test]
-    fn test_intfastfield_null_amplitude() {
+    fn test_intfastfield_null_amplitude() -> crate::Result<()> {
         let path = Path::new("test");
-        let mut directory: RAMDirectory = RAMDirectory::create();
+        let directory: RamDirectory = RamDirectory::create();
 
         {
             let write: WritePtr = directory.open_write(Path::new("test")).unwrap();
-            let mut serializer = FastFieldSerializer::from_write(write).unwrap();
+            let mut serializer = CompositeFastFieldSerializer::from_write(write).unwrap();
             let mut fast_field_writers = FastFieldsWriter::from_schema(&SCHEMA);
             for _ in 0..10_000 {
                 fast_field_writers.add_document(&doc!(*FIELD=>100_000u64));
             }
             fast_field_writers
-                .serialize(&mut serializer, &HashMap::new())
+                .serialize(&mut serializer, &HashMap::new(), None)
                 .unwrap();
             serializer.close().unwrap();
         }
-        let source = directory.open_read(&path).unwrap();
+        let file = directory.open_read(&path).unwrap();
+        assert_eq!(file.len(), 34 as usize);
         {
-            assert_eq!(source.len(), 34 as usize);
+            let fast_fields_composite = CompositeFile::open(&file).unwrap();
-        }
-        {
-            let fast_fields_composite = CompositeFile::open(&source).unwrap();
             let data = fast_fields_composite.open_read(*FIELD).unwrap();
-            let fast_field_reader = FastFieldReader::<u64>::open(data);
+            let fast_field_reader = BitpackedFastFieldReader::<u64>::open(data)?;
             for doc in 0..10_000 {
                 assert_eq!(fast_field_reader.get(doc), 100_000u64);
             }
         }
+        Ok(())
     }
 
     #[test]
-    fn test_intfastfield_large_numbers() {
+    fn test_intfastfield_large_numbers() -> crate::Result<()> {
         let path = Path::new("test");
-        let mut directory: RAMDirectory = RAMDirectory::create();
+        let directory: RamDirectory = RamDirectory::create();
 
         {
             let write: WritePtr = directory.open_write(Path::new("test")).unwrap();
-            let mut serializer = FastFieldSerializer::from_write(write).unwrap();
+            let mut serializer = CompositeFastFieldSerializer::from_write(write).unwrap();
             let mut fast_field_writers = FastFieldsWriter::from_schema(&SCHEMA);
             // forcing the amplitude to be high
             fast_field_writers.add_document(&doc!(*FIELD=>0u64));
@@ -296,18 +362,16 @@ mod tests {
                 fast_field_writers.add_document(&doc!(*FIELD=>5_000_000_000_000_000_000u64 + i));
             }
             fast_field_writers
-                .serialize(&mut serializer, &HashMap::new())
+                .serialize(&mut serializer, &HashMap::new(), None)
                 .unwrap();
             serializer.close().unwrap();
         }
-        let source = directory.open_read(&path).unwrap();
+        let file = directory.open_read(&path).unwrap();
+        assert_eq!(file.len(), 80042 as usize);
         {
-            assert_eq!(source.len(), 80042 as usize);
+            let fast_fields_composite = CompositeFile::open(&file)?;
-        }
-        {
-            let fast_fields_composite = CompositeFile::open(&source).unwrap();
             let data = fast_fields_composite.open_read(*FIELD).unwrap();
-            let fast_field_reader = FastFieldReader::<u64>::open(data);
+            let fast_field_reader = BitpackedFastFieldReader::<u64>::open(data)?;
             assert_eq!(fast_field_reader.get(0), 0u64);
             for doc in 1..10_001 {
                 assert_eq!(
@@ -316,19 +380,20 @@ mod tests {
                 );
             }
         }
+        Ok(())
     }
 
     #[test]
-    fn test_signed_intfastfield() {
+    fn test_signed_intfastfield() -> crate::Result<()> {
         let path = Path::new("test");
-        let mut directory: RAMDirectory = RAMDirectory::create();
+        let directory: RamDirectory = RamDirectory::create();
         let mut schema_builder = Schema::builder();
 
         let i64_field = schema_builder.add_i64_field("field", FAST);
         let schema = schema_builder.build();
         {
             let write: WritePtr = directory.open_write(Path::new("test")).unwrap();
-            let mut serializer = FastFieldSerializer::from_write(write).unwrap();
+            let mut serializer = CompositeFastFieldSerializer::from_write(write).unwrap();
             let mut fast_field_writers = FastFieldsWriter::from_schema(&schema);
             for i in -100i64..10_000i64 {
                 let mut doc = Document::default();
@@ -336,18 +401,16 @@ mod tests {
                 fast_field_writers.add_document(&doc);
             }
             fast_field_writers
-                .serialize(&mut serializer, &HashMap::new())
+                .serialize(&mut serializer, &HashMap::new(), None)
                 .unwrap();
             serializer.close().unwrap();
         }
-        let source = directory.open_read(&path).unwrap();
+        let file = directory.open_read(&path).unwrap();
+        assert_eq!(file.len(), 17709 as usize);
         {
-            assert_eq!(source.len(), 17709 as usize);
+            let fast_fields_composite = CompositeFile::open(&file)?;
-        }
-        {
-            let fast_fields_composite = CompositeFile::open(&source).unwrap();
             let data = fast_fields_composite.open_read(i64_field).unwrap();
-            let fast_field_reader = FastFieldReader::<i64>::open(data);
+            let fast_field_reader = BitpackedFastFieldReader::<i64>::open(data)?;
 
             assert_eq!(fast_field_reader.min_value(), -100i64);
             assert_eq!(fast_field_reader.max_value(), 9_999i64);
@@ -360,35 +423,37 @@ mod tests {
                 assert_eq!(buffer[i], -100i64 + 53i64 + i as i64);
             }
         }
+        Ok(())
     }
 
     #[test]
-    fn test_signed_intfastfield_default_val() {
+    fn test_signed_intfastfield_default_val() -> crate::Result<()> {
         let path = Path::new("test");
-        let mut directory: RAMDirectory = RAMDirectory::create();
+        let directory: RamDirectory = RamDirectory::create();
         let mut schema_builder = Schema::builder();
         let i64_field = schema_builder.add_i64_field("field", FAST);
         let schema = schema_builder.build();
 
         {
             let write: WritePtr = directory.open_write(Path::new("test")).unwrap();
-            let mut serializer = FastFieldSerializer::from_write(write).unwrap();
+            let mut serializer = CompositeFastFieldSerializer::from_write(write).unwrap();
             let mut fast_field_writers = FastFieldsWriter::from_schema(&schema);
             let doc = Document::default();
             fast_field_writers.add_document(&doc);
             fast_field_writers
-                .serialize(&mut serializer, &HashMap::new())
+                .serialize(&mut serializer, &HashMap::new(), None)
                 .unwrap();
             serializer.close().unwrap();
         }
 
-        let source = directory.open_read(&path).unwrap();
+        let file = directory.open_read(&path).unwrap();
         {
-            let fast_fields_composite = CompositeFile::open(&source).unwrap();
+            let fast_fields_composite = CompositeFile::open(&file).unwrap();
             let data = fast_fields_composite.open_read(i64_field).unwrap();
-            let fast_field_reader = FastFieldReader::<i64>::open(data);
+            let fast_field_reader = BitpackedFastFieldReader::<i64>::open(data)?;
             assert_eq!(fast_field_reader.get(0u32), 0i64);
         }
+        Ok(())
     }
 
     // Warning: this generates the same permutation at each call
@@ -399,28 +464,26 @@ mod tests {
     }
 
     #[test]
-    fn test_intfastfield_permutation() {
+    fn test_intfastfield_permutation() -> crate::Result<()> {
         let path = Path::new("test");
         let permutation = generate_permutation();
         let n = permutation.len();
-        let mut directory = RAMDirectory::create();
+        let directory = RamDirectory::create();
         {
-            let write: WritePtr = directory.open_write(Path::new("test")).unwrap();
+            let write: WritePtr = directory.open_write(Path::new("test"))?;
-            let mut serializer = FastFieldSerializer::from_write(write).unwrap();
+            let mut serializer = CompositeFastFieldSerializer::from_write(write)?;
             let mut fast_field_writers = FastFieldsWriter::from_schema(&SCHEMA);
             for &x in &permutation {
                 fast_field_writers.add_document(&doc!(*FIELD=>x));
             }
-            fast_field_writers
+            fast_field_writers.serialize(&mut serializer, &HashMap::new(), None)?;
-                .serialize(&mut serializer, &HashMap::new())
+            serializer.close()?;
-                .unwrap();
-            serializer.close().unwrap();
         }
-        let source = directory.open_read(&path).unwrap();
+        let file = directory.open_read(&path)?;
         {
-            let fast_fields_composite = CompositeFile::open(&source).unwrap();
+            let fast_fields_composite = CompositeFile::open(&file)?;
             let data = fast_fields_composite.open_read(*FIELD).unwrap();
-            let fast_field_reader = FastFieldReader::<u64>::open(data);
+            let fast_field_reader = BitpackedFastFieldReader::<u64>::open(data)?;
 
             let mut a = 0u64;
             for _ in 0..n {
@@ -428,8 +491,95 @@ mod tests {
                 a = fast_field_reader.get(a as u32);
             }
         }
+        Ok(())
     }
 
+    #[test]
+    fn test_merge_missing_date_fast_field() {
+        let mut schema_builder = Schema::builder();
+        let date_field = schema_builder.add_date_field("date", FAST);
+        let schema = schema_builder.build();
+        let index = Index::create_in_ram(schema);
+        let mut index_writer = index.writer_for_tests().unwrap();
+        index_writer.set_merge_policy(Box::new(NoMergePolicy));
+        index_writer.add_document(doc!(date_field =>crate::chrono::prelude::Utc::now()));
+        index_writer.commit().unwrap();
+        index_writer.add_document(doc!());
+        index_writer.commit().unwrap();
+        let reader = index.reader().unwrap();
+        let segment_ids: Vec<SegmentId> = reader
+            .searcher()
+            .segment_readers()
+            .iter()
+            .map(SegmentReader::segment_id)
+            .collect();
+        assert_eq!(segment_ids.len(), 2);
+        let merge_future = index_writer.merge(&segment_ids[..]);
+        let merge_res = futures::executor::block_on(merge_future);
+        assert!(merge_res.is_ok());
+        assert!(reader.reload().is_ok());
+        assert_eq!(reader.searcher().segment_readers().len(), 1);
+    }
+
+    #[test]
+    fn test_default_datetime() {
+        assert_eq!(crate::DateTime::make_zero().timestamp(), 0i64);
+    }
+
+    #[test]
+    fn test_datefastfield() {
+        use crate::fastfield::FastValue;
+        let mut schema_builder = Schema::builder();
+        let date_field = schema_builder.add_date_field("date", FAST);
+        let multi_date_field = schema_builder.add_date_field(
+            "multi_date",
+            IntOptions::default().set_fast(Cardinality::MultiValues),
+        );
+        let schema = schema_builder.build();
+        let index = Index::create_in_ram(schema);
+        let mut index_writer = index.writer_for_tests().unwrap();
+        index_writer.set_merge_policy(Box::new(NoMergePolicy));
+        index_writer.add_document(doc!(
+            date_field => crate::DateTime::from_u64(1i64.to_u64()),
+            multi_date_field => crate::DateTime::from_u64(2i64.to_u64()),
+            multi_date_field => crate::DateTime::from_u64(3i64.to_u64())
+        ));
+        index_writer.add_document(doc!(
+            date_field => crate::DateTime::from_u64(4i64.to_u64())
+        ));
+        index_writer.add_document(doc!(
+            multi_date_field => crate::DateTime::from_u64(5i64.to_u64()),
+            multi_date_field => crate::DateTime::from_u64(6i64.to_u64())
+        ));
+        index_writer.commit().unwrap();
+        let reader = index.reader().unwrap();
+        let searcher = reader.searcher();
+        assert_eq!(searcher.segment_readers().len(), 1);
+        let segment_reader = searcher.segment_reader(0);
+        let fast_fields = segment_reader.fast_fields();
+        let date_fast_field = fast_fields.date(date_field).unwrap();
+        let dates_fast_field = fast_fields.dates(multi_date_field).unwrap();
+        let mut dates = vec![];
+        {
+            assert_eq!(date_fast_field.get(0u32).timestamp(), 1i64);
+            dates_fast_field.get_vals(0u32, &mut dates);
+            assert_eq!(dates.len(), 2);
+            assert_eq!(dates[0].timestamp(), 2i64);
+            assert_eq!(dates[1].timestamp(), 3i64);
+        }
+        {
+            assert_eq!(date_fast_field.get(1u32).timestamp(), 4i64);
+            dates_fast_field.get_vals(1u32, &mut dates);
+            assert!(dates.is_empty());
+        }
+        {
+            assert_eq!(date_fast_field.get(2u32).timestamp(), 0i64);
+            dates_fast_field.get_vals(2u32, &mut dates);
+            assert_eq!(dates.len(), 2);
+            assert_eq!(dates[0].timestamp(), 5i64);
+            assert_eq!(dates[1].timestamp(), 6i64);
+        }
+    }
 }
 
 #[cfg(all(test, feature = "unstable"))]
@@ -437,9 +587,9 @@ mod bench {
     use super::tests::FIELD;
     use super::tests::{generate_permutation, SCHEMA};
     use super::*;
-    use common::CompositeFile;
+    use crate::common::CompositeFile;
-    use directory::{Directory, RAMDirectory, WritePtr};
+    use crate::directory::{Directory, RamDirectory, WritePtr};
-    use fastfield::FastFieldReader;
+    use crate::fastfield::FastFieldReader;
     use std::collections::HashMap;
     use std::path::Path;
     use test::{self, Bencher};
@@ -474,7 +624,7 @@ mod bench {
     fn bench_intfastfield_linear_fflookup(b: &mut Bencher) {
         let path = Path::new("test");
         let permutation = generate_permutation();
-        let mut directory: RAMDirectory = RAMDirectory::create();
+        let directory: RamDirectory = RamDirectory::create();
         {
             let write: WritePtr = directory.open_write(Path::new("test")).unwrap();
             let mut serializer = FastFieldSerializer::from_write(write).unwrap();
@@ -483,15 +633,15 @@ mod bench {
                 fast_field_writers.add_document(&doc!(*FIELD=>x));
             }
             fast_field_writers
-                .serialize(&mut serializer, &HashMap::new())
+                .serialize(&mut serializer, &HashMap::new(), None)
                 .unwrap();
             serializer.close().unwrap();
         }
-        let source = directory.open_read(&path).unwrap();
+        let file = directory.open_read(&path).unwrap();
         {
-            let fast_fields_composite = CompositeFile::open(&source).unwrap();
+            let fast_fields_composite = CompositeFile::open(&file).unwrap();
             let data = fast_fields_composite.open_read(*FIELD).unwrap();
-            let fast_field_reader = FastFieldReader::<u64>::open(data);
+            let fast_field_reader = FastFieldReader::<u64>::open(data).unwrap();
 
             b.iter(|| {
                 let n = test::black_box(7000u32);
@@ -508,7 +658,7 @@ mod bench {
     fn bench_intfastfield_fflookup(b: &mut Bencher) {
         let path = Path::new("test");
         let permutation = generate_permutation();
-        let mut directory: RAMDirectory = RAMDirectory::create();
+        let directory: RamDirectory = RamDirectory::create();
         {
             let write: WritePtr = directory.open_write(Path::new("test")).unwrap();
             let mut serializer = FastFieldSerializer::from_write(write).unwrap();
@@ -517,15 +667,15 @@ mod bench {
                 fast_field_writers.add_document(&doc!(*FIELD=>x));
             }
             fast_field_writers
-                .serialize(&mut serializer, &HashMap::new())
+                .serialize(&mut serializer, &HashMap::new(), None)
                 .unwrap();
             serializer.close().unwrap();
         }
-        let source = directory.open_read(&path).unwrap();
+        let file = directory.open_read(&path).unwrap();
         {
-            let fast_fields_composite = CompositeFile::open(&source).unwrap();
+            let fast_fields_composite = CompositeFile::open(&file).unwrap();
             let data = fast_fields_composite.open_read(*FIELD).unwrap();
-            let fast_field_reader = FastFieldReader::<u64>::open(data);
+            let fast_field_reader = FastFieldReader::<u64>::open(data).unwrap();
 
             b.iter(|| {
                 let n = test::black_box(1000u32);
@@ -537,5 +687,4 @@ mod bench {
             });
         }
     }
-
 }

src/fastfield/multivalued/mod.rs

@@ -1,22 +1,21 @@
 mod reader;
 mod writer;
 
-pub use self::reader::MultiValueIntFastFieldReader;
+pub use self::reader::MultiValuedFastFieldReader;
-pub use self::writer::MultiValueIntFastFieldWriter;
+pub use self::writer::MultiValuedFastFieldWriter;
 
 #[cfg(test)]
 mod tests {
 
-    use time;
-
-    use self::time::Duration;
     use crate::collector::TopDocs;
     use crate::query::QueryParser;
     use crate::schema::Cardinality;
     use crate::schema::Facet;
     use crate::schema::IntOptions;
     use crate::schema::Schema;
+    use crate::schema::INDEXED;
     use crate::Index;
+    use chrono::Duration;
 
     #[test]
     fn test_multivalued_u64() {
@@ -27,7 +26,7 @@ mod tests {
         );
         let schema = schema_builder.build();
         let index = Index::create_in_ram(schema);
-        let mut index_writer = index.writer_with_num_threads(1, 3_000_000).unwrap();
+        let mut index_writer = index.writer_for_tests().unwrap();
         index_writer.add_document(doc!(field=>1u64, field=>3u64));
         index_writer.add_document(doc!());
         index_writer.add_document(doc!(field=>4u64));
@@ -66,7 +65,7 @@ mod tests {
             schema_builder.add_i64_field("time_stamp_i", IntOptions::default().set_stored());
         let schema = schema_builder.build();
         let index = Index::create_in_ram(schema);
-        let mut index_writer = index.writer_with_num_threads(1, 3_000_000).unwrap();
+        let mut index_writer = index.writer_for_tests().unwrap();
         let first_time_stamp = chrono::Utc::now();
         index_writer.add_document(
             doc!(date_field=>first_time_stamp, date_field=>first_time_stamp, time_i=>1i64),
@@ -78,12 +77,15 @@ mod tests {
         // add another second
         let two_secs_ahead = first_time_stamp + Duration::seconds(2);
         index_writer.add_document(doc!(date_field=>two_secs_ahead, date_field=>two_secs_ahead,date_field=>two_secs_ahead, time_i=>3i64));
+        // add three seconds
+        index_writer
+            .add_document(doc!(date_field=>first_time_stamp + Duration::seconds(3), time_i=>4i64));
         assert!(index_writer.commit().is_ok());
 
         let reader = index.reader().unwrap();
         let searcher = reader.searcher();
         let reader = searcher.segment_reader(0);
-        assert_eq!(reader.num_docs(), 4);
+        assert_eq!(reader.num_docs(), 5);
 
         {
             let parser = QueryParser::for_index(&index, vec![date_field]);
@@ -102,6 +104,7 @@ mod tests {
                         .get_first(date_field)
                         .expect("cannot find value")
                         .date_value()
+                        .unwrap()
                         .timestamp(),
                     first_time_stamp.timestamp()
                 );
@@ -110,7 +113,7 @@ mod tests {
                         .get_first(time_i)
                         .expect("cannot find value")
                         .i64_value(),
-                    1i64
+                    Some(1i64)
                 );
             }
         }
@@ -133,6 +136,7 @@ mod tests {
                         .get_first(date_field)
                         .expect("cannot find value")
                         .date_value()
+                        .unwrap()
                         .timestamp(),
                     two_secs_ahead.timestamp()
                 );
@@ -141,42 +145,55 @@ mod tests {
                         .get_first(time_i)
                         .expect("cannot find value")
                         .i64_value(),
-                    3i64
+                    Some(3i64)
                 );
             }
         }
 
-        // TODO: support Date range queries
+        {
-        //        {
+            let parser = QueryParser::for_index(&index, vec![date_field]);
-        //            let parser = QueryParser::for_index(&index, vec![date_field]);
+            let range_q = format!(
-        //            let range_q = format!("\"{}\"..\"{}\"",
+                "[{} TO {}}}",
-        //                                  (first_time_stamp + Duration::seconds(1)).to_rfc3339(),
+                (first_time_stamp + Duration::seconds(1)).to_rfc3339(),
-        //                                  (first_time_stamp + Duration::seconds(3)).to_rfc3339()
+                (first_time_stamp + Duration::seconds(3)).to_rfc3339()
-        //            );
+            );
-        //            let query = parser.parse_query(&range_q)
+            let query = parser.parse_query(&range_q).expect("could not parse query");
-        //                .expect("could not parse query");
+            let results = searcher
-        //            let results = searcher.search(&query, &TopDocs::with_limit(5))
+                .search(&query, &TopDocs::with_limit(5))
-        //                .expect("could not query index");
+                .expect("could not query index");
-        //
+
-        //
+            assert_eq!(results.len(), 2);
-        //            assert_eq!(results.len(), 2);
+            for (i, doc_pair) in results.iter().enumerate() {
-        //            for (i, doc_pair) in results.iter().enumerate() {
+                let retrieved_doc = searcher.doc(doc_pair.1).expect("cannot fetch doc");
-        //                let retrieved_doc = searcher.doc(doc_pair.1).expect("cannot fetch doc");
+                let offset_sec = match i {
-        //                let offset_sec = match i {
+                    0 => 1,
-        //                    0 => 1,
+                    1 => 2,
-        //                    1 => 3,
+                    _ => panic!("should not have more than 2 docs"),
-        //                    _ => panic!("should not have more than 2 docs")
+                };
-        //                };
+                let time_i_val = match i {
-        //                let time_i_val = match i {
+                    0 => 2,
-        //                    0 => 2,
+                    1 => 3,
-        //                    1 => 3,
+                    _ => panic!("should not have more than 2 docs"),
-        //                    _ => panic!("should not have more than 2 docs")
+                };
-        //                };
+                assert_eq!(
-        //                assert_eq!(retrieved_doc.get_first(date_field).expect("cannot find value").date_value().timestamp(),
+                    retrieved_doc
-        //                           (first_time_stamp + Duration::seconds(offset_sec)).timestamp());
+                        .get_first(date_field)
-        //                assert_eq!(retrieved_doc.get_first(time_i).expect("cannot find value").i64_value(), time_i_val);
+                        .expect("cannot find value")
-        //            }
+                        .date_value()
-        //        }
+                        .expect("value not of Date type")
+                        .timestamp(),
+                    (first_time_stamp + Duration::seconds(offset_sec)).timestamp()
+                );
+                assert_eq!(
+                    retrieved_doc
+                        .get_first(time_i)
+                        .expect("cannot find value")
+                        .i64_value()
+                        .expect("value not of i64 type"),
+                    time_i_val
+                );
+            }
+        }
     }
 
     #[test]
@@ -188,7 +205,7 @@ mod tests {
         );
         let schema = schema_builder.build();
         let index = Index::create_in_ram(schema);
-        let mut index_writer = index.writer_with_num_threads(1, 3_000_000).unwrap();
+        let mut index_writer = index.writer_for_tests().unwrap();
         index_writer.add_document(doc!(field=> 1i64, field => 3i64));
         index_writer.add_document(doc!());
         index_writer.add_document(doc!(field=> -4i64));
@@ -199,31 +216,23 @@ mod tests {
         let segment_reader = searcher.segment_reader(0);
         let mut vals = Vec::new();
         let multi_value_reader = segment_reader.fast_fields().i64s(field).unwrap();
-        {
+        multi_value_reader.get_vals(2, &mut vals);
-            multi_value_reader.get_vals(2, &mut vals);
+        assert_eq!(&vals, &[-4i64]);
-            assert_eq!(&vals, &[-4i64]);
+        multi_value_reader.get_vals(0, &mut vals);
-        }
+        assert_eq!(&vals, &[1i64, 3i64]);
-        {
+        multi_value_reader.get_vals(1, &mut vals);
-            multi_value_reader.get_vals(0, &mut vals);
+        assert!(vals.is_empty());
-            assert_eq!(&vals, &[1i64, 3i64]);
+        multi_value_reader.get_vals(3, &mut vals);
-        }
+        assert_eq!(&vals, &[-5i64, -20i64, 1i64]);
-        {
-            multi_value_reader.get_vals(1, &mut vals);
-            assert!(vals.is_empty());
-        }
-        {
-            multi_value_reader.get_vals(3, &mut vals);
-            assert_eq!(&vals, &[-5i64, -20i64, 1i64]);
-        }
     }
     #[test]
     #[ignore]
     fn test_many_facets() {
         let mut schema_builder = Schema::builder();
-        let field = schema_builder.add_facet_field("facetfield");
+        let field = schema_builder.add_facet_field("facetfield", INDEXED);
         let schema = schema_builder.build();
         let index = Index::create_in_ram(schema);
-        let mut index_writer = index.writer_with_num_threads(1, 3_000_000).unwrap();
+        let mut index_writer = index.writer_for_tests().unwrap();
         for i in 0..100_000 {
             index_writer.add_document(doc!(field=> Facet::from(format!("/lang/{}", i).as_str())));
         }

src/fastfield/multivalued/reader.rs

@@ -1,4 +1,6 @@
-use crate::fastfield::{FastFieldReader, FastValue};
+use std::ops::Range;
+
+use crate::fastfield::{BitpackedFastFieldReader, FastFieldReader, FastValue, MultiValueLength};
 use crate::DocId;
 
 /// Reader for a multivalued `u64` fast field.
@@ -10,49 +12,42 @@ use crate::DocId;
 /// The `idx_reader` associated, for each document, the index of its first value.
 ///
 #[derive(Clone)]
-pub struct MultiValueIntFastFieldReader<Item: FastValue> {
+pub struct MultiValuedFastFieldReader<Item: FastValue> {
-    idx_reader: FastFieldReader<u64>,
+    idx_reader: BitpackedFastFieldReader<u64>,
-    vals_reader: FastFieldReader<Item>,
+    vals_reader: BitpackedFastFieldReader<Item>,
 }
 
-impl<Item: FastValue> MultiValueIntFastFieldReader<Item> {
+impl<Item: FastValue> MultiValuedFastFieldReader<Item> {
     pub(crate) fn open(
-        idx_reader: FastFieldReader<u64>,
+        idx_reader: BitpackedFastFieldReader<u64>,
-        vals_reader: FastFieldReader<Item>,
+        vals_reader: BitpackedFastFieldReader<Item>,
-    ) -> MultiValueIntFastFieldReader<Item> {
+    ) -> MultiValuedFastFieldReader<Item> {
-        MultiValueIntFastFieldReader {
+        MultiValuedFastFieldReader {
             idx_reader,
             vals_reader,
         }
     }
 
-    pub(crate) fn into_u64s_reader(self) -> MultiValueIntFastFieldReader<u64> {
-        MultiValueIntFastFieldReader {
-            idx_reader: self.idx_reader,
-            vals_reader: self.vals_reader.into_u64_reader(),
-        }
-    }
-
     /// Returns `(start, stop)`, such that the values associated
     /// to the given document are `start..stop`.
-    fn range(&self, doc: DocId) -> (u64, u64) {
+    fn range(&self, doc: DocId) -> Range<u64> {
         let start = self.idx_reader.get(doc);
         let stop = self.idx_reader.get(doc + 1);
-        (start, stop)
+        start..stop
     }
 
     /// Returns the array of values associated to the given `doc`.
     pub fn get_vals(&self, doc: DocId, vals: &mut Vec<Item>) {
-        let (start, stop) = self.range(doc);
+        let range = self.range(doc);
-        let len = (stop - start) as usize;
+        let len = (range.end - range.start) as usize;
-        vals.resize(len, Item::default());
+        vals.resize(len, Item::make_zero());
-        self.vals_reader.get_range_u64(start, &mut vals[..]);
+        self.vals_reader.get_range_u64(range.start, &mut vals[..]);
     }
 
     /// Returns the number of values associated with the document `DocId`.
     pub fn num_vals(&self, doc: DocId) -> usize {
-        let (start, stop) = self.range(doc);
+        let range = self.range(doc);
-        (stop - start) as usize
+        (range.end - range.start) as usize
     }
 
     /// Returns the overall number of values in this field  .
@@ -61,20 +56,29 @@ impl<Item: FastValue> MultiValueIntFastFieldReader<Item> {
     }
 }
 
+impl<Item: FastValue> MultiValueLength for MultiValuedFastFieldReader<Item> {
+    fn get_len(&self, doc_id: DocId) -> u64 {
+        self.num_vals(doc_id) as u64
+    }
+
+    fn get_total_len(&self) -> u64 {
+        self.total_num_vals() as u64
+    }
+}
 #[cfg(test)]
 mod tests {
 
     use crate::core::Index;
-    use crate::schema::{Facet, Schema};
+    use crate::schema::{Facet, Schema, INDEXED};
 
     #[test]
     fn test_multifastfield_reader() {
         let mut schema_builder = Schema::builder();
-        let facet_field = schema_builder.add_facet_field("facets");
+        let facet_field = schema_builder.add_facet_field("facets", INDEXED);
         let schema = schema_builder.build();
         let index = Index::create_in_ram(schema);
         let mut index_writer = index
-            .writer_with_num_threads(1, 30_000_000)
+            .writer_for_tests()
             .expect("Failed to create index writer.");
         index_writer.add_document(doc!(
             facet_field => Facet::from("/category/cat2"),

src/fastfield/multivalued/writer.rs

@@ -1,13 +1,14 @@
-use crate::fastfield::serializer::FastSingleFieldSerializer;
+use crate::fastfield::serializer::DynamicFastFieldSerializer;
-use crate::fastfield::value_to_u64;
+use crate::fastfield::serializer::FastFieldSerializer;
-use crate::fastfield::FastFieldSerializer;
+use crate::fastfield::CompositeFastFieldSerializer;
 use crate::postings::UnorderedTermId;
 use crate::schema::{Document, Field};
 use crate::termdict::TermOrdinal;
 use crate::DocId;
-use itertools::Itertools;
+use crate::{fastfield::value_to_u64, indexer::doc_id_mapping::DocIdMapping};
-use std::collections::HashMap;
+use fnv::FnvHashMap;
 use std::io;
+use tantivy_bitpacker::minmax;
 
 /// Writer for multi-valued (as in, more than one value per document)
 /// int fast field.
@@ -19,7 +20,7 @@ use std::io;
 /// in your schema
 /// - add your document simply by calling `.add_document(...)`.
 ///
-/// The `MultiValueIntFastFieldWriter` can be acquired from the
+/// The `MultiValuedFastFieldWriter` can be acquired from the
 /// fastfield writer, by calling [`.get_multivalue_writer(...)`](./struct.FastFieldsWriter.html#method.get_multivalue_writer).
 ///
 /// Once acquired, writing is done by calling calls to
@@ -30,17 +31,17 @@ use std::io;
 /// This makes it possible to push unordered term ids,
 /// during indexing and remap them to their respective
 /// term ids when the segment is getting serialized.
-pub struct MultiValueIntFastFieldWriter {
+pub struct MultiValuedFastFieldWriter {
     field: Field,
     vals: Vec<UnorderedTermId>,
     doc_index: Vec<u64>,
     is_facet: bool,
 }
 
-impl MultiValueIntFastFieldWriter {
+impl MultiValuedFastFieldWriter {
     /// Creates a new `IntFastFieldWriter`
     pub(crate) fn new(field: Field, is_facet: bool) -> Self {
-        MultiValueIntFastFieldWriter {
+        MultiValuedFastFieldWriter {
             field,
             vals: Vec::new(),
             doc_index: Vec::new(),
@@ -48,7 +49,13 @@ impl MultiValueIntFastFieldWriter {
         }
     }
 
-    /// Access the field associated to the `MultiValueIntFastFieldWriter`
+    /// The memory used (inclusive childs)
+    pub fn mem_usage(&self) -> usize {
+        self.vals.capacity() * std::mem::size_of::<UnorderedTermId>()
+            + self.doc_index.capacity() * std::mem::size_of::<u64>()
+    }
+
+    /// Access the field associated to the `MultiValuedFastFieldWriter`
     pub fn field(&self) -> Field {
         self.field
     }
@@ -87,7 +94,34 @@ impl MultiValueIntFastFieldWriter {
         self.vals.extend_from_slice(vals);
         doc
     }
+    /// Returns an iterator over values per doc_id in ascending doc_id order.
+    ///
+    /// Normally the order is simply iterating self.doc_id_index.
+    /// With doc_id_map it accounts for the new mapping, returning values in the order of the
+    /// new doc_ids.
+    fn get_ordered_values<'a: 'b, 'b>(
+        &'a self,
+        doc_id_map: Option<&'b DocIdMapping>,
+    ) -> impl Iterator<Item = &'b [u64]> {
+        let doc_id_iter: Box<dyn Iterator<Item = u32>> = if let Some(doc_id_map) = doc_id_map {
+            Box::new(doc_id_map.iter_old_doc_ids())
+        } else {
+            let max_doc = self.doc_index.len() as DocId;
+            Box::new(0..max_doc)
+        };
+        doc_id_iter.map(move |doc_id| self.get_values_for_doc_id(doc_id))
+    }
 
+    /// returns all values for a doc_ids
+    fn get_values_for_doc_id(&self, doc_id: u32) -> &[u64] {
+        let start_pos = self.doc_index[doc_id as usize] as usize;
+        let end_pos = self
+            .doc_index
+            .get(doc_id as usize + 1)
+            .cloned()
+            .unwrap_or(self.vals.len() as u64) as usize; // special case, last doc_id has no offset information
+        &self.vals[start_pos..end_pos]
+    }
     /// Serializes fast field values by pushing them to the `FastFieldSerializer`.
     ///
     /// If a mapping is given, the values are remapped *and sorted* before serialization.
@@ -101,22 +135,27 @@ impl MultiValueIntFastFieldWriter {
     ///
     pub fn serialize(
         &self,
-        serializer: &mut FastFieldSerializer,
+        serializer: &mut CompositeFastFieldSerializer,
-        mapping_opt: Option<&HashMap<UnorderedTermId, TermOrdinal>>,
+        mapping_opt: Option<&FnvHashMap<UnorderedTermId, TermOrdinal>>,
+        doc_id_map: Option<&DocIdMapping>,
     ) -> io::Result<()> {
         {
             // writing the offset index
             let mut doc_index_serializer =
                 serializer.new_u64_fast_field_with_idx(self.field, 0, self.vals.len() as u64, 0)?;
-            for &offset in &self.doc_index {
+
+            let mut offset = 0;
+            for vals in self.get_ordered_values(doc_id_map) {
                 doc_index_serializer.add_val(offset)?;
+                offset += vals.len() as u64;
             }
             doc_index_serializer.add_val(self.vals.len() as u64)?;
+
             doc_index_serializer.close_field()?;
         }
         {
             // writing the values themselves.
-            let mut value_serializer: FastSingleFieldSerializer<'_, _>;
+            let mut value_serializer: DynamicFastFieldSerializer<'_, _>;
             match mapping_opt {
                 Some(mapping) => {
                     value_serializer = serializer.new_u64_fast_field_with_idx(
@@ -126,37 +165,29 @@ impl MultiValueIntFastFieldWriter {
                         1,
                     )?;
 
-                    let last_interval = (
-                        self.doc_index.last().cloned().unwrap(),
-                        self.vals.len() as u64,
-                    );
-
                     let mut doc_vals: Vec<u64> = Vec::with_capacity(100);
-                    for (start, stop) in self
+                    for vals in self.get_ordered_values(doc_id_map) {
-                        .doc_index
-                        .windows(2)
-                        .map(|interval| (interval[0], interval[1]))
-                        .chain(Some(last_interval).into_iter())
-                        .map(|(start, stop)| (start as usize, stop as usize))
-                    {
                         doc_vals.clear();
-                        let remapped_vals = self.vals[start..stop]
+                        let remapped_vals = vals
                             .iter()
                             .map(|val| *mapping.get(val).expect("Missing term ordinal"));
                         doc_vals.extend(remapped_vals);
-                        doc_vals.sort();
+                        doc_vals.sort_unstable();
                         for &val in &doc_vals {
                             value_serializer.add_val(val)?;
                         }
                     }
                 }
                 None => {
-                    let val_min_max = self.vals.iter().cloned().minmax();
+                    let val_min_max = minmax(self.vals.iter().cloned());
-                    let (val_min, val_max) = val_min_max.into_option().unwrap_or((0u64, 0u64));
+                    let (val_min, val_max) = val_min_max.unwrap_or((0u64, 0u64));
                     value_serializer =
                         serializer.new_u64_fast_field_with_idx(self.field, val_min, val_max, 1)?;
-                    for &val in &self.vals {
+                    for vals in self.get_ordered_values(doc_id_map) {
-                        value_serializer.add_val(val)?;
+                        // sort values in case of remapped doc_ids?
+                        for &val in vals {
+                            value_serializer.add_val(val)?;
+                        }
                     }
                 }
             }

src/fastfield/reader.rs

@@ -1,64 +1,21 @@
 use super::FastValue;
-use crate::common::bitpacker::BitUnpacker;
-use crate::common::compute_num_bits;
 use crate::common::BinarySerializable;
 use crate::common::CompositeFile;
-use crate::directory::ReadOnlySource;
+use crate::directory::FileSlice;
-use crate::directory::{Directory, RAMDirectory, WritePtr};
+use crate::directory::OwnedBytes;
-use crate::fastfield::{FastFieldSerializer, FastFieldsWriter};
+use crate::directory::{Directory, RamDirectory, WritePtr};
+use crate::fastfield::{CompositeFastFieldSerializer, FastFieldsWriter};
 use crate::schema::Schema;
 use crate::schema::FAST;
 use crate::DocId;
-use owning_ref::OwningRef;
 use std::collections::HashMap;
 use std::marker::PhantomData;
 use std::path::Path;
+use tantivy_bitpacker::compute_num_bits;
+use tantivy_bitpacker::BitUnpacker;
 
-/// Trait for accessing a fastfield.
+/// FastFieldReader is the trait to access fast field data.
-///
+pub trait FastFieldReader<Item: FastValue>: Clone {
-/// Depending on the field type, a different
-/// fast field is required.
-#[derive(Clone)]
-pub struct FastFieldReader<Item: FastValue> {
-    bit_unpacker: BitUnpacker<OwningRef<ReadOnlySource, [u8]>>,
-    min_value_u64: u64,
-    max_value_u64: u64,
-    _phantom: PhantomData<Item>,
-}
-
-impl<Item: FastValue> FastFieldReader<Item> {
-    /// Opens a fast field given a source.
-    pub fn open(data: ReadOnlySource) -> Self {
-        let min_value: u64;
-        let amplitude: u64;
-        {
-            let mut cursor = data.as_slice();
-            min_value =
-                u64::deserialize(&mut cursor).expect("Failed to read the min_value of fast field.");
-            amplitude =
-                u64::deserialize(&mut cursor).expect("Failed to read the amplitude of fast field.");
-        }
-        let max_value = min_value + amplitude;
-        let num_bits = compute_num_bits(amplitude);
-        let owning_ref = OwningRef::new(data).map(|data| &data[16..]);
-        let bit_unpacker = BitUnpacker::new(owning_ref, num_bits);
-        FastFieldReader {
-            min_value_u64: min_value,
-            max_value_u64: max_value,
-            bit_unpacker,
-            _phantom: PhantomData,
-        }
-    }
-
-    pub(crate) fn into_u64_reader(self) -> FastFieldReader<u64> {
-        FastFieldReader {
-            bit_unpacker: self.bit_unpacker,
-            min_value_u64: self.min_value_u64,
-            max_value_u64: self.max_value_u64,
-            _phantom: PhantomData,
-        }
-    }
-
     /// Return the value associated to the given document.
     ///
     /// This accessor should return as fast as possible.
@@ -66,13 +23,111 @@ impl<Item: FastValue> FastFieldReader<Item> {
     /// # Panics
     ///
     /// May panic if `doc` is greater than the segment
-    // `maxdoc`.
+    fn get(&self, doc: DocId) -> Item;
-    pub fn get(&self, doc: DocId) -> Item {
-        self.get_u64(u64::from(doc))
-    }
 
+    /// Fills an output buffer with the fast field values
+    /// associated with the `DocId` going from
+    /// `start` to `start + output.len()`.
+    ///
+    /// Regardless of the type of `Item`, this method works
+    /// - transmuting the output array
+    /// - extracting the `Item`s as if they were `u64`
+    /// - possibly converting the `u64` value to the right type.
+    ///
+    /// # Panics
+    ///
+    /// May panic if `start + output.len()` is greater than
+    /// the segment's `maxdoc`.
+    fn get_range(&self, start: DocId, output: &mut [Item]);
+
+    /// Returns the minimum value for this fast field.
+    ///
+    /// The max value does not take in account of possible
+    /// deleted document, and should be considered as an upper bound
+    /// of the actual maximum value.
+    fn min_value(&self) -> Item;
+
+    /// Returns the maximum value for this fast field.
+    ///
+    /// The max value does not take in account of possible
+    /// deleted document, and should be considered as an upper bound
+    /// of the actual maximum value.
+    fn max_value(&self) -> Item;
+}
+
+#[derive(Clone)]
+/// DynamicFastFieldReader wraps different readers to access
+/// the various encoded fastfield data
+///
+pub enum DynamicFastFieldReader<Item: FastValue> {
+    /// Bitpacked compressed fastfield data.
+    Bitpacked(BitpackedFastFieldReader<Item>),
+}
+
+impl<Item: FastValue> DynamicFastFieldReader<Item> {
+    /// Returns correct the reader wrapped in the `DynamicFastFieldReader` enum for the data.
+    pub fn open(file: FileSlice) -> crate::Result<DynamicFastFieldReader<Item>> {
+        Ok(DynamicFastFieldReader::Bitpacked(
+            BitpackedFastFieldReader::open(file)?,
+        ))
+    }
+}
+
+impl<Item: FastValue> FastFieldReader<Item> for DynamicFastFieldReader<Item> {
+    fn get(&self, doc: DocId) -> Item {
+        match self {
+            Self::Bitpacked(reader) => reader.get(doc),
+        }
+    }
+    fn get_range(&self, start: DocId, output: &mut [Item]) {
+        match self {
+            Self::Bitpacked(reader) => reader.get_range(start, output),
+        }
+    }
+    fn min_value(&self) -> Item {
+        match self {
+            Self::Bitpacked(reader) => reader.min_value(),
+        }
+    }
+    fn max_value(&self) -> Item {
+        match self {
+            Self::Bitpacked(reader) => reader.max_value(),
+        }
+    }
+}
+
+/// Trait for accessing a fastfield.
+///
+/// Depending on the field type, a different
+/// fast field is required.
+#[derive(Clone)]
+pub struct BitpackedFastFieldReader<Item: FastValue> {
+    bytes: OwnedBytes,
+    bit_unpacker: BitUnpacker,
+    min_value_u64: u64,
+    max_value_u64: u64,
+    _phantom: PhantomData<Item>,
+}
+
+impl<Item: FastValue> BitpackedFastFieldReader<Item> {
+    /// Opens a fast field given a file.
+    pub fn open(file: FileSlice) -> crate::Result<Self> {
+        let mut bytes = file.read_bytes()?;
+        let min_value = u64::deserialize(&mut bytes)?;
+        let amplitude = u64::deserialize(&mut bytes)?;
+        let max_value = min_value + amplitude;
+        let num_bits = compute_num_bits(amplitude);
+        let bit_unpacker = BitUnpacker::new(num_bits);
+        Ok(BitpackedFastFieldReader {
+            bytes,
+            min_value_u64: min_value,
+            max_value_u64: max_value,
+            bit_unpacker,
+            _phantom: PhantomData,
+        })
+    }
     pub(crate) fn get_u64(&self, doc: u64) -> Item {
-        Item::from_u64(self.min_value_u64 + self.bit_unpacker.get(doc))
+        Item::from_u64(self.min_value_u64 + self.bit_unpacker.get(doc, &self.bytes))
     }
 
     /// Internally `multivalued` also use SingleValue Fast fields.
@@ -92,6 +147,20 @@ impl<Item: FastValue> FastFieldReader<Item> {
             *out = self.get_u64(start + (i as u64));
         }
     }
+}
+
+impl<Item: FastValue> FastFieldReader<Item> for BitpackedFastFieldReader<Item> {
+    /// Return the value associated to the given document.
+    ///
+    /// This accessor should return as fast as possible.
+    ///
+    /// # Panics
+    ///
+    /// May panic if `doc` is greater than the segment
+    // `maxdoc`.
+    fn get(&self, doc: DocId) -> Item {
+        self.get_u64(u64::from(doc))
+    }
 
     /// Fills an output buffer with the fast field values
     /// associated with the `DocId` going from
@@ -106,7 +175,7 @@ impl<Item: FastValue> FastFieldReader<Item> {
     ///
     /// May panic if `start + output.len()` is greater than
     /// the segment's `maxdoc`.
-    pub fn get_range(&self, start: DocId, output: &mut [Item]) {
+    fn get_range(&self, start: DocId, output: &mut [Item]) {
         self.get_range_u64(u64::from(start), output);
     }
 
@@ -115,7 +184,7 @@ impl<Item: FastValue> FastFieldReader<Item> {
     /// The max value does not take in account of possible
     /// deleted document, and should be considered as an upper bound
     /// of the actual maximum value.
-    pub fn min_value(&self) -> Item {
+    fn min_value(&self) -> Item {
         Item::from_u64(self.min_value_u64)
     }
 
@@ -124,45 +193,44 @@ impl<Item: FastValue> FastFieldReader<Item> {
     /// The max value does not take in account of possible
     /// deleted document, and should be considered as an upper bound
     /// of the actual maximum value.
-    pub fn max_value(&self) -> Item {
+    fn max_value(&self) -> Item {
         Item::from_u64(self.max_value_u64)
     }
 }
 
-impl<Item: FastValue> From<Vec<Item>> for FastFieldReader<Item> {
+impl<Item: FastValue> From<Vec<Item>> for BitpackedFastFieldReader<Item> {
-    fn from(vals: Vec<Item>) -> FastFieldReader<Item> {
+    fn from(vals: Vec<Item>) -> BitpackedFastFieldReader<Item> {
         let mut schema_builder = Schema::builder();
         let field = schema_builder.add_u64_field("field", FAST);
         let schema = schema_builder.build();
         let path = Path::new("__dummy__");
-        let mut directory: RAMDirectory = RAMDirectory::create();
+        let directory: RamDirectory = RamDirectory::create();
         {
             let write: WritePtr = directory
                 .open_write(path)
-                .expect("With a RAMDirectory, this should never fail.");
+                .expect("With a RamDirectory, this should never fail.");
-            let mut serializer = FastFieldSerializer::from_write(write)
+            let mut serializer = CompositeFastFieldSerializer::from_write(write)
-                .expect("With a RAMDirectory, this should never fail.");
+                .expect("With a RamDirectory, this should never fail.");
             let mut fast_field_writers = FastFieldsWriter::from_schema(&schema);
             {
                 let fast_field_writer = fast_field_writers
-                    .get_field_writer(field)
+                    .get_field_writer_mut(field)
-                    .expect("With a RAMDirectory, this should never fail.");
+                    .expect("With a RamDirectory, this should never fail.");
                 for val in vals {
                     fast_field_writer.add_val(val.to_u64());
                 }
             }
             fast_field_writers
-                .serialize(&mut serializer, &HashMap::new())
+                .serialize(&mut serializer, &HashMap::new(), None)
                 .unwrap();
             serializer.close().unwrap();
         }
 
-        let source = directory.open_read(path).expect("Failed to open the file");
+        let file = directory.open_read(path).expect("Failed to open the file");
-        let composite_file =
+        let composite_file = CompositeFile::open(&file).expect("Failed to read the composite file");
-            CompositeFile::open(&source).expect("Failed to read the composite file");
+        let field_file = composite_file
-        let field_source = composite_file
             .open_read(field)
             .expect("File component not found");
-        FastFieldReader::open(field_source)
+        BitpackedFastFieldReader::open(field_file).unwrap()
     }
 }

src/fastfield/readers.rs

@@ -1,31 +1,30 @@
 use crate::common::CompositeFile;
-use crate::fastfield::BytesFastFieldReader;
+use crate::directory::FileSlice;
-use crate::fastfield::MultiValueIntFastFieldReader;
+use crate::fastfield::MultiValuedFastFieldReader;
-use crate::fastfield::{FastFieldNotAvailableError, FastFieldReader};
+use crate::fastfield::{BitpackedFastFieldReader, FastFieldNotAvailableError};
+use crate::fastfield::{BytesFastFieldReader, FastValue};
 use crate::schema::{Cardinality, Field, FieldType, Schema};
 use crate::space_usage::PerFieldSpaceUsage;
-use crate::Result;
+use crate::TantivyError;
-use std::collections::HashMap;
 
-/// Provides access to all of the FastFieldReader.
+use super::reader::DynamicFastFieldReader;
+use super::FastFieldReader;
+
+/// Provides access to all of the BitpackedFastFieldReader.
 ///
 /// Internally, `FastFieldReaders` have preloaded fast field readers,
 /// and just wraps several `HashMap`.
+#[derive(Clone)]
 pub struct FastFieldReaders {
-    fast_field_i64: HashMap<Field, FastFieldReader<i64>>,
+    schema: Schema,
-    fast_field_u64: HashMap<Field, FastFieldReader<u64>>,
-    fast_field_f64: HashMap<Field, FastFieldReader<f64>>,
-    fast_field_i64s: HashMap<Field, MultiValueIntFastFieldReader<i64>>,
-    fast_field_u64s: HashMap<Field, MultiValueIntFastFieldReader<u64>>,
-    fast_field_f64s: HashMap<Field, MultiValueIntFastFieldReader<f64>>,
-    fast_bytes: HashMap<Field, BytesFastFieldReader>,
     fast_fields_composite: CompositeFile,
 }
-
+#[derive(Eq, PartialEq, Debug)]
 enum FastType {
     I64,
     U64,
     F64,
+    Date,
 }
 
 fn type_and_cardinality(field_type: &FieldType) -> Option<(FastType, Cardinality)> {
@@ -39,191 +38,189 @@ fn type_and_cardinality(field_type: &FieldType) -> Option<(FastType, Cardinality
         FieldType::F64(options) => options
             .get_fastfield_cardinality()
             .map(|cardinality| (FastType::F64, cardinality)),
-        FieldType::HierarchicalFacet => Some((FastType::U64, Cardinality::MultiValues)),
+        FieldType::Date(options) => options
+            .get_fastfield_cardinality()
+            .map(|cardinality| (FastType::Date, cardinality)),
+        FieldType::HierarchicalFacet(_) => Some((FastType::U64, Cardinality::MultiValues)),
         _ => None,
     }
 }
 
 impl FastFieldReaders {
-    pub(crate) fn load_all(
+    pub(crate) fn new(schema: Schema, fast_fields_composite: CompositeFile) -> FastFieldReaders {
-        schema: &Schema,
+        FastFieldReaders {
-        fast_fields_composite: &CompositeFile,
+            schema,
-    ) -> Result<FastFieldReaders> {
+            fast_fields_composite,
-        let mut fast_field_readers = FastFieldReaders {
-            fast_field_i64: Default::default(),
-            fast_field_u64: Default::default(),
-            fast_field_f64: Default::default(),
-            fast_field_i64s: Default::default(),
-            fast_field_u64s: Default::default(),
-            fast_field_f64s: Default::default(),
-            fast_bytes: Default::default(),
-            fast_fields_composite: fast_fields_composite.clone(),
-        };
-        for (field_id, field_entry) in schema.fields().iter().enumerate() {
-            let field = Field(field_id as u32);
-            let field_type = field_entry.field_type();
-            if field_type == &FieldType::Bytes {
-                let idx_reader = fast_fields_composite
-                    .open_read_with_idx(field, 0)
-                    .ok_or_else(|| FastFieldNotAvailableError::new(field_entry))
-                    .map(FastFieldReader::open)?;
-                let data = fast_fields_composite
-                    .open_read_with_idx(field, 1)
-                    .ok_or_else(|| FastFieldNotAvailableError::new(field_entry))?;
-                fast_field_readers
-                    .fast_bytes
-                    .insert(field, BytesFastFieldReader::open(idx_reader, data));
-            } else if let Some((fast_type, cardinality)) = type_and_cardinality(field_type) {
-                match cardinality {
-                    Cardinality::SingleValue => {
-                        if let Some(fast_field_data) = fast_fields_composite.open_read(field) {
-                            match fast_type {
-                                FastType::U64 => {
-                                    let fast_field_reader = FastFieldReader::open(fast_field_data);
-                                    fast_field_readers
-                                        .fast_field_u64
-                                        .insert(field, fast_field_reader);
-                                }
-                                FastType::I64 => {
-                                    fast_field_readers.fast_field_i64.insert(
-                                        field,
-                                        FastFieldReader::open(fast_field_data.clone()),
-                                    );
-                                }
-                                FastType::F64 => {
-                                    fast_field_readers.fast_field_f64.insert(
-                                        field,
-                                        FastFieldReader::open(fast_field_data.clone()),
-                                    );
-                                }
-                            }
-                        } else {
-                            return Err(From::from(FastFieldNotAvailableError::new(field_entry)));
-                        }
-                    }
-                    Cardinality::MultiValues => {
-                        let idx_opt = fast_fields_composite.open_read_with_idx(field, 0);
-                        let data_opt = fast_fields_composite.open_read_with_idx(field, 1);
-                        if let (Some(fast_field_idx), Some(fast_field_data)) = (idx_opt, data_opt) {
-                            let idx_reader = FastFieldReader::open(fast_field_idx);
-                            match fast_type {
-                                FastType::I64 => {
-                                    let vals_reader = FastFieldReader::open(fast_field_data);
-                                    let multivalued_int_fast_field =
-                                        MultiValueIntFastFieldReader::open(idx_reader, vals_reader);
-                                    fast_field_readers
-                                        .fast_field_i64s
-                                        .insert(field, multivalued_int_fast_field);
-                                }
-                                FastType::U64 => {
-                                    let vals_reader = FastFieldReader::open(fast_field_data);
-                                    let multivalued_int_fast_field =
-                                        MultiValueIntFastFieldReader::open(idx_reader, vals_reader);
-                                    fast_field_readers
-                                        .fast_field_u64s
-                                        .insert(field, multivalued_int_fast_field);
-                                }
-                                FastType::F64 => {
-                                    let vals_reader = FastFieldReader::open(fast_field_data);
-                                    let multivalued_int_fast_field =
-                                        MultiValueIntFastFieldReader::open(idx_reader, vals_reader);
-                                    fast_field_readers
-                                        .fast_field_f64s
-                                        .insert(field, multivalued_int_fast_field);
-                                }
-                            }
-                        } else {
-                            return Err(From::from(FastFieldNotAvailableError::new(field_entry)));
-                        }
-                    }
-                }
-            }
         }
-        Ok(fast_field_readers)
     }
 
     pub(crate) fn space_usage(&self) -> PerFieldSpaceUsage {
         self.fast_fields_composite.space_usage()
     }
 
-    /// Returns the `u64` fast field reader reader associated to `field`.
+    fn fast_field_data(&self, field: Field, idx: usize) -> crate::Result<FileSlice> {
-    ///
+        self.fast_fields_composite
-    /// If `field` is not a u64 fast field, this method returns `None`.
+            .open_read_with_idx(field, idx)
-    pub fn u64(&self, field: Field) -> Option<FastFieldReader<u64>> {
+            .ok_or_else(|| {
-        self.fast_field_u64.get(&field).cloned()
+                let field_name = self.schema.get_field_entry(field).name();
+                TantivyError::SchemaError(format!("Field({}) data was not found", field_name))
+            })
     }
 
-    /// If the field is a u64-fast field return the associated reader.
+    fn check_type(
-    /// If the field is a i64-fast field, return the associated u64 reader. Values are
+        &self,
-    /// mapped from i64 to u64 using a (well the, it is unique) monotonic mapping.    ///
+        field: Field,
-    ///
+        expected_fast_type: FastType,
-    ///TODO should it also be lenient with f64?
+        expected_cardinality: Cardinality,
-    ///
+    ) -> crate::Result<()> {
-    /// This method is useful when merging segment reader.
+        let field_entry = self.schema.get_field_entry(field);
-    pub(crate) fn u64_lenient(&self, field: Field) -> Option<FastFieldReader<u64>> {
+        let (fast_type, cardinality) =
-        if let Some(u64_ff_reader) = self.u64(field) {
+            type_and_cardinality(field_entry.field_type()).ok_or_else(|| {
-            return Some(u64_ff_reader);
+                crate::TantivyError::SchemaError(format!(
+                    "Field {:?} is not a fast field.",
+                    field_entry.name()
+                ))
+            })?;
+        if fast_type != expected_fast_type {
+            return Err(crate::TantivyError::SchemaError(format!(
+                "Field {:?} is of type {:?}, expected {:?}.",
+                field_entry.name(),
+                fast_type,
+                expected_fast_type
+            )));
         }
-        if let Some(i64_ff_reader) = self.i64(field) {
+        if cardinality != expected_cardinality {
-            return Some(i64_ff_reader.into_u64_reader());
+            return Err(crate::TantivyError::SchemaError(format!(
+                "Field {:?} is of cardinality {:?}, expected {:?}.",
+                field_entry.name(),
+                cardinality,
+                expected_cardinality
+            )));
         }
-        None
+        Ok(())
+    }
+
+    pub(crate) fn typed_fast_field_reader<TFastValue: FastValue>(
+        &self,
+        field: Field,
+    ) -> crate::Result<DynamicFastFieldReader<TFastValue>> {
+        let fast_field_slice = self.fast_field_data(field, 0)?;
+        DynamicFastFieldReader::open(fast_field_slice)
+    }
+
+    pub(crate) fn typed_fast_field_multi_reader<TFastValue: FastValue>(
+        &self,
+        field: Field,
+    ) -> crate::Result<MultiValuedFastFieldReader<TFastValue>> {
+        let fast_field_slice_idx = self.fast_field_data(field, 0)?;
+        let fast_field_slice_vals = self.fast_field_data(field, 1)?;
+        let idx_reader = BitpackedFastFieldReader::open(fast_field_slice_idx)?;
+        let vals_reader: BitpackedFastFieldReader<TFastValue> =
+            BitpackedFastFieldReader::open(fast_field_slice_vals)?;
+        Ok(MultiValuedFastFieldReader::open(idx_reader, vals_reader))
+    }
+
+    /// Returns the `u64` fast field reader reader associated to `field`.
+    ///
+    /// If `field` is not a u64 fast field, this method returns an Error.
+    pub fn u64(&self, field: Field) -> crate::Result<DynamicFastFieldReader<u64>> {
+        self.check_type(field, FastType::U64, Cardinality::SingleValue)?;
+        self.typed_fast_field_reader(field)
+    }
+
+    /// Returns the `u64` fast field reader reader associated to `field`, regardless of whether the given
+    /// field is effectively of type `u64` or not.
+    ///
+    /// If not, the fastfield reader will returns the u64-value associated to the original FastValue.
+    pub fn u64_lenient(&self, field: Field) -> crate::Result<DynamicFastFieldReader<u64>> {
+        self.typed_fast_field_reader(field)
     }
 
     /// Returns the `i64` fast field reader reader associated to `field`.
     ///
-    /// If `field` is not a i64 fast field, this method returns `None`.
+    /// If `field` is not a i64 fast field, this method returns an Error.
-    pub fn i64(&self, field: Field) -> Option<FastFieldReader<i64>> {
+    pub fn i64(&self, field: Field) -> crate::Result<impl FastFieldReader<i64>> {
-        self.fast_field_i64.get(&field).cloned()
+        self.check_type(field, FastType::I64, Cardinality::SingleValue)?;
+        self.typed_fast_field_reader(field)
+    }
+
+    /// Returns the `i64` fast field reader reader associated to `field`.
+    ///
+    /// If `field` is not a i64 fast field, this method returns an Error.
+    pub fn date(&self, field: Field) -> crate::Result<impl FastFieldReader<crate::DateTime>> {
+        self.check_type(field, FastType::Date, Cardinality::SingleValue)?;
+        self.typed_fast_field_reader(field)
     }
 
     /// Returns the `f64` fast field reader reader associated to `field`.
     ///
-    /// If `field` is not a f64 fast field, this method returns `None`.
+    /// If `field` is not a f64 fast field, this method returns an Error.
-    pub fn f64(&self, field: Field) -> Option<FastFieldReader<f64>> {
+    pub fn f64(&self, field: Field) -> crate::Result<impl FastFieldReader<f64>> {
-        self.fast_field_f64.get(&field).cloned()
+        self.check_type(field, FastType::F64, Cardinality::SingleValue)?;
+        self.typed_fast_field_reader(field)
     }
 
     /// Returns a `u64s` multi-valued fast field reader reader associated to `field`.
     ///
-    /// If `field` is not a u64 multi-valued fast field, this method returns `None`.
+    /// If `field` is not a u64 multi-valued fast field, this method returns an Error.
-    pub fn u64s(&self, field: Field) -> Option<MultiValueIntFastFieldReader<u64>> {
+    pub fn u64s(&self, field: Field) -> crate::Result<MultiValuedFastFieldReader<u64>> {
-        self.fast_field_u64s.get(&field).cloned()
+        self.check_type(field, FastType::U64, Cardinality::MultiValues)?;
+        self.typed_fast_field_multi_reader(field)
     }
 
-    /// If the field is a u64s-fast field return the associated reader.
+    /// Returns a `u64s` multi-valued fast field reader reader associated to `field`, regardless of whether the given
-    /// If the field is a i64s-fast field, return the associated u64s reader. Values are
+    /// field is effectively of type `u64` or not.
-    /// mapped from i64 to u64 using a (well the, it is unique) monotonic mapping.
     ///
-    /// This method is useful when merging segment reader.
+    /// If `field` is not a u64 multi-valued fast field, this method returns an Error.
-    pub(crate) fn u64s_lenient(&self, field: Field) -> Option<MultiValueIntFastFieldReader<u64>> {
+    pub fn u64s_lenient(&self, field: Field) -> crate::Result<MultiValuedFastFieldReader<u64>> {
-        if let Some(u64s_ff_reader) = self.u64s(field) {
+        self.typed_fast_field_multi_reader(field)
-            return Some(u64s_ff_reader);
-        }
-        if let Some(i64s_ff_reader) = self.i64s(field) {
-            return Some(i64s_ff_reader.into_u64s_reader());
-        }
-        None
     }
 
     /// Returns a `i64s` multi-valued fast field reader reader associated to `field`.
     ///
-    /// If `field` is not a i64 multi-valued fast field, this method returns `None`.
+    /// If `field` is not a i64 multi-valued fast field, this method returns an Error.
-    pub fn i64s(&self, field: Field) -> Option<MultiValueIntFastFieldReader<i64>> {
+    pub fn i64s(&self, field: Field) -> crate::Result<MultiValuedFastFieldReader<i64>> {
-        self.fast_field_i64s.get(&field).cloned()
+        self.check_type(field, FastType::I64, Cardinality::MultiValues)?;
+        self.typed_fast_field_multi_reader(field)
     }
 
     /// Returns a `f64s` multi-valued fast field reader reader associated to `field`.
     ///
-    /// If `field` is not a f64 multi-valued fast field, this method returns `None`.
+    /// If `field` is not a f64 multi-valued fast field, this method returns an Error.
-    pub fn f64s(&self, field: Field) -> Option<MultiValueIntFastFieldReader<f64>> {
+    pub fn f64s(&self, field: Field) -> crate::Result<MultiValuedFastFieldReader<f64>> {
-        self.fast_field_f64s.get(&field).cloned()
+        self.check_type(field, FastType::F64, Cardinality::MultiValues)?;
+        self.typed_fast_field_multi_reader(field)
+    }
+
+    /// Returns a `crate::DateTime` multi-valued fast field reader reader associated to `field`.
+    ///
+    /// If `field` is not a `crate::DateTime` multi-valued fast field, this method returns an Error.
+    pub fn dates(
+        &self,
+        field: Field,
+    ) -> crate::Result<MultiValuedFastFieldReader<crate::DateTime>> {
+        self.check_type(field, FastType::Date, Cardinality::MultiValues)?;
+        self.typed_fast_field_multi_reader(field)
     }
 
     /// Returns the `bytes` fast field reader associated to `field`.
     ///
-    /// If `field` is not a bytes fast field, returns `None`.
+    /// If `field` is not a bytes fast field, returns an Error.
-    pub fn bytes(&self, field: Field) -> Option<BytesFastFieldReader> {
+    pub fn bytes(&self, field: Field) -> crate::Result<BytesFastFieldReader> {
-        self.fast_bytes.get(&field).cloned()
+        let field_entry = self.schema.get_field_entry(field);
+        if let FieldType::Bytes(bytes_option) = field_entry.field_type() {
+            if !bytes_option.is_fast() {
+                return Err(crate::TantivyError::SchemaError(format!(
+                    "Field {:?} is not a fast field.",
+                    field_entry.name()
+                )));
+            }
+            let fast_field_idx_file = self.fast_field_data(field, 0)?;
+            let idx_reader = BitpackedFastFieldReader::open(fast_field_idx_file)?;
+            let data = self.fast_field_data(field, 1)?;
+            BytesFastFieldReader::open(idx_reader, data)
+        } else {
+            Err(FastFieldNotAvailableError::new(field_entry).into())
+        }
     }
 }

src/fastfield/serializer.rs

@@ -1,16 +1,16 @@
-use crate::common::bitpacker::BitPacker;
-use crate::common::compute_num_bits;
 use crate::common::BinarySerializable;
 use crate::common::CompositeWrite;
 use crate::common::CountingWriter;
 use crate::directory::WritePtr;
 use crate::schema::Field;
 use std::io::{self, Write};
+use tantivy_bitpacker::compute_num_bits;
+use tantivy_bitpacker::BitPacker;
 
-/// `FastFieldSerializer` is in charge of serializing
+/// `CompositeFastFieldSerializer` is in charge of serializing
 /// fastfields on disk.
 ///
-/// Fast fields are encoded using bit-packing.
+/// Fast fields have differnt encodings like bit-packing.
 ///
 /// `FastFieldWriter`s are in charge of pushing the data to
 /// the serializer.
@@ -27,16 +27,16 @@ use std::io::{self, Write};
 /// * ...
 /// * `close_field()`
 /// * `close()`
-pub struct FastFieldSerializer {
+pub struct CompositeFastFieldSerializer {
     composite_write: CompositeWrite<WritePtr>,
 }
 
-impl FastFieldSerializer {
+impl CompositeFastFieldSerializer {
     /// Constructor
-    pub fn from_write(write: WritePtr) -> io::Result<FastFieldSerializer> {
+    pub fn from_write(write: WritePtr) -> io::Result<CompositeFastFieldSerializer> {
         // just making room for the pointer to header.
         let composite_write = CompositeWrite::wrap(write);
-        Ok(FastFieldSerializer { composite_write })
+        Ok(CompositeFastFieldSerializer { composite_write })
     }
 
     /// Start serializing a new u64 fast field
@@ -45,7 +45,7 @@ impl FastFieldSerializer {
         field: Field,
         min_value: u64,
         max_value: u64,
-    ) -> io::Result<FastSingleFieldSerializer<'_, CountingWriter<WritePtr>>> {
+    ) -> io::Result<DynamicFastFieldSerializer<'_, CountingWriter<WritePtr>>> {
         self.new_u64_fast_field_with_idx(field, min_value, max_value, 0)
     }
 
@@ -56,9 +56,9 @@ impl FastFieldSerializer {
         min_value: u64,
         max_value: u64,
         idx: usize,
-    ) -> io::Result<FastSingleFieldSerializer<'_, CountingWriter<WritePtr>>> {
+    ) -> io::Result<DynamicFastFieldSerializer<'_, CountingWriter<WritePtr>>> {
         let field_write = self.composite_write.for_field_with_idx(field, idx);
-        FastSingleFieldSerializer::open(field_write, min_value, max_value)
+        DynamicFastFieldSerializer::open(field_write, min_value, max_value)
     }
 
     /// Start serializing a new [u8] fast field
@@ -66,9 +66,9 @@ impl FastFieldSerializer {
         &mut self,
         field: Field,
         idx: usize,
-    ) -> io::Result<FastBytesFieldSerializer<'_, CountingWriter<WritePtr>>> {
+    ) -> FastBytesFieldSerializer<'_, CountingWriter<WritePtr>> {
         let field_write = self.composite_write.for_field_with_idx(field, idx);
-        FastBytesFieldSerializer::open(field_write)
+        FastBytesFieldSerializer { write: field_write }
     }
 
     /// Closes the serializer
@@ -79,14 +79,111 @@ impl FastFieldSerializer {
     }
 }
 
-pub struct FastSingleFieldSerializer<'a, W: Write> {
+#[derive(Debug, Clone)]
+pub struct EstimationStats {
+    min_value: u64,
+    max_value: u64,
+}
+/// The FastFieldSerializer trait is the common interface
+/// implemented by every fastfield serializer variant.
+///
+/// `DynamicFastFieldSerializer` is the enum wrapping all variants.
+/// It is used to create an serializer instance.
+pub trait FastFieldSerializer {
+    /// add value to serializer
+    fn add_val(&mut self, val: u64) -> io::Result<()>;
+    /// finish serializing a field.
+    fn close_field(self) -> io::Result<()>;
+}
+
+/// The FastFieldSerializerEstimate trait is required on all variants
+/// of fast field compressions, to decide which one to choose.
+pub trait FastFieldSerializerEstimate {
+    /// returns an estimate of the compression ratio.
+    fn estimate(
+        /*fastfield_accessor: impl FastFieldReader<u64>,*/ stats: EstimationStats,
+    ) -> (f32, &'static str);
+    /// the unique name of the compressor
+    fn name() -> &'static str;
+}
+
+pub enum DynamicFastFieldSerializer<'a, W: Write> {
+    Bitpacked(BitpackedFastFieldSerializer<'a, W>),
+}
+
+impl<'a, W: Write> DynamicFastFieldSerializer<'a, W> {
+    /// Creates a new fast field serializer.
+    ///
+    /// The serializer in fact encode the values by bitpacking
+    /// `(val - min_value)`.
+    ///
+    /// It requires a `min_value` and a `max_value` to compute
+    /// compute the minimum number of bits required to encode
+    /// values.
+    pub fn open(
+        write: &'a mut W,
+        min_value: u64,
+        max_value: u64,
+    ) -> io::Result<DynamicFastFieldSerializer<'a, W>> {
+        let stats = EstimationStats {
+            min_value,
+            max_value,
+        };
+        let (_ratio, name) = (
+            BitpackedFastFieldSerializer::<Vec<u8>>::estimate(stats),
+            BitpackedFastFieldSerializer::<Vec<u8>>::name(),
+        );
+        Self::open_from_name(write, min_value, max_value, name)
+    }
+
+    /// Creates a new fast field serializer.
+    ///
+    /// The serializer in fact encode the values by bitpacking
+    /// `(val - min_value)`.
+    ///
+    /// It requires a `min_value` and a `max_value` to compute
+    /// compute the minimum number of bits required to encode
+    /// values.
+    pub fn open_from_name(
+        write: &'a mut W,
+        min_value: u64,
+        max_value: u64,
+        name: &str,
+    ) -> io::Result<DynamicFastFieldSerializer<'a, W>> {
+        // Weirdly the W generic on BitpackedFastFieldSerializer needs to be set,
+        // although name() doesn't use it
+        let variant = if name == BitpackedFastFieldSerializer::<Vec<u8>>::name() {
+            DynamicFastFieldSerializer::Bitpacked(BitpackedFastFieldSerializer::open(
+                write, min_value, max_value,
+            )?)
+        } else {
+            panic!("unknown fastfield serializer {}", name);
+        };
+
+        Ok(variant)
+    }
+}
+impl<'a, W: Write> FastFieldSerializer for DynamicFastFieldSerializer<'a, W> {
+    fn add_val(&mut self, val: u64) -> io::Result<()> {
+        match self {
+            Self::Bitpacked(serializer) => serializer.add_val(val),
+        }
+    }
+    fn close_field(self) -> io::Result<()> {
+        match self {
+            Self::Bitpacked(serializer) => serializer.close_field(),
+        }
+    }
+}
+
+pub struct BitpackedFastFieldSerializer<'a, W: Write> {
     bit_packer: BitPacker,
     write: &'a mut W,
     min_value: u64,
     num_bits: u8,
 }
 
-impl<'a, W: Write> FastSingleFieldSerializer<'a, W> {
+impl<'a, W: Write> BitpackedFastFieldSerializer<'a, W> {
     /// Creates a new fast field serializer.
     ///
     /// The serializer in fact encode the values by bitpacking
@@ -99,43 +196,56 @@ impl<'a, W: Write> FastSingleFieldSerializer<'a, W> {
         write: &'a mut W,
         min_value: u64,
         max_value: u64,
-    ) -> io::Result<FastSingleFieldSerializer<'a, W>> {
+    ) -> io::Result<BitpackedFastFieldSerializer<'a, W>> {
         assert!(min_value <= max_value);
         min_value.serialize(write)?;
         let amplitude = max_value - min_value;
         amplitude.serialize(write)?;
         let num_bits = compute_num_bits(amplitude);
         let bit_packer = BitPacker::new();
-        Ok(FastSingleFieldSerializer {
+        Ok(BitpackedFastFieldSerializer {
-            write,
             bit_packer,
+            write,
             min_value,
             num_bits,
         })
     }
+}
 
+impl<'a, W: 'a + Write> FastFieldSerializer for BitpackedFastFieldSerializer<'a, W> {
     /// Pushes a new value to the currently open u64 fast field.
-    pub fn add_val(&mut self, val: u64) -> io::Result<()> {
+    fn add_val(&mut self, val: u64) -> io::Result<()> {
         let val_to_write: u64 = val - self.min_value;
         self.bit_packer
             .write(val_to_write, self.num_bits, &mut self.write)?;
         Ok(())
     }
-
+    fn close_field(mut self) -> io::Result<()> {
-    pub fn close_field(mut self) -> io::Result<()> {
         self.bit_packer.close(&mut self.write)
     }
 }
 
+impl<'a, W: 'a + Write> FastFieldSerializerEstimate for BitpackedFastFieldSerializer<'a, W> {
+    fn estimate(
+        /*_fastfield_accessor: impl FastFieldReader<u64>, */ stats: EstimationStats,
+    ) -> (f32, &'static str) {
+        let amplitude = stats.max_value - stats.min_value;
+        let num_bits = compute_num_bits(amplitude);
+        let num_bits_uncompressed = 64;
+        let ratio = num_bits as f32 / num_bits_uncompressed as f32;
+        let name = Self::name();
+        (ratio, name)
+    }
+    fn name() -> &'static str {
+        "Bitpacked"
+    }
+}
+
 pub struct FastBytesFieldSerializer<'a, W: Write> {
     write: &'a mut W,
 }
 
 impl<'a, W: Write> FastBytesFieldSerializer<'a, W> {
-    fn open(write: &'a mut W) -> io::Result<FastBytesFieldSerializer<'a, W>> {
-        Ok(FastBytesFieldSerializer { write })
-    }
-
     pub fn write_all(&mut self, vals: &[u8]) -> io::Result<()> {
         self.write.write_all(vals)
     }

src/fastfield/writer.rs

@@ -1,21 +1,31 @@
-use super::multivalued::MultiValueIntFastFieldWriter;
+use super::multivalued::MultiValuedFastFieldWriter;
 use crate::common;
-use crate::common::BinarySerializable;
+use crate::fastfield::serializer::FastFieldSerializer;
-use crate::common::VInt;
+use crate::fastfield::{BytesFastFieldWriter, CompositeFastFieldSerializer};
-use crate::fastfield::{BytesFastFieldWriter, FastFieldSerializer};
+use crate::indexer::doc_id_mapping::DocIdMapping;
 use crate::postings::UnorderedTermId;
-use crate::schema::{Cardinality, Document, Field, FieldType, Schema};
+use crate::schema::{Cardinality, Document, Field, FieldEntry, FieldType, Schema};
 use crate::termdict::TermOrdinal;
+use fnv::FnvHashMap;
 use std::collections::HashMap;
 use std::io;
+use tantivy_bitpacker::BlockedBitpacker;
 
 /// The fastfieldswriter regroup all of the fast field writers.
 pub struct FastFieldsWriter {
     single_value_writers: Vec<IntFastFieldWriter>,
-    multi_values_writers: Vec<MultiValueIntFastFieldWriter>,
+    multi_values_writers: Vec<MultiValuedFastFieldWriter>,
     bytes_value_writers: Vec<BytesFastFieldWriter>,
 }
 
+fn fast_field_default_value(field_entry: &FieldEntry) -> u64 {
+    match *field_entry.field_type() {
+        FieldType::I64(_) | FieldType::Date(_) => common::i64_to_u64(0i64),
+        FieldType::F64(_) => common::f64_to_u64(0.0f64),
+        _ => 0u64,
+    }
+}
+
 impl FastFieldsWriter {
     /// Create all `FastFieldWriter` required by the schema.
     pub fn from_schema(schema: &Schema) -> FastFieldsWriter {
@@ -23,35 +33,35 @@ impl FastFieldsWriter {
         let mut multi_values_writers = Vec::new();
         let mut bytes_value_writers = Vec::new();
 
-        for (field_id, field_entry) in schema.fields().iter().enumerate() {
+        for (field, field_entry) in schema.fields() {
-            let field = Field(field_id as u32);
+            match field_entry.field_type() {
-            let default_value = match *field_entry.field_type() {
+                FieldType::I64(ref int_options)
-                FieldType::I64(_) => common::i64_to_u64(0i64),
+                | FieldType::U64(ref int_options)
-                FieldType::F64(_) => common::f64_to_u64(0.0f64),
+                | FieldType::F64(ref int_options)
-                _ => 0u64,
+                | FieldType::Date(ref int_options) => {
-            };
-            match *field_entry.field_type() {
-                FieldType::I64(ref int_options) | FieldType::U64(ref int_options) | FieldType::F64(ref int_options) => {
                     match int_options.get_fastfield_cardinality() {
                         Some(Cardinality::SingleValue) => {
                             let mut fast_field_writer = IntFastFieldWriter::new(field);
+                            let default_value = fast_field_default_value(field_entry);
                             fast_field_writer.set_val_if_missing(default_value);
                             single_value_writers.push(fast_field_writer);
                         }
                         Some(Cardinality::MultiValues) => {
-                            let fast_field_writer = MultiValueIntFastFieldWriter::new(field, false);
+                            let fast_field_writer = MultiValuedFastFieldWriter::new(field, false);
                             multi_values_writers.push(fast_field_writer);
                         }
                         None => {}
                     }
                 }
-                FieldType::HierarchicalFacet => {
+                FieldType::HierarchicalFacet(_) => {
-                    let fast_field_writer = MultiValueIntFastFieldWriter::new(field, true);
+                    let fast_field_writer = MultiValuedFastFieldWriter::new(field, true);
                     multi_values_writers.push(fast_field_writer);
                 }
-                FieldType::Bytes => {
+                FieldType::Bytes(bytes_option) => {
-                    let fast_field_writer = BytesFastFieldWriter::new(field);
+                    if bytes_option.is_fast() {
-                    bytes_value_writers.push(fast_field_writer);
+                        let fast_field_writer = BytesFastFieldWriter::new(field);
+                        bytes_value_writers.push(fast_field_writer);
+                    }
                 }
                 _ => {}
             }
@@ -63,8 +73,34 @@ impl FastFieldsWriter {
         }
     }
 
+    /// The memory used (inclusive childs)
+    pub fn mem_usage(&self) -> usize {
+        self.single_value_writers
+            .iter()
+            .map(|w| w.mem_usage())
+            .sum::<usize>()
+            + self
+                .multi_values_writers
+                .iter()
+                .map(|w| w.mem_usage())
+                .sum::<usize>()
+            + self
+                .bytes_value_writers
+                .iter()
+                .map(|w| w.mem_usage())
+                .sum::<usize>()
+    }
+
     /// Get the `FastFieldWriter` associated to a field.
-    pub fn get_field_writer(&mut self, field: Field) -> Option<&mut IntFastFieldWriter> {
+    pub fn get_field_writer(&self, field: Field) -> Option<&IntFastFieldWriter> {
+        // TODO optimize
+        self.single_value_writers
+            .iter()
+            .find(|field_writer| field_writer.field() == field)
+    }
+
+    /// Get the `FastFieldWriter` associated to a field.
+    pub fn get_field_writer_mut(&mut self, field: Field) -> Option<&mut IntFastFieldWriter> {
         // TODO optimize
         self.single_value_writers
             .iter_mut()
@@ -75,10 +111,10 @@ impl FastFieldsWriter {
     ///
     /// Returns None if the field does not exist, or is not
     /// configured as a multivalued fastfield in the schema.
-    pub fn get_multivalue_writer(
+    pub fn get_multivalue_writer_mut(
         &mut self,
         field: Field,
-    ) -> Option<&mut MultiValueIntFastFieldWriter> {
+    ) -> Option<&mut MultiValuedFastFieldWriter> {
         // TODO optimize
         self.multi_values_writers
             .iter_mut()
@@ -89,7 +125,7 @@ impl FastFieldsWriter {
     ///
     /// Returns None if the field does not exist, or is not
     /// configured as a bytes fastfield in the schema.
-    pub fn get_bytes_writer(&mut self, field: Field) -> Option<&mut BytesFastFieldWriter> {
+    pub fn get_bytes_writer_mut(&mut self, field: Field) -> Option<&mut BytesFastFieldWriter> {
         // TODO optimize
         self.bytes_value_writers
             .iter_mut()
@@ -113,18 +149,20 @@ impl FastFieldsWriter {
     /// order to the fast field serializer.
     pub fn serialize(
         &self,
-        serializer: &mut FastFieldSerializer,
+        serializer: &mut CompositeFastFieldSerializer,
-        mapping: &HashMap<Field, HashMap<UnorderedTermId, TermOrdinal>>,
+        mapping: &HashMap<Field, FnvHashMap<UnorderedTermId, TermOrdinal>>,
+        doc_id_map: Option<&DocIdMapping>,
     ) -> io::Result<()> {
         for field_writer in &self.single_value_writers {
-            field_writer.serialize(serializer)?;
+            field_writer.serialize(serializer, doc_id_map)?;
         }
+
         for field_writer in &self.multi_values_writers {
             let field = field_writer.field();
-            field_writer.serialize(serializer, mapping.get(&field))?;
+            field_writer.serialize(serializer, mapping.get(&field), doc_id_map)?;
         }
         for field_writer in &self.bytes_value_writers {
-            field_writer.serialize(serializer)?;
+            field_writer.serialize(serializer, doc_id_map)?;
         }
         Ok(())
     }
@@ -147,7 +185,7 @@ impl FastFieldsWriter {
 /// using `common::i64_to_u64` and `common::f64_to_u64`.
 pub struct IntFastFieldWriter {
     field: Field,
-    vals: Vec<u8>,
+    vals: BlockedBitpacker,
     val_count: usize,
     val_if_missing: u64,
     val_min: u64,
@@ -159,7 +197,7 @@ impl IntFastFieldWriter {
     pub fn new(field: Field) -> IntFastFieldWriter {
         IntFastFieldWriter {
             field,
-            vals: Vec::new(),
+            vals: BlockedBitpacker::new(),
             val_count: 0,
             val_if_missing: 0u64,
             val_min: u64::max_value(),
@@ -167,6 +205,11 @@ impl IntFastFieldWriter {
         }
     }
 
+    /// The memory used (inclusive childs)
+    pub fn mem_usage(&self) -> usize {
+        self.vals.mem_usage()
+    }
+
     /// Returns the field that this writer is targetting.
     pub fn field(&self) -> Field {
         self.field
@@ -186,9 +229,7 @@ impl IntFastFieldWriter {
     /// associated to the document with the `DocId` n.
     /// (Well, `n-1` actually because of 0-indexing)
     pub fn add_val(&mut self, val: u64) {
-        VInt(val)
+        self.vals.add(val);
-            .serialize(&mut self.vals)
-            .expect("unable to serialize VInt to Vec");
 
         if val > self.val_max {
             self.val_max = val;
@@ -224,20 +265,32 @@ impl IntFastFieldWriter {
         self.add_val(val);
     }
 
+    /// Extract the stored data
+    pub(crate) fn get_data(&self) -> Vec<u64> {
+        self.vals.iter().collect::<Vec<u64>>()
+    }
+
     /// Push the fast fields value to the `FastFieldWriter`.
-    pub fn serialize(&self, serializer: &mut FastFieldSerializer) -> io::Result<()> {
+    pub fn serialize(
+        &self,
+        serializer: &mut CompositeFastFieldSerializer,
+        doc_id_map: Option<&DocIdMapping>,
+    ) -> io::Result<()> {
         let (min, max) = if self.val_min > self.val_max {
             (0, 0)
         } else {
             (self.val_min, self.val_max)
         };
-
         let mut single_field_serializer = serializer.new_u64_fast_field(self.field, min, max)?;
-
+        if let Some(doc_id_map) = doc_id_map {
-        let mut cursor = self.vals.as_slice();
+            for doc_id in doc_id_map.iter_old_doc_ids() {
-        while let Ok(VInt(val)) = VInt::deserialize(&mut cursor) {
+                single_field_serializer.add_val(self.vals.get(doc_id as usize))?;
-            single_field_serializer.add_val(val)?;
+            }
-        }
+        } else {
+            for val in self.vals.iter() {
+                single_field_serializer.add_val(val)?;
+            }
+        };
 
         single_field_serializer.close_field()
     }

src/fieldnorm/code.rs

@@ -1,9 +1,9 @@
-#[inline(always)]
+#[inline]
 pub fn id_to_fieldnorm(id: u8) -> u32 {
     FIELD_NORMS_TABLE[id as usize]
 }
 
-#[inline(always)]
+#[inline]
 pub fn fieldnorm_to_id(fieldnorm: u32) -> u8 {
     FIELD_NORMS_TABLE
         .binary_search(&fieldnorm)

src/fieldnorm/mod.rs

@@ -15,13 +15,13 @@
 //! precompute computationally expensive functions of the fieldnorm
 //! in a very short array.
 //!
-//! This trick is used by the BM25 similarity.
+//! This trick is used by the Bm25 similarity.
 mod code;
 mod reader;
 mod serializer;
 mod writer;
 
-pub use self::reader::FieldNormReader;
+pub use self::reader::{FieldNormReader, FieldNormReaders};
 pub use self::serializer::FieldNormsSerializer;
 pub use self::writer::FieldNormsWriter;
 

src/fieldnorm/reader.rs

@@ -1,6 +1,47 @@
 use super::{fieldnorm_to_id, id_to_fieldnorm};
-use crate::directory::ReadOnlySource;
+use crate::common::CompositeFile;
+use crate::directory::FileSlice;
+use crate::directory::OwnedBytes;
+use crate::schema::Field;
+use crate::space_usage::PerFieldSpaceUsage;
 use crate::DocId;
+use std::sync::Arc;
+
+/// Reader for the fieldnorm (for each document, the number of tokens indexed in the
+/// field) of all indexed fields in the index.
+///
+/// Each fieldnorm is approximately compressed over one byte. We refer to this byte as
+/// `fieldnorm_id`.
+/// The mapping from `fieldnorm` to `fieldnorm_id` is given by monotonic.
+#[derive(Clone)]
+pub struct FieldNormReaders {
+    data: Arc<CompositeFile>,
+}
+
+impl FieldNormReaders {
+    /// Creates a field norm reader.
+    pub fn open(file: FileSlice) -> crate::Result<FieldNormReaders> {
+        let data = CompositeFile::open(&file)?;
+        Ok(FieldNormReaders {
+            data: Arc::new(data),
+        })
+    }
+
+    /// Returns the FieldNormReader for a specific field.
+    pub fn get_field(&self, field: Field) -> crate::Result<Option<FieldNormReader>> {
+        if let Some(file) = self.data.open_read(field) {
+            let fieldnorm_reader = FieldNormReader::open(file)?;
+            Ok(Some(fieldnorm_reader))
+        } else {
+            Ok(None)
+        }
+    }
+
+    /// Return a break down of the space usage per field.
+    pub fn space_usage(&self) -> PerFieldSpaceUsage {
+        self.data.space_usage()
+    }
+}
 
 /// Reads the fieldnorm associated to a document.
 /// The fieldnorm represents the length associated to
@@ -19,61 +60,144 @@ use crate::DocId;
 /// Apart from compression, this scale also makes it possible to
 /// precompute computationally expensive functions of the fieldnorm
 /// in a very short array.
-pub struct FieldNormReader {
+#[derive(Clone)]
-    data: ReadOnlySource,
+pub struct FieldNormReader(ReaderImplEnum);
+
+impl From<ReaderImplEnum> for FieldNormReader {
+    fn from(reader_enum: ReaderImplEnum) -> FieldNormReader {
+        FieldNormReader(reader_enum)
+    }
+}
+
+#[derive(Clone)]
+enum ReaderImplEnum {
+    FromData(OwnedBytes),
+    Const {
+        num_docs: u32,
+        fieldnorm_id: u8,
+        fieldnorm: u32,
+    },
 }
 
 impl FieldNormReader {
-    /// Opens a field norm reader given its data source.
+    /// Creates a `FieldNormReader` with a constant fieldnorm.
-    pub fn open(data: ReadOnlySource) -> Self {
+    ///
-        FieldNormReader { data }
+    /// The fieldnorm will be subjected to compression as if it was coming
+    /// from an array-backed fieldnorm reader.
+    pub fn constant(num_docs: u32, fieldnorm: u32) -> FieldNormReader {
+        let fieldnorm_id = fieldnorm_to_id(fieldnorm);
+        let fieldnorm = id_to_fieldnorm(fieldnorm_id);
+        ReaderImplEnum::Const {
+            num_docs,
+            fieldnorm_id,
+            fieldnorm,
+        }
+        .into()
+    }
+
+    /// Opens a field norm reader given its file.
+    pub fn open(fieldnorm_file: FileSlice) -> crate::Result<Self> {
+        let data = fieldnorm_file.read_bytes()?;
+        Ok(FieldNormReader::new(data))
+    }
+
+    fn new(data: OwnedBytes) -> Self {
+        ReaderImplEnum::FromData(data).into()
+    }
+
+    /// Returns the number of documents in this segment.
+    pub fn num_docs(&self) -> u32 {
+        match &self.0 {
+            ReaderImplEnum::FromData(data) => data.len() as u32,
+            ReaderImplEnum::Const { num_docs, .. } => *num_docs,
+        }
     }
 
     /// Returns the `fieldnorm` associated to a doc id.
     /// The fieldnorm is a value approximating the number
     /// of tokens in a given field of the `doc_id`.
     ///
-    /// It is imprecise, and always lower than the actual
+    /// It is imprecise, and equal or lower than
-    /// number of tokens.
+    /// the actual number of tokens.
     ///
     /// The fieldnorm is effectively decoded from the
     /// `fieldnorm_id` by doing a simple table lookup.
     pub fn fieldnorm(&self, doc_id: DocId) -> u32 {
-        let fieldnorm_id = self.fieldnorm_id(doc_id);
+        match &self.0 {
-        id_to_fieldnorm(fieldnorm_id)
+            ReaderImplEnum::FromData(data) => {
+                let fieldnorm_id = data.as_slice()[doc_id as usize];
+                id_to_fieldnorm(fieldnorm_id)
+            }
+            ReaderImplEnum::Const { fieldnorm, .. } => *fieldnorm,
+        }
     }
 
     /// Returns the `fieldnorm_id` associated to a document.
-    #[inline(always)]
+    #[inline]
     pub fn fieldnorm_id(&self, doc_id: DocId) -> u8 {
-        let fielnorms_data = self.data.as_slice();
+        match &self.0 {
-        fielnorms_data[doc_id as usize]
+            ReaderImplEnum::FromData(data) => {
+                let fieldnorm_id = data.as_slice()[doc_id as usize];
+                fieldnorm_id
+            }
+            ReaderImplEnum::Const { fieldnorm_id, .. } => *fieldnorm_id,
+        }
     }
 
     /// Converts a `fieldnorm_id` into a fieldnorm.
-    #[inline(always)]
+    #[inline]
     pub fn id_to_fieldnorm(id: u8) -> u32 {
         id_to_fieldnorm(id)
     }
 
     /// Converts a `fieldnorm` into a `fieldnorm_id`.
     /// (This function is not injective).
-    #[inline(always)]
+    #[inline]
     pub fn fieldnorm_to_id(fieldnorm: u32) -> u8 {
         fieldnorm_to_id(fieldnorm)
     }
+
+    #[cfg(test)]
+    pub fn for_test(field_norms: &[u32]) -> FieldNormReader {
+        let field_norms_id = field_norms
+            .iter()
+            .cloned()
+            .map(FieldNormReader::fieldnorm_to_id)
+            .collect::<Vec<u8>>();
+        let field_norms_data = OwnedBytes::new(field_norms_id);
+        FieldNormReader::new(field_norms_data)
+    }
 }
 
 #[cfg(test)]
-impl From<Vec<u32>> for FieldNormReader {
+mod tests {
-    fn from(field_norms: Vec<u32>) -> FieldNormReader {
+    use crate::fieldnorm::FieldNormReader;
-        let field_norms_id = field_norms
+
-            .into_iter()
+    #[test]
-            .map(FieldNormReader::fieldnorm_to_id)
+    fn test_from_fieldnorms_array() {
-            .collect::<Vec<u8>>();
+        let fieldnorms = &[1, 2, 3, 4, 1_000_000];
-        let field_norms_data = ReadOnlySource::from(field_norms_id);
+        let fieldnorm_reader = FieldNormReader::for_test(fieldnorms);
-        FieldNormReader {
+        assert_eq!(fieldnorm_reader.num_docs(), 5);
-            data: field_norms_data,
+        assert_eq!(fieldnorm_reader.fieldnorm(0), 1);
-        }
+        assert_eq!(fieldnorm_reader.fieldnorm(1), 2);
+        assert_eq!(fieldnorm_reader.fieldnorm(2), 3);
+        assert_eq!(fieldnorm_reader.fieldnorm(3), 4);
+        assert_eq!(fieldnorm_reader.fieldnorm(4), 983_064);
+    }
+
+    #[test]
+    fn test_const_fieldnorm_reader_small_fieldnorm_id() {
+        let fieldnorm_reader = FieldNormReader::constant(1_000_000u32, 10u32);
+        assert_eq!(fieldnorm_reader.num_docs(), 1_000_000u32);
+        assert_eq!(fieldnorm_reader.fieldnorm(0u32), 10u32);
+        assert_eq!(fieldnorm_reader.fieldnorm_id(0u32), 10u8);
+    }
+
+    #[test]
+    fn test_const_fieldnorm_reader_large_fieldnorm_id() {
+        let fieldnorm_reader = FieldNormReader::constant(1_000_000u32, 300u32);
+        assert_eq!(fieldnorm_reader.num_docs(), 1_000_000u32);
+        assert_eq!(fieldnorm_reader.fieldnorm(0u32), 280u32);
+        assert_eq!(fieldnorm_reader.fieldnorm_id(0u32), 72u8);
     }
 }

src/fieldnorm/writer.rs

@@ -1,10 +1,10 @@
-use crate::DocId;
+use crate::{indexer::doc_id_mapping::DocIdMapping, DocId};
 
 use super::fieldnorm_to_id;
 use super::FieldNormsSerializer;
 use crate::schema::Field;
 use crate::schema::Schema;
-use std::io;
+use std::{io, iter};
 
 /// The `FieldNormsWriter` is in charge of tracking the fieldnorm byte
 /// of each document for each field with field norms.
@@ -22,11 +22,14 @@ impl FieldNormsWriter {
     pub(crate) fn fields_with_fieldnorm(schema: &Schema) -> Vec<Field> {
         schema
             .fields()
-            .iter()
+            .filter_map(|(field, field_entry)| {
-            .enumerate()
+                if field_entry.is_indexed() {
-            .filter(|&(_, field_entry)| field_entry.is_indexed())
+                    Some(field)
-            .map(|(field, _)| Field(field as u32))
+                } else {
-            .collect::<Vec<Field>>()
+                    None
+                }
+            })
+            .collect::<Vec<_>>()
     }
 
     /// Initialize with state for tracking the field norm fields
@@ -35,23 +38,32 @@ impl FieldNormsWriter {
         let fields = FieldNormsWriter::fields_with_fieldnorm(schema);
         let max_field = fields
             .iter()
-            .map(|field| field.0)
+            .map(Field::field_id)
             .max()
             .map(|max_field_id| max_field_id as usize + 1)
             .unwrap_or(0);
         FieldNormsWriter {
             fields,
-            fieldnorms_buffer: (0..max_field).map(|_| Vec::new()).collect::<Vec<_>>(),
+            fieldnorms_buffer: iter::repeat_with(Vec::new)
+                .take(max_field)
+                .collect::<Vec<_>>(),
         }
     }
 
+    /// The memory used inclusive childs
+    pub fn mem_usage(&self) -> usize {
+        self.fieldnorms_buffer
+            .iter()
+            .map(|buf| buf.capacity())
+            .sum()
+    }
     /// Ensure that all documents in 0..max_doc have a byte associated with them
     /// in each of the fieldnorm vectors.
     ///
     /// Will extend with 0-bytes for documents that have not been seen.
     pub fn fill_up_to_max_doc(&mut self, max_doc: DocId) {
-        for &field in self.fields.iter() {
+        for field in self.fields.iter() {
-            self.fieldnorms_buffer[field.0 as usize].resize(max_doc as usize, 0u8);
+            self.fieldnorms_buffer[field.field_id() as usize].resize(max_doc as usize, 0u8);
         }
     }
 
@@ -64,7 +76,7 @@ impl FieldNormsWriter {
     /// * field     - the field being set
     /// * fieldnorm - the number of terms present in document `doc` in field `field`
     pub fn record(&mut self, doc: DocId, field: Field, fieldnorm: u32) {
-        let fieldnorm_buffer: &mut Vec<u8> = &mut self.fieldnorms_buffer[field.0 as usize];
+        let fieldnorm_buffer: &mut Vec<u8> = &mut self.fieldnorms_buffer[field.field_id() as usize];
         assert!(
             fieldnorm_buffer.len() <= doc as usize,
             "Cannot register a given fieldnorm twice"
@@ -75,11 +87,25 @@ impl FieldNormsWriter {
     }
 
     /// Serialize the seen fieldnorm values to the serializer for all fields.
-    pub fn serialize(&self, fieldnorms_serializer: &mut FieldNormsSerializer) -> io::Result<()> {
+    pub fn serialize(
+        &self,
+        mut fieldnorms_serializer: FieldNormsSerializer,
+        doc_id_map: Option<&DocIdMapping>,
+    ) -> io::Result<()> {
         for &field in self.fields.iter() {
-            let fieldnorm_values: &[u8] = &self.fieldnorms_buffer[field.0 as usize][..];
+            let fieldnorm_values: &[u8] = &self.fieldnorms_buffer[field.field_id() as usize][..];
-            fieldnorms_serializer.serialize_field(field, fieldnorm_values)?;
+            if let Some(doc_id_map) = doc_id_map {
+                let mut mapped_fieldnorm_values = vec![];
+                mapped_fieldnorm_values.resize(fieldnorm_values.len(), 0u8);
+                for (new_doc_id, old_doc_id) in doc_id_map.iter_old_doc_ids().enumerate() {
+                    mapped_fieldnorm_values[new_doc_id] = fieldnorm_values[old_doc_id as usize];
+                }
+                fieldnorms_serializer.serialize_field(field, &mapped_fieldnorm_values)?;
+            } else {
+                fieldnorms_serializer.serialize_field(field, fieldnorm_values)?;
+            }
         }
+        fieldnorms_serializer.close()?;
         Ok(())
     }
 }