Bugfix position broken.

For Field with several FieldValues, with a
value that contained no token at all, the token position
was reinitialized to 0.

As a result, PhraseQueries can show some false positives.
In addition, after the computation of the position delta, we can
underflow u32, and end up with gigantic delta.

We haven't been able to actually explain the bug in 1629, but it
is assumed that in some corner case these delta can cause a panic.

Closes #1629

.gitattributes

@@ -1 +0,0 @@
-cpp/* linguist-vendored

.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/ISSUE_TEMPLATE/bug_report.md

@@ -0,0 +1,19 @@
+---
+name: Bug report
+about: Create a report to help us improve
+
+---
+
+**Describe the bug**
+- What did you do?
+- What happened?
+- What was expected?
+
+**Which version of tantivy are you using?**
+If "master",  ideally give the specific sha1 revision.
+
+**To Reproduce**
+
+If your bug is deterministic, can you give a minimal reproducing code?
+Some bugs are not deterministic. Can you describe with precision in which context it happened?
+If this is possible, can you share your code?

.github/ISSUE_TEMPLATE/feature_request.md

@@ -0,0 +1,14 @@
+---
+name: Feature request
+about: Suggest an idea for this project
+
+---
+
+**Is your feature request related to a problem? Please describe.**
+A clear and concise description of what the problem is. Ex. I'm always frustrated when [...]
+
+**Describe the solution you'd like**
+A clear and concise description of what you want to happen.
+
+**[Optional] describe alternatives you've considered**
+A clear and concise description of any alternative solutions or features you've considered.

.github/ISSUE_TEMPLATE/question.md

@@ -0,0 +1,7 @@
+---
+name: Question
+about: Ask any question about tantivy's usage...
+
+---
+
+Try to be specific about your use case...

.github/dependabot.yml

@@ -0,0 +1,15 @@
+version: 2
+updates:
+- package-ecosystem: cargo
+  directory: "/"
+  schedule:
+    interval: daily
+    time: "20:00"
+  open-pull-requests-limit: 10
+
+- package-ecosystem: "github-actions"
+  directory: "/"
+  schedule:
+    interval: daily
+    time: "20:00"
+  open-pull-requests-limit: 10

.github/workflows/coverage.yml

@@ -0,0 +1,26 @@
+name: Coverage
+
+on:
+  push:
+    branches: [ main ]
+  pull_request:
+    branches: [ main ]
+
+jobs:
+  coverage:
+    runs-on: ubuntu-latest
+    steps:
+      - uses: actions/checkout@v3
+      - name: Install Rust
+        run: rustup toolchain install nightly --profile minimal --component llvm-tools-preview
+      - uses: Swatinem/rust-cache@v2
+      - uses: taiki-e/install-action@cargo-llvm-cov
+      - name: Generate code coverage
+        run: cargo +nightly llvm-cov --all-features --workspace --lcov --output-path lcov.info
+      - name: Upload coverage to Codecov
+        uses: codecov/codecov-action@v3
+        continue-on-error: true
+        with:
+          token: ${{ secrets.CODECOV_TOKEN }} # not required for public repos
+          files: lcov.info
+          fail_ci_if_error: true

.github/workflows/long_running.yml

@@ -0,0 +1,28 @@
+name: Long running tests
+
+on:
+  push:
+    branches: [ main ]
+
+env:
+  CARGO_TERM_COLOR: always
+  NUM_FUNCTIONAL_TEST_ITERATIONS: 20000
+
+jobs:
+  test:
+
+    runs-on: ubuntu-latest
+
+    steps:
+    - uses: actions/checkout@v3
+    - name: Install stable
+      uses: actions-rs/toolchain@v1
+      with:
+          toolchain: stable
+          profile: minimal
+          override: true
+
+    - name: Run indexing_unsorted
+      run: cargo test indexing_unsorted -- --ignored
+    - name: Run indexing_sorted
+      run: cargo test indexing_sorted -- --ignored

.github/workflows/test.yml

@@ -0,0 +1,74 @@
+name: Unit tests
+
+on:
+  push:
+    branches: [ main ]
+  pull_request:
+    branches: [ main ]
+
+env:
+  CARGO_TERM_COLOR: always
+
+jobs:
+  check:
+
+    runs-on: ubuntu-latest
+
+    steps:
+    - uses: actions/checkout@v3
+
+    - name: Install nightly
+      uses: actions-rs/toolchain@v1
+      with:
+            toolchain: nightly
+            profile: minimal
+            components: rustfmt
+    - name: Install stable
+      uses: actions-rs/toolchain@v1
+      with:
+            toolchain: stable
+            profile: minimal
+            components: clippy
+
+    - uses: Swatinem/rust-cache@v2
+
+    - name: Check Formatting
+      run: cargo +nightly fmt --all -- --check
+
+    - uses: actions-rs/clippy-check@v1
+      with:
+        toolchain: stable
+        token: ${{ secrets.GITHUB_TOKEN }}
+        args: --tests
+
+  test:
+
+    runs-on: ubuntu-latest
+
+    strategy:
+      matrix:
+        features: [
+            { label: "all", flags: "mmap,brotli-compression,lz4-compression,snappy-compression,zstd-compression,failpoints" },
+            { label: "quickwit", flags: "mmap,quickwit,failpoints" }
+        ]
+
+    name: test-${{ matrix.features.label}}
+
+    steps:
+    - uses: actions/checkout@v3
+
+    - name: Install stable
+      uses: actions-rs/toolchain@v1
+      with:
+            toolchain: stable
+            profile: minimal
+            override: true
+
+    - uses: taiki-e/install-action@nextest
+    - uses: Swatinem/rust-cache@v2
+
+    - name: Run tests
+      run: cargo +stable nextest run --features ${{ matrix.features.flags }} --verbose --workspace
+
+    - name: Run doctests
+      run: cargo +stable test --doc --features ${{ matrix.features.flags }} --verbose --workspace

.gitignore

@@ -1,3 +1,6 @@
+tantivy.iml
+.cargo
+proptest-regressions
 *.swp
 target
 target/debug
@@ -6,7 +9,7 @@ target/release
 Cargo.lock
 benchmark
 .DS_Store
-cpp/simdcomp/bitpackingbenchmark
 *.bk
 .idea
 trace.dat
+cargo-timing*

.travis.yml

@@ -1,37 +0,0 @@
-language: rust
-sudo: required
-cache: cargo
-rust:
-  - nightly
-env:
-  global:
-    - CC=gcc-4.8
-    - CXX=g++-4.8
-    - TRAVIS_CARGO_NIGHTLY_FEATURE=""
-    - secure: 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
-addons:
-  apt:
-    sources:
-      - ubuntu-toolchain-r-test
-      - kalakris-cmake
-    packages:
-      - gcc-4.8
-      - g++-4.8
-      - libcurl4-openssl-dev
-      - libelf-dev
-      - libdw-dev
-      - binutils-dev
-      - cmake
-before_script:
-  - export PATH=$HOME/.cargo/bin:$PATH
-  - cargo install cargo-update || echo "cargo-update already installed"
-  - cargo install cargo-travis || echo "cargo-travis already installed"
-script:
-  - cargo build
-  - cargo test
-  - cargo test -- --ignored
-  - cargo run --example simple_search
-  - cargo doc
-after_success:
-  - cargo coveralls --exclude-pattern src/functional_test.rs
-  - cargo doc-upload

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 remove from [alive bitset file](src/fastfield/alive_bitset.rs) that represents the bitset of the alive document ids.
+Like all segment files, this file is immutable. Because it is possible to have more than one alive bitset file at a given instant, the alive bitset 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
+
+```rust
+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 through 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.

AUTHORS

@@ -0,0 +1,11 @@
+# This is the list of authors of tantivy for copyright purposes.
+Paul Masurel
+Laurentiu Nicola
+Dru Sellers
+Ashley Mannix
+Michael J. Curry
+Jason Wolfe
+# As an employee of Google I am required to add Google LLC
+# in the list of authors, but this project is not affiliated to Google
+# in any other way.
+Google LLC 

CHANGELOG.md

@@ -1,46 +1,457 @@
-Tantivy 0.5.2
+Tantivy 0.19
+================================
+
+- Major bugfix: Fix missing fieldnorms for u64, i64, f64, bool, bytes and date [#1620](https://github.com/quickwit-oss/tantivy/pull/1620) (@PSeitz)
+- Updated [Date Field Type](https://github.com/quickwit-oss/tantivy/pull/1396)
+  The `DateTime` type has been updated to hold timestamps with microseconds precision.
+  `DateOptions` and `DatePrecision` have been added to configure Date fields. The precision is used to hint on fast values compression. Otherwise, seconds precision is used everywhere else (i.e terms, indexing). (@evanxg852000)
+- Add IP address field type [#1553](https://github.com/quickwit-oss/tantivy/pull/1553) (@PSeitz)
+- Add boolean field type [#1382](https://github.com/quickwit-oss/tantivy/pull/1382) (@boraarslan)
+- Remove Searcher pool and make `Searcher` cloneable. (@PSeitz)
+- Validate settings on create [#1570](https://github.com/quickwit-oss/tantivy/pull/1570 (@PSeitz)
+- Fix interpolation overflow in linear interpolation fastfield codec [#1480](https://github.com/quickwit-oss/tantivy/pull/1480 (@PSeitz @fulmicoton)
+- Detect and apply gcd on fastfield codecs [#1418](https://github.com/quickwit-oss/tantivy/pull/1418) (@PSeitz)
+- Doc store
+  - use separate thread to compress block store [#1389](https://github.com/quickwit-oss/tantivy/pull/1389) [#1510](https://github.com/quickwit-oss/tantivy/pull/1510 (@PSeitz @fulmicoton)
+  - Expose doc store cache size [#1403](https://github.com/quickwit-oss/tantivy/pull/1403) (@PSeitz)
+  - Enable compression levels for doc store [#1378](https://github.com/quickwit-oss/tantivy/pull/1378) (@PSeitz)
+  - Make block size configurable [#1374](https://github.com/quickwit-oss/tantivy/pull/1374) (@kryesh)
+- Make `tantivy::TantivyError` cloneable [#1402](https://github.com/quickwit-oss/tantivy/pull/1402) (@PSeitz)
+- Add support for phrase slop in query language [#1393](https://github.com/quickwit-oss/tantivy/pull/1393) (@saroh)
+- Aggregation
+  - Add support for keyed parameter in range and histgram aggregations [#1424](https://github.com/quickwit-oss/tantivy/pull/1424) (@k-yomo)
+  - Add aggregation bucket limit [#1363](https://github.com/quickwit-oss/tantivy/pull/1363) (@PSeitz)
+- Faster indexing
+  - [#1610](https://github.com/quickwit-oss/tantivy/pull/1610 (@PSeitz)
+  - [#1594](https://github.com/quickwit-oss/tantivy/pull/1594 (@PSeitz)
+  - [#1582](https://github.com/quickwit-oss/tantivy/pull/1582 (@PSeitz)
+  - [#1611](https://github.com/quickwit-oss/tantivy/pull/1611 (@PSeitz)
+
+
+Tantivy 0.18
+================================
+
+- For date values `chrono` has been replaced with `time` (@uklotzde) #1304 :
+  - The `time` crate is re-exported as `tantivy::time` instead of `tantivy::chrono`.
+  - The type alias `tantivy::DateTime` has been removed.
+  - `Value::Date` wraps `time::PrimitiveDateTime` without time zone information.
+  - Internally date/time values are stored as seconds since UNIX epoch in UTC.
+  - Converting a `time::OffsetDateTime` to `Value::Date` implicitly converts the value into UTC.
+    If this is not desired do the time zone conversion yourself and use `time::PrimitiveDateTime`
+    directly instead.
+- Add [histogram](https://github.com/quickwit-oss/tantivy/pull/1306) aggregation (@PSeitz)
+- Add support for fastfield on text fields (@PSeitz)
+- Add terms aggregation (@PSeitz)
+- Add support for zstd compression (@kryesh)
+
+Tantivy 0.17
+================================
+
+- LogMergePolicy now triggers merges if the ratio of deleted documents reaches a threshold (@shikhar @fulmicoton) [#115](https://github.com/quickwit-oss/tantivy/issues/115)
+- Adds a searcher Warmer API (@shikhar @fulmicoton)
+- Change to non-strict schema. Ignore fields in data which are not defined in schema. Previously this returned an error. #1211
+- Facets are necessarily indexed. Existing index with indexed facets should work out of the box. Index without facets that are marked with index: false should be broken (but they were already broken in a sense). (@fulmicoton) #1195 .
+- Bugfix that could in theory impact durability in theory on some filesystems [#1224](https://github.com/quickwit-oss/tantivy/issues/1224)
+- Schema now offers not indexing fieldnorms (@lpouget) [#922](https://github.com/quickwit-oss/tantivy/issues/922)
+- Reduce the number of fsync calls [#1225](https://github.com/quickwit-oss/tantivy/issues/1225)
+- Fix opening bytes index with dynamic codec (@PSeitz) [#1278](https://github.com/quickwit-oss/tantivy/issues/1278)
+- Added an aggregation collector for range, average and stats compatible with Elasticsearch. (@PSeitz)
+- Added a JSON schema type @fulmicoton [#1251](https://github.com/quickwit-oss/tantivy/issues/1251)
+- Added support for slop in phrase queries @halvorboe [#1068](https://github.com/quickwit-oss/tantivy/issues/1068)
+
+Tantivy 0.16.2
+================================
+
+- Bugfix in FuzzyTermQuery. (transposition_cost_one was not doing anything)
+
+Tantivy 0.16.1
+========================
+
+- Major Bugfix on multivalued fastfield.  #1151
+- Demux operation (@PSeitz)
+
+Tantivy 0.16.0
+=========================
+
+- Bugfix in the filesum check. (@evanxg852000) #1127
+- Bugfix in positions when the index is sorted by a field. (@appaquet) #1125
+
+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/quickwit-oss/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
+=====================
+
+- Closes #544.  A few users experienced problems with the directory watching system.
+Avoid watching the mmap directory until someone effectively creates a reader that uses
+this functionality.
+
+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` (@fulmicoton)
+- Added an ASCII folding filter (@drusellers)
+- Bugfix in `query.count` in presence of deletes (@fulmicoton)
+- 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.
+
+Minor
+---------
+
+- Switched to Rust 2018 (@uvd)
+- Small simplification of the code.
+Calling .freq() or .doc() when .advance() has never been called
+on segment postings should panic from now on.
+- Tokens exceeding `u16::max_value() - 4` chars are discarded silently instead of panicking.
+- Fast fields are now preloaded when the `SegmentReader` is created.
+- `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. (@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
+- Compilation resources improved (@fdb-hiroshima)
+
+## How to update?
+
+Your program should be usable as is.
+
+### Fast fields
+
+Fast fields used to be accessed directly from the `SegmentReader`.
+The API changed, you are now required to acquire your fast field reader via the
+`segment_reader.fast_fields()`, and use one of the typed method:
+
+- `.u64()`, `.i64()` if your field is single-valued ;
+- `.u64s()`, `.i64s()` if your field is multi-valued ;
+- `.bytes()` if your field is bytes fast field.
+
+Tantivy 0.9.0
+=====================
+
+*0.9.0 index format is not compatible with the
+previous index format.*
+
+- MAJOR BUGFIX :
+  Some `Mmap` objects were being leaked, and would never get released. (@fulmicoton)
+- Removed most unsafe (@fulmicoton)
+- Indexer memory footprint improved. (VInt comp, inlining the first block. (@fulmicoton)
+- Stemming in other language possible (@pentlander)
+- Segments with no docs are deleted earlier (@barrotsteindev)
+- Added grouped add and delete operations.
+  They are guaranteed to happen together (i.e. they cannot be split by a commit).
+  In addition, adds are guaranteed to happen on the same segment. (@elbow-jason)
+- Removed `INT_STORED` and `INT_INDEXED`. It is now possible to use `STORED` and `INDEXED`
+  for int fields. (@fulmicoton)
+- Added DateTime field (@barrotsteindev)
+- Added IndexReader. By default, index is reloaded automatically upon new commits (@fulmicoton)
+- SIMD linear search within blocks (@fulmicoton)
+
+## How to update ?
+
+tantivy 0.9 brought some API breaking change.
+To update from tantivy 0.8, you will need to go through the following steps.
+
+- `schema::INT_INDEXED` and `schema::INT_STORED`  should be replaced by `schema::INDEXED` and `schema::INT_STORED`.
+- The index now does not hold the pool of searcher anymore. You are required to create an intermediary object called
+`IndexReader` for this.
+
+    ```rust
+    // create the reader. You typically need to create 1 reader for the entire
+    // lifetime of you program.
+    let reader = index.reader()?;
+
+    // Acquire a searcher (previously `index.searcher()`) is now written:
+    let searcher = reader.searcher();
+
+    // With the default setting of the reader, you are not required to
+    // call `index.load_searchers()` anymore.
+    //
+    // The IndexReader will pick up that change automatically, regardless
+    // of whether the update was done in a different process or not.
+    // If this behavior is not wanted, you can create your reader with
+    // the `ReloadPolicy::Manual`, and manually decide when to reload the index
+    // by calling `reader.reload()?`.
+
+    ```
+
+Tantivy 0.8.2
+=====================
+
+Fixing build for x86_64 platforms. (#496)
+No need to update from 0.8.1 if tantivy
+is building on your platform.
+
+Tantivy 0.8.1
+=====================
+
+Hotfix of #476.
+
+Merge was reflecting deletes before commit was passed.
+Thanks @barrotsteindev  for reporting the bug.
+
+Tantivy 0.8.0
+=====================
+
+*No change in the index format*
+
+- API Breaking change in the collector API. (@jwolfe, @fulmicoton)
+- Multithreaded search (@jwolfe, @fulmicoton)
+
+Tantivy 0.7.1
+=====================
+
+*No change in the index format*
+
+- Bugfix: NGramTokenizer panics on non ascii chars
+- Added a space usage API
+
+Tantivy 0.7
+=====================
+
+- Skip data for doc ids and positions (@fulmicoton),
+  greatly improving performance
+- Tantivy error now rely on the failure crate (@drusellers)
+- Added support for `AND`, `OR`, `NOT` syntax in addition to the `+`,`-` syntax
+- Added a snippet generator with highlight (@vigneshsarma, @fulmicoton)
+- Added a `TopFieldCollector` (@pentlander)
+
+Tantivy 0.6.1
+=========================
+
+- Bugfix #324. GC removing was removing file that were still in useful
+- Added support for parsing AllQuery and RangeQuery via QueryParser
+  - AllQuery: `*`
+  - RangeQuery:
+    - Inclusive `field:[startIncl to endIncl]`
+    - Exclusive `field:{startExcl to endExcl}`
+    - Mixed `field:[startIncl to endExcl}` and vice versa
+    - Unbounded `field:[start to *]`, `field:[* to end]`
+
+Tantivy 0.6
 ==========================
 
-- Removed C code. Tantivy is now pure Rust.
-- BM25
-- Approximate field norms encoded over 1 byte.
+Special thanks to @drusellers and @jason-wolfe for their contributions
+to this release!
+
+- Removed C code. Tantivy is now pure Rust. (@fulmicoton)
+- BM25 (@fulmicoton)
+- Approximate field norms encoded over 1 byte. (@fulmicoton)
+- 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.
+- Add NGram token support (@drusellers)
+- Add Stopword Filter support (@drusellers)
+- Add a FuzzyTermQuery (@drusellers)
+- Add a RegexQuery (@drusellers)
+- Various performance improvements (@fulmicoton)_
+
+Tantivy 0.5.2
+===========================
+
+- bugfix #274
+- bugfix #280
+- bugfix #289
 
 Tantivy 0.5.1
 ==========================
-- bugfix #254 : tantivy failed if no documents in a segment contained a specific field.
 
+- bugfix #254 : tantivy failed if no documents in a segment contained a specific field.
 
 Tantivy 0.5
 ==========================
+
 - Faceting
 - RangeQuery
 - Configurable tokenization pipeline
 - Bugfix in PhraseQuery
 - Various query optimisation
 - Allowing very large indexes
-    - 64 bits file address
-    - Smarter encoding of the `TermInfo` objects
-
-
+  - 64 bits file address
+  - Smarter encoding of the `TermInfo` objects
 
 Tantivy 0.4.3
 ==========================
 
 - Bugfix race condition when deleting files. (#198)
 
-
 Tantivy 0.4.2
 ==========================
 
 - Prevent usage of AVX2 instructions (#201)
 
-
 Tantivy 0.4.1
 ==========================
 
 - Bugfix for non-indexed fields. (#199)
 
-
 Tantivy 0.4.0
 ==========================
 
@@ -55,58 +466,49 @@ Tantivy 0.4.0
   - Searching for a non-indexed field returns an explicit Error
   - Phrase query for non-tokenized field are not tokenized by the query parser.
 - Faster/Better indexing (@fulmicoton)
-    - using murmurhash2
-    - faster merging
-    - more memory efficient fast field writer (@lnicola )
-    - better handling of collisions
-    - lesser memory usage
+  - using murmurhash2
+  - faster merging
+  - more memory efficient fast field writer (@lnicola )
+  - better handling of collisions
+  - lesser memory usage
 - Added API, most notably to iterate over ranges of terms (@fulmicoton)
 - Bugfix that was preventing to unmap segment files, on index drop (@fulmicoton)
 - Made the doc! macro public (@fulmicoton)
 - Added an alternative implementation of the streaming dictionary (@fulmicoton)
 
-
-
 Tantivy 0.3.1
 ==========================
 
 - Expose a method to trigger files garbage collection
 
-
-
 Tantivy 0.3
 ==========================
 
-
 Special thanks to @Kodraus @lnicola @Ameobea @manuel-woelker @celaus
 for their contribution to this release.
 
-Thanks also to everyone in tantivy gitter chat 
+Thanks also to everyone in tantivy gitter chat
 for their advise and company :)
 
-https://gitter.im/tantivy-search/tantivy
-
+<https://gitter.im/tantivy-search/tantivy>
 
 Warning:
 
-Tantivy 0.3 is NOT backward compatible with tantivy 0.2 
+Tantivy 0.3 is NOT backward compatible with tantivy 0.2
 code and index format.
-You should not expect backward compatibility before 
+You should not expect backward compatibility before
 tantivy 1.0.
 
-
-
 New Features
 ------------
 
 - Delete. You can now delete documents from an index.
 - Support for windows (Thanks to @lnicola)
 
-
 Various Bugfixes & small improvements
 ----------------------------------------
 
-- Added CI for Windows (https://ci.appveyor.com/project/fulmicoton/tantivy)
+- Added CI for Windows (<https://ci.appveyor.com/project/fulmicoton/tantivy>)
 Thanks to @KodrAus ! (#108)
 - Various dependy version update (Thanks to @Ameobea) #76
 - Fixed several race conditions in `Index.wait_merge_threads`
@@ -117,8 +519,4 @@ Thanks to @KodrAus ! (#108)
   the natural ordering.
 - Building binary targets for tantivy-cli (Thanks to @KodrAus)
 - Misc invisible bug fixes, and code cleanup.
-- Use 
-
-
-
-
+- Use

Cargo.toml

@@ -1,70 +1,128 @@
 [package]
 name = "tantivy"
-version = "0.5.1"
+version = "0.18.0"
 authors = ["Paul Masurel <paul.masurel@gmail.com>"]
 license = "MIT"
 categories = ["database-implementations", "data-structures"]
-description = """Tantivy is a 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"
+description = """Search engine library"""
+documentation = "https://docs.rs/tantivy/"
+homepage = "https://github.com/quickwit-oss/tantivy"
+repository = "https://github.com/quickwit-oss/tantivy"
 readme = "README.md"
 keywords = ["search", "information", "retrieval"]
+edition = "2021"
+rust-version = "1.62"
 
 [dependencies]
-byteorder = "1.0"
-lazy_static = "0.2.1"
-tinysegmenter = "0.1.0"
-regex = "0.2"
-fst = {version="0.2", default-features=false}
-atomicwrites = {version="0.1", optional=true}
-tempfile = "2.1"
-log = "0.3.6"
-combine = "2.2"
-tempdir = "0.3"
-serde = "1.0"
-serde_derive = "1.0"
-serde_json = "1.0"
-num_cpus = "1.2"
-itertools = "0.5.9"
-lz4 = "1.20"
-bit-set = "0.4.0"
-uuid = { version = "0.6", features = ["v4", "serde"] }
-chan = "0.1"
-crossbeam = "0.3"
-futures = "0.1"
-futures-cpupool = "0.1"
-error-chain = "0.8"
-owning_ref = "0.3"
-stable_deref_trait = "1.0.0"
-rust-stemmers = "0.1.0"
-downcast = { version="0.9", features = ["nightly"]}
-matches = "0.1"
-bitpacking = "0.3"
+oneshot = "0.1.3"
+base64 = "0.13.0"
+byteorder = "1.4.3"
+crc32fast = "1.3.2"
+once_cell = "1.10.0"
+regex = { version = "1.5.5", default-features = false, features = ["std", "unicode"] }
+tantivy-fst = "0.4.0"
+memmap2 = { version = "0.5.3", optional = true }
+lz4_flex = { version = "0.9.2", default-features = false, features = ["checked-decode"], optional = true }
+brotli = { version = "3.3.4", optional = true }
+zstd = { version = "0.11", optional = true, default-features = false }
+snap = { version = "1.0.5", optional = true }
+tempfile = { version = "3.3.0", optional = true }
+log = "0.4.16"
+serde = { version = "1.0.136", features = ["derive"] }
+serde_json = "1.0.79"
+num_cpus = "1.13.1"
+fs2 = { version = "0.4.3", optional = true }
+levenshtein_automata = "0.2.1"
+uuid = { version = "1.0.0", features = ["v4", "serde"] }
+crossbeam-channel = "0.5.4"
+tantivy-query-grammar = { version="0.18.0", path="./query-grammar" }
+tantivy-bitpacker = { version="0.2", path="./bitpacker" }
+common = { version = "0.3", path = "./common/", package = "tantivy-common" }
+fastfield_codecs = { version="0.2", path="./fastfield_codecs", default-features = false }
+ownedbytes = { version="0.3", path="./ownedbytes" }
+stable_deref_trait = "1.2.0"
+rust-stemmers = "1.2.0"
+downcast-rs = "1.2.0"
+bitpacking = { version = "0.8.4", default-features = false, features = ["bitpacker4x"] }
+census = "0.4.0"
+fnv = "1.0.7"
+thiserror = "1.0.30"
+htmlescape = "0.3.1"
+fail = "0.5.0"
+murmurhash32 = "0.2.0"
+time = { version = "0.3.10", features = ["serde-well-known"] }
+smallvec = "1.8.0"
+rayon = "1.5.2"
+lru = "0.7.5"
+fastdivide = "0.4.0"
+itertools = "0.10.3"
+measure_time = "0.8.2"
+ciborium = { version = "0.2", optional = true}
+async-trait = "0.1.53"
+arc-swap = "1.5.0"
 
 [target.'cfg(windows)'.dependencies]
-winapi = "0.2"
+winapi = "0.3.9"
 
 [dev-dependencies]
-rand = "0.3"
-env_logger = "0.4"
+rand = "0.8.5"
+maplit = "1.0.2"
+matches = "0.1.9"
+pretty_assertions = "1.2.1"
+proptest = "1.0.0"
+criterion = "0.3.5"
+test-log = "0.2.10"
+env_logger = "0.9.0"
+pprof = { version = "0.10.0", features = ["flamegraph", "criterion"] }
+futures = "0.3.21"
+
+[dev-dependencies.fail]
+version = "0.5.0"
+features = ["failpoints"]
 
 [profile.release]
 opt-level = 3
 debug = false
-lto = true
 debug-assertions = false
 
+[profile.test]
+debug-assertions = true
+overflow-checks = true
 
 [features]
-default = ["mmap"]
-streamdict = []
-mmap = ["fst/mmap", "atomicwrites"]
+default = ["mmap", "lz4-compression" ]
+mmap = ["fs2", "tempfile", "memmap2"]
 
+brotli-compression = ["brotli"]
+lz4-compression = ["lz4_flex"]
+snappy-compression = ["snap"]
+zstd-compression = ["zstd"]
 
-[badges]
-travis-ci = { repository = "tantivy-search/tantivy" }
+failpoints = ["fail/failpoints"]
+unstable = [] # useful for benches.
+
+quickwit = ["ciborium"]
+
+[workspace]
+members = ["query-grammar", "bitpacker", "common", "fastfield_codecs", "ownedbytes"]
+
+# Following the "fail" crate best practises, we isolate
+# tests that define specific behavior in fail check points
+# in a different binary.
+#
+# We do that because, fail rely on a global definition of
+# failpoints behavior and hence, it is incompatible with
+# multithreading.
+[[test]]
+name = "failpoints"
+path = "tests/failpoints/mod.rs"
+required-features = ["fail/failpoints"]
+
+[[bench]]
+name = "analyzer"
+harness = false
+
+[[bench]]
+name = "index-bench"
+harness = false
 
-[[example]]
-name = "simple_search"
-required-features = ["mmap"]

LICENSE

@@ -1,4 +1,4 @@
-Copyright (c) 2016 Paul Masurel
+Copyright (c) 2018 by the project authors, as listed in the AUTHORS file. 
 
 Permission is hereby granted, free of charge, to any person obtaining a copy of this software and associated documentation files (the "Software"), to deal in the Software without restriction, including without limitation the rights to use, copy, modify, merge, publish, distribute, sublicense, and/or sell copies of the Software, and to permit persons to whom the Software is furnished to do so, subject to the following conditions:
 

Makefile

@@ -0,0 +1,6 @@
+test:
+	echo "Run test only... No examples."
+	cargo test --tests --lib
+
+fmt:
+	cargo +nightly fmt --all

README.md

@@ -1,85 +1,173 @@
+[![Docs](https://docs.rs/tantivy/badge.svg)](https://docs.rs/crate/tantivy/)
+[![Build Status](https://github.com/quickwit-oss/tantivy/actions/workflows/test.yml/badge.svg)](https://github.com/quickwit-oss/tantivy/actions/workflows/test.yml)
+[![codecov](https://codecov.io/gh/quickwit-oss/tantivy/branch/main/graph/badge.svg)](https://codecov.io/gh/quickwit-oss/tantivy)
+[![Join the chat at https://discord.gg/MT27AG5EVE](https://shields.io/discord/908281611840282624?label=chat%20on%20discord)](https://discord.gg/MT27AG5EVE)
+[![License: MIT](https://img.shields.io/badge/License-MIT-yellow.svg)](https://opensource.org/licenses/MIT)
+[![Crates.io](https://img.shields.io/crates/v/tantivy.svg)](https://crates.io/crates/tantivy)
+
 ![Tantivy](https://tantivy-search.github.io/logo/tantivy-logo.png)
 
-[![Build Status](https://travis-ci.org/tantivy-search/tantivy.svg?branch=master)](https://travis-ci.org/tantivy-search/tantivy)
-[![Coverage Status](https://coveralls.io/repos/github/tantivy-search/tantivy/badge.svg?branch=master&refresh1)](https://coveralls.io/github/tantivy-search/tantivy?branch=master)
-[![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?svg=true)](https://ci.appveyor.com/project/fulmicoton/tantivy)
+**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) 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.
 
-It is strongly inspired by Lucene's design.
+Tantivy is, in fact, strongly inspired by Lucene's design.
+
+If you are looking for an alternative to Elasticsearch or Apache Solr, check out [Quickwit](https://github.com/quickwit-oss/quickwit), our search engine built on top of Tantivy.
+
+# Benchmark
+
+The following [benchmark](https://tantivy-search.github.io/bench/) breakdowns
+performance for different types of queries/collections.
+
+Your mileage WILL vary depending on the nature of queries and their load.
+
+<img src="doc/assets/images/searchbenchmark.png">
 
 # Features
 
-- Tiny startup time (<10ms), perfect for command line tools
-- tf-idf scoring
-- Basic query language
-- Phrase queries
+- Full-text search
+- 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), [Vaporetto](https://crates.io/crates/vaporetto_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)
+- Natural query language (e.g. `(michael AND jackson) OR "king of pop"`)
+- Phrase queries search (e.g. `"michael jackson"`)
 - Incremental indexing
 - Multithreaded indexing (indexing English Wikipedia takes < 3 minutes on my desktop)
 - Mmap directory
-- optional SIMD integer compression
-- Single valued and multivalued u64 and i64 fast fields (equivalent of doc values in Lucene)
+- 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)
+- `&[u8]` fast fields
+- Text, i64, u64, f64, dates, and hierarchical facet fields
 - LZ4 compressed document store
 - Range queries
-- Faceting
-- configurable indexing (optional term frequency and position indexing
+- Faceted search
+- Configurable indexing (optional term frequency and position indexing)
+- JSON Field
+- Aggregation Collector: range buckets, average, and stats metrics
+- LogMergePolicy with deletes
+- Searcher Warmer API
 - Cheesy logo with a horse
 
-Tantivy supports Linux, MacOS and Windows.
+## Non-features
 
+Distributed search is out of the scope of Tantivy, but if you are looking for this feature, check out [Quickwit](https://github.com/quickwit-oss/quickwit/).
 
 # Getting started
 
-- [tantivy's usage example](http://fulmicoton.com/tantivy-examples/simple_search.html)
-- [tantivy-cli and its tutorial](https://github.com/tantivy-search/tantivy-cli).
-It will walk 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/)
-    - [For the last master branch](https://tantivy-search.github.io/tantivy/tantivy/index.html)
+Tantivy works on stable Rust and supports Linux, macOS, and Windows.
 
-# Compiling
+- [Tantivy's simple search example](https://tantivy-search.github.io/examples/basic_search.html)
+- [tantivy-cli and its tutorial](https://github.com/quickwit-oss/tantivy-cli) - `tantivy-cli` is an actual command-line interface that makes it easy for you to create a search engine,
+index documents, and search via the CLI or a small server with a REST API.
+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/)
 
-## Development
+# How can I support this project?
 
-Tantivy requires Rust Nightly because it uses requires the features [`box_syntax`](https://doc.rust-lang.org/stable/unstable-book/language-features/box-syntax.html), [`optin_builtin_traits`](https://github.com/rust-lang/rfcs/blob/master/text/0019-opt-in-builtin-traits.md), [`conservative_impl_trait`](https://github.com/rust-lang/rfcs/blob/master/text/1522-conservative-impl-trait.md),
-and [simd](https://github.com/rust-lang/rust/issues/27731).
+There are many ways to support this project.
 
+- Use Tantivy and tell us about your experience on [Discord](https://discord.gg/MT27AG5EVE) 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 Discord server](https://discord.gg/MT27AG5EVE))
+- Talk about Tantivy around you
 
-To check out and run test, you can simply run :
+# Contributing code
 
-    git clone git@github.com:tantivy-search/tantivy.git
+We use the GitHub Pull Request workflow: reference a GitHub ticket and/or include a comprehensive commit message when opening a PR.
+
+## Minimum supported Rust version
+
+Tantivy currently requires at least Rust 1.62 or later to compile.
+
+## Clone and build locally
+
+Tantivy compiles on stable Rust.
+To check out and run tests, you can simply run:
+
+```bash
+    git clone https://github.com/quickwit-oss/tantivy.git
     cd tantivy
-    cargo +nightly build
+    cargo build
+```
 
+## Run tests
 
-## Note on release build and performance
+Some tests will not run with just `cargo test` because of `fail-rs`.
+To run the tests exhaustively, run `./run-tests.sh`.
 
-If your project depends on `tantivy`, for better performance, make sure to enable
-`sse3` instructions using a RUSTFLAGS. (This instruction set is likely to
-be available on most `x86_64` CPUs you will encounter).
+## Debug
 
-For instance,
+You might find it useful to step through the programme with a debugger.
 
-    RUSTFLAGS='-C target-feature=+sse3'
+### A failing test
 
-Or, if you are targetting a specific cpu
+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`:
 
-    RUSTFLAGS='-C target-cpu=native' build --release
+```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
+```
 
-Regardless of the flags you pass, by default `tantivy` will contain `SSE3` instructions.
-If you want to disable those, you can run the following command :
+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:
 
-    cargo build --no-default-features
+```bash
+$gdb run --test-threads 1 --test $NAME_OF_TEST
+```
 
-Alternatively, if you are trying to compile `tantivy` without simd compression,
-you can disable this functionality. In this case, this submodule is not required
-and you can compile tantivy by using the `--no-default-features` flag.
+### An example
 
-    cargo build --no-default-features
+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/$EXAMPLE_NAME
+$ gdb run
+```
 
-# Contribute
+# Companies Using Tantivy
 
-Send me an email (paul.masurel at gmail.com) if you want to contribute to tantivy.
+<p align="left">
+<img align="center" src="doc/assets/images/etsy.png" alt="Etsy" height="25" width="auto" />&nbsp;
+<img align="center" src="doc/assets/images/Nuclia.png#gh-light-mode-only" alt="Nuclia" height="25" width="auto" /> &nbsp;
+<img align="center" src="doc/assets/images/humanfirst.png#gh-light-mode-only" alt="Humanfirst.ai" height="30" width="auto" />
+<img align="center" src="doc/assets/images/element.io.svg#gh-light-mode-only" alt="Element.io" height="25" width="auto" />
+<img align="center" src="doc/assets/images/nuclia-dark-theme.png#gh-dark-mode-only" alt="Nuclia" height="35" width="auto" /> &nbsp;
+<img align="center" src="doc/assets/images/humanfirst.ai-dark-theme.png#gh-dark-mode-only" alt="Humanfirst.ai" height="25" width="auto" />&nbsp; &nbsp;
+<img align="center" src="doc/assets/images/element-dark-theme.png#gh-dark-mode-only" alt="Element.io" height="25" width="auto" />
+</p>
+
+# FAQ
+
+### Can I use Tantivy in other languages?
+
+- Python → [tantivy-py](https://github.com/quickwit-oss/tantivy-py)
+- Ruby → [tantiny](https://github.com/baygeldin/tantiny)
+
+You can also find other bindings on [GitHub](https://github.com/search?q=tantivy) but they may be less maintained.
+
+### What are some examples of Tantivy use?
+
+- [seshat](https://github.com/matrix-org/seshat/): A matrix message database/indexer
+- [tantiny](https://github.com/baygeldin/tantiny): Tiny full-text search for Ruby
+- [lnx](https://github.com/lnx-search/lnx): adaptable, typo tolerant search engine with a REST API
+- and [more](https://github.com/search?q=tantivy)!
+
+### On average, how much faster is Tantivy compared to Lucene?
+
+- According to our [search latency benchmark](https://tantivy-search.github.io/bench/), Tantivy is approximately 2x faster than Lucene.
+
+### Does tantivy support incremental indexing?
+
+- Yes.
+
+### How can I edit documents?
+
+- Data in tantivy is immutable. To edit a document, the document needs to be deleted and reindexed.
+
+### When will my documents be searchable during indexing?
+
+- Documents will be searchable after a `commit` is called on an `IndexWriter`. Existing `IndexReader`s will also need to be reloaded in order to reflect the changes. Finally, changes are only visible to newly acquired `Searcher`.

appveyor.yml

@@ -4,11 +4,8 @@
 os: Visual Studio 2015
 environment:
   matrix:
-    - channel: nightly
+    - channel: stable
       target: x86_64-pc-windows-msvc
-    - channel: nightly
-      target: x86_64-pc-windows-gnu
-      msys_bits: 64
 
 install:
   - appveyor DownloadFile https://win.rustup.rs/ -FileName rustup-init.exe
@@ -21,5 +18,6 @@ install:
 build: false
 
 test_script:
-  - REM SET RUST_LOG=tantivy,test & cargo test --verbose
-  - REM SET RUST_BACKTRACE=1 & cargo run --example simple_search
+  - 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: &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);

benches/index-bench.rs

@@ -0,0 +1,121 @@
+use criterion::{criterion_group, criterion_main, Criterion};
+use pprof::criterion::{Output, PProfProfiler};
+use tantivy::schema::{INDEXED, STORED, STRING, TEXT};
+use tantivy::Index;
+
+const HDFS_LOGS: &str = include_str!("hdfs.json");
+const NUM_REPEATS: usize = 2;
+
+pub fn hdfs_index_benchmark(c: &mut Criterion) {
+    let schema = {
+        let mut schema_builder = tantivy::schema::SchemaBuilder::new();
+        schema_builder.add_u64_field("timestamp", INDEXED);
+        schema_builder.add_text_field("body", TEXT);
+        schema_builder.add_text_field("severity", STRING);
+        schema_builder.build()
+    };
+    let schema_with_store = {
+        let mut schema_builder = tantivy::schema::SchemaBuilder::new();
+        schema_builder.add_u64_field("timestamp", INDEXED | STORED);
+        schema_builder.add_text_field("body", TEXT | STORED);
+        schema_builder.add_text_field("severity", STRING | STORED);
+        schema_builder.build()
+    };
+    let dynamic_schema = {
+        let mut schema_builder = tantivy::schema::SchemaBuilder::new();
+        schema_builder.add_json_field("json", TEXT);
+        schema_builder.build()
+    };
+
+    let mut group = c.benchmark_group("index-hdfs");
+    group.sample_size(20);
+    group.bench_function("index-hdfs-no-commit", |b| {
+        b.iter(|| {
+            let index = Index::create_in_ram(schema.clone());
+            let index_writer = index.writer_with_num_threads(1, 100_000_000).unwrap();
+            for _ in 0..NUM_REPEATS {
+                for doc_json in HDFS_LOGS.trim().split("\n") {
+                    let doc = schema.parse_document(doc_json).unwrap();
+                    index_writer.add_document(doc).unwrap();
+                }
+            }
+        })
+    });
+    group.bench_function("index-hdfs-with-commit", |b| {
+        b.iter(|| {
+            let index = Index::create_in_ram(schema.clone());
+            let mut index_writer = index.writer_with_num_threads(1, 100_000_000).unwrap();
+            for _ in 0..NUM_REPEATS {
+                for doc_json in HDFS_LOGS.trim().split("\n") {
+                    let doc = schema.parse_document(doc_json).unwrap();
+                    index_writer.add_document(doc).unwrap();
+                }
+            }
+            index_writer.commit().unwrap();
+        })
+    });
+    group.bench_function("index-hdfs-no-commit-with-docstore", |b| {
+        b.iter(|| {
+            let index = Index::create_in_ram(schema_with_store.clone());
+            let index_writer = index.writer_with_num_threads(1, 100_000_000).unwrap();
+            for _ in 0..NUM_REPEATS {
+                for doc_json in HDFS_LOGS.trim().split("\n") {
+                    let doc = schema.parse_document(doc_json).unwrap();
+                    index_writer.add_document(doc).unwrap();
+                }
+            }
+        })
+    });
+    group.bench_function("index-hdfs-with-commit-with-docstore", |b| {
+        b.iter(|| {
+            let index = Index::create_in_ram(schema_with_store.clone());
+            let mut index_writer = index.writer_with_num_threads(1, 100_000_000).unwrap();
+            for _ in 0..NUM_REPEATS {
+                for doc_json in HDFS_LOGS.trim().split("\n") {
+                    let doc = schema.parse_document(doc_json).unwrap();
+                    index_writer.add_document(doc).unwrap();
+                }
+            }
+            index_writer.commit().unwrap();
+        })
+    });
+    group.bench_function("index-hdfs-no-commit-json-without-docstore", |b| {
+        b.iter(|| {
+            let index = Index::create_in_ram(dynamic_schema.clone());
+            let json_field = dynamic_schema.get_field("json").unwrap();
+            let mut index_writer = index.writer_with_num_threads(1, 100_000_000).unwrap();
+            for _ in 0..NUM_REPEATS {
+                for doc_json in HDFS_LOGS.trim().split("\n") {
+                    let json_val: serde_json::Map<String, serde_json::Value> =
+                        serde_json::from_str(doc_json).unwrap();
+                    let doc = tantivy::doc!(json_field=>json_val);
+                    index_writer.add_document(doc).unwrap();
+                }
+            }
+            index_writer.commit().unwrap();
+        })
+    });
+    group.bench_function("index-hdfs-with-commit-json-without-docstore", |b| {
+        b.iter(|| {
+            let index = Index::create_in_ram(dynamic_schema.clone());
+            let json_field = dynamic_schema.get_field("json").unwrap();
+            let mut index_writer = index.writer_with_num_threads(1, 100_000_000).unwrap();
+            for _ in 0..NUM_REPEATS {
+                for doc_json in HDFS_LOGS.trim().split("\n") {
+                    let json_val: serde_json::Map<String, serde_json::Value> =
+                        serde_json::from_str(doc_json).unwrap();
+                    let doc = tantivy::doc!(json_field=>json_val);
+                    index_writer.add_document(doc).unwrap();
+                }
+            }
+            index_writer.commit().unwrap();
+        })
+    });
+}
+
+criterion_group! {
+    name = benches;
+    config = Criterion::default().with_profiler(PProfProfiler::new(100, Output::Flamegraph(None)));
+    targets = hdfs_index_benchmark
+}
+criterion_main!(benches);

bitpacker/Cargo.toml

@@ -0,0 +1,15 @@
+[package]
+name = "tantivy-bitpacker"
+version = "0.2.0"
+edition = "2021"
+authors = ["Paul Masurel <paul.masurel@gmail.com>"]
+license = "MIT"
+categories = []
+description = """Tantivy-sub crate: bitpacking"""
+repository = "https://github.com/quickwit-oss/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 +1,35 @@
+#![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

@@ -0,0 +1,145 @@
+use std::convert::TryInto;
+use std::io;
+
+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 {
+            mini_buffer: 0u64,
+            mini_buffer_written: 0,
+        }
+    }
+
+    #[inline]
+    pub fn write<TWrite: io::Write>(
+        &mut self,
+        val: u64,
+        num_bits: u8,
+        output: &mut TWrite,
+    ) -> io::Result<()> {
+        let val_u64 = val as u64;
+        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_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_all(self.mini_buffer.to_le_bytes().as_ref())?;
+                self.mini_buffer_written = 0;
+                self.mini_buffer = 0u64;
+            }
+        }
+        Ok(())
+    }
+
+    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 bytes = self.mini_buffer.to_le_bytes();
+            output.write_all(&bytes[..num_bytes])?;
+            self.mini_buffer_written = 0;
+            self.mini_buffer = 0;
+        }
+        Ok(())
+    }
+
+    pub fn close<TWrite: io::Write>(&mut self, output: &mut TWrite) -> io::Result<()> {
+        self.flush(output)?;
+        // Padding the write file to simplify reads.
+        output.write_all(&[0u8; 7])?;
+        Ok(())
+    }
+}
+
+#[derive(Clone, Debug, Default)]
+pub struct BitUnpacker {
+    num_bits: u64,
+    mask: u64,
+}
+
+impl BitUnpacker {
+    pub fn new(num_bits: u8) -> BitUnpacker {
+        let mask: u64 = if num_bits == 64 {
+            !0u64
+        } else {
+            (1u64 << num_bits) - 1u64
+        };
+        BitUnpacker {
+            num_bits: u64::from(num_bits),
+            mask,
+        }
+    }
+
+    pub fn bit_width(&self) -> u8 {
+        self.num_bits as u8
+    }
+
+    #[inline]
+    pub fn get(&self, idx: u64, data: &[u8]) -> u64 {
+        if self.num_bits == 0 {
+            return 0u64;
+        }
+        let addr_in_bits = idx * self.num_bits;
+        let addr = addr_in_bits >> 3;
+        let bit_shift = addr_in_bits & 7;
+        debug_assert!(
+            addr + 8 <= data.len() as u64,
+            "The fast field field should have been padded with 7 bytes."
+        );
+        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 & self.mask
+    }
+}
+
+#[cfg(test)]
+mod test {
+    use super::{BitPacker, BitUnpacker};
+
+    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;
+        let vals: Vec<u64> = (0u64..len as u64)
+            .map(|i| if max_val == 0 { 0 } else { i % max_val })
+            .collect();
+        for &val in &vals {
+            bitpacker.write(val, num_bits, &mut data).unwrap();
+        }
+        bitpacker.close(&mut data).unwrap();
+        assert_eq!(data.len(), ((num_bits as usize) * len + 7) / 8 + 7);
+        let bitunpacker = BitUnpacker::new(num_bits);
+        (bitunpacker, vals, data)
+    }
+
+    fn test_bitpacker_util(len: usize, num_bits: u8) {
+        let (bitunpacker, vals, data) = create_fastfield_bitpacker(len, num_bits);
+        for (i, val) in vals.iter().enumerate() {
+            assert_eq!(bitunpacker.get(i as u64, &data), *val);
+        }
+    }
+
+    #[test]
+    fn test_bitpacker() {
+        test_bitpacker_util(10, 3);
+        test_bitpacker_util(10, 0);
+        test_bitpacker_util(10, 1);
+        test_bitpacker_util(6, 14);
+        test_bitpacker_util(1000, 14);
+    }
+}

bitpacker/src/blocked_bitpacker.rs

@@ -0,0 +1,179 @@
+use super::bitpacker::BitPacker;
+use super::compute_num_bits;
+use crate::{minmax, BitUnpacker};
+
+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);
+}
+
+fn mem_usage<T>(items: &Vec<T>) -> usize {
+    items.capacity() * std::mem::size_of::<T>()
+}
+
+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()
+            + mem_usage(&self.offset_and_bits)
+            + mem_usage(&self.buffer)
+    }
+
+    #[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 +1,80 @@
+mod bitpacker;
+mod blocked_bitpacker;
+
+pub use crate::bitpacker::{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
+}
+
+#[test]
+fn test_compute_num_bits() {
+    assert_eq!(compute_num_bits(1), 1u8);
+    assert_eq!(compute_num_bits(0), 0u8);
+    assert_eq!(compute_num_bits(2), 2u8);
+    assert_eq!(compute_num_bits(3), 2u8);
+    assert_eq!(compute_num_bits(4), 3u8);
+    assert_eq!(compute_num_bits(255), 8u8);
+    assert_eq!(compute_num_bits(256), 9u8);
+    assert_eq!(compute_num_bits(5_000_000_000), 33u8);
+}
+
+#[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)));
+}

ci/before_deploy.ps1

@@ -0,0 +1,23 @@
+# This script takes care of packaging the build artifacts that will go in the
+# release zipfile
+
+$SRC_DIR = $PWD.Path
+$STAGE = [System.Guid]::NewGuid().ToString()
+
+Set-Location $ENV:Temp
+New-Item -Type Directory -Name $STAGE
+Set-Location $STAGE
+
+$ZIP = "$SRC_DIR\$($Env:CRATE_NAME)-$($Env:APPVEYOR_REPO_TAG_NAME)-$($Env:TARGET).zip"
+
+# TODO Update this to package the right artifacts
+Copy-Item "$SRC_DIR\target\$($Env:TARGET)\release\hello.exe" '.\'
+
+7z a "$ZIP" *
+
+Push-AppveyorArtifact "$ZIP"
+
+Remove-Item *.* -Force
+Set-Location ..
+Remove-Item $STAGE
+Set-Location $SRC_DIR

ci/before_deploy.sh

@@ -0,0 +1,33 @@
+# This script takes care of building your crate and packaging it for release
+
+set -ex
+
+main() {
+    local src=$(pwd) \
+          stage=
+
+    case $TRAVIS_OS_NAME in
+        linux)
+            stage=$(mktemp -d)
+            ;;
+        osx)
+            stage=$(mktemp -d -t tmp)
+            ;;
+    esac
+
+    test -f Cargo.lock || cargo generate-lockfile
+
+    # TODO Update this to build the artifacts that matter to you
+    cross rustc --bin hello --target $TARGET --release -- -C lto
+
+    # TODO Update this to package the right artifacts
+    cp target/$TARGET/release/hello $stage/
+
+    cd $stage
+    tar czf $src/$CRATE_NAME-$TRAVIS_TAG-$TARGET.tar.gz *
+    cd $src
+
+    rm -rf $stage
+}
+
+main

ci/install.sh

@@ -0,0 +1,47 @@
+set -ex
+
+main() {
+    local target=
+    if [ $TRAVIS_OS_NAME = linux ]; then
+        target=x86_64-unknown-linux-musl
+        sort=sort
+    else
+        target=x86_64-apple-darwin
+        sort=gsort  # for `sort --sort-version`, from brew's coreutils.
+    fi
+
+    # Builds for iOS are done on OSX, but require the specific target to be
+    # installed.
+    case $TARGET in
+        aarch64-apple-ios)
+            rustup target install aarch64-apple-ios
+            ;;
+        armv7-apple-ios)
+            rustup target install armv7-apple-ios
+            ;;
+        armv7s-apple-ios)
+            rustup target install armv7s-apple-ios
+            ;;
+        i386-apple-ios)
+            rustup target install i386-apple-ios
+            ;;
+        x86_64-apple-ios)
+            rustup target install x86_64-apple-ios
+            ;;
+    esac
+
+    # This fetches latest stable release
+    local tag=$(git ls-remote --tags --refs --exit-code https://github.com/japaric/cross \
+                       | cut -d/ -f3 \
+                       | grep -E '^v[0.1.0-9.]+$' \
+                       | $sort --version-sort \
+                       | tail -n1)
+    curl -LSfs https://japaric.github.io/trust/install.sh | \
+        sh -s -- \
+           --force \
+           --git japaric/cross \
+           --tag $tag \
+           --target $target
+}
+
+main

ci/script.sh

@@ -0,0 +1,30 @@
+#!/usr/bin/env bash
+
+# This script takes care of testing your crate
+
+set -ex
+
+main() {
+    if [ ! -z $CODECOV ]; then
+        echo "Codecov"
+        cargo build --verbose && cargo coverage --verbose --all && bash <(curl -s https://codecov.io/bash) -s target/kcov
+    else
+        echo "Build"
+        cross build --target $TARGET
+        if [ ! -z $DISABLE_TESTS ]; then
+            return
+        fi
+        echo "Test"
+        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
+        cargo run --example  $(basename $example .rs)
+    done
+}
+
+# we don't run the "test phase" when doing deploys
+if [ -z $TRAVIS_TAG ]; then
+    main
+fi

common/Cargo.toml

@@ -0,0 +1,17 @@
+[package]
+name = "tantivy-common"
+version = "0.3.0"
+authors = ["Paul Masurel <paul@quickwit.io>", "Pascal Seitz <pascal@quickwit.io>"]
+license = "MIT"
+edition = "2021"
+description = "common traits and utility functions used by multiple tantivy subcrates"
+
+# See more keys and their definitions at https://doc.rust-lang.org/cargo/reference/manifest.html
+
+[dependencies]
+byteorder = "1.4.3"
+ownedbytes = { version="0.3", path="../ownedbytes" }
+
+[dev-dependencies]
+proptest = "1.0.0"
+rand = "0.8.4"

common/src/bitset.rs

@@ -0,0 +1,737 @@
+use std::convert::TryInto;
+use std::io::Write;
+use std::{fmt, io, u64};
+
+use ownedbytes::OwnedBytes;
+
+#[derive(Clone, Copy, Eq, PartialEq)]
+pub struct TinySet(u64);
+
+impl fmt::Debug for TinySet {
+    fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
+        self.into_iter().collect::<Vec<u32>>().fmt(f)
+    }
+}
+
+pub struct TinySetIterator(TinySet);
+impl Iterator for TinySetIterator {
+    type Item = u32;
+
+    #[inline]
+    fn next(&mut self) -> Option<Self::Item> {
+        self.0.pop_lowest()
+    }
+}
+
+impl IntoIterator for TinySet {
+    type Item = u32;
+    type IntoIter = TinySetIterator;
+    fn into_iter(self) -> Self::IntoIter {
+        TinySetIterator(self)
+    }
+}
+
+impl TinySet {
+    pub fn serialize<T: Write>(&self, writer: &mut T) -> io::Result<()> {
+        writer.write_all(self.0.to_le_bytes().as_ref())
+    }
+
+    pub fn into_bytes(self) -> [u8; 8] {
+        self.0.to_le_bytes()
+    }
+
+    #[inline]
+    pub fn deserialize(data: [u8; 8]) -> Self {
+        let val: u64 = u64::from_le_bytes(data);
+        TinySet(val)
+    }
+
+    /// Returns an empty `TinySet`.
+    #[inline]
+    pub fn empty() -> TinySet {
+        TinySet(0u64)
+    }
+
+    /// Returns a full `TinySet`.
+    #[inline]
+    pub fn full() -> TinySet {
+        TinySet::empty().complement()
+    }
+
+    pub fn clear(&mut self) {
+        self.0 = 0u64;
+    }
+
+    /// Returns the complement of the set in `[0, 64[`.
+    ///
+    /// Careful on making this function public, as it will break the padding handling in the last
+    /// bucket.
+    #[inline]
+    fn complement(self) -> TinySet {
+        TinySet(!self.0)
+    }
+
+    /// Returns true iff the `TinySet` contains the element `el`.
+    #[inline]
+    pub fn contains(self, el: u32) -> bool {
+        !self.intersect(TinySet::singleton(el)).is_empty()
+    }
+
+    /// Returns the number of elements in the TinySet.
+    #[inline]
+    pub fn len(self) -> u32 {
+        self.0.count_ones()
+    }
+
+    /// Returns the intersection of `self` and `other`
+    #[inline]
+    #[must_use]
+    pub fn intersect(self, other: TinySet) -> TinySet {
+        TinySet(self.0 & other.0)
+    }
+
+    /// Creates a new `TinySet` containing only one element
+    /// within `[0; 64[`
+    #[inline]
+    pub fn singleton(el: u32) -> TinySet {
+        TinySet(1u64 << u64::from(el))
+    }
+
+    /// Insert a new element within [0..64)
+    #[inline]
+    #[must_use]
+    pub fn insert(self, el: u32) -> TinySet {
+        self.union(TinySet::singleton(el))
+    }
+
+    /// Removes an element within [0..64)
+    #[inline]
+    #[must_use]
+    pub fn remove(self, el: u32) -> TinySet {
+        self.intersect(TinySet::singleton(el).complement())
+    }
+
+    /// Insert a new element within [0..64)
+    ///
+    /// returns true if the set changed
+    #[inline]
+    pub fn insert_mut(&mut self, el: u32) -> bool {
+        let old = *self;
+        *self = old.insert(el);
+        old != *self
+    }
+
+    /// Remove a element within [0..64)
+    ///
+    /// returns true if the set changed
+    #[inline]
+    pub fn remove_mut(&mut self, el: u32) -> bool {
+        let old = *self;
+        *self = old.remove(el);
+        old != *self
+    }
+
+    /// Returns the union of two tinysets
+    #[inline]
+    #[must_use]
+    pub fn union(self, other: TinySet) -> TinySet {
+        TinySet(self.0 | other.0)
+    }
+
+    /// Returns true iff the `TinySet` is empty.
+    #[inline]
+    pub fn is_empty(self) -> bool {
+        self.0 == 0u64
+    }
+
+    /// Returns the lowest element in the `TinySet`
+    /// and removes it.
+    #[inline]
+    pub fn pop_lowest(&mut self) -> Option<u32> {
+        if self.is_empty() {
+            None
+        } else {
+            let lowest = self.0.trailing_zeros() as u32;
+            self.0 ^= TinySet::singleton(lowest).0;
+            Some(lowest)
+        }
+    }
+
+    /// Returns a `TinySet` than contains all values up
+    /// to limit excluded.
+    ///
+    /// The limit is assumed to be strictly lower than 64.
+    pub fn range_lower(upper_bound: u32) -> TinySet {
+        TinySet((1u64 << u64::from(upper_bound % 64u32)) - 1u64)
+    }
+
+    /// Returns a `TinySet` that contains all values greater
+    /// or equal to the given limit, included. (and up to 63)
+    ///
+    /// The limit is assumed to be strictly lower than 64.
+    pub fn range_greater_or_equal(from_included: u32) -> TinySet {
+        TinySet::range_lower(from_included).complement()
+    }
+}
+
+#[derive(Clone)]
+pub struct BitSet {
+    tinysets: Box<[TinySet]>,
+    len: u64,
+    max_value: u32,
+}
+
+fn num_buckets(max_val: u32) -> u32 {
+    (max_val + 63u32) / 64u32
+}
+
+impl BitSet {
+    /// serialize a `BitSet`.
+    pub fn serialize<T: Write>(&self, writer: &mut T) -> io::Result<()> {
+        writer.write_all(self.max_value.to_le_bytes().as_ref())?;
+        for tinyset in self.tinysets.iter().cloned() {
+            writer.write_all(&tinyset.into_bytes())?;
+        }
+        writer.flush()?;
+        Ok(())
+    }
+
+    /// Create a new `BitSet` that may contain elements
+    /// within `[0, max_val)`.
+    pub fn with_max_value(max_value: u32) -> BitSet {
+        let num_buckets = num_buckets(max_value);
+        let tinybitsets = vec![TinySet::empty(); num_buckets as usize].into_boxed_slice();
+        BitSet {
+            tinysets: tinybitsets,
+            len: 0,
+            max_value,
+        }
+    }
+
+    /// Create a new `BitSet` that may contain elements. Initially all values will be set.
+    /// within `[0, max_val)`.
+    pub fn with_max_value_and_full(max_value: u32) -> BitSet {
+        let num_buckets = num_buckets(max_value);
+        let mut tinybitsets = vec![TinySet::full(); num_buckets as usize].into_boxed_slice();
+
+        // Fix padding
+        let lower = max_value % 64u32;
+        if lower != 0 {
+            tinybitsets[tinybitsets.len() - 1] = TinySet::range_lower(lower);
+        }
+        BitSet {
+            tinysets: tinybitsets,
+            len: max_value as u64,
+            max_value,
+        }
+    }
+
+    /// Removes all elements from the `BitSet`.
+    pub fn clear(&mut self) {
+        for tinyset in self.tinysets.iter_mut() {
+            *tinyset = TinySet::empty();
+        }
+    }
+
+    /// Intersect with serialized bitset
+    pub fn intersect_update(&mut self, other: &ReadOnlyBitSet) {
+        self.intersect_update_with_iter(other.iter_tinysets());
+    }
+
+    /// Intersect with tinysets
+    fn intersect_update_with_iter(&mut self, other: impl Iterator<Item = TinySet>) {
+        self.len = 0;
+        for (left, right) in self.tinysets.iter_mut().zip(other) {
+            *left = left.intersect(right);
+            self.len += left.len() as u64;
+        }
+    }
+
+    /// Returns the number of elements in the `BitSet`.
+    #[inline]
+    pub fn len(&self) -> usize {
+        self.len as usize
+    }
+
+    /// Inserts an element in the `BitSet`
+    #[inline]
+    pub fn insert(&mut self, el: u32) {
+        // we do not check saturated els.
+        let higher = el / 64u32;
+        let lower = el % 64u32;
+        self.len += u64::from(self.tinysets[higher as usize].insert_mut(lower));
+    }
+
+    /// Inserts an element in the `BitSet`
+    #[inline]
+    pub fn remove(&mut self, el: u32) {
+        // we do not check saturated els.
+        let higher = el / 64u32;
+        let lower = el % 64u32;
+        self.len -= u64::from(self.tinysets[higher as usize].remove_mut(lower));
+    }
+
+    /// Returns true iff the elements is in the `BitSet`.
+    #[inline]
+    pub fn contains(&self, el: u32) -> bool {
+        self.tinyset(el / 64u32).contains(el % 64)
+    }
+
+    /// Returns the first non-empty `TinySet` associated with a bucket lower
+    /// or greater than bucket.
+    ///
+    /// Reminder: the tiny set with the bucket `bucket`, represents the
+    /// elements from `bucket * 64` to `(bucket+1) * 64`.
+    pub fn first_non_empty_bucket(&self, bucket: u32) -> Option<u32> {
+        self.tinysets[bucket as usize..]
+            .iter()
+            .cloned()
+            .position(|tinyset| !tinyset.is_empty())
+            .map(|delta_bucket| bucket + delta_bucket as u32)
+    }
+
+    #[inline]
+    pub fn max_value(&self) -> u32 {
+        self.max_value
+    }
+
+    /// Returns the tiny bitset representing the
+    /// the set restricted to the number range from
+    /// `bucket * 64` to `(bucket + 1) * 64`.
+    pub fn tinyset(&self, bucket: u32) -> TinySet {
+        self.tinysets[bucket as usize]
+    }
+}
+
+/// Serialized BitSet.
+#[derive(Clone)]
+pub struct ReadOnlyBitSet {
+    data: OwnedBytes,
+    max_value: u32,
+}
+
+pub fn intersect_bitsets(left: &ReadOnlyBitSet, other: &ReadOnlyBitSet) -> ReadOnlyBitSet {
+    assert_eq!(left.max_value(), other.max_value());
+    assert_eq!(left.data.len(), other.data.len());
+    let union_tinyset_it = left
+        .iter_tinysets()
+        .zip(other.iter_tinysets())
+        .map(|(left_tinyset, right_tinyset)| left_tinyset.intersect(right_tinyset));
+    let mut output_dataset: Vec<u8> = Vec::with_capacity(left.data.len());
+    for tinyset in union_tinyset_it {
+        output_dataset.extend_from_slice(&tinyset.into_bytes());
+    }
+    ReadOnlyBitSet {
+        data: OwnedBytes::new(output_dataset),
+        max_value: left.max_value(),
+    }
+}
+
+impl ReadOnlyBitSet {
+    pub fn open(data: OwnedBytes) -> Self {
+        let (max_value_data, data) = data.split(4);
+        assert_eq!(data.len() % 8, 0);
+        let max_value: u32 = u32::from_le_bytes(max_value_data.as_ref().try_into().unwrap());
+        ReadOnlyBitSet { data, max_value }
+    }
+
+    /// Number of elements in the bitset.
+    #[inline]
+    pub fn len(&self) -> usize {
+        self.iter_tinysets()
+            .map(|tinyset| tinyset.len() as usize)
+            .sum()
+    }
+
+    /// Iterate the tinyset on the fly from serialized data.
+    #[inline]
+    fn iter_tinysets(&self) -> impl Iterator<Item = TinySet> + '_ {
+        self.data.chunks_exact(8).map(move |chunk| {
+            let tinyset: TinySet = TinySet::deserialize(chunk.try_into().unwrap());
+            tinyset
+        })
+    }
+
+    /// Iterate over the positions of the elements.
+    #[inline]
+    pub fn iter(&self) -> impl Iterator<Item = u32> + '_ {
+        self.iter_tinysets()
+            .enumerate()
+            .flat_map(move |(chunk_num, tinyset)| {
+                let chunk_base_val = chunk_num as u32 * 64;
+                tinyset
+                    .into_iter()
+                    .map(move |val| val + chunk_base_val)
+                    .take_while(move |doc| *doc < self.max_value)
+            })
+    }
+
+    /// Returns true iff the elements is in the `BitSet`.
+    #[inline]
+    pub fn contains(&self, el: u32) -> bool {
+        let byte_offset = el / 8u32;
+        let b: u8 = self.data[byte_offset as usize];
+        let shift = (el % 8) as u8;
+        b & (1u8 << shift) != 0
+    }
+
+    /// Maximum value the bitset may contain.
+    /// (Note this is not the maximum value contained in the set.)
+    ///
+    /// A bitset has an intrinsic capacity.
+    /// It only stores elements within [0..max_value).
+    #[inline]
+    pub fn max_value(&self) -> u32 {
+        self.max_value
+    }
+
+    /// Number of bytes used in the bitset representation.
+    pub fn num_bytes(&self) -> usize {
+        self.data.len()
+    }
+}
+
+impl<'a> From<&'a BitSet> for ReadOnlyBitSet {
+    fn from(bitset: &'a BitSet) -> ReadOnlyBitSet {
+        let mut buffer = Vec::with_capacity(bitset.tinysets.len() * 8 + 4);
+        bitset
+            .serialize(&mut buffer)
+            .expect("serializing into a buffer should never fail");
+        ReadOnlyBitSet::open(OwnedBytes::new(buffer))
+    }
+}
+
+#[cfg(test)]
+mod tests {
+
+    use std::collections::HashSet;
+
+    use ownedbytes::OwnedBytes;
+    use rand::distributions::Bernoulli;
+    use rand::rngs::StdRng;
+    use rand::{Rng, SeedableRng};
+
+    use super::{BitSet, ReadOnlyBitSet, TinySet};
+
+    #[test]
+    fn test_read_serialized_bitset_full_multi() {
+        for i in 0..1000 {
+            let bitset = BitSet::with_max_value_and_full(i);
+            let mut out = vec![];
+            bitset.serialize(&mut out).unwrap();
+
+            let bitset = ReadOnlyBitSet::open(OwnedBytes::new(out));
+            assert_eq!(bitset.len() as usize, i as usize);
+        }
+    }
+
+    #[test]
+    fn test_read_serialized_bitset_full_block() {
+        let bitset = BitSet::with_max_value_and_full(64);
+        let mut out = vec![];
+        bitset.serialize(&mut out).unwrap();
+
+        let bitset = ReadOnlyBitSet::open(OwnedBytes::new(out));
+        assert_eq!(bitset.len() as usize, 64);
+    }
+
+    #[test]
+    fn test_read_serialized_bitset_full() {
+        let mut bitset = BitSet::with_max_value_and_full(5);
+        bitset.remove(3);
+        let mut out = vec![];
+        bitset.serialize(&mut out).unwrap();
+
+        let bitset = ReadOnlyBitSet::open(OwnedBytes::new(out));
+        assert_eq!(bitset.len(), 4);
+    }
+
+    #[test]
+    fn test_bitset_intersect() {
+        let bitset_serialized = {
+            let mut bitset = BitSet::with_max_value_and_full(5);
+            bitset.remove(1);
+            bitset.remove(3);
+            let mut out = vec![];
+            bitset.serialize(&mut out).unwrap();
+
+            ReadOnlyBitSet::open(OwnedBytes::new(out))
+        };
+
+        let mut bitset = BitSet::with_max_value_and_full(5);
+        bitset.remove(1);
+        bitset.intersect_update(&bitset_serialized);
+
+        assert!(bitset.contains(0));
+        assert!(!bitset.contains(1));
+        assert!(bitset.contains(2));
+        assert!(!bitset.contains(3));
+        assert!(bitset.contains(4));
+
+        bitset.intersect_update_with_iter(vec![TinySet::singleton(0)].into_iter());
+
+        assert!(bitset.contains(0));
+        assert!(!bitset.contains(1));
+        assert!(!bitset.contains(2));
+        assert!(!bitset.contains(3));
+        assert!(!bitset.contains(4));
+        assert_eq!(bitset.len(), 1);
+
+        bitset.intersect_update_with_iter(vec![TinySet::singleton(1)].into_iter());
+        assert!(!bitset.contains(0));
+        assert!(!bitset.contains(1));
+        assert!(!bitset.contains(2));
+        assert!(!bitset.contains(3));
+        assert!(!bitset.contains(4));
+        assert_eq!(bitset.len(), 0);
+    }
+
+    #[test]
+    fn test_read_serialized_bitset_empty() {
+        let mut bitset = BitSet::with_max_value(5);
+        bitset.insert(3);
+        let mut out = vec![];
+        bitset.serialize(&mut out).unwrap();
+
+        let bitset = ReadOnlyBitSet::open(OwnedBytes::new(out));
+        assert_eq!(bitset.len(), 1);
+
+        {
+            let bitset = BitSet::with_max_value(5);
+            let mut out = vec![];
+            bitset.serialize(&mut out).unwrap();
+            let bitset = ReadOnlyBitSet::open(OwnedBytes::new(out));
+            assert_eq!(bitset.len(), 0);
+        }
+    }
+
+    #[test]
+    fn test_tiny_set_remove() {
+        {
+            let mut u = TinySet::empty().insert(63u32).insert(5).remove(63u32);
+            assert_eq!(u.pop_lowest(), Some(5u32));
+            assert!(u.pop_lowest().is_none());
+        }
+        {
+            let mut u = TinySet::empty()
+                .insert(63u32)
+                .insert(1)
+                .insert(5)
+                .remove(63u32);
+            assert_eq!(u.pop_lowest(), Some(1u32));
+            assert_eq!(u.pop_lowest(), Some(5u32));
+            assert!(u.pop_lowest().is_none());
+        }
+        {
+            let mut u = TinySet::empty().insert(1).remove(63u32);
+            assert_eq!(u.pop_lowest(), Some(1u32));
+            assert!(u.pop_lowest().is_none());
+        }
+        {
+            let mut u = TinySet::empty().insert(1).remove(1u32);
+            assert!(u.pop_lowest().is_none());
+        }
+    }
+    #[test]
+    fn test_tiny_set() {
+        assert!(TinySet::empty().is_empty());
+        {
+            let mut u = TinySet::empty().insert(1u32);
+            assert_eq!(u.pop_lowest(), Some(1u32));
+            assert!(u.pop_lowest().is_none())
+        }
+        {
+            let mut u = TinySet::empty().insert(1u32).insert(1u32);
+            assert_eq!(u.pop_lowest(), Some(1u32));
+            assert!(u.pop_lowest().is_none())
+        }
+        {
+            let mut u = TinySet::empty().insert(2u32);
+            assert_eq!(u.pop_lowest(), Some(2u32));
+            u.insert_mut(1u32);
+            assert_eq!(u.pop_lowest(), Some(1u32));
+            assert!(u.pop_lowest().is_none());
+        }
+        {
+            let mut u = TinySet::empty().insert(63u32);
+            assert_eq!(u.pop_lowest(), Some(63u32));
+            assert!(u.pop_lowest().is_none());
+        }
+        {
+            let mut u = TinySet::empty().insert(63u32).insert(5);
+            assert_eq!(u.pop_lowest(), Some(5u32));
+            assert_eq!(u.pop_lowest(), Some(63u32));
+            assert!(u.pop_lowest().is_none());
+        }
+        {
+            let original = TinySet::empty().insert(63u32).insert(5);
+            let after_serialize_deserialize = TinySet::deserialize(original.into_bytes());
+            assert_eq!(original, after_serialize_deserialize);
+        }
+    }
+
+    #[test]
+    fn test_bitset() {
+        let test_against_hashset = |els: &[u32], max_value: u32| {
+            let mut hashset: HashSet<u32> = HashSet::new();
+            let mut bitset = BitSet::with_max_value(max_value);
+            for &el in els {
+                assert!(el < max_value);
+                hashset.insert(el);
+                bitset.insert(el);
+            }
+            for el in 0..max_value {
+                assert_eq!(hashset.contains(&el), bitset.contains(el));
+            }
+            assert_eq!(bitset.max_value(), max_value);
+
+            // test deser
+            let mut data = vec![];
+            bitset.serialize(&mut data).unwrap();
+            let ro_bitset = ReadOnlyBitSet::open(OwnedBytes::new(data));
+            for el in 0..max_value {
+                assert_eq!(hashset.contains(&el), ro_bitset.contains(el));
+            }
+            assert_eq!(ro_bitset.max_value(), max_value);
+            assert_eq!(ro_bitset.len(), els.len());
+        };
+
+        test_against_hashset(&[], 0);
+        test_against_hashset(&[], 1);
+        test_against_hashset(&[0u32], 1);
+        test_against_hashset(&[0u32], 100);
+        test_against_hashset(&[1u32, 2u32], 4);
+        test_against_hashset(&[99u32], 100);
+        test_against_hashset(&[63u32], 64);
+        test_against_hashset(&[62u32, 63u32], 64);
+    }
+
+    #[test]
+    fn test_bitset_num_buckets() {
+        use super::num_buckets;
+        assert_eq!(num_buckets(0u32), 0);
+        assert_eq!(num_buckets(1u32), 1);
+        assert_eq!(num_buckets(64u32), 1);
+        assert_eq!(num_buckets(65u32), 2);
+        assert_eq!(num_buckets(128u32), 2);
+        assert_eq!(num_buckets(129u32), 3);
+    }
+
+    #[test]
+    fn test_tinyset_range() {
+        assert_eq!(
+            TinySet::range_lower(3).into_iter().collect::<Vec<u32>>(),
+            [0, 1, 2]
+        );
+        assert!(TinySet::range_lower(0).is_empty());
+        assert_eq!(
+            TinySet::range_lower(63).into_iter().collect::<Vec<u32>>(),
+            (0u32..63u32).collect::<Vec<_>>()
+        );
+        assert_eq!(
+            TinySet::range_lower(1).into_iter().collect::<Vec<u32>>(),
+            [0]
+        );
+        assert_eq!(
+            TinySet::range_lower(2).into_iter().collect::<Vec<u32>>(),
+            [0, 1]
+        );
+        assert_eq!(
+            TinySet::range_greater_or_equal(3)
+                .into_iter()
+                .collect::<Vec<u32>>(),
+            (3u32..64u32).collect::<Vec<_>>()
+        );
+    }
+
+    #[test]
+    fn test_bitset_len() {
+        let mut bitset = BitSet::with_max_value(1_000);
+        assert_eq!(bitset.len(), 0);
+        bitset.insert(3u32);
+        assert_eq!(bitset.len(), 1);
+        bitset.insert(103u32);
+        assert_eq!(bitset.len(), 2);
+        bitset.insert(3u32);
+        assert_eq!(bitset.len(), 2);
+        bitset.insert(103u32);
+        assert_eq!(bitset.len(), 2);
+        bitset.insert(104u32);
+        assert_eq!(bitset.len(), 3);
+        bitset.remove(105u32);
+        assert_eq!(bitset.len(), 3);
+        bitset.remove(104u32);
+        assert_eq!(bitset.len(), 2);
+        bitset.remove(3u32);
+        assert_eq!(bitset.len(), 1);
+        bitset.remove(103u32);
+        assert_eq!(bitset.len(), 0);
+    }
+
+    pub fn sample_with_seed(n: u32, ratio: f64, seed_val: u8) -> Vec<u32> {
+        StdRng::from_seed([seed_val; 32])
+            .sample_iter(&Bernoulli::new(ratio).unwrap())
+            .take(n as usize)
+            .enumerate()
+            .filter_map(|(val, keep)| if keep { Some(val as u32) } else { None })
+            .collect()
+    }
+
+    pub fn sample(n: u32, ratio: f64) -> Vec<u32> {
+        sample_with_seed(n, ratio, 4)
+    }
+
+    #[test]
+    fn test_bitset_clear() {
+        let mut bitset = BitSet::with_max_value(1_000);
+        let els = sample(1_000, 0.01f64);
+        for &el in &els {
+            bitset.insert(el);
+        }
+        assert!(els.iter().all(|el| bitset.contains(*el)));
+        bitset.clear();
+        for el in 0u32..1000u32 {
+            assert!(!bitset.contains(el));
+        }
+    }
+}
+
+#[cfg(all(test, feature = "unstable"))]
+mod bench {
+
+    use test;
+
+    use super::{BitSet, TinySet};
+
+    #[bench]
+    fn bench_tinyset_pop(b: &mut test::Bencher) {
+        b.iter(|| {
+            let mut tinyset = TinySet::singleton(test::black_box(31u32));
+            tinyset.pop_lowest();
+            tinyset.pop_lowest();
+            tinyset.pop_lowest();
+            tinyset.pop_lowest();
+            tinyset.pop_lowest();
+            tinyset.pop_lowest();
+        });
+    }
+
+    #[bench]
+    fn bench_tinyset_sum(b: &mut test::Bencher) {
+        let tiny_set = TinySet::empty().insert(10u32).insert(14u32).insert(21u32);
+        b.iter(|| {
+            assert_eq!(test::black_box(tiny_set).into_iter().sum::<u32>(), 45u32);
+        });
+    }
+
+    #[bench]
+    fn bench_tinyarr_sum(b: &mut test::Bencher) {
+        let v = [10u32, 14u32, 21u32];
+        b.iter(|| test::black_box(v).iter().cloned().sum::<u32>());
+    }
+
+    #[bench]
+    fn bench_bitset_initialize(b: &mut test::Bencher) {
+        b.iter(|| BitSet::with_max_value(1_000_000));
+    }
+}

common/src/lib.rs

@@ -0,0 +1,171 @@
+#![allow(clippy::len_without_is_empty)]
+
+use std::ops::Deref;
+
+pub use byteorder::LittleEndian as Endianness;
+
+mod bitset;
+mod serialize;
+mod vint;
+mod writer;
+
+pub use bitset::*;
+pub use serialize::{BinarySerializable, DeserializeFrom, FixedSize};
+pub use vint::{
+    deserialize_vint_u128, read_u32_vint, read_u32_vint_no_advance, serialize_vint_u128,
+    serialize_vint_u32, write_u32_vint, VInt, VIntU128,
+};
+pub use writer::{AntiCallToken, CountingWriter, TerminatingWrite};
+
+/// Has length trait
+pub trait HasLen {
+    /// Return length
+    fn len(&self) -> usize;
+
+    /// Returns true iff empty.
+    fn is_empty(&self) -> bool {
+        self.len() == 0
+    }
+}
+
+impl<T: Deref<Target = [u8]>> HasLen for T {
+    fn len(&self) -> usize {
+        self.deref().len()
+    }
+}
+
+const HIGHEST_BIT: u64 = 1 << 63;
+
+/// Maps a `i64` to `u64`
+///
+/// For simplicity, tantivy internally handles `i64` as `u64`.
+/// The mapping is defined by this function.
+///
+/// Maps `i64` to `u64` so that
+/// `-2^63 .. 2^63-1` is mapped
+///     to
+/// `0 .. 2^64-1`
+/// in that order.
+///
+/// This is more suited than simply casting (`val as u64`)
+/// because of bitpacking.
+///
+/// Imagine a list of `i64` ranging from -10 to 10.
+/// When casting negative values, the negative values are projected
+/// to values over 2^63, and all values end up requiring 64 bits.
+///
+/// # See also
+/// The reverse mapping is [`u64_to_i64()`].
+#[inline]
+pub fn i64_to_u64(val: i64) -> u64 {
+    (val as u64) ^ HIGHEST_BIT
+}
+
+/// Reverse the mapping given by [`i64_to_u64()`].
+#[inline]
+pub fn u64_to_i64(val: u64) -> i64 {
+    (val ^ HIGHEST_BIT) as i64
+}
+
+/// Maps a `f64` to `u64`
+///
+/// For simplicity, tantivy internally handles `f64` as `u64`.
+/// The mapping is defined by this function.
+///
+/// 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()`].
+#[inline]
+pub fn f64_to_u64(val: f64) -> u64 {
+    let bits = val.to_bits();
+    if val.is_sign_positive() {
+        bits ^ HIGHEST_BIT
+    } else {
+        !bits
+    }
+}
+
+/// Reverse the mapping given by [`f64_to_u64()`].
+#[inline]
+pub fn u64_to_f64(val: u64) -> f64 {
+    f64::from_bits(if val & HIGHEST_BIT != 0 {
+        val ^ HIGHEST_BIT
+    } else {
+        !val
+    })
+}
+
+#[cfg(test)]
+pub mod test {
+
+    use proptest::prelude::*;
+
+    use super::{f64_to_u64, i64_to_u64, u64_to_f64, u64_to_i64, BinarySerializable, FixedSize};
+
+    fn test_i64_converter_helper(val: i64) {
+        assert_eq!(u64_to_i64(i64_to_u64(val)), val);
+    }
+
+    fn test_f64_converter_helper(val: f64) {
+        assert_eq!(u64_to_f64(f64_to_u64(val)), val);
+    }
+
+    pub fn fixed_size_test<O: BinarySerializable + FixedSize + Default>() {
+        let mut buffer = Vec::new();
+        O::default().serialize(&mut buffer).unwrap();
+        assert_eq!(buffer.len(), O::SIZE_IN_BYTES);
+    }
+
+    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), u64::MIN);
+        assert_eq!(i64_to_u64(i64::MAX), u64::MAX);
+        test_i64_converter_helper(0i64);
+        test_i64_converter_helper(i64::MIN);
+        test_i64_converter_helper(i64::MAX);
+        for i in -1000i64..1000i64 {
+            test_i64_converter_helper(i);
+        }
+    }
+
+    #[test]
+    fn test_f64_converter() {
+        test_f64_converter_helper(f64::INFINITY);
+        test_f64_converter_helper(f64::NEG_INFINITY);
+        test_f64_converter_helper(0.0);
+        test_f64_converter_helper(-0.0);
+        test_f64_converter_helper(1.0);
+        test_f64_converter_helper(-1.0);
+    }
+
+    #[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(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
+        assert!(f64_to_u64(1.0) > f64_to_u64(-1.0)); // pos > neg
+        assert!(f64_to_u64(-1.5) < f64_to_u64(-1.0));
+        assert!(f64_to_u64(-2.0) < f64_to_u64(1.0));
+        assert!(f64_to_u64(-2.0) < f64_to_u64(-1.5));
+    }
+}

common/src/serialize.rs

@@ -1,10 +1,9 @@
+use std::io::{Read, Write};
+use std::{fmt, io};
+
 use byteorder::{ReadBytesExt, WriteBytesExt};
-use common::Endianness;
-use std::fmt;
-use std::io::Write;
-use std::io::Read;
-use std::io;
-use common::VInt;
+
+use crate::{Endianness, VInt};
 
 /// Trait for a simple binary serialization.
 pub trait BinarySerializable: fmt::Debug + Sized {
@@ -14,6 +13,20 @@ 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 preferable, 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 +74,11 @@ 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 +107,32 @@ impl FixedSize for u64 {
     const SIZE_IN_BYTES: usize = 8;
 }
 
+impl BinarySerializable for u128 {
+    fn serialize<W: Write>(&self, writer: &mut W) -> io::Result<()> {
+        writer.write_u128::<Endianness>(*self)
+    }
+    fn deserialize<R: Read>(reader: &mut R) -> io::Result<Self> {
+        reader.read_u128::<Endianness>()
+    }
+}
+
+impl FixedSize for u128 {
+    const SIZE_IN_BYTES: usize = 16;
+}
+
+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)
@@ -102,6 +146,19 @@ impl FixedSize for i64 {
     const SIZE_IN_BYTES: usize = 8;
 }
 
+impl BinarySerializable for f64 {
+    fn serialize<W: Write>(&self, writer: &mut W) -> io::Result<()> {
+        writer.write_f64::<Endianness>(*self)
+    }
+    fn deserialize<R: Read>(reader: &mut R) -> io::Result<Self> {
+        reader.read_f64::<Endianness>()
+    }
+}
+
+impl FixedSize for f64 {
+    const SIZE_IN_BYTES: usize = 8;
+}
+
 impl BinarySerializable for u8 {
     fn serialize<W: Write>(&self, writer: &mut W) -> io::Result<()> {
         writer.write_u8(*self)
@@ -115,6 +172,27 @@ impl FixedSize for u8 {
     const SIZE_IN_BYTES: usize = 1;
 }
 
+impl BinarySerializable for bool {
+    fn serialize<W: Write>(&self, writer: &mut W) -> io::Result<()> {
+        writer.write_u8(u8::from(*self))
+    }
+    fn deserialize<R: Read>(reader: &mut R) -> io::Result<bool> {
+        let val = reader.read_u8()?;
+        match val {
+            0 => Ok(false),
+            1 => Ok(true),
+            _ => Err(io::Error::new(
+                io::ErrorKind::InvalidData,
+                "invalid bool value on deserialization, data corrupted",
+            )),
+        }
+    }
+}
+
+impl FixedSize for bool {
+    const SIZE_IN_BYTES: usize = 1;
+}
+
 impl BinarySerializable for String {
     fn serialize<W: Write>(&self, writer: &mut W) -> io::Result<()> {
         let data: &[u8] = self.as_bytes();
@@ -135,9 +213,8 @@ impl BinarySerializable for String {
 #[cfg(test)]
 pub mod test {
 
-    use common::VInt;
-    use super::*;
-
+    use super::{VInt, *};
+    use crate::serialize::BinarySerializable;
     pub fn fixed_size_test<O: BinarySerializable + FixedSize + Default>() {
         let mut buffer = Vec::new();
         O::default().serialize(&mut buffer).unwrap();
@@ -164,7 +241,7 @@ pub mod test {
         fixed_size_test::<u32>();
         assert_eq!(4, serialize_test(3u32));
         assert_eq!(4, serialize_test(5u32));
-        assert_eq!(4, serialize_test(u32::max_value()));
+        assert_eq!(4, serialize_test(u32::MAX));
     }
 
     #[test]
@@ -172,19 +249,26 @@ pub mod test {
         fixed_size_test::<i64>();
     }
 
+    #[test]
+    fn test_serialize_f64() {
+        fixed_size_test::<f64>();
+    }
+
     #[test]
     fn test_serialize_u64() {
         fixed_size_test::<u64>();
     }
 
+    #[test]
+    fn test_serialize_bool() {
+        fixed_size_test::<bool>();
+    }
+
     #[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!(
-            serialize_test(String::from("富士さん見える。")),
-            1 + 3 * 8
-        );
+        assert_eq!(serialize_test(String::from("富士さん見える。")), 1 + 3 * 8);
     }
 
     #[test]
@@ -205,6 +289,6 @@ pub mod test {
         assert_eq!(serialize_test(VInt(1234u64)), 2);
         assert_eq!(serialize_test(VInt(16_383u64)), 2);
         assert_eq!(serialize_test(VInt(16_384u64)), 3);
-        assert_eq!(serialize_test(VInt(u64::max_value())), 10);
+        assert_eq!(serialize_test(VInt(u64::MAX)), 10);
     }
 }

common/src/vint.rs

@@ -0,0 +1,323 @@
+use std::io;
+use std::io::{Read, Write};
+
+use byteorder::{ByteOrder, LittleEndian};
+
+use super::BinarySerializable;
+
+/// Variable int serializes a u128 number
+pub fn serialize_vint_u128(mut val: u128, output: &mut Vec<u8>) {
+    loop {
+        let next_byte: u8 = (val % 128u128) as u8;
+        val /= 128u128;
+        if val == 0 {
+            output.push(next_byte | STOP_BIT);
+            return;
+        } else {
+            output.push(next_byte);
+        }
+    }
+}
+
+/// Deserializes a u128 number
+///
+/// Returns the number and the slice after the vint
+pub fn deserialize_vint_u128(data: &[u8]) -> io::Result<(u128, &[u8])> {
+    let mut result = 0u128;
+    let mut shift = 0u64;
+    for i in 0..19 {
+        let b = data[i];
+        result |= u128::from(b % 128u8) << shift;
+        if b >= STOP_BIT {
+            return Ok((result, &data[i + 1..]));
+        }
+        shift += 7;
+    }
+    Err(io::Error::new(
+        io::ErrorKind::InvalidData,
+        "Failed to deserialize u128 vint",
+    ))
+}
+
+///   Wrapper over a `u128` that serializes as a variable int.
+#[derive(Clone, Copy, Debug, Eq, PartialEq)]
+pub struct VIntU128(pub u128);
+
+impl BinarySerializable for VIntU128 {
+    fn serialize<W: Write>(&self, writer: &mut W) -> io::Result<()> {
+        let mut buffer = vec![];
+        serialize_vint_u128(self.0, &mut buffer);
+        writer.write_all(&buffer)
+    }
+
+    fn deserialize<R: Read>(reader: &mut R) -> io::Result<Self> {
+        let mut bytes = reader.bytes();
+        let mut result = 0u128;
+        let mut shift = 0u64;
+        loop {
+            match bytes.next() {
+                Some(Ok(b)) => {
+                    result |= u128::from(b % 128u8) << shift;
+                    if b >= STOP_BIT {
+                        return Ok(VIntU128(result));
+                    }
+                    shift += 7;
+                }
+                _ => {
+                    return Err(io::Error::new(
+                        io::ErrorKind::InvalidData,
+                        "Reach end of buffer while reading VInt",
+                    ));
+                }
+            }
+        }
+    }
+}
+
+///   Wrapper over a `u64` that serializes as a variable int.
+#[derive(Clone, Copy, Debug, Eq, PartialEq)]
+pub struct VInt(pub u64);
+
+const STOP_BIT: u8 = 128;
+
+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;
+    const START_5: u64 = 1 << 28;
+
+    const STOP_1: u64 = START_2 - 1;
+    const STOP_2: u64 = START_3 - 1;
+    const STOP_3: u64 = START_4 - 1;
+    const STOP_4: u64 = START_5 - 1;
+
+    const MASK_1: u64 = 127;
+    const MASK_2: u64 = MASK_1 << 7;
+    const MASK_3: u64 = MASK_2 << 7;
+    const MASK_4: u64 = MASK_3 << 7;
+    const MASK_5: u64 = MASK_4 << 7;
+
+    let val = u64::from(val);
+    const STOP_BIT: u64 = 128u64;
+    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)),
+            2,
+        ),
+        START_3..=STOP_3 => (
+            (val & MASK_1) | ((val & MASK_2) << 1) | ((val & MASK_3) << 2) | (STOP_BIT << (8 * 2)),
+            3,
+        ),
+        START_4..=STOP_4 => (
+            (val & MASK_1)
+                | ((val & MASK_2) << 1)
+                | ((val & MASK_3) << 2)
+                | ((val & MASK_4) << 3)
+                | (STOP_BIT << (8 * 3)),
+            4,
+        ),
+        _ => (
+            (val & MASK_1)
+                | ((val & MASK_2) << 1)
+                | ((val & MASK_3) << 2)
+                | ((val & MASK_4) << 3)
+                | ((val & MASK_5) << 4)
+                | (STOP_BIT << (8 * 4)),
+            5,
+        ),
+    };
+    LittleEndian::write_u64(&mut buf[..], res);
+    &buf[0..num_bytes]
+}
+
+/// Returns the number of bytes covered by a
+/// serialized vint `u32`.
+///
+/// Expects a buffer data that starts
+/// by the serialized `vint`, scans at most 5 bytes ahead until
+/// it finds the vint final byte.
+///
+/// # May Panic
+/// If the payload does not start by a valid `vint`
+fn vint_len(data: &[u8]) -> usize {
+    for (i, &val) in data.iter().enumerate().take(5) {
+        if val >= STOP_BIT {
+            return i + 1;
+        }
+    }
+    panic!("Corrupted data. Invalid VInt 32");
+}
+
+/// Reads a vint `u32` from a buffer, and
+/// consumes its payload data.
+///
+/// # Panics
+///
+/// If the buffer does not start by a valid
+/// vint payload
+pub fn read_u32_vint(data: &mut &[u8]) -> u32 {
+    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;
+    }
+    (result, vlen)
+}
+/// Write a `u32` as a vint payload.
+pub fn write_u32_vint<W: io::Write>(val: u32, writer: &mut W) -> io::Result<()> {
+    let mut buf = [0u8; 8];
+    let data = serialize_vint_u32(val, &mut buf);
+    writer.write_all(data)
+}
+
+impl VInt {
+    pub fn val(&self) -> u64 {
+        self.0
+    }
+
+    pub fn deserialize_u64<R: Read>(reader: &mut R) -> io::Result<u64> {
+        VInt::deserialize(reader).map(|vint| vint.0)
+    }
+
+    pub fn serialize_into_vec(&self, output: &mut Vec<u8>) {
+        let mut buffer = [0u8; 10];
+        let num_bytes = self.serialize_into(&mut buffer);
+        output.extend(&buffer[0..num_bytes]);
+    }
+
+    pub fn serialize_into(&self, buffer: &mut [u8; 10]) -> usize {
+        let mut remaining = self.0;
+        for (i, b) in buffer.iter_mut().enumerate() {
+            let next_byte: u8 = (remaining % 128u64) as u8;
+            remaining /= 128u64;
+            if remaining == 0u64 {
+                *b = next_byte | STOP_BIT;
+                return i + 1;
+            } else {
+                *b = next_byte;
+            }
+        }
+        unreachable!();
+    }
+}
+
+impl BinarySerializable for VInt {
+    fn serialize<W: Write>(&self, writer: &mut W) -> io::Result<()> {
+        let mut buffer = [0u8; 10];
+        let num_bytes = self.serialize_into(&mut buffer);
+        writer.write_all(&buffer[0..num_bytes])
+    }
+
+    fn deserialize<R: Read>(reader: &mut R) -> io::Result<Self> {
+        let mut bytes = reader.bytes();
+        let mut result = 0u64;
+        let mut shift = 0u64;
+        loop {
+            match bytes.next() {
+                Some(Ok(b)) => {
+                    result |= u64::from(b % 128u8) << shift;
+                    if b >= STOP_BIT {
+                        return Ok(VInt(result));
+                    }
+                    shift += 7;
+                }
+                _ => {
+                    return Err(io::Error::new(
+                        io::ErrorKind::InvalidData,
+                        "Reach end of buffer while reading VInt",
+                    ));
+                }
+            }
+        }
+    }
+}
+
+#[cfg(test)]
+mod tests {
+
+    use super::{serialize_vint_u32, BinarySerializable, VInt};
+    use crate::vint::{deserialize_vint_u128, serialize_vint_u128, VIntU128};
+
+    fn aux_test_vint(val: u64) {
+        let mut v = [14u8; 10];
+        let num_bytes = VInt(val).serialize_into(&mut v);
+        for el in &v[num_bytes..10] {
+            assert_eq!(el, &14u8);
+        }
+        assert!(num_bytes > 0);
+        if num_bytes < 10 {
+            assert!(1u64 << (7 * num_bytes) > val);
+        }
+        if num_bytes > 1 {
+            assert!(1u64 << (7 * (num_bytes - 1)) <= val);
+        }
+        let serdeser_val = VInt::deserialize(&mut &v[..]).unwrap();
+        assert_eq!(val, serdeser_val.0);
+    }
+
+    #[test]
+    fn test_vint() {
+        aux_test_vint(0);
+        aux_test_vint(1);
+        aux_test_vint(5);
+        aux_test_vint(u64::MAX);
+        for i in 1..9 {
+            let power_of_128 = 1u64 << (7 * i);
+            aux_test_vint(power_of_128 - 1u64);
+            aux_test_vint(power_of_128);
+            aux_test_vint(power_of_128 + 1u64);
+        }
+        aux_test_vint(10);
+    }
+
+    fn aux_test_serialize_vint_u32(val: u32) {
+        let mut buffer = [0u8; 10];
+        let mut buffer2 = [0u8; 8];
+        let len_vint = VInt(val as u64).serialize_into(&mut buffer);
+        let res2 = serialize_vint_u32(val, &mut buffer2);
+        assert_eq!(&buffer[..len_vint], res2, "array wrong for {}", val);
+    }
+
+    fn aux_test_vint_u128(val: u128) {
+        let mut data = vec![];
+        serialize_vint_u128(val, &mut data);
+        let (deser_val, _data) = deserialize_vint_u128(&data).unwrap();
+        assert_eq!(val, deser_val);
+
+        let mut out = vec![];
+        VIntU128(val).serialize(&mut out).unwrap();
+        let deser_val = VIntU128::deserialize(&mut &out[..]).unwrap();
+        assert_eq!(val, deser_val.0);
+    }
+
+    #[test]
+    fn test_vint_u128() {
+        aux_test_vint_u128(0);
+        aux_test_vint_u128(1);
+        aux_test_vint_u128(u128::MAX / 3);
+        aux_test_vint_u128(u128::MAX);
+    }
+
+    #[test]
+    fn test_vint_u32() {
+        aux_test_serialize_vint_u32(0);
+        aux_test_serialize_vint_u32(1);
+        aux_test_serialize_vint_u32(5);
+        for i in 1..3 {
+            let power_of_128 = 1u32 << (7 * i);
+            aux_test_serialize_vint_u32(power_of_128 - 1u32);
+            aux_test_serialize_vint_u32(power_of_128);
+            aux_test_serialize_vint_u32(power_of_128 + 1u32);
+        }
+        aux_test_serialize_vint_u32(u32::MAX);
+    }
+}

common/src/writer.rs

@@ -0,0 +1,114 @@
+use std::io::{self, BufWriter, Write};
+
+pub struct CountingWriter<W> {
+    underlying: W,
+    written_bytes: u64,
+}
+
+impl<W: Write> CountingWriter<W> {
+    pub fn wrap(underlying: W) -> CountingWriter<W> {
+        CountingWriter {
+            underlying,
+            written_bytes: 0,
+        }
+    }
+
+    #[inline]
+    pub fn written_bytes(&self) -> u64 {
+        self.written_bytes
+    }
+
+    /// Returns the underlying write object.
+    /// Note that this method does not trigger any flushing.
+    #[inline]
+    pub fn finish(self) -> W {
+        self.underlying
+    }
+}
+
+impl<W: Write> Write for CountingWriter<W> {
+    #[inline]
+    fn write(&mut self, buf: &[u8]) -> io::Result<usize> {
+        let written_size = self.underlying.write(buf)?;
+        self.written_bytes += written_size as u64;
+        Ok(written_size)
+    }
+
+    #[inline]
+    fn write_all(&mut self, buf: &[u8]) -> io::Result<()> {
+        self.underlying.write_all(buf)?;
+        self.written_bytes += buf.len() as u64;
+        Ok(())
+    }
+
+    #[inline]
+    fn flush(&mut self) -> io::Result<()> {
+        self.underlying.flush()
+    }
+}
+
+impl<W: TerminatingWrite> TerminatingWrite for CountingWriter<W> {
+    #[inline]
+    fn terminate_ref(&mut self, token: AntiCallToken) -> io::Result<()> {
+        self.underlying.terminate_ref(token)
+    }
+}
+
+/// Struct used to prevent from calling
+/// [`terminate_ref`](TerminatingWrite::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 + Send + Sync {
+    /// 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)
+    }
+}
+
+impl<'a> TerminatingWrite for &'a mut Vec<u8> {
+    fn terminate_ref(&mut self, _a: AntiCallToken) -> io::Result<()> {
+        self.flush()
+    }
+}
+
+#[cfg(test)]
+mod test {
+
+    use std::io::Write;
+
+    use super::CountingWriter;
+
+    #[test]
+    fn test_counting_writer() {
+        let buffer: Vec<u8> = vec![];
+        let mut counting_writer = CountingWriter::wrap(buffer);
+        let bytes = (0u8..10u8).collect::<Vec<u8>>();
+        counting_writer.write_all(&bytes).unwrap();
+        let len = counting_writer.written_bytes();
+        let buffer_restituted: Vec<u8> = counting_writer.finish();
+        assert_eq!(len, 10u64);
+        assert_eq!(buffer_restituted.len(), 10);
+    }
+}

doc/.gitignore

@@ -0,0 +1 @@
+book

doc/assets/images/element.io.svg

@@ -0,0 +1,8 @@
+<svg width="518" height="112" viewBox="0 0 518 112" fill="none" xmlns="http://www.w3.org/2000/svg">
+<path fill-rule="evenodd" clip-rule="evenodd" d="M56 112C86.9279 112 112 86.9279 112 56C112 25.0721 86.9279 0 56 0C25.0721 0 0 25.0721 0 56C0 86.9279 25.0721 112 56 112Z" fill="#0DBD8B"/>
+<path fill-rule="evenodd" clip-rule="evenodd" d="M45.7615 26.093C45.7615 23.8325 47.5977 22.0001 49.8629 22.0001C65.2154 22.0001 77.6611 34.4199 77.6611 49.7406C77.6611 52.001 75.8248 53.8335 73.5597 53.8335C71.2945 53.8335 69.4583 52.001 69.4583 49.7406C69.4583 38.9408 60.6851 30.1859 49.8629 30.1859C47.5977 30.1859 45.7615 28.3534 45.7615 26.093Z" fill="white"/>
+<path fill-rule="evenodd" clip-rule="evenodd" d="M85.8986 45.6477C88.1637 45.6477 89.9999 47.4801 89.9999 49.7406C89.9999 65.0612 77.5543 77.4811 62.2017 77.4811C59.9366 77.4811 58.1003 75.6486 58.1003 73.3882C58.1003 71.1277 59.9366 69.2953 62.2017 69.2953C73.024 69.2953 81.7972 60.5403 81.7972 49.7406C81.7972 47.4801 83.6334 45.6477 85.8986 45.6477Z" fill="white"/>
+<path fill-rule="evenodd" clip-rule="evenodd" d="M66.3031 85.907C66.3031 88.1675 64.4668 89.9999 62.2017 89.9999C46.8492 89.9999 34.4035 77.58 34.4035 62.2594C34.4035 59.9989 36.2398 58.1665 38.5049 58.1665C40.77 58.1665 42.6063 59.9989 42.6063 62.2594C42.6063 73.0592 51.3795 81.8141 62.2017 81.8141C64.4668 81.8141 66.3031 83.6466 66.3031 85.907Z" fill="white"/>
+<path fill-rule="evenodd" clip-rule="evenodd" d="M26.1014 66.3523C23.8363 66.3523 22.0001 64.5199 22.0001 62.2594C22 46.9388 34.4457 34.5189 49.7983 34.5189C52.0634 34.5189 53.8997 36.3514 53.8997 38.6118C53.8997 40.8723 52.0634 42.7047 49.7983 42.7047C38.976 42.7047 30.2028 51.4597 30.2028 62.2594C30.2028 64.5199 28.3666 66.3523 26.1014 66.3523Z" fill="white"/>
+<path d="M197 63.5H157.5C157.967 67.6333 159.467 70.9333 162 73.4C164.533 75.8 167.867 77 172 77C174.733 77 177.2 76.3333 179.4 75C181.6 73.6667 183.167 71.8667 184.1 69.6H196.1C194.5 74.8667 191.5 79.1333 187.1 82.4C182.767 85.6 177.633 87.2 171.7 87.2C163.967 87.2 157.7 84.6333 152.9 79.5C148.167 74.3667 145.8 67.8667 145.8 60C145.8 52.3333 148.2 45.9 153 40.7C157.8 35.5 164 32.9 171.6 32.9C179.2 32.9 185.333 35.4667 190 40.6C194.733 45.6667 197.1 52.0667 197.1 59.8L197 63.5ZM171.6 42.6C167.867 42.6 164.767 43.7 162.3 45.9C159.833 48.1 158.3 51.0333 157.7 54.7H185.3C184.767 51.0333 183.3 48.1 180.9 45.9C178.5 43.7 175.4 42.6 171.6 42.6ZM205.289 70.5V11H217.189V70.7C217.189 73.3667 218.656 74.7 221.589 74.7L223.689 74.6V85.9C222.556 86.1 221.356 86.2 220.089 86.2C214.956 86.2 211.189 84.9 208.789 82.3C206.456 79.7 205.289 75.7667 205.289 70.5ZM279.109 63.5H239.609C240.076 67.6333 241.576 70.9333 244.109 73.4C246.643 75.8 249.976 77 254.109 77C256.843 77 259.309 76.3333 261.509 75C263.709 73.6667 265.276 71.8667 266.209 69.6H278.209C276.609 74.8667 273.609 79.1333 269.209 82.4C264.876 85.6 259.743 87.2 253.809 87.2C246.076 87.2 239.809 84.6333 235.009 79.5C230.276 74.3667 227.909 67.8667 227.909 60C227.909 52.3333 230.309 45.9 235.109 40.7C239.909 35.5 246.109 32.9 253.709 32.9C261.309 32.9 267.443 35.4667 272.109 40.6C276.843 45.6667 279.209 52.0667 279.209 59.8L279.109 63.5ZM253.709 42.6C249.976 42.6 246.876 43.7 244.409 45.9C241.943 48.1 240.409 51.0333 239.809 54.7H267.409C266.876 51.0333 265.409 48.1 263.009 45.9C260.609 43.7 257.509 42.6 253.709 42.6ZM332.798 56.2V86H320.898V54.9C320.898 47.0333 317.632 43.1 311.098 43.1C307.565 43.1 304.732 44.2333 302.598 46.5C300.532 48.7667 299.498 51.8667 299.498 55.8V86H287.598V34.1H298.598V41C299.865 38.6667 301.798 36.7333 304.398 35.2C306.998 33.6667 310.232 32.9 314.098 32.9C321.298 32.9 326.498 35.6333 329.698 41.1C334.098 35.6333 339.965 32.9 347.298 32.9C353.365 32.9 358.032 34.8 361.298 38.6C364.565 42.3333 366.198 47.2667 366.198 53.4V86H354.298V54.9C354.298 47.0333 351.032 43.1 344.498 43.1C340.898 43.1 338.032 44.2667 335.898 46.6C333.832 48.8667 332.798 52.0667 332.798 56.2ZM425.379 63.5H385.879C386.346 67.6333 387.846 70.9333 390.379 73.4C392.912 75.8 396.246 77 400.379 77C403.112 77 405.579 76.3333 407.779 75C409.979 73.6667 411.546 71.8667 412.479 69.6H424.479C422.879 74.8667 419.879 79.1333 415.479 82.4C411.146 85.6 406.012 87.2 400.079 87.2C392.346 87.2 386.079 84.6333 381.279 79.5C376.546 74.3667 374.179 67.8667 374.179 60C374.179 52.3333 376.579 45.9 381.379 40.7C386.179 35.5 392.379 32.9 399.979 32.9C407.579 32.9 413.712 35.4667 418.379 40.6C423.112 45.6667 425.479 52.0667 425.479 59.8L425.379 63.5ZM399.979 42.6C396.246 42.6 393.146 43.7 390.679 45.9C388.212 48.1 386.679 51.0333 386.079 54.7H413.679C413.146 51.0333 411.679 48.1 409.279 45.9C406.879 43.7 403.779 42.6 399.979 42.6ZM444.868 34.1V41C446.068 38.7333 448.035 36.8333 450.768 35.3C453.568 33.7 456.935 32.9 460.868 32.9C467.001 32.9 471.735 34.7667 475.068 38.5C478.468 42.2333 480.168 47.2 480.168 53.4V86H468.268V54.9C468.268 51.2333 467.401 48.3667 465.668 46.3C464.001 44.1667 461.435 43.1 457.968 43.1C454.168 43.1 451.168 44.2333 448.968 46.5C446.835 48.7667 445.768 51.9 445.768 55.9V86H433.868V34.1H444.868ZM514.922 75.4V85.7C513.455 86.1 511.389 86.3 508.722 86.3C498.589 86.3 493.522 81.2 493.522 71V43.6H485.622V34.1H493.522V20.6H505.422V34.1H515.122V43.6H505.422V69.8C505.422 73.8667 507.355 75.9 511.222 75.9L514.922 75.4Z" fill="black"/>
+</svg>

doc/book.toml

@@ -0,0 +1,5 @@
+[book]
+authors = ["Paul Masurel"]
+multilingual = false
+src = "src"
+title = "Tantivy, the user guide"

doc/src/SUMMARY.md

@@ -0,0 +1,14 @@
+# Summary
+
+[Avant Propos](./avant-propos.md)
+
+- [Segments](./basis.md)
+- [Defining your schema](./schema.md)
+- [Facetting](./facetting.md)
+- [Index Sorting](./index_sorting.md)
+- [Innerworkings](./innerworkings.md)
+  - [Inverted index](./inverted_index.md)
+- [Best practise](./inverted_index.md)
+
+[Frequently Asked Questions](./faq.md)
+[Examples](./examples.md)

doc/src/avant-propos.md

@@ -0,0 +1,34 @@
+# Foreword, what is the scope of tantivy?
+
+> Tantivy is a **search** engine **library** for Rust.
+
+If you are familiar with Lucene, it's an excellent approximation to consider tantivy as Lucene for rust. tantivy is heavily inspired by Lucene's design and
+they both have the same scope and targeted use cases.
+
+If you are not familiar with Lucene, let's break down our little tagline.
+
+- **Search** here means full-text search : fundamentally, tantivy is here to help you
+identify efficiently what are the documents matching a given query in your corpus.
+But modern search UI are so much more : text processing, facetting, autocomplete, fuzzy search, good
+relevancy, collapsing, highlighting, spatial search.
+
+  While some of these features are not available in tantivy yet, all of these are relevant
+  feature requests. Tantivy's objective is to offer a solid toolbox to create the best search
+  experience. But keep in mind this is just a toolbox.
+  Which bring us to the second keyword...
+
+- **Library** means that you will have to write code. tantivy is not an *all-in-one* server solution like elastic search for instance.
+
+  Sometimes a functionality will not be available in tantivy because it is too
+  specific to your use case. By design, tantivy should make it possible to extend
+  the available set of features using the existing rock-solid datastructures.
+
+  Most frequently this will mean writing your own `Collector`, your own `Scorer` or your own
+  `TokenFilter`... Some of your requirements may also be related to
+  something closer to architecture or operations. For instance, you may
+  want to build a large corpus on Hadoop, fine-tune the merge policy to keep your
+  index sharded in a time-wise fashion, or you may want to convert and existing
+  index from a different format.
+
+  Tantivy exposes a lot of low level API to do all of these things.
+  

doc/src/basis.md

@@ -0,0 +1,68 @@
+# Anatomy of an index
+
+## Straight from disk
+
+Tantivy accesses its data using an abstracting trait called `Directory`.
+In theory, one can come and override the data access logic. In practise, the
+trait somewhat assumes that your data can be mapped to memory, and tantivy
+seems deeply married to using `mmap` for its io [^1], and the only persisting
+directory shipped with tantivy is the `MmapDirectory`.
+
+While this design has some downsides, this greatly simplifies the source code of
+tantivy. Caching is also entirely delegated to the OS.
+
+`tantivy` works entirely (or almost) by directly reading the datastructures as they are laid on disk. As a result, the act of opening an indexing does not involve loading different datastructures from the disk into random access memory : starting a process, opening an index, and performing your first query can typically be done in a matter of milliseconds.
+
+This is an interesting property for a command line search engine, or for some multi-tenant log search engine : spawning a new process for each new query can be a perfectly sensible solution in some use case.
+
+In later chapters, we will discuss tantivy's inverted index data layout.
+One key take away is that to achieve great performance, search indexes are extremely compact.
+Of course this is crucial to reduce IO, and ensure that as much of our index can sit in RAM.
+
+Also, whenever possible its data is accessed sequentially. Of course, this is an amazing property when tantivy needs to access the data from your spinning hard disk, but this is also
+critical for performance, if your data is read from and an `SSD` or even already in your pagecache.
+
+## Segments, and the log method
+
+That kind of compact layout comes at one cost: it prevents our datastructures from being dynamic.
+In fact, the `Directory` trait does not even allow you to modify part of a file.
+
+To allow the addition / deletion of documents, and create the illusion that
+your index is dynamic (i.e.: adding and deleting documents), tantivy uses a common database trick sometimes referred to as the *log method*.
+
+Let's forget about deletes for a moment.
+
+As you add documents, these documents are processed and stored in a dedicated datastructure, in a `RAM` buffer. This datastructure is not ready for search, but it is useful to receive your data and rearrange it very rapidly.
+
+As you add documents, this buffer will reach its capacity and tantivy will transparently stop adding document to it and start converting this datastructure to its final read-only format on disk. Once written, an brand empty buffer is available to resume adding documents.
+
+The resulting chunk of index obtained after this serialization is called a `Segment`.
+
+> A segment is a self-contained atomic piece of index. It is identified with a UUID, and all of its files are identified using the naming scheme : `<UUID>.*`.
+
+Which brings us to the nature of a tantivy `Index`.
+
+> A tantivy `Index` is a collection of `Segments`.
+
+Physically, this really just means and index is a bunch of segment files in a given `Directory`,
+linked together by a `meta.json` file. This transparency can become extremely handy
+to get tantivy to fit your use case:
+
+*Example 1* You could for instance use hadoop to build a very large search index in a timely manner, copy all of the resulting segment files in the same directory and edit the `meta.json` to get a functional index.[^2]
+
+*Example 2* You could also disable your merge policy and enforce daily segments. Removing data after one week can then be done very efficiently by just editing the `meta.json` and deleting the files associated with segment `D-7`.
+
+## Merging
+
+As you index more and more data, your index will accumulate more and more segments.
+Having a lot of small segments is not really optimal. There is a bit of redundancy in having
+all these term dictionary. Also when searching, we will need to do term lookups as many times as we have segments.  It can hurt search performance a bit.
+
+That's where merging or compacting comes into place. Tantivy will continuously consider merge
+opportunities and start merging segments in the background.
+
+## Indexing throughput, number of indexing threads
+
+[^1]: This may eventually change.
+
+[^2]: Be careful however. By default these files will not be considered as *managed* by tantivy. This means they will never be garbage collected by tantivy, regardless of whether they become obsolete or not.

doc/src/examples.md

@@ -0,0 +1,3 @@
+# Examples
+
+- [Basic search](/examples/basic_search.html)

doc/src/facetting.md

@@ -0,0 +1,5 @@
+# Facetting
+
+wewew
+
+## weeewe

doc/src/index_sorting.md

@@ -0,0 +1,62 @@
+
+- [Index Sorting](#index-sorting)
+  - [Why Sorting](#why-sorting)
+    - [Compression](#compression)
+    - [Top-N Optimization](#top-n-optimization)
+    - [Pruning](#pruning)
+    - [Other](#other)
+  - [Usage](#usage)
+
+# Index Sorting
+
+Tantivy allows you to sort the index according to a property.
+
+## Why Sorting
+
+Presorting an index has several advantages:
+
+### Compression
+
+When data is sorted it is easier to compress the data. E.g. the numbers sequence [5, 2, 3, 1, 4] would be sorted to [1, 2, 3, 4, 5].
+If we apply delta encoding this list would be unsorted [5, -3, 1, -2, 3] vs. [1, 1, 1, 1, 1].
+Compression ratio is mainly affected on the fast field of the sorted property, every thing else is likely unaffected.
+
+### Top-N Optimization
+
+When data is presorted by a field and search queries request sorting by the same field, we can leverage the natural order of the documents.
+E.g. if the data is sorted by timestamp and want the top n newest docs containing a term, we can simply leveraging the order of the docids.
+
+Note: Tantivy 0.16 does not do this optimization yet.
+
+### Pruning
+
+Let's say we want all documents and want to apply the filter `>= 2010-08-11`. When the data is sorted, we could make a lookup in the fast field to find the docid range and use this as the filter.
+
+Note: Tantivy 0.16 does not do this optimization yet.
+
+### Other?
+
+In principle there are many algorithms possible that exploit the monotonically increasing nature. (aggregations maybe?)
+
+## Usage
+
+The index sorting can be configured setting [`sort_by_field`](https://github.com/quickwit-oss/tantivy/blob/000d76b11a139a84b16b9b95060a1c93e8b9851c/src/core/index_meta.rs#L238) on `IndexSettings` and passing it to a `IndexBuilder`. As of Tantivy 0.16 only fast fields are allowed to be used.
+
+```rust
+let settings = IndexSettings {
+    sort_by_field: Some(IndexSortByField {
+        field: "intval".to_string(),
+        order: Order::Desc,
+    }),
+    ..Default::default()
+};
+let mut index_builder = Index::builder().schema(schema);
+index_builder = index_builder.settings(settings);
+let index = index_builder.create_in_ram().unwrap();
+```
+
+## Implementation details
+
+Sorting an index is applied in the serialization step. In general there are two serialization steps: [Finishing a single segment](https://github.com/quickwit-oss/tantivy/blob/000d76b11a139a84b16b9b95060a1c93e8b9851c/src/indexer/segment_writer.rs#L338) and [merging multiple segments](https://github.com/quickwit-oss/tantivy/blob/000d76b11a139a84b16b9b95060a1c93e8b9851c/src/indexer/merger.rs#L1073).
+
+In both cases we generate a docid mapping reflecting the sort. This mapping is used when serializing the different components (doc store, fastfields, posting list, normfield, facets).

doc/src/innerworkings.md

@@ -0,0 +1 @@
+# Innerworkings

doc/src/inverted_index.md

@@ -0,0 +1 @@
+# Inverted index

doc/src/json.md

@@ -0,0 +1,130 @@
+# Json
+
+As of tantivy 0.17, tantivy supports a json object type.
+This type can be used to allow for a schema-less search index.
+
+When indexing a json object, we "flatten" the JSON. This operation emits terms that represent a triplet `(json_path, value_type, value)`
+
+For instance,  if user is a json field, the following document:
+
+```json
+{
+    "user": {
+        "name": "Paul Masurel",
+        "address": {
+            "city": "Tokyo",
+            "country": "Japan"
+        },
+        "created_at": "2018-11-12T23:20:50.52Z"
+    }
+}
+```
+
+emits the following tokens:
+
+- ("name", Text, "Paul")
+- ("name", Text, "Masurel")
+- ("address.city", Text, "Tokyo")
+- ("address.country", Text, "Japan")
+- ("created_at", Date, 15420648505)
+
+## Bytes-encoding and lexicographical sort
+
+Like any other terms, these triplets are encoded into a binary format as follows.
+
+- `json_path`: the json path is a sequence of "segments". In the example above, `address.city`
+is just a debug representation of the json path `["address", "city"]`.
+Its representation is done by separating segments by a unicode char `\x01`, and ending the path by `\x00`.
+- `value type`: One byte represents the `Value` type.
+- `value`: The value representation is just the regular Value representation.
+
+This representation is designed to align the natural sort of Terms with the lexicographical sort
+of their binary representation (Tantivy's dictionary (whether fst or sstable) is sorted and does prefix encoding).
+
+In the example above, the terms will be sorted as
+
+- ("address.city", Text, "Tokyo")
+- ("address.country", Text, "Japan")
+- ("name", Text, "Masurel")
+- ("name", Text, "Paul")
+- ("created_at", Date, 15420648505)
+
+As seen in "pitfalls", we may end up having to search for a value for a same path in several different fields. Putting the field code after the path makes it maximizes compression opportunities but also increases the chances for the two terms to end up in the actual same term dictionary block.
+
+## Pitfalls, limitation and corner cases
+
+Json gives very little information about the type of the literals it stores.
+All numeric types end up mapped as a "Number" and there are no types for dates.
+
+At indexing, tantivy will try to interpret number and strings as different type with a
+priority order.
+
+Numbers will be interpreted as u64, i64 and f64 in that order.
+Strings will be interpreted as rfc3999 dates or simple strings.
+
+The first working type is picked and is the only term that is emitted for indexing.
+Note this interpretation happens on a per-document basis, and there is no effort to try to sniff
+a consistent field type at the scale of a segment.
+
+On the query parser side on the other hand, we may end up emitting more than one type.
+For instance, we do not even know if the type is a number or string based.
+
+So the query
+
+```rust
+my_path.my_segment:233
+```
+
+Will be interpreted as
+
+```rust
+(my_path.my_segment, String, 233) or (my_path.my_segment, u64, 233)
+```
+
+Likewise, we need to emit two tokens if the query contains an rfc3999 date.
+Indeed the date could have been actually a single token inside the text of a document at ingestion time. Generally speaking, we will always at least emit a string token in query parsing, and sometimes more.
+
+If one more json field is defined, things get even more complicated.
+
+## Default json field
+
+If the schema contains a text field called "text" and a json field that is set as a default field:
+`text:hello` could be reasonably interpreted as targeting the text field or as targeting the json field called `json_dynamic` with the json_path "text".
+
+If there is such an ambiguity, we decide to only search in the "text" field: `text:hello`.
+
+In other words, the parser will not search in default json fields if there is a schema hit.
+This is a product decision.
+
+The user can still target the JSON field by specifying its name explicitly:
+`json_dynamic.text:hello`.
+
+## Range queries are not supported
+
+Json field do not support range queries.
+
+## Arrays do not work like nested object
+
+If json object contains an array, a search query might return more documents
+than what might be expected.
+
+Let's take an example.
+
+```json
+{
+    "cart_id": 3234234 ,
+    "cart": [
+        {"product_type": "sneakers", "attributes": {"color": "white"} },
+        {"product_type": "t-shirt", "attributes": {"color": "red"}},
+    ]
+}
+```
+
+Despite the array structure, a document in tantivy is a bag of terms.
+The query:
+
+```rust
+cart.product_type:sneakers AND cart.attributes.color:red
+```
+
+Actually match the document above.

doc/src/schema.md

@@ -0,0 +1 @@
+# Defining your schema

examples/aggregation.rs

@@ -0,0 +1,130 @@
+// # Aggregation example
+//
+// This example shows how you can use built-in aggregations.
+// We will use range buckets and compute the average in each bucket.
+//
+
+use serde_json::Value;
+use tantivy::aggregation::agg_req::{
+    Aggregation, Aggregations, BucketAggregation, BucketAggregationType, MetricAggregation,
+    RangeAggregation,
+};
+use tantivy::aggregation::agg_result::AggregationResults;
+use tantivy::aggregation::metric::AverageAggregation;
+use tantivy::aggregation::AggregationCollector;
+use tantivy::query::TermQuery;
+use tantivy::schema::{self, Cardinality, IndexRecordOption, Schema, TextFieldIndexing};
+use tantivy::{doc, Index, Term};
+
+fn main() -> tantivy::Result<()> {
+    let mut schema_builder = Schema::builder();
+    let text_fieldtype = schema::TextOptions::default()
+        .set_indexing_options(
+            TextFieldIndexing::default().set_index_option(IndexRecordOption::WithFreqs),
+        )
+        .set_stored();
+    let text_field = schema_builder.add_text_field("text", text_fieldtype);
+    let score_fieldtype =
+        crate::schema::NumericOptions::default().set_fast(Cardinality::SingleValue);
+    let highscore_field = schema_builder.add_f64_field("highscore", score_fieldtype.clone());
+    let price_field = schema_builder.add_f64_field("price", score_fieldtype.clone());
+
+    let schema = schema_builder.build();
+
+    // # Indexing documents
+    //
+    // Lets index a bunch of documents for this example.
+    let index = Index::create_in_ram(schema);
+
+    let mut index_writer = index.writer(50_000_000)?;
+    // writing the segment
+    index_writer.add_document(doc!(
+        text_field => "cool",
+        highscore_field => 1f64,
+        price_field => 0f64,
+    ))?;
+    index_writer.add_document(doc!(
+        text_field => "cool",
+        highscore_field => 3f64,
+        price_field => 1f64,
+    ))?;
+    index_writer.add_document(doc!(
+        text_field => "cool",
+        highscore_field => 5f64,
+        price_field => 1f64,
+    ))?;
+    index_writer.add_document(doc!(
+        text_field => "nohit",
+        highscore_field => 6f64,
+        price_field => 2f64,
+    ))?;
+    index_writer.add_document(doc!(
+        text_field => "cool",
+        highscore_field => 7f64,
+        price_field => 2f64,
+    ))?;
+    index_writer.commit()?;
+    index_writer.add_document(doc!(
+        text_field => "cool",
+        highscore_field => 11f64,
+        price_field => 10f64,
+    ))?;
+    index_writer.add_document(doc!(
+        text_field => "cool",
+        highscore_field => 14f64,
+        price_field => 15f64,
+    ))?;
+
+    index_writer.add_document(doc!(
+        text_field => "cool",
+        highscore_field => 15f64,
+        price_field => 20f64,
+    ))?;
+
+    index_writer.commit()?;
+
+    let reader = index.reader()?;
+    let text_field = reader.searcher().schema().get_field("text").unwrap();
+
+    let term_query = TermQuery::new(
+        Term::from_field_text(text_field, "cool"),
+        IndexRecordOption::Basic,
+    );
+
+    let sub_agg_req_1: Aggregations = vec![(
+        "average_price".to_string(),
+        Aggregation::Metric(MetricAggregation::Average(
+            AverageAggregation::from_field_name("price".to_string()),
+        )),
+    )]
+    .into_iter()
+    .collect();
+
+    let agg_req_1: Aggregations = vec![(
+        "score_ranges".to_string(),
+        Aggregation::Bucket(BucketAggregation {
+            bucket_agg: BucketAggregationType::Range(RangeAggregation {
+                field: "highscore".to_string(),
+                ranges: vec![
+                    (-1f64..9f64).into(),
+                    (9f64..14f64).into(),
+                    (14f64..20f64).into(),
+                ],
+                ..Default::default()
+            }),
+            sub_aggregation: sub_agg_req_1.clone(),
+        }),
+    )]
+    .into_iter()
+    .collect();
+
+    let collector = AggregationCollector::from_aggs(agg_req_1, None);
+
+    let searcher = reader.searcher();
+    let agg_res: AggregationResults = searcher.search(&term_query, &collector).unwrap();
+
+    let res: Value = serde_json::to_value(&agg_res)?;
+    println!("{}", serde_json::to_string_pretty(&res)?);
+
+    Ok(())
+}

examples/basic_search.rs

@@ -0,0 +1,225 @@
+// # Basic Example
+//
+// This example covers the basic functionalities of
+// tantivy.
+//
+// We will :
+// - define our schema
+// - create an index in a directory
+// - index a few documents into our index
+// - search for the best document matching a basic query
+// - retrieve the best document's original content.
+
+// ---
+// Importing tantivy...
+use tantivy::collector::TopDocs;
+use tantivy::query::QueryParser;
+use tantivy::schema::*;
+use tantivy::{doc, Index, ReloadPolicy};
+use tempfile::TempDir;
+
+fn main() -> tantivy::Result<()> {
+    // Let's create a temporary directory for the
+    // sake of this example
+    let index_path = TempDir::new()?;
+
+    // # Defining the schema
+    //
+    // The Tantivy index requires a very strict schema.
+    // The schema declares which fields are in the index,
+    // and for each field, its type and "the way it should
+    // be indexed".
+
+    // First we need to define a schema ...
+    let mut schema_builder = Schema::builder();
+
+    // Our first field is title.
+    // We want full-text search for it, and we also want
+    // to be able to retrieve the document after the search.
+    //
+    // `TEXT | STORED` is some syntactic sugar to describe
+    // that.
+    //
+    // `TEXT` means the field should be tokenized and indexed,
+    // along with its term frequency and term positions.
+    //
+    // `STORED` means that the field will also be saved
+    // in a compressed, row-oriented key-value store.
+    // This store is useful for reconstructing the
+    // documents that were selected during the search phase.
+    schema_builder.add_text_field("title", TEXT | STORED);
+
+    // Our second field is body.
+    // We want full-text search for it, but we do not
+    // need to be able to be able to retrieve it
+    // for our application.
+    //
+    // We can make our index lighter by omitting the `STORED` flag.
+    schema_builder.add_text_field("body", TEXT);
+
+    let schema = schema_builder.build();
+
+    // # Indexing documents
+    //
+    // Let's create a brand new index.
+    //
+    // This will actually just save a meta.json
+    // with our schema in the directory.
+    let index = Index::create_in_dir(&index_path, schema.clone())?;
+
+    // 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.
+    //
+    // Here we give tantivy a budget of `50MB`.
+    // Using a bigger memory_arena for the indexer may increase
+    // throughput, but 50 MB is already plenty.
+    let mut index_writer = index.writer(50_000_000)?;
+
+    // Let's index our documents!
+    // We first need a handle on the title and the body field.
+
+    // ### Adding documents
+    //
+    // 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 mut old_man_doc = Document::default();
+    old_man_doc.add_text(title, "The Old Man and the Sea");
+    old_man_doc.add_text(
+        body,
+        "He was an old man who fished alone in a skiff in the Gulf Stream and he had gone \
+         eighty-four days now without taking a fish.",
+    );
+
+    // ... and add it to the `IndexWriter`.
+    index_writer.add_document(old_man_doc)?;
+
+    // For convenience, tantivy also comes with a macro to
+    // reduce the boilerplate above.
+    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",
+    title => "The Modern Prometheus",
+    body => "You will rejoice to hear that no disaster has accompanied the commencement of an \
+             enterprise which you have regarded with such evil forebodings.  I arrived here \
+             yesterday, and my first task is to assure my dear sister of my welfare and \
+             increasing confidence in the success of my undertaking."
+    ))?;
+
+    // This is an example, so we will only index 3 documents
+    // here. You can check out tantivy's tutorial to index
+    // the English wikipedia. Tantivy's indexing is rather fast.
+    // Indexing 5 million articles of the English wikipedia takes
+    // around 3 minutes on my computer!
+
+    // ### Committing
+    //
+    // At this point our documents are not searchable.
+    //
+    //
+    // We need to call `.commit()` explicitly to force the
+    // `index_writer` to finish processing the documents in the queue,
+    // flush the current index to the disk, and advertise
+    // the existence of new documents.
+    //
+    // This call is blocking.
+    index_writer.commit()?;
+
+    // If `.commit()` returns correctly, then all of the
+    // documents that have been added are guaranteed to be
+    // persistently indexed.
+    //
+    // In the scenario of a crash or a power failure,
+    // tantivy behaves as if it has rolled back to its last
+    // commit.
+
+    // # Searching
+    //
+    // ### Searcher
+    //
+    // A reader is required first in order to search an index.
+    // It acts as a `Searcher` pool that reloads itself,
+    // depending on a `ReloadPolicy`.
+    //
+    // For a search server you will typically create one reader for the entire lifetime of your
+    // program, and acquire a new searcher for every single request.
+    //
+    // In the code below, we rely on the 'ON_COMMIT' policy: the reader
+    // will reload the index automatically after each commit.
+    let reader = index
+        .reader_builder()
+        .reload_policy(ReloadPolicy::OnCommit)
+        .try_into()?;
+
+    // We now need to acquire a searcher.
+    //
+    // 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
+    // same version of the index.
+    //
+    // Acquiring a `searcher` is very cheap.
+    //
+    // You should acquire a searcher every time you start processing a request and
+    // and release it right after your query is finished.
+    let searcher = reader.searcher();
+
+    // ### Query
+
+    // The query parser can interpret human queries.
+    // Here, if the user does not specify which
+    // field they want to search, tantivy will search
+    // 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
+    // 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")?;
+
+    // A query defines a set of documents, as
+    // well as the way they should be scored.
+    //
+    // A query created by the query parser is scored according
+    // to a metric called Tf-Idf, and will consider
+    // any document matching at least one of our terms.
+
+    // ### Collectors
+    //
+    // 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` collector.
+
+    // We can now perform our query.
+    let top_docs = searcher.search(&query, &TopDocs::with_limit(10))?;
+
+    // The actual documents still need to be
+    // retrieved from Tantivy's store.
+    //
+    // Since the body field was not configured as stored,
+    // the document returned will only contain
+    // a title.
+    for (_score, doc_address) in top_docs {
+        let retrieved_doc = searcher.doc(doc_address)?;
+        println!("{}", schema.to_json(&retrieved_doc));
+    }
+
+    Ok(())
+}

examples/custom_collector.rs

@@ -0,0 +1,181 @@
+// # 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.
+
+use std::sync::Arc;
+
+use fastfield_codecs::Column;
+// ---
+// Importing tantivy...
+use tantivy::collector::{Collector, SegmentCollector};
+use tantivy::query::QueryParser;
+use tantivy::schema::{Field, Schema, FAST, INDEXED, TEXT};
+use tantivy::{doc, Index, Score, SegmentReader};
+
+#[derive(Default)]
+struct Stats {
+    count: usize,
+    sum: f64,
+    squared_sum: f64,
+}
+
+impl Stats {
+    pub fn count(&self) -> usize {
+        self.count
+    }
+
+    pub fn mean(&self) -> f64 {
+        self.sum / (self.count as f64)
+    }
+
+    fn square_mean(&self) -> f64 {
+        self.squared_sum / (self.count as f64)
+    }
+
+    pub fn standard_deviation(&self) -> f64 {
+        let mean = self.mean();
+        (self.square_mean() - mean * mean).sqrt()
+    }
+
+    fn non_zero_count(self) -> Option<Stats> {
+        if self.count == 0 {
+            None
+        } else {
+            Some(self)
+        }
+    }
+}
+
+struct StatsCollector {
+    field: Field,
+}
+
+impl StatsCollector {
+    fn with_field(field: Field) -> StatsCollector {
+        StatsCollector { field }
+    }
+}
+
+impl Collector for StatsCollector {
+    // That's the type of our result.
+    // Our standard deviation will be a float.
+    type Fruit = Option<Stats>;
+
+    type Child = StatsSegmentCollector;
+
+    fn for_segment(
+        &self,
+        _segment_local_id: u32,
+        segment_reader: &SegmentReader,
+    ) -> tantivy::Result<StatsSegmentCollector> {
+        let fast_field_reader = segment_reader.fast_fields().u64(self.field)?;
+        Ok(StatsSegmentCollector {
+            fast_field_reader,
+            stats: Stats::default(),
+        })
+    }
+
+    fn requires_scoring(&self) -> bool {
+        // this collector does not care about score.
+        false
+    }
+
+    fn merge_fruits(&self, segment_stats: Vec<Option<Stats>>) -> tantivy::Result<Option<Stats>> {
+        let mut stats = Stats::default();
+        for segment_stats in segment_stats.into_iter().flatten() {
+            stats.count += segment_stats.count;
+            stats.sum += segment_stats.sum;
+            stats.squared_sum += segment_stats.squared_sum;
+        }
+        Ok(stats.non_zero_count())
+    }
+}
+
+struct StatsSegmentCollector {
+    fast_field_reader: Arc<dyn Column<u64>>,
+    stats: Stats,
+}
+
+impl SegmentCollector for StatsSegmentCollector {
+    type Fruit = Option<Stats>;
+
+    fn collect(&mut self, doc: u32, _score: Score) {
+        let value = self.fast_field_reader.get_val(doc as u64) as f64;
+        self.stats.count += 1;
+        self.stats.sum += value;
+        self.stats.squared_sum += value * value;
+    }
+
+    fn harvest(self) -> <Self as SegmentCollector>::Fruit {
+        self.stats.non_zero_count()
+    }
+}
+
+fn main() -> tantivy::Result<()> {
+    // # Defining the schema
+    //
+    // The Tantivy index requires a very strict schema.
+    // The schema declares which fields are in the index,
+    // and for each field, its type and "the way it should
+    // be indexed".
+
+    // first we need to define a schema ...
+    let mut schema_builder = Schema::builder();
+
+    // We'll assume a fictional index containing
+    // products, and with a name, a description, and a price.
+    let product_name = schema_builder.add_text_field("name", TEXT);
+    let product_description = schema_builder.add_text_field("description", TEXT);
+    let price = schema_builder.add_u64_field("price", INDEXED | FAST);
+    let schema = schema_builder.build();
+
+    // # Indexing documents
+    //
+    // Lets index a bunch of fake documents for the sake of
+    // this example.
+    let index = Index::create_in_ram(schema);
+
+    let mut index_writer = index.writer(50_000_000)?;
+    index_writer.add_document(doc!(
+        product_name => "Super Broom 2000",
+        product_description => "While it is ok for short distance travel, this broom \
+        was designed quiditch. It will up your game.",
+        price => 30_200u64
+    ))?;
+    index_writer.add_document(doc!(
+        product_name => "Turbulobroom",
+        product_description => "You might have heard of this broom before : it is the sponsor of the Wales team.\
+            You'll enjoy its sharp turns, and rapid acceleration",
+        price => 29_240u64
+    ))?;
+    index_writer.add_document(doc!(
+        product_name => "Broomio",
+        product_description => "Great value for the price. This broom is a market favorite",
+        price => 21_240u64
+    ))?;
+    index_writer.add_document(doc!(
+        product_name => "Whack a Mole",
+        product_description => "Prime quality bat.",
+        price => 5_200u64
+    ))?;
+    index_writer.commit()?;
+
+    let reader = index.reader()?;
+    let searcher = reader.searcher();
+    let query_parser = QueryParser::for_index(&index, vec![product_name, product_description]);
+
+    // here we want to get a hit on the 'ken' in Frankenstein
+    let query = query_parser.parse_query("broom")?;
+    if let Some(stats) = searcher.search(&query, &StatsCollector::with_field(price))? {
+        println!("count: {}", stats.count());
+        println!("mean: {}", stats.mean());
+        println!("standard deviation: {}", stats.standard_deviation());
+    }
+
+    Ok(())
+}

examples/custom_tokenizer.rs

@@ -0,0 +1,111 @@
+// # Defining a tokenizer pipeline
+//
+// In this example, we'll see how to define a tokenizer pipeline
+// by aligning a bunch of `TokenFilter`.
+use tantivy::collector::TopDocs;
+use tantivy::query::QueryParser;
+use tantivy::schema::*;
+use tantivy::tokenizer::NgramTokenizer;
+use tantivy::{doc, Index};
+
+fn main() -> tantivy::Result<()> {
+    // # Defining the schema
+    //
+    // The Tantivy index requires a very strict schema.
+    // The schema declares which fields are in the index,
+    // and for each field, its type and "the way it should
+    // be indexed".
+
+    // first we need to define a schema ...
+    let mut schema_builder = Schema::builder();
+
+    // Our first field is title.
+    // In this example we want to use NGram searching
+    // we will set that to 3 characters, so any three
+    // char in the title should be findable.
+    let text_field_indexing = TextFieldIndexing::default()
+        .set_tokenizer("ngram3")
+        .set_index_option(IndexRecordOption::WithFreqsAndPositions);
+    let text_options = TextOptions::default()
+        .set_indexing_options(text_field_indexing)
+        .set_stored();
+    let title = schema_builder.add_text_field("title", text_options);
+
+    // Our second field is body.
+    // We want full-text search for it, but we do not
+    // need to be able to be able to retrieve it
+    // for our application.
+    //
+    // We can make our index lighter by omitting the `STORED` flag.
+    let body = schema_builder.add_text_field("body", TEXT);
+
+    let schema = schema_builder.build();
+
+    // # Indexing documents
+    //
+    // Let's create a brand new index.
+    // To simplify we will work entirely in RAM.
+    // This is not what you want in reality, but it is very useful
+    // for your unit tests... Or this example.
+    let index = Index::create_in_ram(schema.clone());
+
+    // here we are registering our custom tokenizer
+    // this will store tokens of 3 characters each
+    index
+        .tokenizers()
+        .register("ngram3", NgramTokenizer::new(3, 3, false));
+
+    // To insert document we need an index writer.
+    // There must be only one writer at a time.
+    // This single `IndexWriter` is already
+    // multithreaded.
+    //
+    // Here we use a buffer of 50MB per thread. Using a bigger
+    // memory arena for the indexer can increase its throughput.
+    let mut index_writer = index.writer(50_000_000)?;
+    index_writer.add_document(doc!(
+    title => "The Old Man and the Sea",
+    body => "He was an old man who fished alone in a skiff in the Gulf Stream and \
+     he had gone eighty-four days now without taking a fish."
+    ))?;
+    index_writer.add_document(doc!(
+    title => "Of Mice and Men",
+       body => r#"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"#
+    ))?;
+    index_writer.add_document(doc!(
+    title => "Frankenstein",
+        body => r#"You will rejoice to hear that no disaster has accompanied the commencement of an
+                enterprise which you have regarded with such evil forebodings.  I arrived here
+                yesterday, and my first task is to assure my dear sister of my welfare and
+                increasing confidence in the success of my undertaking."#
+    ))?;
+    index_writer.commit()?;
+
+    let reader = index.reader()?;
+    let searcher = reader.searcher();
+
+    // The query parser can interpret human queries.
+    // Here, if the user does not specify which
+    // field they want to search, tantivy will search
+    // in both title and body.
+    let query_parser = QueryParser::for_index(&index, vec![title, body]);
+
+    // here we want to get a hit on the 'ken' in Frankenstein
+    let query = query_parser.parse_query("ken")?;
+
+    let top_docs = searcher.search(&query, &TopDocs::with_limit(10))?;
+
+    for (_, doc_address) in top_docs {
+        let retrieved_doc = searcher.doc(doc_address)?;
+        println!("{}", schema.to_json(&retrieved_doc));
+    }
+
+    Ok(())
+}

examples/date_time_field.rs

@@ -0,0 +1,69 @@
+// # DateTime field example
+//
+// This example shows how the DateTime field can be used
+
+use tantivy::collector::TopDocs;
+use tantivy::query::QueryParser;
+use tantivy::schema::{Cardinality, DateOptions, Schema, Value, INDEXED, STORED, STRING};
+use tantivy::Index;
+
+fn main() -> tantivy::Result<()> {
+    // # Defining the schema
+    let mut schema_builder = Schema::builder();
+    let opts = DateOptions::from(INDEXED)
+        .set_stored()
+        .set_fast(Cardinality::SingleValue)
+        .set_precision(tantivy::DatePrecision::Seconds);
+    let occurred_at = schema_builder.add_date_field("occurred_at", opts);
+    let event_type = schema_builder.add_text_field("event", STRING | STORED);
+    let schema = schema_builder.build();
+
+    // # Indexing documents
+    let index = Index::create_in_ram(schema.clone());
+
+    let mut index_writer = index.writer(50_000_000)?;
+    let doc = schema.parse_document(
+        r#"{
+        "occurred_at": "2022-06-22T12:53:50.53Z",
+        "event": "pull-request"
+    }"#,
+    )?;
+    index_writer.add_document(doc)?;
+    let doc = schema.parse_document(
+        r#"{
+        "occurred_at": "2022-06-22T13:00:00.22Z",
+        "event": "comment"
+    }"#,
+    )?;
+    index_writer.add_document(doc)?;
+    index_writer.commit()?;
+
+    let reader = index.reader()?;
+    let searcher = reader.searcher();
+
+    // # Default fields: event_type
+    let query_parser = QueryParser::for_index(&index, vec![event_type]);
+    {
+        let query = query_parser.parse_query("event:comment")?;
+        let count_docs = searcher.search(&*query, &TopDocs::with_limit(5))?;
+        assert_eq!(count_docs.len(), 1);
+    }
+    {
+        let query = query_parser
+            .parse_query(r#"occurred_at:[2022-06-22T12:58:00Z TO 2022-06-23T00:00:00Z}"#)?;
+        let count_docs = searcher.search(&*query, &TopDocs::with_limit(4))?;
+        assert_eq!(count_docs.len(), 1);
+        for (_score, doc_address) in count_docs {
+            let retrieved_doc = searcher.doc(doc_address)?;
+            assert!(matches!(
+                retrieved_doc.get_first(occurred_at),
+                Some(Value::Date(_))
+            ));
+            assert_eq!(
+                schema.to_json(&retrieved_doc),
+                r#"{"event":["comment"],"occurred_at":["2022-06-22T13:00:00.22Z"]}"#
+            );
+        }
+    }
+    Ok(())
+}

examples/deleting_updating_documents.rs

@@ -0,0 +1,144 @@
+// # Deleting and Updating (?) documents
+//
+// This example explains how to delete and update documents.
+// In fact there is actually no such thing as an update in tantivy.
+//
+// To update a document, you need to delete a document and then reinsert
+// its new version.
+//
+// ---
+// Importing tantivy...
+use tantivy::collector::TopDocs;
+use tantivy::query::TermQuery;
+use tantivy::schema::*;
+use tantivy::{doc, Index, IndexReader};
+
+// A simple helper function to fetch a single document
+// given its id from our index.
+// It will be helpful to check our work.
+fn extract_doc_given_isbn(
+    reader: &IndexReader,
+    isbn_term: &Term,
+) -> tantivy::Result<Option<Document>> {
+    let searcher = reader.searcher();
+
+    // This is the simplest query you can think of.
+    // It matches all of the documents containing a specific term.
+    //
+    // The second argument is here to tell we don't care about decoding positions,
+    // or term frequencies.
+    let term_query = TermQuery::new(isbn_term.clone(), IndexRecordOption::Basic);
+    let top_docs = searcher.search(&term_query, &TopDocs::with_limit(1))?;
+
+    if let Some((_score, doc_address)) = top_docs.first() {
+        let doc = searcher.doc(*doc_address)?;
+        Ok(Some(doc))
+    } else {
+        // no doc matching this ID.
+        Ok(None)
+    }
+}
+
+fn main() -> tantivy::Result<()> {
+    // # Defining the schema
+    //
+    // Check out the *basic_search* example if this makes
+    // small sense to you.
+    let mut schema_builder = Schema::builder();
+
+    // Tantivy does not really have a notion of primary id.
+    // This may change in the future.
+    //
+    // Still, we can create a `isbn` field and use it as an id. This
+    // field can be `u64` or a `text`, depending on your use case.
+    // It just needs to be indexed.
+    //
+    // If it is `text`, let's make sure to keep it `raw` and let's avoid
+    // running any text processing on it.
+    // This is done by associating this field to the tokenizer named `raw`.
+    // Rather than building our
+    // [`TextOptions`](//docs.rs/tantivy/~0/tantivy/schema/struct.TextOptions.html) manually, We
+    // use the `STRING` shortcut. `STRING` stands for indexed (without term frequency or positions)
+    // and untokenized.
+    //
+    // Because we also want to be able to see this `id` in our returned documents,
+    // we also mark the field as stored.
+    let isbn = schema_builder.add_text_field("isbn", STRING | STORED);
+    let title = schema_builder.add_text_field("title", TEXT | STORED);
+    let schema = schema_builder.build();
+
+    let index = Index::create_in_ram(schema.clone());
+
+    let mut index_writer = index.writer(50_000_000)?;
+
+    // Let's add a couple of documents, for the sake of the example.
+    let mut old_man_doc = Document::default();
+    old_man_doc.add_text(title, "The Old Man and the Sea");
+    index_writer.add_document(doc!(
+        isbn => "978-0099908401",
+        title => "The old Man and the see"
+    ))?;
+    index_writer.add_document(doc!(
+        isbn => "978-0140177398",
+        title => "Of Mice and Men",
+    ))?;
+    index_writer.add_document(doc!(
+       title => "Frankentein", //< Oops there is a typo here.
+       isbn => "978-9176370711",
+    ))?;
+    index_writer.commit()?;
+    let reader = index.reader()?;
+
+    let frankenstein_isbn = Term::from_field_text(isbn, "978-9176370711");
+
+    // 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),
+        r#"{"isbn":["978-9176370711"],"title":["Frankentein"]}"#,
+    );
+
+    // # Update = Delete + Insert
+    //
+    // Here we will want to update the typo in the `Frankenstein` book.
+    //
+    // Tantivy does not handle updates directly, we need to delete
+    // and reinsert the document.
+    //
+    // This can be complicated as it means you need to have access
+    // to the entire document. It is good practise to integrate tantivy
+    // with a key value store for this reason.
+    //
+    // To remove one of the document, we just call `delete_term`
+    // on its id.
+    //
+    // Note that `tantivy` does nothing to enforce the idea that
+    // there is only one document associated with this id.
+    //
+    // Also you might have noticed that we apply the delete before
+    // having committed. This does not matter really...
+    index_writer.delete_term(frankenstein_isbn.clone());
+
+    // We now need to reinsert our document without the typo.
+    index_writer.add_document(doc!(
+       title => "Frankenstein",
+       isbn => "978-9176370711",
+    ))?;
+
+    // You are guaranteed that your clients will only observe your index in
+    // the state it was in after a commit.
+    // In this example, your search engine will at no point be missing the *Frankenstein* document.
+    // Everything happened as if the document was updated.
+    index_writer.commit()?;
+    // We reload our searcher to make our change available to clients.
+    reader.reload()?;
+
+    // No more typo!
+    let frankenstein_new_doc = extract_doc_given_isbn(&reader, &frankenstein_isbn)?.unwrap();
+    assert_eq!(
+        schema.to_json(&frankenstein_new_doc),
+        r#"{"isbn":["978-9176370711"],"title":["Frankenstein"]}"#,
+    );
+
+    Ok(())
+}

examples/faceted_search.rs

@@ -0,0 +1,112 @@
+// # Basic Example
+//
+// This example covers the basic functionalities of
+// tantivy.
+//
+// We will :
+// - define our schema
+// = create an index in a directory
+// - index few documents in our index
+// - search for the best document matchings "sea whale"
+// - retrieve the best document original content.
+
+// ---
+// Importing tantivy...
+use tantivy::collector::FacetCollector;
+use tantivy::query::{AllQuery, TermQuery};
+use tantivy::schema::*;
+use tantivy::{doc, Index};
+
+fn main() -> tantivy::Result<()> {
+    // Let's create a temporary directory for the sake of this example
+    let mut schema_builder = Schema::builder();
+
+    let name = schema_builder.add_text_field("felin_name", TEXT | STORED);
+    // this is our faceted field: its scientific classification
+    let classification = schema_builder.add_facet_field("classification", FacetOptions::default());
+
+    let schema = schema_builder.build();
+    let index = Index::create_in_ram(schema);
+
+    let mut index_writer = index.writer(30_000_000)?;
+
+    // For convenience, tantivy also comes with a macro to
+    // reduce the boilerplate above.
+    index_writer.add_document(doc!(
+        name => "Cat",
+        classification => Facet::from("/Felidae/Felinae/Felis")
+    ))?;
+    index_writer.add_document(doc!(
+        name => "Canada lynx",
+        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),
+            ]
+        );
+    }
+
+    // Facets are also searchable.
+    //
+    // 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.
+
+    // Check the reference doc for different ways to create a `Facet` object.
+    {
+        let facet = Facet::from("/Felidae/Pantherinae");
+        let facet_term = Term::from_facet(classification, &facet);
+        let facet_term_query = TermQuery::new(facet_term, IndexRecordOption::Basic);
+        let mut facet_collector = FacetCollector::for_field(classification);
+        facet_collector.add_facet("/Felidae/Pantherinae");
+        let facet_counts = searcher.search(&facet_term_query, &facet_collector)?;
+        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(())
+}

examples/faceted_search_with_tweaked_score.rs

@@ -0,0 +1,98 @@
+use std::collections::HashSet;
+
+use tantivy::collector::TopDocs;
+use tantivy::query::BooleanQuery;
+use tantivy::schema::*;
+use tantivy::{doc, 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", FacetOptions::default());
+
+    let schema = schema_builder.build();
+    let index = Index::create_in_ram(schema);
+
+    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()
+                    .as_text()
+                    .unwrap()
+                    .to_owned()
+            })
+            .collect();
+        assert_eq!(titles, vec!["Fried egg", "Egg rolls"]);
+    }
+    Ok(())
+}

examples/generate_html.sh

@@ -1,2 +0,0 @@
-#!/bin/bash
-docco simple_search.rs -o html

examples/html/docco.css

@@ -1,518 +0,0 @@
-/*--------------------- Typography ----------------------------*/
-
-@font-face {
-    font-family: 'aller-light';
-    src: url('public/fonts/aller-light.eot');
-    src: url('public/fonts/aller-light.eot?#iefix') format('embedded-opentype'),
-         url('public/fonts/aller-light.woff') format('woff'),
-         url('public/fonts/aller-light.ttf') format('truetype');
-    font-weight: normal;
-    font-style: normal;
-}
-
-@font-face {
-    font-family: 'aller-bold';
-    src: url('public/fonts/aller-bold.eot');
-    src: url('public/fonts/aller-bold.eot?#iefix') format('embedded-opentype'),
-         url('public/fonts/aller-bold.woff') format('woff'),
-         url('public/fonts/aller-bold.ttf') format('truetype');
-    font-weight: normal;
-    font-style: normal;
-}
-
-@font-face {
-    font-family: 'roboto-black';
-    src: url('public/fonts/roboto-black.eot');
-    src: url('public/fonts/roboto-black.eot?#iefix') format('embedded-opentype'),
-         url('public/fonts/roboto-black.woff') format('woff'),
-         url('public/fonts/roboto-black.ttf') format('truetype');
-    font-weight: normal;
-    font-style: normal;
-}
-
-/*--------------------- Layout ----------------------------*/
-html { height: 100%; }
-body {
-  font-family: "aller-light";
-  font-size: 14px;
-  line-height: 18px;
-  color: #30404f;
-  margin: 0; padding: 0;
-  height:100%;
-}
-#container { min-height: 100%; }
-
-a {
-  color: #000;
-}
-
-b, strong {
-  font-weight: normal;
-  font-family: "aller-bold";
-}
-
-p {
-  margin: 15px 0 0px;
-}
-  .annotation ul, .annotation ol {
-    margin: 25px 0;
-  }
-    .annotation ul li, .annotation ol li {
-      font-size: 14px;
-      line-height: 18px;
-      margin: 10px 0;
-    }
-
-h1, h2, h3, h4, h5, h6 {
-  color: #112233;
-  line-height: 1em;
-  font-weight: normal;
-  font-family: "roboto-black";
-  text-transform: uppercase;
-  margin: 30px 0 15px 0;
-}
-
-h1 {
-  margin-top: 40px;
-}
-h2 {
-  font-size: 1.26em;
-}
-
-hr {
-  border: 0;
-  background: 1px #ddd;
-  height: 1px;
-  margin: 20px 0;
-}
-
-pre, tt, code {
-  font-size: 12px; line-height: 16px;
-  font-family: Menlo, Monaco, Consolas, "Lucida Console", monospace;
-  margin: 0; padding: 0;
-}
-  .annotation pre {
-    display: block;
-    margin: 0;
-    padding: 7px 10px;
-    background: #fcfcfc;
-    -moz-box-shadow:    inset 0 0 10px rgba(0,0,0,0.1);
-    -webkit-box-shadow: inset 0 0 10px rgba(0,0,0,0.1);
-    box-shadow:         inset 0 0 10px rgba(0,0,0,0.1);
-    overflow-x: auto;
-  }
-    .annotation pre code {
-      border: 0;
-      padding: 0;
-      background: transparent;
-    }
-
-
-blockquote {
-  border-left: 5px solid #ccc;
-  margin: 0;
-  padding: 1px 0 1px 1em;
-}
-  .sections blockquote p {
-    font-family: Menlo, Consolas, Monaco, monospace;
-    font-size: 12px; line-height: 16px;
-    color: #999;
-    margin: 10px 0 0;
-    white-space: pre-wrap;
-  }
-
-ul.sections {
-  list-style: none;
-  padding:0 0 5px 0;;
-  margin:0;
-}
-
-/*
-  Force border-box so that % widths fit the parent
-  container without overlap because of margin/padding.
-
-  More Info : http://www.quirksmode.org/css/box.html
-*/
-ul.sections > li > div {
-  -moz-box-sizing: border-box;    /* firefox */
-  -ms-box-sizing: border-box;     /* ie */
-  -webkit-box-sizing: border-box; /* webkit */
-  -khtml-box-sizing: border-box;  /* konqueror */
-  box-sizing: border-box;         /* css3 */
-}
-
-
-/*---------------------- Jump Page -----------------------------*/
-#jump_to, #jump_page {
-  margin: 0;
-  background: white;
-  -webkit-box-shadow: 0 0 25px #777; -moz-box-shadow: 0 0 25px #777;
-  -webkit-border-bottom-left-radius: 5px; -moz-border-radius-bottomleft: 5px;
-  font: 16px Arial;
-  cursor: pointer;
-  text-align: right;
-  list-style: none;
-}
-
-#jump_to a {
-  text-decoration: none;
-}
-
-#jump_to a.large {
-  display: none;
-}
-#jump_to a.small {
-  font-size: 22px;
-  font-weight: bold;
-  color: #676767;
-}
-
-#jump_to, #jump_wrapper {
-  position: fixed;
-  right: 0; top: 0;
-  padding: 10px 15px;
-  margin:0;
-}
-
-#jump_wrapper {
-  display: none;
-  padding:0;
-}
-
-#jump_to:hover #jump_wrapper {
-  display: block;
-}
-
-#jump_page_wrapper{
-  position: fixed;
-  right: 0;
-  top: 0;
-  bottom: 0;
-}
-
-#jump_page {
-  padding: 5px 0 3px;
-  margin: 0 0 25px 25px;
-  max-height: 100%;
-  overflow: auto;
-}
-
-#jump_page .source {
-  display: block;
-  padding: 15px;
-  text-decoration: none;
-  border-top: 1px solid #eee;
-}
-
-#jump_page .source:hover {
-  background: #f5f5ff;
-}
-
-#jump_page .source:first-child {
-}
-
-/*---------------------- Low resolutions (> 320px) ---------------------*/
-@media only screen and (min-width: 320px) {
-  .pilwrap { display: none; }
-
-  ul.sections > li > div {
-    display: block;
-    padding:5px 10px 0 10px;
-  }
-
-  ul.sections > li > div.annotation ul, ul.sections > li > div.annotation ol {
-    padding-left: 30px;
-  }
-
-  ul.sections > li > div.content {
-    overflow-x:auto;
-    -webkit-box-shadow: inset 0 0 5px #e5e5ee;
-    box-shadow: inset 0 0 5px #e5e5ee;
-    border: 1px solid #dedede;
-    margin:5px 10px 5px 10px;
-    padding-bottom: 5px;
-  }
-
-  ul.sections > li > div.annotation pre {
-    margin: 7px 0 7px;
-    padding-left: 15px;
-  }
-
-  ul.sections > li > div.annotation p tt, .annotation code {
-    background: #f8f8ff;
-    border: 1px solid #dedede;
-    font-size: 12px;
-    padding: 0 0.2em;
-  }
-}
-
-/*----------------------  (> 481px) ---------------------*/
-@media only screen and (min-width: 481px) {
-  #container {
-    position: relative;
-  }
-  body {
-    background-color: #F5F5FF;
-    font-size: 15px;
-    line-height: 21px;
-  }
-  pre, tt, code {
-    line-height: 18px;
-  }
-  p, ul, ol {
-    margin: 0 0 15px;
-  }
-
-
-  #jump_to {
-    padding: 5px 10px;
-  }
-  #jump_wrapper {
-    padding: 0;
-  }
-  #jump_to, #jump_page {
-    font: 10px Arial;
-    text-transform: uppercase;
-  }
-  #jump_page .source {
-    padding: 5px 10px;
-  }
-  #jump_to a.large {
-    display: inline-block;
-  }
-  #jump_to a.small {
-    display: none;
-  }
-
-
-
-  #background {
-    position: absolute;
-    top: 0; bottom: 0;
-    width: 350px;
-    background: #fff;
-    border-right: 1px solid #e5e5ee;
-    z-index: -1;
-  }
-
-  ul.sections > li > div.annotation ul, ul.sections > li > div.annotation ol {
-    padding-left: 40px;
-  }
-
-  ul.sections > li {
-    white-space: nowrap;
-  }
-
-  ul.sections > li > div {
-    display: inline-block;
-  }
-
-  ul.sections > li > div.annotation {
-    max-width: 350px;
-    min-width: 350px;
-    min-height: 5px;
-    padding: 13px;
-    overflow-x: hidden;
-    white-space: normal;
-    vertical-align: top;
-    text-align: left;
-  }
-  ul.sections > li > div.annotation pre {
-    margin: 15px 0 15px;
-    padding-left: 15px;
-  }
-
-  ul.sections > li > div.content {
-    padding: 13px;
-    vertical-align: top;
-    border: none;
-    -webkit-box-shadow: none;
-    box-shadow: none;
-  }
-
-  .pilwrap {
-    position: relative;
-    display: inline;
-  }
-
-  .pilcrow {
-    font: 12px Arial;
-    text-decoration: none;
-    color: #454545;
-    position: absolute;
-    top: 3px; left: -20px;
-    padding: 1px 2px;
-    opacity: 0;
-    -webkit-transition: opacity 0.2s linear;
-  }
-    .for-h1 .pilcrow {
-      top: 47px;
-    }
-    .for-h2 .pilcrow, .for-h3 .pilcrow, .for-h4 .pilcrow {
-      top: 35px;
-    }
-
-  ul.sections > li > div.annotation:hover .pilcrow {
-    opacity: 1;
-  }
-}
-
-/*---------------------- (> 1025px) ---------------------*/
-@media only screen and (min-width: 1025px) {
-
-  body {
-    font-size: 16px;
-    line-height: 24px;
-  }
-
-  #background {
-    width: 525px;
-  }
-  ul.sections > li > div.annotation {
-    max-width: 525px;
-    min-width: 525px;
-    padding: 10px 25px 1px 50px;
-  }
-  ul.sections > li > div.content {
-    padding: 9px 15px 16px 25px;
-  }
-}
-
-/*---------------------- Syntax Highlighting -----------------------------*/
-
-td.linenos { background-color: #f0f0f0; padding-right: 10px; }
-span.lineno { background-color: #f0f0f0; padding: 0 5px 0 5px; }
-/*
-
-github.com style (c) Vasily Polovnyov <vast@whiteants.net>
-
-*/
-
-pre code {
-  display: block; padding: 0.5em;
-  color: #000;
-  background: #f8f8ff
-}
-
-pre .hljs-comment,
-pre .hljs-template_comment,
-pre .hljs-diff .hljs-header,
-pre .hljs-javadoc {
-  color: #408080;
-  font-style: italic
-}
-
-pre .hljs-keyword,
-pre .hljs-assignment,
-pre .hljs-literal,
-pre .hljs-css .hljs-rule .hljs-keyword,
-pre .hljs-winutils,
-pre .hljs-javascript .hljs-title,
-pre .hljs-lisp .hljs-title,
-pre .hljs-subst {
-  color: #954121;
-  /*font-weight: bold*/
-}
-
-pre .hljs-number,
-pre .hljs-hexcolor {
-  color: #40a070
-}
-
-pre .hljs-string,
-pre .hljs-tag .hljs-value,
-pre .hljs-phpdoc,
-pre .hljs-tex .hljs-formula {
-  color: #219161;
-}
-
-pre .hljs-title,
-pre .hljs-id {
-  color: #19469D;
-}
-pre .hljs-params {
-  color: #00F;
-}
-
-pre .hljs-javascript .hljs-title,
-pre .hljs-lisp .hljs-title,
-pre .hljs-subst {
-  font-weight: normal
-}
-
-pre .hljs-class .hljs-title,
-pre .hljs-haskell .hljs-label,
-pre .hljs-tex .hljs-command {
-  color: #458;
-  font-weight: bold
-}
-
-pre .hljs-tag,
-pre .hljs-tag .hljs-title,
-pre .hljs-rules .hljs-property,
-pre .hljs-django .hljs-tag .hljs-keyword {
-  color: #000080;
-  font-weight: normal
-}
-
-pre .hljs-attribute,
-pre .hljs-variable,
-pre .hljs-instancevar,
-pre .hljs-lisp .hljs-body {
-  color: #008080
-}
-
-pre .hljs-regexp {
-  color: #B68
-}
-
-pre .hljs-class {
-  color: #458;
-  font-weight: bold
-}
-
-pre .hljs-symbol,
-pre .hljs-ruby .hljs-symbol .hljs-string,
-pre .hljs-ruby .hljs-symbol .hljs-keyword,
-pre .hljs-ruby .hljs-symbol .hljs-keymethods,
-pre .hljs-lisp .hljs-keyword,
-pre .hljs-tex .hljs-special,
-pre .hljs-input_number {
-  color: #990073
-}
-
-pre .hljs-builtin,
-pre .hljs-constructor,
-pre .hljs-built_in,
-pre .hljs-lisp .hljs-title {
-  color: #0086b3
-}
-
-pre .hljs-preprocessor,
-pre .hljs-pi,
-pre .hljs-doctype,
-pre .hljs-shebang,
-pre .hljs-cdata {
-  color: #999;
-  font-weight: bold
-}
-
-pre .hljs-deletion {
-  background: #fdd
-}
-
-pre .hljs-addition {
-  background: #dfd
-}
-
-pre .hljs-diff .hljs-change {
-  background: #0086b3
-}
-
-pre .hljs-chunk {
-  color: #aaa
-}
-
-pre .hljs-tex .hljs-formula {
-  opacity: 0.5;
-}

examples/html/public/stylesheets/normalize.css

@@ -1,375 +0,0 @@
-/*! normalize.css v2.0.1 | MIT License | git.io/normalize */
-
-/* ==========================================================================
-   HTML5 display definitions
-   ========================================================================== */
-
-/*
- * Corrects `block` display not defined in IE 8/9.
- */
-
-article,
-aside,
-details,
-figcaption,
-figure,
-footer,
-header,
-hgroup,
-nav,
-section,
-summary {
-    display: block;
-}
-
-/*
- * Corrects `inline-block` display not defined in IE 8/9.
- */
-
-audio,
-canvas,
-video {
-    display: inline-block;
-}
-
-/*
- * Prevents modern browsers from displaying `audio` without controls.
- * Remove excess height in iOS 5 devices.
- */
-
-audio:not([controls]) {
-    display: none;
-    height: 0;
-}
-
-/*
- * Addresses styling for `hidden` attribute not present in IE 8/9.
- */
-
-[hidden] {
-    display: none;
-}
-
-/* ==========================================================================
-   Base
-   ========================================================================== */
-
-/*
- * 1. Sets default font family to sans-serif.
- * 2. Prevents iOS text size adjust after orientation change, without disabling
- *    user zoom.
- */
-
-html {
-    font-family: sans-serif; /* 1 */
-    -webkit-text-size-adjust: 100%; /* 2 */
-    -ms-text-size-adjust: 100%; /* 2 */
-}
-
-/*
- * Removes default margin.
- */
-
-body {
-    margin: 0;
-}
-
-/* ==========================================================================
-   Links
-   ========================================================================== */
-
-/*
- * Addresses `outline` inconsistency between Chrome and other browsers.
- */
-
-a:focus {
-    outline: thin dotted;
-}
-
-/*
- * Improves readability when focused and also mouse hovered in all browsers.
- */
-
-a:active,
-a:hover {
-    outline: 0;
-}
-
-/* ==========================================================================
-   Typography
-   ========================================================================== */
-
-/*
- * Addresses `h1` font sizes within `section` and `article` in Firefox 4+,
- * Safari 5, and Chrome.
- */
-
-h1 {
-    font-size: 2em;
-}
-
-/*
- * Addresses styling not present in IE 8/9, Safari 5, and Chrome.
- */
-
-abbr[title] {
-    border-bottom: 1px dotted;
-}
-
-/*
- * Addresses style set to `bolder` in Firefox 4+, Safari 5, and Chrome.
- */
-
-b,
-strong {
-    font-weight: bold;
-}
-
-/*
- * Addresses styling not present in Safari 5 and Chrome.
- */
-
-dfn {
-    font-style: italic;
-}
-
-/*
- * Addresses styling not present in IE 8/9.
- */
-
-mark {
-    background: #ff0;
-    color: #000;
-}
-
-
-/*
- * Corrects font family set oddly in Safari 5 and Chrome.
- */
-
-code,
-kbd,
-pre,
-samp {
-    font-family: monospace, serif;
-    font-size: 1em;
-}
-
-/*
- * Improves readability of pre-formatted text in all browsers.
- */
-
-pre {
-    white-space: pre;
-    white-space: pre-wrap;
-    word-wrap: break-word;
-}
-
-/*
- * Sets consistent quote types.
- */
-
-q {
-    quotes: "\201C" "\201D" "\2018" "\2019";
-}
-
-/*
- * Addresses inconsistent and variable font size in all browsers.
- */
-
-small {
-    font-size: 80%;
-}
-
-/*
- * Prevents `sub` and `sup` affecting `line-height` in all browsers.
- */
-
-sub,
-sup {
-    font-size: 75%;
-    line-height: 0;
-    position: relative;
-    vertical-align: baseline;
-}
-
-sup {
-    top: -0.5em;
-}
-
-sub {
-    bottom: -0.25em;
-}
-
-/* ==========================================================================
-   Embedded content
-   ========================================================================== */
-
-/*
- * Removes border when inside `a` element in IE 8/9.
- */
-
-img {
-    border: 0;
-}
-
-/*
- * Corrects overflow displayed oddly in IE 9.
- */
-
-svg:not(:root) {
-    overflow: hidden;
-}
-
-/* ==========================================================================
-   Figures
-   ========================================================================== */
-
-/*
- * Addresses margin not present in IE 8/9 and Safari 5.
- */
-
-figure {
-    margin: 0;
-}
-
-/* ==========================================================================
-   Forms
-   ========================================================================== */
-
-/*
- * Define consistent border, margin, and padding.
- */
-
-fieldset {
-    border: 1px solid #c0c0c0;
-    margin: 0 2px;
-    padding: 0.35em 0.625em 0.75em;
-}
-
-/*
- * 1. Corrects color not being inherited in IE 8/9.
- * 2. Remove padding so people aren't caught out if they zero out fieldsets.
- */
-
-legend {
-    border: 0; /* 1 */
-    padding: 0; /* 2 */
-}
-
-/*
- * 1. Corrects font family not being inherited in all browsers.
- * 2. Corrects font size not being inherited in all browsers.
- * 3. Addresses margins set differently in Firefox 4+, Safari 5, and Chrome
- */
-
-button,
-input,
-select,
-textarea {
-    font-family: inherit; /* 1 */
-    font-size: 100%; /* 2 */
-    margin: 0; /* 3 */
-}
-
-/*
- * Addresses Firefox 4+ setting `line-height` on `input` using `!important` in
- * the UA stylesheet.
- */
-
-button,
-input {
-    line-height: normal;
-}
-
-/*
- * 1. Avoid the WebKit bug in Android 4.0.* where (2) destroys native `audio`
- *    and `video` controls.
- * 2. Corrects inability to style clickable `input` types in iOS.
- * 3. Improves usability and consistency of cursor style between image-type
- *    `input` and others.
- */
-
-button,
-html input[type="button"], /* 1 */
-input[type="reset"],
-input[type="submit"] {
-    -webkit-appearance: button; /* 2 */
-    cursor: pointer; /* 3 */
-}
-
-/*
- * Re-set default cursor for disabled elements.
- */
-
-button[disabled],
-input[disabled] {
-    cursor: default;
-}
-
-/*
- * 1. Addresses box sizing set to `content-box` in IE 8/9.
- * 2. Removes excess padding in IE 8/9.
- */
-
-input[type="checkbox"],
-input[type="radio"] {
-    box-sizing: border-box; /* 1 */
-    padding: 0; /* 2 */
-}
-
-/*
- * 1. Addresses `appearance` set to `searchfield` in Safari 5 and Chrome.
- * 2. Addresses `box-sizing` set to `border-box` in Safari 5 and Chrome
- *    (include `-moz` to future-proof).
- */
-
-input[type="search"] {
-    -webkit-appearance: textfield; /* 1 */
-    -moz-box-sizing: content-box;
-    -webkit-box-sizing: content-box; /* 2 */
-    box-sizing: content-box;
-}
-
-/*
- * Removes inner padding and search cancel button in Safari 5 and Chrome
- * on OS X.
- */
-
-input[type="search"]::-webkit-search-cancel-button,
-input[type="search"]::-webkit-search-decoration {
-    -webkit-appearance: none;
-}
-
-/*
- * Removes inner padding and border in Firefox 4+.
- */
-
-button::-moz-focus-inner,
-input::-moz-focus-inner {
-    border: 0;
-    padding: 0;
-}
-
-/*
- * 1. Removes default vertical scrollbar in IE 8/9.
- * 2. Improves readability and alignment in all browsers.
- */
-
-textarea {
-    overflow: auto; /* 1 */
-    vertical-align: top; /* 2 */
-}
-
-/* ==========================================================================
-   Tables
-   ========================================================================== */
-
-/*
- * Remove most spacing between table cells.
- */
-
-table {
-    border-collapse: collapse;
-    border-spacing: 0;
-}

examples/html/simple_search.html

@@ -1,542 +0,0 @@
-<!DOCTYPE html>
-
-<html>
-<head>
-  <title>simple_search.rs</title>
-  <meta http-equiv="content-type" content="text/html; charset=UTF-8">
-  <meta name="viewport" content="width=device-width, target-densitydpi=160dpi, initial-scale=1.0; maximum-scale=1.0; user-scalable=0;">
-  <link rel="stylesheet" media="all" href="docco.css" />
-</head>
-<body>
-  <div id="container">
-    <div id="background"></div>
-    
-    <ul class="sections">
-        
-          <li id="title">
-              <div class="annotation">
-                  <h1>simple_search.rs</h1>
-              </div>
-          </li>
-        
-        
-        
-        <li id="section-1">
-            <div class="annotation">
-              
-              <div class="pilwrap ">
-                <a class="pilcrow" href="#section-1">&#182;</a>
-              </div>
-              
-            </div>
-            
-            <div class="content"><div class='highlight'><pre><span class="hljs-keyword">extern</span> <span class="hljs-keyword">crate</span> tantivy;
-<span class="hljs-keyword">extern</span> <span class="hljs-keyword">crate</span> tempdir;
-
-<span class="hljs-meta">#[macro_use]</span>
-<span class="hljs-keyword">extern</span> <span class="hljs-keyword">crate</span> serde_json;
-
-<span class="hljs-keyword">use</span> std::path::Path;
-<span class="hljs-keyword">use</span> tempdir::TempDir;
-<span class="hljs-keyword">use</span> tantivy::Index;
-<span class="hljs-keyword">use</span> tantivy::schema::*;
-<span class="hljs-keyword">use</span> tantivy::collector::TopCollector;
-<span class="hljs-keyword">use</span> tantivy::query::QueryParser;
-
-<span class="hljs-function"><span class="hljs-keyword">fn</span> <span class="hljs-title">main</span></span>() {</pre></div></div>
-            
-        </li>
-        
-        
-        <li id="section-2">
-            <div class="annotation">
-              
-              <div class="pilwrap ">
-                <a class="pilcrow" href="#section-2">&#182;</a>
-              </div>
-              <p>Let’s create a temporary directory for the
-sake of this example</p>
-
-            </div>
-            
-            <div class="content"><div class='highlight'><pre>    <span class="hljs-keyword">if</span> <span class="hljs-keyword">let</span> <span class="hljs-literal">Ok</span>(dir) = TempDir::new(<span class="hljs-string">"tantivy_example_dir"</span>) {
-        run_example(dir.path()).unwrap();
-        dir.close().unwrap();
-    }
-}
-
-
-<span class="hljs-function"><span class="hljs-keyword">fn</span> <span class="hljs-title">run_example</span></span>(index_path: &amp;Path) -&gt; tantivy::<span class="hljs-built_in">Result</span>&lt;()&gt; {</pre></div></div>
-            
-        </li>
-        
-        
-        <li id="section-3">
-            <div class="annotation">
-              
-              <div class="pilwrap ">
-                <a class="pilcrow" href="#section-3">&#182;</a>
-              </div>
-              <h1 id="defining-the-schema">Defining the schema</h1>
-<p>The Tantivy index requires a very strict schema.
-The schema declares which fields are in the index,
-and for each field, its type and “the way it should
-be indexed”.</p>
-
-            </div>
-            
-        </li>
-        
-        
-        <li id="section-4">
-            <div class="annotation">
-              
-              <div class="pilwrap ">
-                <a class="pilcrow" href="#section-4">&#182;</a>
-              </div>
-              <p>first we need to define a schema …</p>
-
-            </div>
-            
-            <div class="content"><div class='highlight'><pre>    <span class="hljs-keyword">let</span> <span class="hljs-keyword">mut</span> schema_builder = SchemaBuilder::<span class="hljs-keyword">default</span>();</pre></div></div>
-            
-        </li>
-        
-        
-        <li id="section-5">
-            <div class="annotation">
-              
-              <div class="pilwrap ">
-                <a class="pilcrow" href="#section-5">&#182;</a>
-              </div>
-              <p>Our first field is title.
-We want full-text search for it, and we also want 
-to be able to retrieve the document after the search.</p>
-<p>TEXT | STORED is some syntactic sugar to describe
-that.</p>
-<p><code>TEXT</code> means the field should be tokenized and indexed,
-along with its term frequency and term positions.</p>
-<p><code>STORED</code> means that the field will also be saved
-in a compressed, row-oriented key-value store.
-This store is useful to reconstruct the
-documents that were selected during the search phase.</p>
-
-            </div>
-            
-            <div class="content"><div class='highlight'><pre>    schema_builder.add_text_field(<span class="hljs-string">"title"</span>, TEXT | STORED);</pre></div></div>
-            
-        </li>
-        
-        
-        <li id="section-6">
-            <div class="annotation">
-              
-              <div class="pilwrap ">
-                <a class="pilcrow" href="#section-6">&#182;</a>
-              </div>
-              <p>Our second field is body.
-We want full-text search for it, but we do not 
-need to be able to be able to retrieve it
-for our application. </p>
-<p>We can make our index lighter and 
-by omitting <code>STORED</code> flag.</p>
-
-            </div>
-            
-            <div class="content"><div class='highlight'><pre>    schema_builder.add_text_field(<span class="hljs-string">"body"</span>, TEXT);
-
-    <span class="hljs-keyword">let</span> schema = schema_builder.build();</pre></div></div>
-            
-        </li>
-        
-        
-        <li id="section-7">
-            <div class="annotation">
-              
-              <div class="pilwrap ">
-                <a class="pilcrow" href="#section-7">&#182;</a>
-              </div>
-              <h1 id="indexing-documents">Indexing documents</h1>
-<p>Let’s create a brand new index.</p>
-<p>This will actually just save a meta.json
-with our schema in the directory.</p>
-
-            </div>
-            
-            <div class="content"><div class='highlight'><pre>    <span class="hljs-keyword">let</span> index = Index::create(index_path, schema.clone())?;</pre></div></div>
-            
-        </li>
-        
-        
-        <li id="section-8">
-            <div class="annotation">
-              
-              <div class="pilwrap ">
-                <a class="pilcrow" href="#section-8">&#182;</a>
-              </div>
-              <p>To insert document we need an index writer.
-There must be only one writer at a time.
-This single <code>IndexWriter</code> is already
-multithreaded.</p>
-<p>Here we use a buffer of 50MB per thread. Using a bigger
-heap for the indexer can increase its throughput.</p>
-
-            </div>
-            
-            <div class="content"><div class='highlight'><pre>    <span class="hljs-keyword">let</span> <span class="hljs-keyword">mut</span> index_writer = index.writer(<span class="hljs-number">50_000_000</span>)?;</pre></div></div>
-            
-        </li>
-        
-        
-        <li id="section-9">
-            <div class="annotation">
-              
-              <div class="pilwrap ">
-                <a class="pilcrow" href="#section-9">&#182;</a>
-              </div>
-              <p>Let’s index our documents!
-We first need a handle on the title and the body field.</p>
-
-            </div>
-            
-        </li>
-        
-        
-        <li id="section-10">
-            <div class="annotation">
-              
-              <div class="pilwrap ">
-                <a class="pilcrow" href="#section-10">&#182;</a>
-              </div>
-              <h3 id="create-a-document-manually-">Create a document “manually”.</h3>
-<p>We can create a document manually, by setting the fields
-one by one in a Document object.</p>
-
-            </div>
-            
-            <div class="content"><div class='highlight'><pre>    <span class="hljs-keyword">let</span> title = schema.get_field(<span class="hljs-string">"title"</span>).unwrap();
-    <span class="hljs-keyword">let</span> body = schema.get_field(<span class="hljs-string">"body"</span>).unwrap();
-
-    <span class="hljs-keyword">let</span> <span class="hljs-keyword">mut</span> old_man_doc = Document::<span class="hljs-keyword">default</span>();
-    old_man_doc.add_text(title, <span class="hljs-string">"The Old Man and the Sea"</span>);
-    old_man_doc.add_text(
-        body,
-        <span class="hljs-string">"He was an old man who fished alone in a skiff in the Gulf Stream and \
-                          he had gone eighty-four days now without taking a fish."</span>,
-    );</pre></div></div>
-            
-        </li>
-        
-        
-        <li id="section-11">
-            <div class="annotation">
-              
-              <div class="pilwrap ">
-                <a class="pilcrow" href="#section-11">&#182;</a>
-              </div>
-              <p>… and add it to the <code>IndexWriter</code>.</p>
-
-            </div>
-            
-            <div class="content"><div class='highlight'><pre>    index_writer.add_document(old_man_doc);</pre></div></div>
-            
-        </li>
-        
-        
-        <li id="section-12">
-            <div class="annotation">
-              
-              <div class="pilwrap ">
-                <a class="pilcrow" href="#section-12">&#182;</a>
-              </div>
-              <h3 id="create-a-document-directly-from-json-">Create a document directly from json.</h3>
-<p>Alternatively, we can use our schema to parse a
-document object directly from json.
-The document is a string, but we use the <code>json</code> macro
-from <code>serde_json</code> for the convenience of multi-line support.</p>
-
-            </div>
-            
-            <div class="content"><div class='highlight'><pre>    <span class="hljs-keyword">let</span> json = json!({
-       <span class="hljs-string">"title"</span>: <span class="hljs-string">"Of Mice and Men"</span>,
-       <span class="hljs-string">"body"</span>: <span class="hljs-string">"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"</span>
-    });
-    <span class="hljs-keyword">let</span> mice_and_men_doc = schema.parse_document(&amp;json.to_string())?;
-
-    index_writer.add_document(mice_and_men_doc);</pre></div></div>
-            
-        </li>
-        
-        
-        <li id="section-13">
-            <div class="annotation">
-              
-              <div class="pilwrap ">
-                <a class="pilcrow" href="#section-13">&#182;</a>
-              </div>
-              <p>Multi-valued field are allowed, they are
-expressed in JSON by an array.
-The following document has two titles.</p>
-
-            </div>
-            
-            <div class="content"><div class='highlight'><pre>    <span class="hljs-keyword">let</span> json = json!({
-       <span class="hljs-string">"title"</span>: [<span class="hljs-string">"Frankenstein"</span>, <span class="hljs-string">"The Modern Prometheus"</span>],
-       <span class="hljs-string">"body"</span>: <span class="hljs-string">"You will rejoice to hear that no disaster has accompanied the commencement of an \
-                enterprise which you have regarded with such evil forebodings.  I arrived here \
-                yesterday, and my first task is to assure my dear sister of my welfare and \
-                increasing confidence in the success of my undertaking."</span>
-    });
-    <span class="hljs-keyword">let</span> frankenstein_doc = schema.parse_document(&amp;json.to_string())?;
-
-    index_writer.add_document(frankenstein_doc);</pre></div></div>
-            
-        </li>
-        
-        
-        <li id="section-14">
-            <div class="annotation">
-              
-              <div class="pilwrap ">
-                <a class="pilcrow" href="#section-14">&#182;</a>
-              </div>
-              <p>This is an example, so we will only index 3 documents
-here. You can check out tantivy’s tutorial to index
-the English wikipedia. Tantivy’s indexing is rather fast.
-Indexing 5 million articles of the English wikipedia takes
-around 4 minutes on my computer!</p>
-
-            </div>
-            
-        </li>
-        
-        
-        <li id="section-15">
-            <div class="annotation">
-              
-              <div class="pilwrap ">
-                <a class="pilcrow" href="#section-15">&#182;</a>
-              </div>
-              <h3 id="committing">Committing</h3>
-<p>At this point our documents are not searchable.</p>
-<p>We need to call .commit() explicitly to force the
-index_writer to finish processing the documents in the queue,
-flush the current index to the disk, and advertise
-the existence of new documents.</p>
-<p>This call is blocking.</p>
-
-            </div>
-            
-            <div class="content"><div class='highlight'><pre>    index_writer.commit()?;</pre></div></div>
-            
-        </li>
-        
-        
-        <li id="section-16">
-            <div class="annotation">
-              
-              <div class="pilwrap ">
-                <a class="pilcrow" href="#section-16">&#182;</a>
-              </div>
-              <p>If <code>.commit()</code> returns correctly, then all of the
-documents that have been added are guaranteed to be
-persistently indexed.</p>
-<p>In the scenario of a crash or a power failure,
-tantivy behaves as if has rolled back to its last
-commit.</p>
-
-            </div>
-            
-        </li>
-        
-        
-        <li id="section-17">
-            <div class="annotation">
-              
-              <div class="pilwrap ">
-                <a class="pilcrow" href="#section-17">&#182;</a>
-              </div>
-              <h1 id="searching">Searching</h1>
-<p>Let’s search our index. Start by reloading
-searchers in the index. This should be done
-after every commit().</p>
-
-            </div>
-            
-            <div class="content"><div class='highlight'><pre>    index.load_searchers()?;</pre></div></div>
-            
-        </li>
-        
-        
-        <li id="section-18">
-            <div class="annotation">
-              
-              <div class="pilwrap ">
-                <a class="pilcrow" href="#section-18">&#182;</a>
-              </div>
-              <p>Afterwards create one (or more) searchers.</p>
-<p>You should create a searcher
-every time you start a “search query”.</p>
-
-            </div>
-            
-            <div class="content"><div class='highlight'><pre>    <span class="hljs-keyword">let</span> searcher = index.searcher();</pre></div></div>
-            
-        </li>
-        
-        
-        <li id="section-19">
-            <div class="annotation">
-              
-              <div class="pilwrap ">
-                <a class="pilcrow" href="#section-19">&#182;</a>
-              </div>
-              <p>The query parser can interpret human queries.
-Here, if the user does not specify which
-field they want to search, tantivy will search
-in both title and body.</p>
-
-            </div>
-            
-            <div class="content"><div class='highlight'><pre>    <span class="hljs-keyword">let</span> <span class="hljs-keyword">mut</span> query_parser = QueryParser::for_index(index, <span class="hljs-built_in">vec!</span>[title, body]);</pre></div></div>
-            
-        </li>
-        
-        
-        <li id="section-20">
-            <div class="annotation">
-              
-              <div class="pilwrap ">
-                <a class="pilcrow" href="#section-20">&#182;</a>
-              </div>
-              <p>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.</p>
-
-            </div>
-            
-            <div class="content"><div class='highlight'><pre>    <span class="hljs-keyword">let</span> query = query_parser.parse_query(<span class="hljs-string">"sea whale"</span>)?;</pre></div></div>
-            
-        </li>
-        
-        
-        <li id="section-21">
-            <div class="annotation">
-              
-              <div class="pilwrap ">
-                <a class="pilcrow" href="#section-21">&#182;</a>
-              </div>
-              <p>A query defines a set of documents, as
-well as the way they should be scored.</p>
-<p>A query created by the query parser is scored according
-to a metric called Tf-Idf, and will consider
-any document matching at least one of our terms.</p>
-
-            </div>
-            
-        </li>
-        
-        
-        <li id="section-22">
-            <div class="annotation">
-              
-              <div class="pilwrap ">
-                <a class="pilcrow" href="#section-22">&#182;</a>
-              </div>
-              <h3 id="collectors">Collectors</h3>
-<p>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 TopCollector.</p>
-
-            </div>
-            
-            <div class="content"><div class='highlight'><pre>    <span class="hljs-keyword">let</span> <span class="hljs-keyword">mut</span> top_collector = TopCollector::with_limit(<span class="hljs-number">10</span>);</pre></div></div>
-            
-        </li>
-        
-        
-        <li id="section-23">
-            <div class="annotation">
-              
-              <div class="pilwrap ">
-                <a class="pilcrow" href="#section-23">&#182;</a>
-              </div>
-              <p>We can now perform our query.</p>
-
-            </div>
-            
-            <div class="content"><div class='highlight'><pre>    searcher.search(&amp;*query, &amp;<span class="hljs-keyword">mut</span> top_collector)?;</pre></div></div>
-            
-        </li>
-        
-        
-        <li id="section-24">
-            <div class="annotation">
-              
-              <div class="pilwrap ">
-                <a class="pilcrow" href="#section-24">&#182;</a>
-              </div>
-              <p>Our top collector now contains the 10
-most relevant doc ids…</p>
-
-            </div>
-            
-            <div class="content"><div class='highlight'><pre>    <span class="hljs-keyword">let</span> doc_addresses = top_collector.docs();</pre></div></div>
-            
-        </li>
-        
-        
-        <li id="section-25">
-            <div class="annotation">
-              
-              <div class="pilwrap ">
-                <a class="pilcrow" href="#section-25">&#182;</a>
-              </div>
-              <p>The actual documents still need to be
-retrieved from Tantivy’s store.</p>
-<p>Since the body field was not configured as stored,
-the document returned will only contain
-a title.</p>
-
-            </div>
-            
-            <div class="content"><div class='highlight'><pre>
-    <span class="hljs-keyword">for</span> doc_address <span class="hljs-keyword">in</span> doc_addresses {
-        <span class="hljs-keyword">let</span> retrieved_doc = searcher.doc(&amp;doc_address)?;
-        <span class="hljs-built_in">println!</span>(<span class="hljs-string">"{}"</span>, schema.to_json(&amp;retrieved_doc));
-    }</pre></div></div>
-            
-        </li>
-        
-        
-        <li id="section-26">
-            <div class="annotation">
-              
-              <div class="pilwrap ">
-                <a class="pilcrow" href="#section-26">&#182;</a>
-              </div>
-              <p>Wait for indexing and merging threads to shut down.
-Usually this isn’t needed, but in <code>main</code> we try to
-delete the temporary directory and that fails on
-Windows if the files are still open.</p>
-
-            </div>
-            
-            <div class="content"><div class='highlight'><pre>    index_writer.wait_merging_threads()?;
-
-    <span class="hljs-literal">Ok</span>(())
-}</pre></div></div>
-            
-        </li>
-        
-    </ul>
-  </div>
-</body>
-</html>

examples/integer_range_search.rs

@@ -0,0 +1,35 @@
+// # Searching a range on an indexed int field.
+//
+// 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.
+use tantivy::collector::Count;
+use tantivy::query::RangeQuery;
+use tantivy::schema::{Schema, INDEXED};
+use tantivy::{doc, Index, Result};
+
+fn main() -> Result<()> {
+    // For the sake of simplicity, this schema will only have 1 field
+    let mut schema_builder = Schema::builder();
+
+    // `INDEXED` is a short-hand to indicate that our field should be "searchable".
+    let year_field = schema_builder.add_u64_field("year", INDEXED);
+    let schema = schema_builder.build();
+    let index = Index::create_in_ram(schema);
+    let reader = index.reader()?;
+    {
+        let mut index_writer = index.writer_with_num_threads(1, 6_000_000)?;
+        for year in 1950u64..2019u64 {
+            index_writer.add_document(doc!(year_field => year))?;
+        }
+        index_writer.commit()?;
+        // The index will be a range of years
+    }
+    reader.reload()?;
+    let searcher = reader.searcher();
+    // The end is excluded i.e. here we are searching up to 1969
+    let docs_in_the_sixties = RangeQuery::new_u64(year_field, 1960..1970);
+    // Uses a Count collector to sum the total number of docs in the range
+    let num_60s_books = searcher.search(&docs_in_the_sixties, &Count)?;
+    assert_eq!(num_60s_books, 10);
+    Ok(())
+}

examples/iterating_docs_and_positions.rs

@@ -0,0 +1,135 @@
+// # Iterating docs and positions.
+//
+// At its core of tantivy, relies on a data structure
+// called an inverted index.
+//
+// This example shows how to manually iterate through
+// the list of documents containing a term, getting
+// its term frequency, and accessing its positions.
+
+// ---
+// Importing tantivy...
+use tantivy::schema::*;
+use tantivy::{doc, DocSet, Index, Postings, TERMINATED};
+
+fn main() -> tantivy::Result<()> {
+    // We first create a schema for the sake of the
+    // example. Check the `basic_search` example for more information.
+    let mut schema_builder = Schema::builder();
+
+    // For this example, we need to make sure to index positions for our title
+    // field. `TEXT` precisely does this.
+    let title = schema_builder.add_text_field("title", TEXT | STORED);
+    let schema = schema_builder.build();
+
+    let index = Index::create_in_ram(schema);
+
+    let mut index_writer = index.writer_with_num_threads(1, 50_000_000)?;
+    index_writer.add_document(doc!(title => "The Old Man and the Sea"))?;
+    index_writer.add_document(doc!(title => "Of Mice and Men"))?;
+    index_writer.add_document(doc!(title => "The modern Promotheus"))?;
+    index_writer.commit()?;
+
+    let reader = index.reader()?;
+
+    let searcher = reader.searcher();
+
+    // A tantivy index is actually a collection of segments.
+    // Similarly, a searcher just wraps a list `segment_reader`.
+    //
+    // (Because we indexed a very small number of documents over one thread
+    // there is actually only one segment here, but let's iterate through the list
+    // anyway)
+    for segment_reader in searcher.segment_readers() {
+        // A segment contains different data structure.
+        // Inverted index stands for the combination of
+        // - the term dictionary
+        // - the inverted lists associated with each terms and their positions
+        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
+        let term_the = Term::from_field_text(title, "the");
+
+        // 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 and positions.
+        //
+        // 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)?
+        {
+            // this buffer will be used to request for positions
+            let mut positions: Vec<u32> = Vec::with_capacity(100);
+            let mut doc_id = segment_postings.doc();
+            while doc_id != TERMINATED {
+                // This MAY contains deleted documents as well.
+                if segment_reader.is_deleted(doc_id) {
+                    doc_id = segment_postings.advance();
+                    continue;
+                }
+
+                // the number of time the term appears in the document.
+                let term_freq: u32 = segment_postings.term_freq();
+                // accessing positions is slightly expensive and lazy, do not request
+                // for them if you don't need them for some documents.
+                segment_postings.positions(&mut positions);
+
+                // By definition we should have `term_freq` positions.
+                assert_eq!(positions.len(), term_freq as usize);
+
+                // This prints:
+                // ```
+                // Doc 0: TermFreq 2: [0, 4]
+                // Doc 2: TermFreq 1: [0]
+                // ```
+                println!("Doc {}: TermFreq {}: {:?}", doc_id, term_freq, positions);
+                doc_id = segment_postings.advance();
+            }
+        }
+    }
+
+    // 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
+    let term_the = Term::from_field_text(title, "the");
+
+    // Some other powerful operations (especially `.skip_to`) may be useful to consume these
+    // posting lists rapidly.
+    // You can check for them in the [`DocSet`](https://docs.rs/tantivy/~0/tantivy/trait.DocSet.html) trait
+    // and the [`Postings`](https://docs.rs/tantivy/~0/tantivy/trait.Postings.html) trait
+
+    // Also, for some VERY specific high performance use case like an OLAP analysis of logs,
+    // you can get better performance by accessing directly the blocks of doc ids.
+    for segment_reader in searcher.segment_readers() {
+        // A segment contains different data structure.
+        // Inverted index stands for the combination of
+        // - the term dictionary
+        // - the inverted lists associated with each terms and their positions
+        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 and positions.
+        //
+        // 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)?
+        {
+            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();
+            }
+        }
+    }
+
+    Ok(())
+}

examples/json_field.rs

@@ -0,0 +1,105 @@
+// # Json field example
+//
+// This example shows how the json field can be used
+// to make tantivy partially schemaless by setting it as
+// default query parser field.
+
+use tantivy::collector::{Count, TopDocs};
+use tantivy::query::QueryParser;
+use tantivy::schema::{Schema, FAST, STORED, STRING, TEXT};
+use tantivy::Index;
+
+fn main() -> tantivy::Result<()> {
+    // # Defining the schema
+    let mut schema_builder = Schema::builder();
+    schema_builder.add_date_field("timestamp", FAST | STORED);
+    let event_type = schema_builder.add_text_field("event_type", STRING | STORED);
+    let attributes = schema_builder.add_json_field("attributes", STORED | TEXT);
+    let schema = schema_builder.build();
+
+    // # Indexing documents
+    let index = Index::create_in_ram(schema.clone());
+
+    let mut index_writer = index.writer(50_000_000)?;
+    let doc = schema.parse_document(
+        r#"{
+        "timestamp": "2022-02-22T23:20:50.53Z",
+        "event_type": "click",
+        "attributes": {
+            "target": "submit-button",
+            "cart": {"product_id": 103},
+            "description": "the best vacuum cleaner ever"
+        }
+    }"#,
+    )?;
+    index_writer.add_document(doc)?;
+    let doc = schema.parse_document(
+        r#"{
+        "timestamp": "2022-02-22T23:20:51.53Z",
+        "event_type": "click",
+        "attributes": {
+            "target": "submit-button",
+            "cart": {"product_id": 133},
+            "description": "das keyboard",
+            "event_type": "holiday-sale"
+        }
+    }"#,
+    )?;
+    index_writer.add_document(doc)?;
+    index_writer.commit()?;
+
+    let reader = index.reader()?;
+    let searcher = reader.searcher();
+
+    // # Default fields: event_type and attributes
+    // By setting attributes as a default field it allows omitting attributes itself, e.g. "target",
+    // instead of "attributes.target"
+    let query_parser = QueryParser::for_index(&index, vec![event_type, attributes]);
+    {
+        let query = query_parser.parse_query("target:submit-button")?;
+        let count_docs = searcher.search(&*query, &TopDocs::with_limit(2))?;
+        assert_eq!(count_docs.len(), 2);
+    }
+    {
+        let query = query_parser.parse_query("target:submit")?;
+        let count_docs = searcher.search(&*query, &TopDocs::with_limit(2))?;
+        assert_eq!(count_docs.len(), 2);
+    }
+    {
+        let query = query_parser.parse_query("cart.product_id:103")?;
+        let count_docs = searcher.search(&*query, &Count)?;
+        assert_eq!(count_docs, 1);
+    }
+    {
+        let query = query_parser.parse_query("click AND cart.product_id:133")?;
+        let hits = searcher.search(&*query, &TopDocs::with_limit(2))?;
+        assert_eq!(hits.len(), 1);
+    }
+    {
+        // The sub-fields in the json field marked as default field still need to be explicitly
+        // addressed
+        let query = query_parser.parse_query("click AND 133")?;
+        let hits = searcher.search(&*query, &TopDocs::with_limit(2))?;
+        assert_eq!(hits.len(), 0);
+    }
+    {
+        // Default json fields are ignored if they collide with the schema
+        let query = query_parser.parse_query("event_type:holiday-sale")?;
+        let hits = searcher.search(&*query, &TopDocs::with_limit(2))?;
+        assert_eq!(hits.len(), 0);
+    }
+    // # Query via full attribute path
+    {
+        // This only searches in our schema's `event_type` field
+        let query = query_parser.parse_query("event_type:click")?;
+        let hits = searcher.search(&*query, &TopDocs::with_limit(2))?;
+        assert_eq!(hits.len(), 2);
+    }
+    {
+        // Default json fields can still be accessed by full path
+        let query = query_parser.parse_query("attributes.event_type:holiday-sale")?;
+        let hits = searcher.search(&*query, &TopDocs::with_limit(2))?;
+        assert_eq!(hits.len(), 1);
+    }
+    Ok(())
+}

examples/multiple_producer.rs

@@ -0,0 +1,104 @@
+// # Indexing from different threads.
+//
+// It is fairly common to have to index from different threads.
+// Tantivy forbids to create more than one `IndexWriter` at a time.
+//
+// This `IndexWriter` itself has its own multithreaded layer, so managing your own
+// indexing threads will not help. However, it can still be useful for some applications.
+//
+// For instance, if preparing documents to send to tantivy before indexing is the bottleneck of
+// your application, it is reasonable to have multiple threads.
+//
+// Another very common reason to want to index from multiple threads, is implementing a webserver
+// with CRUD capabilities. The server framework will most likely handle request from
+// different threads.
+//
+// The recommended way to address both of these use case is to wrap your `IndexWriter` into a
+// `Arc<RwLock<IndexWriter>>`.
+//
+// While this is counterintuitive, adding and deleting documents do not require mutability
+// over the `IndexWriter`, so several threads will be able to do this operation concurrently.
+//
+// The example below does not represent an actual real-life use case (who would spawn thread to
+// index a single document?), but aims at demonstrating the mechanism that makes indexing
+// from several threads possible.
+
+// ---
+// Importing tantivy...
+use std::sync::{Arc, RwLock};
+use std::thread;
+use std::time::Duration;
+
+use tantivy::schema::{Schema, STORED, TEXT};
+use tantivy::{doc, Index, IndexWriter, Opstamp, TantivyError};
+
+fn main() -> tantivy::Result<()> {
+    // # Defining the schema
+    let mut schema_builder = Schema::builder();
+    let title = schema_builder.add_text_field("title", TEXT | STORED);
+    let body = schema_builder.add_text_field("body", TEXT);
+    let schema = schema_builder.build();
+
+    let index = Index::create_in_ram(schema);
+    let index_writer: Arc<RwLock<IndexWriter>> = Arc::new(RwLock::new(index.writer(50_000_000)?));
+
+    // # First indexing thread.
+    let index_writer_clone_1 = index_writer.clone();
+    thread::spawn(move || {
+        // we index 100 times the document... for the sake of the example.
+        for i in 0..100 {
+            let opstamp = index_writer_clone_1
+                .read().unwrap() //< A read lock is sufficient here.
+                .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"
+                    ))?;
+            println!("add doc {} from thread 1 - opstamp {}", i, opstamp);
+            thread::sleep(Duration::from_millis(20));
+        }
+        Result::<(), TantivyError>::Ok(())
+    });
+
+    // # Second indexing thread.
+    let index_writer_clone_2 = index_writer.clone();
+    // For convenience, tantivy also comes with a macro to
+    // reduce the boilerplate above.
+    thread::spawn(move || {
+        // we index 100 times the document... for the sake of the example.
+        for i in 0..100 {
+            // A read lock is sufficient here.
+            let opstamp = {
+                let index_writer_rlock = index_writer_clone_2.read().unwrap();
+                index_writer_rlock.add_document(doc!(
+                    title => "Manufacturing consent",
+                    body => "Some great book description..."
+                ))?
+            };
+            println!("add doc {} from thread 2 - opstamp {}", i, opstamp);
+            thread::sleep(Duration::from_millis(10));
+        }
+        Result::<(), TantivyError>::Ok(())
+    });
+
+    // # In the main thread, we commit 10 times, once every 500ms.
+    for _ in 0..10 {
+        let opstamp: Opstamp = {
+            // Committing or rollbacking on the other hand requires write lock. This will block
+            // other threads.
+            let mut index_writer_wlock = index_writer.write().unwrap();
+            index_writer_wlock.commit()?
+        };
+        println!("committed with opstamp {}", opstamp);
+        thread::sleep(Duration::from_millis(500));
+    }
+
+    Ok(())
+}

examples/pre_tokenized_text.rs

@@ -0,0 +1,135 @@
+// # 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))?;
+
+    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))?;
+
+    assert_eq!(count, 0);
+
+    Ok(())
+}

examples/simple_search.rs

@@ -1,225 +0,0 @@
-extern crate tantivy;
-extern crate tempdir;
-
-#[macro_use]
-extern crate serde_json;
-
-use std::path::Path;
-use tempdir::TempDir;
-use tantivy::Index;
-use tantivy::schema::*;
-use tantivy::collector::TopCollector;
-use tantivy::query::QueryParser;
-
-fn main() {
-    // Let's create a temporary directory for the
-    // sake of this example
-    if let Ok(dir) = TempDir::new("tantivy_example_dir") {
-        run_example(dir.path()).unwrap();
-        dir.close().unwrap();
-    }
-}
-
-fn run_example(index_path: &Path) -> tantivy::Result<()> {
-    // # Defining the schema
-    //
-    // The Tantivy index requires a very strict schema.
-    // The schema declares which fields are in the index,
-    // and for each field, its type and "the way it should
-    // be indexed".
-
-    // first we need to define a schema ...
-    let mut schema_builder = SchemaBuilder::default();
-
-    // Our first field is title.
-    // We want full-text search for it, and we also want
-    // to be able to retrieve the document after the search.
-    //
-    // TEXT | STORED is some syntactic sugar to describe
-    // that.
-    //
-    // `TEXT` means the field should be tokenized and indexed,
-    // along with its term frequency and term positions.
-    //
-    // `STORED` means that the field will also be saved
-    // in a compressed, row-oriented key-value store.
-    // This store is useful to reconstruct the
-    // documents that were selected during the search phase.
-    schema_builder.add_text_field("title", TEXT | STORED);
-
-    // Our second field is body.
-    // We want full-text search for it, but we do not
-    // need to be able to be able to retrieve it
-    // for our application.
-    //
-    // We can make our index lighter and
-    // by omitting `STORED` flag.
-    schema_builder.add_text_field("body", TEXT);
-
-    let schema = schema_builder.build();
-
-    // # Indexing documents
-    //
-    // Let's create a brand new index.
-    //
-    // This will actually just save a meta.json
-    // with our schema in the directory.
-    let index = Index::create(index_path, schema.clone())?;
-
-    // To insert document we need an index writer.
-    // There must be only one writer at a time.
-    // This single `IndexWriter` is already
-    // multithreaded.
-    //
-    // Here we use a buffer of 50MB per thread. Using a bigger
-    // heap for the indexer can increase its throughput.
-    let mut index_writer = index.writer(50_000_000)?;
-
-    // Let's index our documents!
-    // We first need a handle on the title and the body field.
-
-    // ### Create a document "manually".
-    //
-    // 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 mut old_man_doc = Document::default();
-    old_man_doc.add_text(title, "The Old Man and the Sea");
-    old_man_doc.add_text(
-        body,
-        "He was an old man who fished alone in a skiff in the Gulf Stream and \
-         he had gone eighty-four days now without taking a fish.",
-    );
-
-    // ... and add it to the `IndexWriter`.
-    index_writer.add_document(old_man_doc);
-
-    // ### Create a document directly from json.
-    //
-    // Alternatively, we can use our schema to parse a
-    // document object directly from json.
-    // The document is a string, but we use the `json` macro
-    // from `serde_json` for the convenience of multi-line support.
-    let json = json!({
-       "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"
-    });
-    let mice_and_men_doc = schema.parse_document(&json.to_string())?;
-
-    index_writer.add_document(mice_and_men_doc);
-
-    // Multi-valued field are allowed, they are
-    // expressed in JSON by an array.
-    // The following document has two titles.
-    let json = json!({
-       "title": ["Frankenstein", "The Modern Prometheus"],
-       "body": "You will rejoice to hear that no disaster has accompanied the commencement of an \
-                enterprise which you have regarded with such evil forebodings.  I arrived here \
-                yesterday, and my first task is to assure my dear sister of my welfare and \
-                increasing confidence in the success of my undertaking."
-    });
-    let frankenstein_doc = schema.parse_document(&json.to_string())?;
-
-    index_writer.add_document(frankenstein_doc);
-
-    // This is an example, so we will only index 3 documents
-    // here. You can check out tantivy's tutorial to index
-    // the English wikipedia. Tantivy's indexing is rather fast.
-    // Indexing 5 million articles of the English wikipedia takes
-    // around 4 minutes on my computer!
-
-    // ### Committing
-    //
-    // At this point our documents are not searchable.
-    //
-    //
-    // We need to call .commit() explicitly to force the
-    // index_writer to finish processing the documents in the queue,
-    // flush the current index to the disk, and advertise
-    // the existence of new documents.
-    //
-    // This call is blocking.
-    index_writer.commit()?;
-
-    // If `.commit()` returns correctly, then all of the
-    // documents that have been added are guaranteed to be
-    // persistently indexed.
-    //
-    // In the scenario of a crash or a power failure,
-    // tantivy behaves as if has rolled back to its last
-    // commit.
-
-    // # Searching
-    //
-    // Let's search our index. Start by reloading
-    // searchers in the index. This should be done
-    // after every commit().
-    index.load_searchers()?;
-
-    // Afterwards create one (or more) searchers.
-    //
-    // You should create a searcher
-    // every time you start a "search query".
-    let searcher = index.searcher();
-
-    // The query parser can interpret human queries.
-    // Here, if the user does not specify which
-    // field they want to search, tantivy will search
-    // 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
-    // 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")?;
-
-    // A query defines a set of documents, as
-    // well as the way they should be scored.
-    //
-    // A query created by the query parser is scored according
-    // to a metric called Tf-Idf, and will consider
-    // any document matching at least one of our terms.
-
-    // ### Collectors
-    //
-    // 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 TopCollector.
-    let mut top_collector = TopCollector::with_limit(10);
-
-    // We can now perform our query.
-    searcher.search(&*query, &mut top_collector)?;
-
-    // Our top collector now contains the 10
-    // most relevant doc ids...
-    let doc_addresses = top_collector.docs();
-
-    // The actual documents still need to be
-    // retrieved from Tantivy's store.
-    //
-    // Since the body field was not configured as stored,
-    // the document returned will only contain
-    // a title.
-
-    for doc_address in doc_addresses {
-        let retrieved_doc = searcher.doc(&doc_address)?;
-        println!("{}", schema.to_json(&retrieved_doc));
-    }
-
-    // Wait for indexing and merging threads to shut down.
-    // Usually this isn't needed, but in `main` we try to
-    // delete the temporary directory and that fails on
-    // Windows if the files are still open.
-    index_writer.wait_merging_threads()?;
-
-    Ok(())
-}

examples/snippet.rs

@@ -0,0 +1,85 @@
+// # Snippet example
+//
+// This example shows how to return a representative snippet of
+// your hit result.
+// Snippet are an extracted of a target document, and returned in HTML format.
+// The keyword searched by the user are highlighted with a `<b>` tag.
+
+// ---
+// Importing tantivy...
+use tantivy::collector::TopDocs;
+use tantivy::query::QueryParser;
+use tantivy::schema::*;
+use tantivy::{doc, Index, Snippet, SnippetGenerator};
+use tempfile::TempDir;
+
+fn main() -> tantivy::Result<()> {
+    // Let's create a temporary directory for the
+    // sake of this example
+    let index_path = TempDir::new()?;
+
+    // # Defining the schema
+    let mut schema_builder = Schema::builder();
+    let title = schema_builder.add_text_field("title", TEXT | STORED);
+    let body = schema_builder.add_text_field("body", TEXT | STORED);
+    let schema = schema_builder.build();
+
+    // # Indexing documents
+    let index = Index::create_in_dir(&index_path, schema)?;
+
+    let mut index_writer = index.writer(50_000_000)?;
+
+    // we'll only need one doc for this example.
+    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"
+    ))?;
+    // ...
+    index_writer.commit()?;
+
+    let reader = index.reader()?;
+    let searcher = reader.searcher();
+    let query_parser = QueryParser::for_index(&index, vec![title, body]);
+    let query = query_parser.parse_query("sycamore spring")?;
+
+    let top_docs = searcher.search(&query, &TopDocs::with_limit(10))?;
+
+    let snippet_generator = SnippetGenerator::create(&searcher, &*query, body)?;
+
+    for (score, doc_address) in top_docs {
+        let doc = searcher.doc(doc_address)?;
+        let snippet = snippet_generator.snippet_from_doc(&doc);
+        println!("Document score {}:", score);
+        println!(
+            "title: {}",
+            doc.get_first(title).unwrap().as_text().unwrap()
+        );
+        println!("snippet: {}", snippet.to_html());
+        println!("custom highlighting: {}", highlight(snippet));
+    }
+
+    Ok(())
+}
+
+fn highlight(snippet: Snippet) -> String {
+    let mut result = String::new();
+    let mut start_from = 0;
+
+    for fragment_range in snippet.highlighted() {
+        result.push_str(&snippet.fragment()[start_from..fragment_range.start]);
+        result.push_str(" --> ");
+        result.push_str(&snippet.fragment()[fragment_range.clone()]);
+        result.push_str(" <-- ");
+        start_from = fragment_range.end;
+    }
+
+    result.push_str(&snippet.fragment()[start_from..]);
+    result
+}

examples/stop_words.rs

@@ -0,0 +1,113 @@
+// # Stop Words Example
+//
+// This example covers the basic usage of stop words
+// with tantivy
+//
+// We will :
+// - define our schema
+// - create an index in a directory
+// - add a few stop words
+// - index few documents in our index
+
+// ---
+// Importing tantivy...
+use tantivy::collector::TopDocs;
+use tantivy::query::QueryParser;
+use tantivy::schema::*;
+use tantivy::tokenizer::*;
+use tantivy::{doc, Index};
+
+fn main() -> tantivy::Result<()> {
+    // this example assumes you understand the content in `basic_search`
+    let mut schema_builder = Schema::builder();
+
+    // This configures your custom options for how tantivy will
+    // store and process your content in the index; The key
+    // to note is that we are setting the tokenizer to `stoppy`
+    // which will be defined and registered below.
+    let text_field_indexing = TextFieldIndexing::default()
+        .set_tokenizer("stoppy")
+        .set_index_option(IndexRecordOption::WithFreqsAndPositions);
+    let text_options = TextOptions::default()
+        .set_indexing_options(text_field_indexing)
+        .set_stored();
+
+    // Our first field is title.
+    schema_builder.add_text_field("title", text_options);
+
+    // Our second field is body.
+    let text_field_indexing = TextFieldIndexing::default()
+        .set_tokenizer("stoppy")
+        .set_index_option(IndexRecordOption::WithFreqsAndPositions);
+    let text_options = TextOptions::default()
+        .set_indexing_options(text_field_indexing)
+        .set_stored();
+    schema_builder.add_text_field("body", text_options);
+
+    let schema = schema_builder.build();
+
+    let index = Index::create_in_ram(schema.clone());
+
+    // 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 = TextAnalyzer::from(SimpleTokenizer)
+        .filter(LowerCaser)
+        .filter(StopWordFilter::remove(vec![
+            "the".to_string(),
+            "and".to_string(),
+        ]));
+
+    index.tokenizers().register("stoppy", tokenizer);
+
+    let mut index_writer = index.writer(50_000_000)?;
+
+    let title = schema.get_field("title").unwrap();
+    let body = schema.get_field("body").unwrap();
+
+    index_writer.add_document(doc!(
+    title => "The Old Man and the Sea",
+    body => "He was an old man who fished alone in a skiff in the Gulf Stream and \
+     he had gone eighty-four days now without taking a fish."
+    ))?;
+
+    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"
+    ))?;
+
+    index_writer.add_document(doc!(
+    title => "Frankenstein",
+    body => "You will rejoice to hear that no disaster has accompanied the commencement of an \
+             enterprise which you have regarded with such evil forebodings.  I arrived here \
+             yesterday, and my first task is to assure my dear sister of my welfare and \
+             increasing confidence in the success of my undertaking."
+    ))?;
+
+    index_writer.commit()?;
+
+    let reader = index.reader()?;
+
+    let searcher = reader.searcher();
+
+    let query_parser = QueryParser::for_index(&index, vec![title, body]);
+
+    // stop words are applied on the query as well.
+    // The following will be equivalent to `title:frankenstein`
+    let query = query_parser.parse_query("title:\"the Frankenstein\"")?;
+    let top_docs = searcher.search(&query, &TopDocs::with_limit(10))?;
+
+    for (score, doc_address) in top_docs {
+        let retrieved_doc = searcher.doc(doc_address)?;
+        println!("\n==\nDocument score {}:", score);
+        println!("{}", schema.to_json(&retrieved_doc));
+    }
+
+    Ok(())
+}

examples/warmer.rs

@@ -0,0 +1,219 @@
+use std::cmp::Reverse;
+use std::collections::{HashMap, HashSet};
+use std::sync::{Arc, RwLock, Weak};
+
+use tantivy::collector::TopDocs;
+use tantivy::query::QueryParser;
+use tantivy::schema::{Field, Schema, FAST, TEXT};
+use tantivy::{
+    doc, DocAddress, DocId, Index, IndexReader, Opstamp, Searcher, SearcherGeneration, SegmentId,
+    SegmentReader, Warmer,
+};
+
+// This example shows how warmers can be used to
+// load a values from an external sources using the Warmer API.
+//
+// In this example, we assume an e-commerce search engine.
+
+type ProductId = u64;
+
+/// Price
+type Price = u32;
+
+pub trait PriceFetcher: Send + Sync + 'static {
+    fn fetch_prices(&self, product_ids: &[ProductId]) -> Vec<Price>;
+}
+
+struct DynamicPriceColumn {
+    field: Field,
+    price_cache: RwLock<HashMap<(SegmentId, Option<Opstamp>), Arc<Vec<Price>>>>,
+    price_fetcher: Box<dyn PriceFetcher>,
+}
+
+impl DynamicPriceColumn {
+    pub fn with_product_id_field<T: PriceFetcher>(field: Field, price_fetcher: T) -> Self {
+        DynamicPriceColumn {
+            field,
+            price_cache: Default::default(),
+            price_fetcher: Box::new(price_fetcher),
+        }
+    }
+
+    pub fn price_for_segment(&self, segment_reader: &SegmentReader) -> Option<Arc<Vec<Price>>> {
+        let segment_key = (segment_reader.segment_id(), segment_reader.delete_opstamp());
+        self.price_cache.read().unwrap().get(&segment_key).cloned()
+    }
+}
+impl Warmer for DynamicPriceColumn {
+    fn warm(&self, searcher: &Searcher) -> tantivy::Result<()> {
+        for segment in searcher.segment_readers() {
+            let key = (segment.segment_id(), segment.delete_opstamp());
+            let product_id_reader = segment.fast_fields().u64(self.field)?;
+            let product_ids: Vec<ProductId> = segment
+                .doc_ids_alive()
+                .map(|doc| product_id_reader.get_val(doc as u64))
+                .collect();
+            let mut prices_it = self.price_fetcher.fetch_prices(&product_ids).into_iter();
+            let mut price_vals: Vec<Price> = Vec::new();
+            for doc in 0..segment.max_doc() {
+                if segment.is_deleted(doc) {
+                    price_vals.push(0);
+                } else {
+                    price_vals.push(prices_it.next().unwrap())
+                }
+            }
+            self.price_cache
+                .write()
+                .unwrap()
+                .insert(key, Arc::new(price_vals));
+        }
+        Ok(())
+    }
+
+    fn garbage_collect(&self, live_generations: &[&SearcherGeneration]) {
+        let live_segment_id_and_delete_ops: HashSet<(SegmentId, Option<Opstamp>)> =
+            live_generations
+                .iter()
+                .flat_map(|gen| gen.segments())
+                .map(|(&segment_id, &opstamp)| (segment_id, opstamp))
+                .collect();
+        let mut price_cache_wrt = self.price_cache.write().unwrap();
+        // let price_cache = std::mem::take(&mut *price_cache_wrt);
+        // Drain would be nicer here.
+        *price_cache_wrt = std::mem::take(&mut *price_cache_wrt)
+            .into_iter()
+            .filter(|(seg_id_and_op, _)| !live_segment_id_and_delete_ops.contains(seg_id_and_op))
+            .collect();
+    }
+}
+
+/// For the sake of this example, the table is just an editable HashMap behind a RwLock.
+/// This map represents a map (ProductId -> Price)
+///
+/// In practise, it could be fetching things from an external service, like a SQL table.
+#[derive(Default, Clone)]
+pub struct ExternalPriceTable {
+    prices: Arc<RwLock<HashMap<ProductId, Price>>>,
+}
+
+impl ExternalPriceTable {
+    pub fn update_price(&self, product_id: ProductId, price: Price) {
+        let mut prices_wrt = self.prices.write().unwrap();
+        prices_wrt.insert(product_id, price);
+    }
+}
+
+impl PriceFetcher for ExternalPriceTable {
+    fn fetch_prices(&self, product_ids: &[ProductId]) -> Vec<Price> {
+        let prices_read = self.prices.read().unwrap();
+        product_ids
+            .iter()
+            .map(|product_id| prices_read.get(product_id).cloned().unwrap_or(0))
+            .collect()
+    }
+}
+
+fn main() -> tantivy::Result<()> {
+    // Declaring our schema.
+    let mut schema_builder = Schema::builder();
+    // The product id is assumed to be a primary id for our external price source.
+    let product_id = schema_builder.add_u64_field("product_id", FAST);
+    let text = schema_builder.add_text_field("text", TEXT);
+    let schema: Schema = schema_builder.build();
+
+    let price_table = ExternalPriceTable::default();
+    let price_dynamic_column = Arc::new(DynamicPriceColumn::with_product_id_field(
+        product_id,
+        price_table.clone(),
+    ));
+    price_table.update_price(OLIVE_OIL, 12);
+    price_table.update_price(GLOVES, 13);
+    price_table.update_price(SNEAKERS, 80);
+
+    const OLIVE_OIL: ProductId = 323423;
+    const GLOVES: ProductId = 3966623;
+    const SNEAKERS: ProductId = 23222;
+
+    let index = Index::create_in_ram(schema);
+    let mut writer = index.writer_with_num_threads(1, 10_000_000)?;
+    writer.add_document(doc!(product_id=>OLIVE_OIL, text=>"cooking olive oil from greece"))?;
+    writer.add_document(doc!(product_id=>GLOVES, text=>"kitchen gloves, perfect for cooking"))?;
+    writer.add_document(doc!(product_id=>SNEAKERS, text=>"uber sweet sneakers"))?;
+    writer.commit()?;
+
+    let warmers: Vec<Weak<dyn Warmer>> = vec![Arc::downgrade(
+        &(price_dynamic_column.clone() as Arc<dyn Warmer>),
+    )];
+    let reader: IndexReader = index.reader_builder().warmers(warmers).try_into()?;
+    reader.reload()?;
+
+    let query_parser = QueryParser::for_index(&index, vec![text]);
+    let query = query_parser.parse_query("cooking")?;
+
+    let searcher = reader.searcher();
+    let score_by_price = move |segment_reader: &SegmentReader| {
+        let price = price_dynamic_column
+            .price_for_segment(segment_reader)
+            .unwrap();
+        move |doc_id: DocId| Reverse(price[doc_id as usize])
+    };
+
+    let most_expensive_first = TopDocs::with_limit(10).custom_score(score_by_price);
+
+    let hits = searcher.search(&query, &most_expensive_first)?;
+    assert_eq!(
+        &hits,
+        &[
+            (
+                Reverse(12u32),
+                DocAddress {
+                    segment_ord: 0,
+                    doc_id: 0u32
+                }
+            ),
+            (
+                Reverse(13u32),
+                DocAddress {
+                    segment_ord: 0,
+                    doc_id: 1u32
+                }
+            ),
+        ]
+    );
+
+    // Olive oil just got more expensive!
+    price_table.update_price(OLIVE_OIL, 15);
+
+    // The price update are directly reflected on `reload`.
+    //
+    // Be careful here though!...
+    // You may have spotted that we are still using the same `Searcher`.
+    //
+    // It is up to the `Warmer` implementer to decide how
+    // to control this behavior.
+
+    reader.reload()?;
+
+    let hits_with_new_prices = searcher.search(&query, &most_expensive_first)?;
+    assert_eq!(
+        &hits_with_new_prices,
+        &[
+            (
+                Reverse(13u32),
+                DocAddress {
+                    segment_ord: 0,
+                    doc_id: 1u32
+                }
+            ),
+            (
+                Reverse(15u32),
+                DocAddress {
+                    segment_ord: 0,
+                    doc_id: 0u32
+                }
+            ),
+        ]
+    );
+
+    Ok(())
+}

examples/working_with_json.rs

@@ -0,0 +1,40 @@
+use tantivy::schema::*;
+
+// # Document from json
+//
+// For convenience, `Document` can be parsed directly from json.
+fn main() -> tantivy::Result<()> {
+    // Let's first define a schema and an index.
+    // Check out the basic example if this is confusing to you.
+    //
+    // first we need to define a schema ...
+    let mut schema_builder = Schema::builder();
+    schema_builder.add_text_field("title", TEXT | STORED);
+    schema_builder.add_text_field("body", TEXT);
+    schema_builder.add_u64_field("year", INDEXED);
+    let schema = schema_builder.build();
+
+    // Let's assume we have a json-serialized document.
+    let mice_and_men_doc_json = r#"{
+       "title": "Of Mice and Men",
+       "year": 1937
+    }"#;
+
+    // We can parse our document
+    let _mice_and_men_doc = schema.parse_document(mice_and_men_doc_json)?;
+
+    // Multi-valued field are allowed, they are
+    // expressed in JSON by an array.
+    // The following document has two titles.
+    let frankenstein_json = r#"{
+       "title": ["Frankenstein", "The Modern Prometheus"],
+       "year": 1818
+    }"#;
+    let _frankenstein_doc = schema.parse_document(frankenstein_json)?;
+
+    // Note that the schema is saved in your index directory.
+    //
+    // As a result, Indexes are aware of their schema, and you can use this feature
+    // just by opening an existing `Index`, and calling `index.schema()..parse_document(json)`.
+    Ok(())
+}

fastfield_codecs/Cargo.toml

@@ -0,0 +1,31 @@
+[package]
+name = "fastfield_codecs"
+version = "0.2.0"
+authors = ["Pascal Seitz <pascal@quickwit.io>"]
+license = "MIT"
+edition = "2021"
+description = "Fast field codecs used by tantivy"
+
+# See more keys and their definitions at https://doc.rust-lang.org/cargo/reference/manifest.html
+
+[dependencies]
+common = { version = "0.3", path = "../common/", package = "tantivy-common" }
+tantivy-bitpacker = { version="0.2", path = "../bitpacker/" }
+ownedbytes = { version = "0.3.0", path = "../ownedbytes" }
+prettytable-rs = {version="0.9.0", optional= true}
+rand = {version="0.8.3", optional= true}
+fastdivide = "0.4"
+log = "0.4"
+itertools = { version = "0.10.3" }
+measure_time = { version="0.8.2", optional=true}
+
+[dev-dependencies]
+more-asserts = "0.3.0"
+proptest = "1.0.0"
+rand = "0.8.3"
+
+[features]
+bin = ["prettytable-rs", "rand", "measure_time"]
+default = ["bin"]
+unstable = []
+

fastfield_codecs/README.md

@@ -0,0 +1,68 @@
+
+
+# Fast Field Codecs
+
+This crate contains various fast field codecs, used to compress/decompress fast field data in tantivy.
+
+## Contributing
+
+Contributing is pretty straightforward. Since the bitpacking is the simplest compressor, you can check it for reference.
+
+A codec needs to implement 2 traits:
+
+- A reader implementing `FastFieldCodecReader` to read the codec.
+- A serializer implementing `FastFieldCodecSerializer` for compression estimation and codec name + id.
+
+### Tests
+
+Once the traits are implemented test and benchmark integration is pretty easy (see `test_with_codec_data_sets` and `bench.rs`).
+
+Make sure to add the codec to the main.rs, which tests the compression ratio and estimation against different data sets. You can run it with:
+```
+cargo run --features bin
+```
+
+### TODO
+- Add real world data sets in comparison
+- Add codec to cover sparse data sets
+
+
+### Codec Comparison
+```
++----------------------------------+-------------------+------------------------+
+|                                  | Compression Ratio | Compression Estimation |
++----------------------------------+-------------------+------------------------+
+| Autoincrement                    |                   |                        |
++----------------------------------+-------------------+------------------------+
+| LinearInterpol                   | 0.000039572664    | 0.000004396963         |
++----------------------------------+-------------------+------------------------+
+| MultiLinearInterpol              | 0.1477348         | 0.17275847             |
++----------------------------------+-------------------+------------------------+
+| Bitpacked                        | 0.28126493        | 0.28125                |
++----------------------------------+-------------------+------------------------+
+| Monotonically increasing concave |                   |                        |
++----------------------------------+-------------------+------------------------+
+| LinearInterpol                   | 0.25003937        | 0.26562938             |
++----------------------------------+-------------------+------------------------+
+| MultiLinearInterpol              | 0.190665          | 0.1883836              |
++----------------------------------+-------------------+------------------------+
+| Bitpacked                        | 0.31251436        | 0.3125                 |
++----------------------------------+-------------------+------------------------+
+| Monotonically increasing convex  |                   |                        |
++----------------------------------+-------------------+------------------------+
+| LinearInterpol                   | 0.25003937        | 0.28125438             |
++----------------------------------+-------------------+------------------------+
+| MultiLinearInterpol              | 0.18676           | 0.2040086              |
++----------------------------------+-------------------+------------------------+
+| Bitpacked                        | 0.31251436        | 0.3125                 |
++----------------------------------+-------------------+------------------------+
+| Almost monotonically increasing  |                   |                        |
++----------------------------------+-------------------+------------------------+
+| LinearInterpol                   | 0.14066513        | 0.1562544              |
++----------------------------------+-------------------+------------------------+
+| MultiLinearInterpol              | 0.16335973        | 0.17275847             |
++----------------------------------+-------------------+------------------------+
+| Bitpacked                        | 0.28126493        | 0.28125                |
++----------------------------------+-------------------+------------------------+
+
+```

fastfield_codecs/benches/bench.rs

@@ -0,0 +1,226 @@
+#![feature(test)]
+
+extern crate test;
+
+#[cfg(test)]
+mod tests {
+    use std::iter;
+    use std::sync::Arc;
+
+    use fastfield_codecs::*;
+    use ownedbytes::OwnedBytes;
+    use rand::prelude::*;
+    use test::Bencher;
+
+    use super::*;
+
+    // Warning: this generates the same permutation at each call
+    fn generate_permutation() -> Vec<u64> {
+        let mut permutation: Vec<u64> = (0u64..100_000u64).collect();
+        permutation.shuffle(&mut StdRng::from_seed([1u8; 32]));
+        permutation
+    }
+
+    fn generate_random() -> Vec<u64> {
+        let mut permutation: Vec<u64> = (0u64..100_000u64)
+            .map(|el| el + random::<u16>() as u64)
+            .collect();
+        permutation.shuffle(&mut StdRng::from_seed([1u8; 32]));
+        permutation
+    }
+
+    // Warning: this generates the same permutation at each call
+    fn generate_permutation_gcd() -> Vec<u64> {
+        let mut permutation: Vec<u64> = (1u64..100_000u64).map(|el| el * 1000).collect();
+        permutation.shuffle(&mut StdRng::from_seed([1u8; 32]));
+        permutation
+    }
+
+    pub fn serialize_and_load<T: MonotonicallyMappableToU64 + Ord + Default>(
+        column: &[T],
+    ) -> Arc<dyn Column<T>> {
+        let mut buffer = Vec::new();
+        serialize(VecColumn::from(&column), &mut buffer, &ALL_CODEC_TYPES).unwrap();
+        open(OwnedBytes::new(buffer)).unwrap()
+    }
+
+    #[bench]
+    fn bench_intfastfield_jumpy_veclookup(b: &mut Bencher) {
+        let permutation = generate_permutation();
+        let n = permutation.len();
+        b.iter(|| {
+            let mut a = 0u64;
+            for _ in 0..n {
+                a = permutation[a as usize];
+            }
+            a
+        });
+    }
+
+    #[bench]
+    fn bench_intfastfield_jumpy_fflookup(b: &mut Bencher) {
+        let permutation = generate_permutation();
+        let n = permutation.len();
+        let column: Arc<dyn Column<u64>> = serialize_and_load(&permutation);
+        b.iter(|| {
+            let mut a = 0u64;
+            for _ in 0..n {
+                a = column.get_val(a as u64);
+            }
+            a
+        });
+    }
+
+    fn get_exp_data() -> Vec<u64> {
+        let mut data = vec![];
+        for i in 0..100 {
+            let num = i * i;
+            data.extend(iter::repeat(i as u64).take(num));
+        }
+        data.shuffle(&mut StdRng::from_seed([1u8; 32]));
+
+        // lengt = 328350
+        data
+    }
+
+    fn get_data_50percent_item() -> (u128, u128, Vec<u128>) {
+        let mut permutation = get_exp_data();
+        let major_item = 20;
+        let minor_item = 10;
+        permutation.extend(iter::repeat(major_item).take(permutation.len()));
+        permutation.shuffle(&mut StdRng::from_seed([1u8; 32]));
+        let permutation = permutation.iter().map(|el| *el as u128).collect::<Vec<_>>();
+        (major_item as u128, minor_item as u128, permutation)
+    }
+    fn get_u128_column_random() -> Arc<dyn Column<u128>> {
+        let permutation = generate_random();
+        let permutation = permutation.iter().map(|el| *el as u128).collect::<Vec<_>>();
+        get_u128_column_from_data(&permutation)
+    }
+
+    fn get_u128_column_from_data(data: &[u128]) -> Arc<dyn Column<u128>> {
+        let mut out = vec![];
+        let iter_gen = || data.iter().cloned();
+        serialize_u128(iter_gen, data.len() as u64, &mut out).unwrap();
+        let out = OwnedBytes::new(out);
+        open_u128::<u128>(out).unwrap()
+    }
+
+    #[bench]
+    fn bench_intfastfield_getrange_u128_50percent_hit(b: &mut Bencher) {
+        let (major_item, _minor_item, data) = get_data_50percent_item();
+        let column = get_u128_column_from_data(&data);
+
+        b.iter(|| column.get_between_vals(major_item..=major_item));
+    }
+
+    #[bench]
+    fn bench_intfastfield_getrange_u128_single_hit(b: &mut Bencher) {
+        let (_major_item, minor_item, data) = get_data_50percent_item();
+        let column = get_u128_column_from_data(&data);
+
+        b.iter(|| column.get_between_vals(minor_item..=minor_item));
+    }
+
+    #[bench]
+    fn bench_intfastfield_getrange_u128_hit_all(b: &mut Bencher) {
+        let (_major_item, _minor_item, data) = get_data_50percent_item();
+        let column = get_u128_column_from_data(&data);
+
+        b.iter(|| column.get_between_vals(0..=u128::MAX));
+    }
+
+    #[bench]
+    fn bench_intfastfield_scan_all_fflookup_u128(b: &mut Bencher) {
+        let column = get_u128_column_random();
+
+        b.iter(|| {
+            let mut a = 0u128;
+            for i in 0u64..column.num_vals() as u64 {
+                a += column.get_val(i);
+            }
+            a
+        });
+    }
+
+    #[bench]
+    fn bench_intfastfield_jumpy_stride5_u128(b: &mut Bencher) {
+        let column = get_u128_column_random();
+
+        b.iter(|| {
+            let n = column.num_vals();
+            let mut a = 0u128;
+            for i in (0..n / 5).map(|val| val * 5) {
+                a += column.get_val(i as u64);
+            }
+            a
+        });
+    }
+
+    #[bench]
+    fn bench_intfastfield_stride7_vec(b: &mut Bencher) {
+        let permutation = generate_permutation();
+        let n = permutation.len();
+        b.iter(|| {
+            let mut a = 0u64;
+            for i in (0..n / 7).map(|val| val * 7) {
+                a += permutation[i as usize];
+            }
+            a
+        });
+    }
+
+    #[bench]
+    fn bench_intfastfield_stride7_fflookup(b: &mut Bencher) {
+        let permutation = generate_permutation();
+        let n = permutation.len();
+        let column: Arc<dyn Column<u64>> = serialize_and_load(&permutation);
+        b.iter(|| {
+            let mut a = 0u64;
+            for i in (0..n / 7).map(|val| val * 7) {
+                a += column.get_val(i as u64);
+            }
+            a
+        });
+    }
+
+    #[bench]
+    fn bench_intfastfield_scan_all_fflookup(b: &mut Bencher) {
+        let permutation = generate_permutation();
+        let n = permutation.len();
+        let column: Arc<dyn Column<u64>> = serialize_and_load(&permutation);
+        b.iter(|| {
+            let mut a = 0u64;
+            for i in 0u64..n as u64 {
+                a += column.get_val(i);
+            }
+            a
+        });
+    }
+
+    #[bench]
+    fn bench_intfastfield_scan_all_fflookup_gcd(b: &mut Bencher) {
+        let permutation = generate_permutation_gcd();
+        let n = permutation.len();
+        let column: Arc<dyn Column<u64>> = serialize_and_load(&permutation);
+        b.iter(|| {
+            let mut a = 0u64;
+            for i in 0..n as u64 {
+                a += column.get_val(i);
+            }
+            a
+        });
+    }
+
+    #[bench]
+    fn bench_intfastfield_scan_all_vec(b: &mut Bencher) {
+        let permutation = generate_permutation();
+        b.iter(|| {
+            let mut a = 0u64;
+            for i in 0..permutation.len() {
+                a += permutation[i as usize] as u64;
+            }
+            a
+        });
+    }
+}

fastfield_codecs/src/bitpacked.rs

@@ -0,0 +1,116 @@
+use std::io::{self, Write};
+
+use ownedbytes::OwnedBytes;
+use tantivy_bitpacker::{compute_num_bits, BitPacker, BitUnpacker};
+
+use crate::serialize::NormalizedHeader;
+use crate::{Column, FastFieldCodec, FastFieldCodecType};
+
+/// Depending on the field type, a different
+/// fast field is required.
+#[derive(Clone)]
+pub struct BitpackedReader {
+    data: OwnedBytes,
+    bit_unpacker: BitUnpacker,
+    normalized_header: NormalizedHeader,
+}
+
+impl Column for BitpackedReader {
+    #[inline]
+    fn get_val(&self, doc: u64) -> u64 {
+        self.bit_unpacker.get(doc, &self.data)
+    }
+    #[inline]
+    fn min_value(&self) -> u64 {
+        // The BitpackedReader assumes a normalized vector.
+        0
+    }
+    #[inline]
+    fn max_value(&self) -> u64 {
+        self.normalized_header.max_value
+    }
+    #[inline]
+    fn num_vals(&self) -> u64 {
+        self.normalized_header.num_vals
+    }
+}
+
+pub struct BitpackedCodec;
+
+impl FastFieldCodec for BitpackedCodec {
+    /// The CODEC_TYPE is an enum value used for serialization.
+    const CODEC_TYPE: FastFieldCodecType = FastFieldCodecType::Bitpacked;
+
+    type Reader = BitpackedReader;
+
+    /// Opens a fast field given a file.
+    fn open_from_bytes(
+        data: OwnedBytes,
+        normalized_header: NormalizedHeader,
+    ) -> io::Result<Self::Reader> {
+        let num_bits = compute_num_bits(normalized_header.max_value);
+        let bit_unpacker = BitUnpacker::new(num_bits);
+        Ok(BitpackedReader {
+            data,
+            bit_unpacker,
+            normalized_header,
+        })
+    }
+
+    /// Serializes data with the BitpackedFastFieldSerializer.
+    ///
+    /// The bitpacker assumes that the column has been normalized.
+    /// i.e. It has already been shifted by its minimum value, so that its
+    /// current minimum value is 0.
+    ///
+    /// Ideally, we made a shift upstream on the column so that `col.min_value() == 0`.
+    fn serialize(column: &dyn Column, write: &mut impl Write) -> io::Result<()> {
+        assert_eq!(column.min_value(), 0u64);
+        let num_bits = compute_num_bits(column.max_value());
+        let mut bit_packer = BitPacker::new();
+        for val in column.iter() {
+            bit_packer.write(val, num_bits, write)?;
+        }
+        bit_packer.close(write)?;
+        Ok(())
+    }
+
+    fn estimate(column: &dyn Column) -> Option<f32> {
+        let num_bits = compute_num_bits(column.max_value());
+        let num_bits_uncompressed = 64;
+        Some(num_bits as f32 / num_bits_uncompressed as f32)
+    }
+}
+
+#[cfg(test)]
+mod tests {
+    use super::*;
+    use crate::tests::get_codec_test_datasets;
+
+    fn create_and_validate(data: &[u64], name: &str) {
+        crate::tests::create_and_validate::<BitpackedCodec>(data, name);
+    }
+
+    #[test]
+    fn test_with_codec_data_sets() {
+        let data_sets = get_codec_test_datasets();
+        for (mut data, name) in data_sets {
+            create_and_validate(&data, name);
+            data.reverse();
+            create_and_validate(&data, name);
+        }
+    }
+
+    #[test]
+    fn bitpacked_fast_field_rand() {
+        for _ in 0..500 {
+            let mut data = (0..1 + rand::random::<u8>() as usize)
+                .map(|_| rand::random::<i64>() as u64 / 2)
+                .collect::<Vec<_>>();
+            create_and_validate(&data, "rand");
+
+            data.reverse();
+            create_and_validate(&data, "rand");
+        }
+    }
+}