rust/tantivy
1
0
Fork 0
mirror of https://github.com/quickwit-oss/tantivy.git synced 2025-12-27 20:42:54 +00:00
Code Issues Packages Projects Releases Wiki Activity

Compare commits

base: rust:issue/396
Branches Tags
rust:main rust:stuhood.upstream-erased-sort-computer rust:trinity.pointard/ff-query rust:bucket_id_agg rust:paradedb/fix-agg-validation rust:flatheadmill-geo rust:paradedb-paradedb/fix-overflow-issue rust:low_card_optimization rust:paul.masurel/clippy rust:fix_size_hint rust:qw-airmail-20250916 rust:unit-test-aggregation-sum-other rust:clippy-and-cleanups rust:trinity.pointard/opti-sstable-sorted-ords-to-term rust:release_0.24 rust:dependabot/cargo/rand-0.9.1 rust:dependabot/cargo/lru-0.14.0 rust:dependabot/cargo/fs4-0.13.1 rust:tagged-user-input-ast rust:1686a/parse-agg-res rust:dependabot/cargo/rand_distr-0.5.1 rust:issue/2411-adding-panic-handler-to-rayon-merge-thread-pool rust:dependabot/cargo/nom-8 rust:upgrade_deps rust:test_parser rust:dependabot/github_actions/codecov/codecov-action-5 rust:binggan-0.12.0 rust:agg_format rust:path_collector rust:test_order_bug rust:test_columnar_multi rust:columnar_optionalmask_in_multivalues rust:prefix-phrase-query-optim rust:tantivy-with-oneshot rust:update_example rust:store-reader-enumerate rust:document-deserialize-seed rust:check_doc_mem rust:experimental-do-not-merge rust:expdonotmerge rust:cherry-pick-vec-instead-of-btreemap rust:validate_length rust:raphael_bench_avx_filter rust:raphael_optim_memset2 rust:query-field-name-avoid-alloc rust:check_bug rust:missing_test rust:list_fields rust:block-cache-control rust:trinity--multilayer-sstable rust:tracing-tantivy2 rust:tracing-tantivy rust:index_writer_err rust:default_fast_field_tokenizer rust:fmassot/add-preserve-original-ascii-folding rust:numeric_with_str_col rust:straightforward-simplify rust:fmassot/fake-test-branch rust:fmassot/dynamic-tokenizer-poc-follow-up rust:fmassot/add-box-token-filter-and-refactor rust:dynamic-tokenizer-proof-of-concept rust:fmassot/add-box-tokenizer-and-token-filter rust:tokenizer-api-0.1.1 rust:addconversion rust:removed-extra-enum-alternative-for-datetime-precision rust:issue/1981-reproduce-bm25-bug rust:tokenizer_docs rust:merge-proptests rust:columnar-todo rust:quickwit-0.5 rust:remove-byteorder rust:remove_dyn rust:pr/alexcole/1855-inject-bm25-statistics rust:evance-expose-build-query-from-user-input-ast rust:missing-specialized-method-arcdyncolumn rust:fast-u64-range-query rust:0.19.2 rust:columnar-cli2 rust:trinity--pub-term-as-slice rust:typed-column3 rust:columnar-merge3 rust:columnar-merge-exploring-fn rust:use_columnar rust:typed-column2 rust:term_agg_bucket_limit rust:range rust:typed-column rust:columnar-main rust:forced-types rust:u128-columnar rust:use_stats rust:truncation_comment rust:0.19.1 rust:trinity/yoke-documents rust:column-handle rust:columnar-cli rust:columnar rust:revert-1711-sparse_dense_index rust:commit-changes rust:sparse_dense_index_headroom rust:optional_column rust:no-column-for-multivalued rust:fmassot--bench-hdfs-with-array rust:bugfix-position-broken rust:float rust:simplify rust:debug_time rust:sparse_codecs rust:fix_estimate rust:column-reader rust:column-reader-b rust:columnreader-simple rust:col-trait-refact rust:ip_fastfield rust:column-trait-2 rust:refact-dyn-column rust:column-trait rust:refact-codec-trait rust:pr/PSeitz/1485-1 rust:fastfieldcodec rust:termmap_per_field rust:fastfield-codecs-tests rust:rayon-indexing-writers rust:issue/1251 rust:index-bench rust:fix_open_bytes_index rust:warming rust:revert-1222-issue/1195 rust:add-random-bench-for-fast-field rust:wasm-friendly-exploratory rust:revert-1159-patch-1 rust:debug-position-panic-0.15.3 rust:fix-issue-1111 rust:remove_rand rust:unused-deps rust:hfb/phrase_match_slop rust:0.15.3 rust:expose-min-max-on-multi-fast-field-reader rust:revert-1094-expose-min-max-on-multi-fast-field-reader rust:hotfix-1088 rust:owned-bytes-instead-of-rawdoc rust:rihardsk-date-ranges rust:issue/997 rust:0.13.3 rust:fail-fast-store-checkpoint rust:debugging-store rust:issue/reproduce-897 rust:issue/969-reproduce rust:troublescooter-static rust:intermediate-mergeable-troublescooter rust:barrotsteindev-filter-collector-tpredicatevalue rust:issue/938b rust:issue/938 rust:barrotsteindev-filter-collector rust:issue/922 rust:issue/896 rust:quickfix-explain rust:slog rust:limit-random-seeks rust:0.13.1 rust:issue/866b rust:bugfix-unittest-macos rust:githubactions rust:fieldnorm_readers rust:blockwand rust:nodeffeatfails rust:removedaliveiterator rust:robyoung-introduce-offset-to-top-docs rust:specialization-of-foreach-for-union-scorer rust:elshize-bmw rust:cutt rust:nrt rust:storewriter-out rust:segment_writer_in_ram2 rust:polling rust:audunhalland-query-boost rust:poll-meta rust:segment_writer_in_ram rust:bundle rust:petr-tik-n662_footer_is_incompatible_impl rust:race-condition-in-tests rust:hotfix-0.10.3-issue681 rust:remove-tru rust:kkoziara-pre-tokenized-text rust:issue/681 rust:issue/680 rust:refactoring/field rust:updating/uuid rust:fix-bench rust:hotfix-656 rust:readd-macos-ci rust:incremental-search rust:crate-division rust:streamer-with-state rust:criterion rust:python-binding-changes rust:bump-version rust:gh-pages rust:issue/554 rust:perf/experimental-union-for_each rust:issue/526b rust:softcommits rust:0.9.1-hotfix rust:revert-521-fix-non-english-stemmers rust:issue/500-hotfix rust:hotfix/0.8.2 rust:issue/483 rust:bug/merge-deted rust:issue/407 rust:issue/457 rust:segment_fruits rust:doc rust:dds/lenient rust:issue/webasm rust:issue/396 rust:broken_collector_with_result rust:dss/automaton-weight-suggestion rust:wasm rust:common-crawl rust:tantivy-imhotep
rust:0.25.0 rust:0.22.1 rust:0.24.2 rust:0.24.1 rust:tantivy-query-grammar-v0.24.0 rust:tantivy-columnar-v0.5.0 rust:tantivy-bitpacker-v0.8.0 rust:ownedbytes-v0.9.0 rust:tantivy-stacker-v0.5.0 rust:tantivy-sstable-v0.5.0 rust:tantivy-common-v0.9.0 rust:tantivy-tokenizer-api-v0.5.0 rust:0.24 rust:qw-airmail-20250224-hotfix-merge-panic rust:qw-airmail-20250219-hotfix-merge-panic rust:qw-airmail-202406 rust:qw-airmail-20240611 rust:qw-airmail-20240606 rust:quickwit-rev rust:0.22.0 rust:tantivy-stacker-v0.3.0 rust:ownedbytes-v0.7.0 rust:tantivy-tokenizer-api-v0.3.0 rust:tantivy-columnar-v0.3.0 rust:tantivy-common-v0.7.0 rust:tantivy-query-grammar-v0.22.0 rust:tantivy-sstable-v0.3.0 rust:tantivy-bitpacker-v0.6.0 rust:0.21.1 rust:0.21 rust:0.20.2 rust:0.20.1 rust:0.20 rust:quickwit-0.5-rev rust:0.19.2 rust:0.19.1 rust:0.19 rust:0.18.1 rust:0.18 rust:0.17 rust:0.16.1 rust:0.15.3 rust:0.15.2 rust:0.15.1 rust:0.15 rust:0.14 rust:0.13.3 rust:0.13.2 rust:0.13.1 rust:0.13 rust:0.12 rust:0.11.3 rust:0.11.1 rust:0.11 rust:0.10.3 rust:0.1 rust:0.10.2 rust:0.10.1 rust:0.10.0 rust:0.9.1 rust:0.9 rust:0.8.2 rust:0.8.2b rust:0.8.1 rust:0.8.0 rust:0.7.2 rust:0.7.1 rust:0.7.0 rust:0.6.1 rust:0.6.0 rust:0.5.2 rust:0.5.1 rust:0.5.0 rust:0.4.3 rust:0.4.2 rust:0.4.0 rust:0.3.1 rust:0.3.0 rust:0.2.0
...
compare: rust:perf/experimental-union-for_each
Branches Tags
rust:stuhood.upstream-erased-sort-computer rust:main rust:trinity.pointard/ff-query rust:bucket_id_agg rust:paradedb/fix-agg-validation rust:flatheadmill-geo rust:paradedb-paradedb/fix-overflow-issue rust:low_card_optimization rust:paul.masurel/clippy rust:fix_size_hint rust:qw-airmail-20250916 rust:unit-test-aggregation-sum-other rust:clippy-and-cleanups rust:trinity.pointard/opti-sstable-sorted-ords-to-term rust:release_0.24 rust:dependabot/cargo/rand-0.9.1 rust:dependabot/cargo/lru-0.14.0 rust:dependabot/cargo/fs4-0.13.1 rust:tagged-user-input-ast rust:1686a/parse-agg-res rust:dependabot/cargo/rand_distr-0.5.1 rust:issue/2411-adding-panic-handler-to-rayon-merge-thread-pool rust:dependabot/cargo/nom-8 rust:upgrade_deps rust:test_parser rust:dependabot/github_actions/codecov/codecov-action-5 rust:binggan-0.12.0 rust:agg_format rust:path_collector rust:test_order_bug rust:test_columnar_multi rust:columnar_optionalmask_in_multivalues rust:prefix-phrase-query-optim rust:tantivy-with-oneshot rust:update_example rust:store-reader-enumerate rust:document-deserialize-seed rust:check_doc_mem rust:experimental-do-not-merge rust:expdonotmerge rust:cherry-pick-vec-instead-of-btreemap rust:validate_length rust:raphael_bench_avx_filter rust:raphael_optim_memset2 rust:query-field-name-avoid-alloc rust:check_bug rust:missing_test rust:list_fields rust:block-cache-control rust:trinity--multilayer-sstable rust:tracing-tantivy2 rust:tracing-tantivy rust:index_writer_err rust:default_fast_field_tokenizer rust:fmassot/add-preserve-original-ascii-folding rust:numeric_with_str_col rust:straightforward-simplify rust:fmassot/fake-test-branch rust:fmassot/dynamic-tokenizer-poc-follow-up rust:fmassot/add-box-token-filter-and-refactor rust:dynamic-tokenizer-proof-of-concept rust:fmassot/add-box-tokenizer-and-token-filter rust:tokenizer-api-0.1.1 rust:addconversion rust:removed-extra-enum-alternative-for-datetime-precision rust:issue/1981-reproduce-bm25-bug rust:tokenizer_docs rust:merge-proptests rust:columnar-todo rust:quickwit-0.5 rust:remove-byteorder rust:remove_dyn rust:pr/alexcole/1855-inject-bm25-statistics rust:evance-expose-build-query-from-user-input-ast rust:missing-specialized-method-arcdyncolumn rust:fast-u64-range-query rust:0.19.2 rust:columnar-cli2 rust:trinity--pub-term-as-slice rust:typed-column3 rust:columnar-merge3 rust:columnar-merge-exploring-fn rust:use_columnar rust:typed-column2 rust:term_agg_bucket_limit rust:range rust:typed-column rust:columnar-main rust:forced-types rust:u128-columnar rust:use_stats rust:truncation_comment rust:0.19.1 rust:trinity/yoke-documents rust:column-handle rust:columnar-cli rust:columnar rust:revert-1711-sparse_dense_index rust:commit-changes rust:sparse_dense_index_headroom rust:optional_column rust:no-column-for-multivalued rust:fmassot--bench-hdfs-with-array rust:bugfix-position-broken rust:float rust:simplify rust:debug_time rust:sparse_codecs rust:fix_estimate rust:column-reader rust:column-reader-b rust:columnreader-simple rust:col-trait-refact rust:ip_fastfield rust:column-trait-2 rust:refact-dyn-column rust:column-trait rust:refact-codec-trait rust:pr/PSeitz/1485-1 rust:fastfieldcodec rust:termmap_per_field rust:fastfield-codecs-tests rust:rayon-indexing-writers rust:issue/1251 rust:index-bench rust:fix_open_bytes_index rust:warming rust:revert-1222-issue/1195 rust:add-random-bench-for-fast-field rust:wasm-friendly-exploratory rust:revert-1159-patch-1 rust:debug-position-panic-0.15.3 rust:fix-issue-1111 rust:remove_rand rust:unused-deps rust:hfb/phrase_match_slop rust:0.15.3 rust:expose-min-max-on-multi-fast-field-reader rust:revert-1094-expose-min-max-on-multi-fast-field-reader rust:hotfix-1088 rust:owned-bytes-instead-of-rawdoc rust:rihardsk-date-ranges rust:issue/997 rust:0.13.3 rust:fail-fast-store-checkpoint rust:debugging-store rust:issue/reproduce-897 rust:issue/969-reproduce rust:troublescooter-static rust:intermediate-mergeable-troublescooter rust:barrotsteindev-filter-collector-tpredicatevalue rust:issue/938b rust:issue/938 rust:barrotsteindev-filter-collector rust:issue/922 rust:issue/896 rust:quickfix-explain rust:slog rust:limit-random-seeks rust:0.13.1 rust:issue/866b rust:bugfix-unittest-macos rust:githubactions rust:fieldnorm_readers rust:blockwand rust:nodeffeatfails rust:removedaliveiterator rust:robyoung-introduce-offset-to-top-docs rust:specialization-of-foreach-for-union-scorer rust:elshize-bmw rust:cutt rust:nrt rust:storewriter-out rust:segment_writer_in_ram2 rust:polling rust:audunhalland-query-boost rust:poll-meta rust:segment_writer_in_ram rust:bundle rust:petr-tik-n662_footer_is_incompatible_impl rust:race-condition-in-tests rust:hotfix-0.10.3-issue681 rust:remove-tru rust:kkoziara-pre-tokenized-text rust:issue/681 rust:issue/680 rust:refactoring/field rust:updating/uuid rust:fix-bench rust:hotfix-656 rust:readd-macos-ci rust:incremental-search rust:crate-division rust:streamer-with-state rust:criterion rust:python-binding-changes rust:bump-version rust:gh-pages rust:issue/554 rust:perf/experimental-union-for_each rust:issue/526b rust:softcommits rust:0.9.1-hotfix rust:revert-521-fix-non-english-stemmers rust:issue/500-hotfix rust:hotfix/0.8.2 rust:issue/483 rust:bug/merge-deted rust:issue/407 rust:issue/457 rust:segment_fruits rust:doc rust:dds/lenient rust:issue/webasm rust:issue/396 rust:broken_collector_with_result rust:dss/automaton-weight-suggestion rust:wasm rust:common-crawl rust:tantivy-imhotep
rust:0.25.0 rust:0.22.1 rust:0.24.2 rust:0.24.1 rust:tantivy-query-grammar-v0.24.0 rust:tantivy-columnar-v0.5.0 rust:tantivy-bitpacker-v0.8.0 rust:ownedbytes-v0.9.0 rust:tantivy-stacker-v0.5.0 rust:tantivy-sstable-v0.5.0 rust:tantivy-common-v0.9.0 rust:tantivy-tokenizer-api-v0.5.0 rust:0.24 rust:qw-airmail-20250224-hotfix-merge-panic rust:qw-airmail-20250219-hotfix-merge-panic rust:qw-airmail-202406 rust:qw-airmail-20240611 rust:qw-airmail-20240606 rust:quickwit-rev rust:0.22.0 rust:tantivy-stacker-v0.3.0 rust:ownedbytes-v0.7.0 rust:tantivy-tokenizer-api-v0.3.0 rust:tantivy-columnar-v0.3.0 rust:tantivy-common-v0.7.0 rust:tantivy-query-grammar-v0.22.0 rust:tantivy-sstable-v0.3.0 rust:tantivy-bitpacker-v0.6.0 rust:0.21.1 rust:0.21 rust:0.20.2 rust:0.20.1 rust:0.20 rust:quickwit-0.5-rev rust:0.19.2 rust:0.19.1 rust:0.19 rust:0.18.1 rust:0.18 rust:0.17 rust:0.16.1 rust:0.15.3 rust:0.15.2 rust:0.15.1 rust:0.15 rust:0.14 rust:0.13.3 rust:0.13.2 rust:0.13.1 rust:0.13 rust:0.12 rust:0.11.3 rust:0.11.1 rust:0.11 rust:0.10.3 rust:0.1 rust:0.10.2 rust:0.10.1 rust:0.10.0 rust:0.9.1 rust:0.9 rust:0.8.2 rust:0.8.2b rust:0.8.1 rust:0.8.0 rust:0.7.2 rust:0.7.1 rust:0.7.0 rust:0.6.1 rust:0.6.0 rust:0.5.2 rust:0.5.1 rust:0.5.0 rust:0.4.3 rust:0.4.2 rust:0.4.0 rust:0.3.1 rust:0.3.0 rust:0.2.0

175 Commits
issue/396 ... perf/exper

Author SHA1 Message Date
Paul Masurel
a7a98b11d7 exploratory 2019-05-22 10:18:53 +09:00
Paul Masurel
a18932165f for_each in union 2019-05-07 08:08:55 +09:00
Paul Masurel
8f82d0b773 Added impl for for_each specific to unions. 2019-05-05 17:31:32 +09:00
Paul Masurel
7102b363f5 Fix build 2019-05-05 14:19:54 +09:00
Paul Masurel
66b4615e4e Issue/542 (#543) 
* Closes 542.

Fast fields are all loaded when the segment reader is created.
 2019-05-05 13:52:43 +09:00
Paul Masurel
3df037961f Added more info to fast fields. 2019-04-30 13:14:01 +09:00
Paul Masurel
dac50c6aeb Dds merged (#539) 
* add ascii folding support

* Minor change and added Changelog.

* add additional tests

* Add tests for ascii folding (#533)

* first tests for ascii folding

* use a `RawTokenizer` for tokens using punctuation

* add test for all (?) folding, inspired by Lucene

* Simplification of the unit test code
 2019-04-26 10:25:08 +09:00
Paul Masurel
31b22c5acc Added logging when token is dropped. (#538) 2019-04-26 09:23:28 +09:00
Paul Masurel
96a4f503ec Closes #526 (#535) 2019-04-24 20:59:48 +09:00
Paul Masurel
9df288b0c9 Merge branch 'master' of github.com:tantivy-search/tantivy 2019-04-24 12:31:47 +09:00
Paul Masurel
b7c2d0de97 Clippy2 (#534) 
* Clippy comments

Clippy complaints that about the cast of &[u32] to a *const __m128i,
because of the lack of alignment constraints.

This commit passes the OutputBuffer object (which enforces proper
    alignment) instead of `&[u32]`.

* Clippy. Block alignment

* Code simplification

* Added comment. Code simplification

* Removed the extraneous freq block len hack.
 2019-04-24 12:31:32 +09:00
Paul Masurel
62445e0ec8 Merge branch 'master' of github.com:tantivy-search/tantivy 2019-04-23 09:55:55 +09:00
Paul Masurel
a228825462 Clippy comments (#532) 
Clippy complaints that about the cast of &[u32] to a *const __m128i,
because of the lack of alignment constraints.

This commit passes the OutputBuffer object (which enforces proper
    alignment) instead of `&[u32]`.
 2019-04-23 09:54:02 +09:00
Paul Masurel
d3eabd14bc Clippy comments 
Clippy complaints that about the cast of &[u32] to a *const __m128i,
because of the lack of alignment constraints.

This commit passes the OutputBuffer object (which enforces proper
    alignment) instead of `&[u32]`.
 2019-04-22 11:16:21 +09:00
petr-tik
c967031d21 Delete files from target/ dir to avoid caching them on CI (#531) 
* Delete files from target/ dir to avoid caching them on CI

idea from here https://github.com/rust-lang/cargo/issues/5885#issuecomment-432723546

* Delete examples
 2019-04-21 08:02:27 +09:00
Paul Masurel
d823163d52 Closes #527. (#529) 
Fixing the bug that affects the result of `query.count()` in presence of
deletes.
 2019-04-19 09:19:50 +09:00
Paul Masurel
c4f59f202d Bumped combine version 2019-04-11 08:33:56 +09:00
Paul Masurel
acd29b535d Fix comment 2019-04-02 10:05:14 +09:00
Panagiotis Ktistakis
2cd31bcda2 Fix non english stemmers (#521) 2019-03-27 08:54:16 +09:00
Paul Masurel
99870de55c 0.10.0-dev 2019-03-25 08:58:26 +09:00
Paul Masurel
cad2d91845 Disabled tests for android 2019-03-24 22:58:46 +09:00
Paul Masurel
79f3cd6cf4 Added instructions to update 2019-03-24 09:10:31 +09:00
Paul Masurel
e3abb4481b broken link 2019-03-22 09:58:28 +09:00
Paul Masurel
bfa61d2f2f Added patreon button 2019-03-22 09:51:00 +09:00
Paul Masurel
6c0e621fdb Added bench info in README 2019-03-21 09:35:04 +09:00
Paul Masurel
a8cc5208f1 Linear simd (#519) 
* linear simd search within block
 2019-03-20 22:10:05 +09:00
Paul Masurel
83eb0d0cb7 Disabling tests on Android 2019-03-20 10:24:17 +09:00
Paul Masurel
ee6e273365 cleanup for nodefaultfeatures 2019-03-20 10:04:42 +09:00
Paul Masurel
6ea34b3d53 Fix version 2019-03-20 09:39:24 +09:00
Paul Masurel
22cf1004bd Reenabled test on android 2019-03-20 08:54:52 +09:00
Paul Masurel
5768d93171 Rename try to attempt as try is becoming a keyword in rust 2019-03-20 08:54:19 +09:00
Paul Masurel
663dd89c05 Feature/reader (#517) 
Adding IndexReader to the API. Making it possible to watch for changes.

* Closes #500
 2019-03-20 08:39:22 +09:00
barrotsteindev
a934577168 WIP: date field (#487) 
* initial version, still a work in progress

* remove redudant or

* add chrono::DateTime and index i64

* add more tests

* fix tests

* pass DateTime by ptr

* remove println!

* document query_parser rfc 3339 date support

* added some more docs about implementation to schema.rs

* enforce DateTime is UTC, and re-export chrono

* added DateField to changelog

* fixed conflict

* use INDEXED instead of INT_INDEXED for date fields
 2019-03-15 22:10:37 +09:00
Paul Masurel
94f1885334 Issue/513 (#514) 
* Closes #513

* Clean up and doc

* Updated changelog
 2019-03-07 09:39:30 +09:00
Jonathan Fok kan
2ccfdb97b5 WIP: compiling to wasm (#512) 
* First work to enable compile to wasm

* Added back fst-regex/mmap to mmap feature

* Removed fst-regex. Forced uuid version 0.7.2.
 2019-03-06 10:40:54 +09:00
Paul Masurel
e67883138d Cargo fmt 2019-03-06 10:31:00 +09:00
Paul Masurel
f5c65f1f60 Added comment on the constructor fo TopDocSByField 2019-03-06 10:30:37 +09:00
Mauri de Souza Nunes
ec73a9a284 Remove note about panicking in get_field docs (#503) 
Since get_field rely on calling get on the underlying InnerSchema HashMap
it shouldn't fail if the field was not found, it simply returns None.
 2019-02-28 09:23:00 +09:00
Thomas Schaller
a814a31f1e Remove semicolon from doc! expansion (#509) 2019-02-28 09:20:43 +09:00
Paul Masurel
9acadb3756 Code cleaning 2019-02-26 10:50:36 +09:00
Paul Masurel
774fcecf23 cargo fmt 2019-02-26 10:44:59 +09:00
Paul Masurel
27c9fa6028 Jannickj prove bug with facets (#508) 
* prove bug with facets

* Closing #505

Introduce a term id in the TermHashMap
 2019-02-25 22:33:17 +09:00
Paul Masurel
fdefea9e26 Removed path reference to tantivy-fst 2019-02-23 10:42:44 +09:00
Paul Masurel
b422f9c389 Partially addresses #500 (#502) 
Using `tantivy_fst`. Storing `Weak<Mmap>` in the Mmap cache.
 2019-02-23 10:33:59 +09:00
petr-tik
9451fd5b09 MsQueue to channel (#495) 
* Format

Made the docstring consistent
remove empty line

* Move matches to dev deps

* Replace MsQueue with an unbounded crossbeam-channel

Questions:
queue.push ignores Result return

How to test pop() calls, if they block

* Format

Made the docstring consistent
remove empty line

* Unwrap the Result of queue.pop

* Addressed Paul's review

wrap the Result-returning send call with expect()

implemented the test not to fail after popping from empty queue

removed references to the Michael-Scott Queue

formatted
 2019-02-23 09:06:50 +09:00
Jason Goldberger
788b3803d9 updated changelog (#501) 
* updated changelog

* Update CHANGELOG.md

* Update CHANGELOG.md
 2019-02-19 00:25:18 +09:00
Paul Masurel
5b11228083 Merge branch 'master' of github.com:tantivy-search/tantivy 2019-02-15 08:30:55 +09:00
Paul Masurel
515adff644 Merge branch 'hotfix/0.8.2' 2019-02-15 08:30:27 +09:00
Paul Masurel
e70a45426a 0.8.2 release 
Backporting a fix for non x86_64 platforms
 2019-02-14 09:16:27 +09:00
Jason Goldberger
e14701e9cd Add grouped operations (#493) 
* [WIP] added UserOperation enum, added IndexWriter.run, and added MultiStamp

* removed MultiStamp in favor of std::ops::Range

* changed IndexWriter::run to return u64, Stamper::stamps to return a Range, added tests, and added docs

* changed delete_cursor skipping to use first operation's opstamp vice last. change index_writer test to use 1 thread

* added test for order batch of operations

* added a test comment
 2019-02-14 08:56:01 +09:00
Paul Masurel
45e62d4329 Code simplification and adding comments 2019-02-06 10:05:15 +09:00
petr-tik
76d2b4dab6 Add integer range search example (#490) 
Copied and simplified the example in the range_query mod
 2019-02-05 23:34:06 +01:00
Paul Masurel
04e9606638 simplification of positions 2019-02-05 15:36:13 +01:00
Paul Masurel
a5c57ebbd9 Positions simplification 2019-02-05 14:50:51 +01:00
Paul Masurel
96eaa5bc63 Positions 2019-02-05 14:50:16 +01:00
Paul Masurel
f1d30ab196 fastfield reader fix 2019-02-05 14:10:16 +01:00
Paul Masurel
4507df9255 Closes #461 (#489) 
Multivalued fast field uses `u64` indexes.
 2019-02-04 13:24:00 +01:00
Paul Masurel
e8625548b7 Closes #461 (#488) 
Multivalued fast field uses `u64` indexes.
 2019-02-04 13:20:20 +01:00
Paul Masurel
50ed6fb534 Code cleanup 
Fixed compilation without the mmap directory
 2019-02-05 12:39:30 +01:00
Panagiotis Ktistakis
76609deadf Add Greek stemmer (#486) 2019-02-01 06:30:49 +01:00
Paul Masurel
749e62c40b renamed 2019-01-30 16:29:17 +01:00
Paul Masurel
259ce567d1 Using linear search 2019-01-29 15:59:24 +01:00
Paul Masurel
4c93b096eb Rustfmt 2019-01-29 11:45:30 +01:00
Paul Masurel
6a547b0b5f Issue/483 (#484) 
* Downcast_ref

* fixing unit test
 2019-01-28 11:43:42 +01:00
Paul Masurel
e99d1a2355 Better exponential search 2019-01-29 11:29:17 +01:00
Paul Masurel
c7bddc5fe3 Inlined exponential search 2019-01-28 17:28:07 +01:00
Paul Masurel
7b97dde335 Clippy + cargo fmt 2019-01-28 12:37:55 +01:00
Paul Masurel
644b4bd0a1 Issue/468b (#482) 
* Moving lock to directory/

* added fs2

* doc

* Using fs2 for locking

* Added unit test

* Fixed error message related unit test

* Fixing location of import
 2019-01-27 12:32:21 +01:00
Paul Masurel
bf94fd77db Issue/471 (#481) 
* Closes 471

Removing writing_segments in the segment manager as it is now useless.
Removing the target merged segment id as it is useless as well.

* RAII for tracking which segment is in merge.

Closes #471

* fmt

* Using Inventory::default().
 2019-01-27 12:18:59 +09:00
Paul Masurel
097eaf4aa6 impl Future as a result of merges 2019-01-28 03:56:43 +01:00
Paul Masurel
1fd46c1e9b Clippy 2019-01-28 03:46:23 +01:00
Paul Masurel
2fb219d017 Changelog 2019-01-24 09:12:07 +09:00
Paul Masurel
63b593bd0a Lower RAM usage in tests. 2019-01-24 09:10:38 +09:00
Paul Masurel
286bb75a0c Updated changelog 2019-01-24 09:03:58 +09:00
barrotsteindev
222b7f2580 Tantivy-288 (#472) 
* add unit test

* improved test

* added SegmentManager#remove_empty_segments

* update old tests for new behaviour

* cleaner filter for empty segments

* PR adjustments

* rename x in closures

* simplify assert_eq!(vec.len(), 0)

* wait_merging_threads

* acquire searchers

* add comments to test

* rebased on latest master

* harden test

* fix merger#test_merge_multivalued_int_fields_all_deleted test
 2019-01-24 08:58:56 +09:00
pentlander
5292e78860 Allow stemmers in languages other than English (#473) 
Allow users to create stemmers for languages other than English. Add a
default stemmer for English.
 2019-01-23 22:24:32 +09:00
Paul Masurel
c0cc6aac83 Updated changelog 2019-01-23 22:22:34 +09:00
Paul Masurel
0b0bf59a32 Allow stemmers in languages other than English (#478) 
Allow users to create stemmers for languages other than English. Add a
default stemmer for English.

Closes #478
 2019-01-23 22:21:00 +09:00
Paul Masurel
74f70a5c2c 32bits platforms 2019-01-23 13:21:31 +09:00
Paul Masurel
1acfb2ebb5 cargo fmt 2019-01-23 10:21:39 +09:00
Paul Masurel
4dfd091e67 Bumped version to 0.8.2-dev 2019-01-23 10:20:59 +09:00
Paul Masurel
8eba4ab807 Merge branch 'hotfix-476' 2019-01-23 10:20:33 +09:00
Paul Masurel
5e8e03882b Merge branch 'bug/476' 2019-01-23 10:18:27 +09:00
Paul Masurel
7df3260a15 Version bump 2019-01-23 10:13:18 +09:00
Paul Masurel
176f67a266 Refactoring 2019-01-23 10:06:40 +09:00
Paul Masurel
19babff849 Closes #476 2019-01-23 10:06:39 +09:00
Paul Masurel
bf2576adf9 Added a broken unit test 2019-01-23 10:04:27 +09:00
Paul Masurel
0e8fcd5727 Plastic surgery 2019-01-19 23:13:27 +09:00
Paul Masurel
f745c83bb7 Closes 466. Removing mentions of the chain collector. (#467) 2019-01-16 10:28:19 +09:00
Paul Masurel
ffb16d9103 More efficient indexing (#463) 
* Using unrolled u32 VInt and caching Vec s

* cargo fmt

* Exposing a io::Write in the Expull thing

* expull as a writer. clippy + format

* inline the first block

* simplified -if let Some-

* vint reader iterator

* blop
 2019-01-13 14:51:18 +09:00
Paul Masurel
98ca703daa More efficient indexing (#462) 
* Using unrolled u32 VInt and caching Vec s

* cargo fmt

* Exposing a io::Write in the Expull thing

* expull as a writer. clippy + format

* inline the first block

* simplified -if let Some-

* vint reader iterator
 2019-01-13 14:41:56 +09:00
Paul Masurel
b9d25cda5d Using LittleEndian explicitely 2019-01-08 12:41:58 +09:00
Paul Masurel
beb4289ec2 Less unsafe 2019-01-08 00:48:14 +09:00
Andrew Banchich
bdd72e4683 Update README.md (#459) 
Fix Elasticsearch spelling
 2018-12-27 07:26:49 +09:00
Paul Masurel
45c3cd19be Fixing README: git clone https... 2018-12-26 21:13:33 +09:00
Paul Masurel
b8241c5603 0.8.0 2018-12-26 10:18:34 +09:00
Paul Masurel
a4745151c0 Version to 0.8 2018-12-26 10:11:06 +09:00
Paul Masurel
e2ce326a8c Merge branch 'issue/457' 2018-12-18 10:35:01 +09:00
Paul Masurel
bb21d12a70 Bumping version 2018-12-18 10:14:12 +09:00
Paul Masurel
4565aba62a Added unit test for exponential search 2018-12-18 09:24:31 +09:00
Paul Masurel
545a7ec8dd Closes #457 2018-12-18 09:18:46 +09:00
Paul Masurel
e68775d71c Format and update murmurhash32 version 2018-12-17 19:12:38 +09:00
Paul Masurel
dcc92d287e Facet remove unsafe (#456) 
* Removing some unsafe

* Removing some unsafe (2)

* Remove murmurhash
 2018-12-17 19:08:48 +09:00
Paul Masurel
b48f81c051 Removing unsafe from bitpacking code (#455) 2018-12-17 19:06:37 +09:00
Paul Masurel
a3042e956b Facet remove unsafe (#454) 
* Removing some unsafe

* Removing some unsafe (2)
 2018-12-17 09:31:09 +09:00
dependabot[bot]
1fa10f0a0b Update itertools requirement from 0.7 to 0.8 (#453) 
Updates the requirements on [itertools](https://github.com/bluss/rust-itertools) to permit the latest version.
- [Release notes](https://github.com/bluss/rust-itertools/releases)
- [Commits](https://github.com/bluss/rust-itertools/commits/0.8.0)

Signed-off-by: dependabot[bot] <support@dependabot.com>
 2018-12-17 09:28:36 +09:00
Paul Masurel
279a9eb5e3 Closes #449 (#450) 
Clippy working on stable.
Clippy warnings addressed
 2018-12-10 12:20:59 +09:00
fdb-hiroshima
21a24672d8 Add accessors for Snippet and HighlightSection (#448) 
* Add accessors for Snippet and HighlightSection

And add an example of custom highlighter

* Remove inline(always) and unnecessary empty lines
 2018-12-02 18:00:16 +09:00
dependabot[bot]
a3f1fbaae6 Update scoped-pool requirement from 0.1 to 1.0 (#447) 
Updates the requirements on [scoped-pool](https://github.com/reem/rust-scoped-pool) to permit the latest version.
- [Release notes](https://github.com/reem/rust-scoped-pool/releases)
- [Commits](https://github.com/reem/rust-scoped-pool/commits/1.0.0)

Signed-off-by: dependabot[bot] <support@dependabot.com>
 2018-12-01 13:54:59 +09:00
Paul Masurel
a6e767c877 Cargo fmt 2018-11-30 22:52:45 +09:00
Paul Masurel
6af0488dbe Executor made sorted 2018-11-30 22:52:26 +09:00
Paul Masurel
07d87e154b Collector refactoring and multithreaded search (#437) 
* Split Collector into an overall Collector and a per-segment SegmentCollector. Precursor to cross-segment parallelism, and as a side benefit cleans up any per-segment fields from being Option<T> to just T.

* Attempt to add MultiCollector back

* working. Chained collector is broken though

* Fix chained collector

* Fix test

* Make Weight Send+Sync for parallelization purposes

* Expose parameters of RangeQuery for external usage

* Removed &mut self

* fixing tests

* Restored TestCollectors

* blop

* multicollector working

* chained collector working

* test broken

* fixing unit test

* blop

* blop

* Blop

* simplifying APi

* blop

* better syntax

* Simplifying top_collector

* refactoring

* blop

* Sync with master

* Added multithread search

* Collector refactoring

* Schema::builder

* CR and rustdoc

* CR comments

* blop

* Added an executor

* Sorted the segment readers in the searcher

* Update searcher.rs

* Fixed unit testst

* changed the place where we have the sort-segment-by-count heuristic

* using crossbeam::channel

* inlining

* Comments about panics propagating

* Added unit test for executor panicking

* Readded default

* Removed Default impl

* Added unit test for executor
 2018-11-30 22:46:59 +09:00
Paul Masurel
8b0b0133dd Importing crossbeam_channel from crossbeam reexport. 2018-11-19 09:19:28 +09:00
dependabot[bot]
7b9752f897 Update crossbeam-channel requirement from 0.2 to 0.3 (#436) 
* Update crossbeam-channel requirement from 0.2 to 0.3

Updates the requirements on [crossbeam-channel](https://github.com/crossbeam-rs/crossbeam-channel) to permit the latest version.
- [Release notes](https://github.com/crossbeam-rs/crossbeam-channel/releases)
- [Changelog](https://github.com/crossbeam-rs/crossbeam-channel/blob/master/CHANGELOG.md)
- [Commits](https://github.com/crossbeam-rs/crossbeam-channel/commits/v0.3.0)

Signed-off-by: dependabot[bot] <support@dependabot.com>

* fixing build
 2018-11-16 14:26:59 +09:00
dependabot[bot]
c92f41aea8 Update rand requirement from 0.5 to 0.6 (#440) 
* Update rand requirement from 0.5 to 0.6

Updates the requirements on [rand](https://github.com/rust-random/rand) to permit the latest version.
- [Release notes](https://github.com/rust-random/rand/releases)
- [Changelog](https://github.com/rust-random/rand/blob/master/CHANGELOG.md)
- [Commits](https://github.com/rust-random/rand/commits)

Signed-off-by: dependabot[bot] <support@dependabot.com>

* Updating rand.
 2018-11-16 12:38:01 +09:00
Do Duy
dea16f1d9d Derive Clone for QueryParser (#442) 2018-11-15 18:45:40 +09:00
dependabot[bot]
236cfbec08 Update crossbeam requirement from 0.4 to 0.5 (#438) 
Updates the requirements on [crossbeam](https://github.com/crossbeam-rs/crossbeam) to permit the latest version.
- [Release notes](https://github.com/crossbeam-rs/crossbeam/releases)
- [Changelog](https://github.com/crossbeam-rs/crossbeam/blob/master/CHANGELOG.md)
- [Commits](https://github.com/crossbeam-rs/crossbeam/commits/crossbeam-0.5.0)

Signed-off-by: dependabot[bot] <support@dependabot.com>
 2018-11-15 06:16:22 +09:00
Paul Masurel
edcafb69bb Fixed benches 2018-11-10 17:04:29 -08:00
Paul Masurel
14908479d5 Release 0.7.1 2018-11-02 17:56:25 +09:00
Dru Sellers
ab4593eeb7 Adds open_or_create method (#428) 
* Change the semantic of Index::create_in_dir.

It should return an error if the directory already contains an Index.

* Index::open_or_create is working

* additional test

* Checking that schema matches on open_or_create.

Simplifying unit tests.

* simplifying Eq
 2018-10-31 08:36:39 +09:00
Dru Sellers
e75bb1d6a1 Fix NGram processing of non-ascii characters (#430) 
* A working version

* optimize the ngram parsing

* Decoding codepoint only once.

* Closes #429

* using leading_zeros to make code less cryptic

* lookup in a table
 2018-10-31 08:35:27 +09:00
dependabot[bot]
63b9d62237 Update base64 requirement from 0.9.1 to 0.10.0 (#433) 
Updates the requirements on [base64](https://github.com/alicemaz/rust-base64) to permit the latest version.
- [Release notes](https://github.com/alicemaz/rust-base64/releases)
- [Changelog](https://github.com/alicemaz/rust-base64/blob/master/RELEASE-NOTES.md)
- [Commits](https://github.com/alicemaz/rust-base64/commits/v0.10.0)

Signed-off-by: dependabot[bot] <support@dependabot.com>
 2018-10-31 08:34:44 +09:00
Jason Wolfe
0098e3d428 Compute space usage of a Searcher / SegmentReader / CompositeFile (#282) 
* Compute space usage of a Searcher / SegmentReader / CompositeFile

* Fix typo

* Add serde Serialize/Deserialize for all the SpaceUsage structs

* Fix indexing

* Public methods for consuming space usage information

* #281: Add a space usage method that takes a SegmentComponent to support code that is unaware of particular segment components, and to make it more likely to update methods when a new component type is added.

* Add support for space usage computation of positions skip index file (#281)

* Add some tests for space usage computation (#281)
 2018-10-15 09:04:36 +09:00
Konstantin Gribov
69d5e4b9b1 Added proper references for Apache Lucene & Solr (#432) 
Also, added links to websites for Lucene, Solr & ElasticSearch
 2018-10-12 08:46:07 +09:00
Paul Masurel
e0cdd3114d Fixing README (#427) 
Closes #424.
 2018-09-17 08:52:29 +09:00
Paul Masurel
f32b4a2ebe Removing release build from ci, disabling lto (#425) 2018-09-17 06:41:40 +09:00
Paul Masurel
6ff60b8ed8 Fixing README (#426) 2018-09-17 06:20:44 +09:00
Paul Masurel
8da28fb6cf Added iml filewq 2018-09-16 13:26:54 +09:00
Paul Masurel
0df2a221da Bump version pre-release 2018-09-16 13:24:14 +09:00
Paul Masurel
5449ec3c11 Snippet term score (#423) 2018-09-16 10:21:02 +09:00
Paul Masurel
10f6c07c53 Clippy (#422) 
* Cargo Format
* Clippy
 2018-09-15 20:20:22 +09:00
Paul Masurel
06e7bd18e7 Clippy (#421) 
* Cargo Format

* Clippy

* bugfix

* still clippy stuff

* clippy step 2
 2018-09-15 14:56:14 +09:00
Paul Masurel
37e4280c0a Cargo Format (#420) 2018-09-15 07:44:22 +09:00
Paul Masurel
0ba1cf93f7 Remove Searcher dereference (#419) 2018-09-14 09:54:26 +09:00
Paul Masurel
21a9940726 Update Changelog with #388 (#418) 2018-09-14 09:31:11 +09:00
pentlander
8600b8ea25 Top collector (#413) 
* Make TopCollector generic

Make TopCollector take a generic type instead of only being tied to
score. This will allow for sharing code between a TopCollector that
sorts results by Score and a TopCollector that sorts documents by a fast
field. This commit makes no functional changes to TopCollector.

* Add TopFieldCollector and TopScoreCollector

Create two new collectors that use the refactored TopCollector.
TopFieldCollector has the same functionality that TopCollector
originally had. TopFieldCollector allows for sorting results by a given
fast field. Closes tantivy-search/tantivy#388

* Make TopCollector private

Make TopCollector package private and export TopFieldCollector as
TopCollector to maintain backwards compatibility. Mark TopCollector
as deprecated to encourage use of the non-aliased TopFieldCollector.
Remove Collector implementation for TopCollector since it is not longer
used.
 2018-09-14 09:22:17 +09:00
Paul Masurel
30f4f85d48 Closes #414. (#417) 
Updating documentation for load_searchers.
 2018-09-14 09:11:07 +09:00
Paul Masurel
82d25b8397 Fixing snippet example 2018-09-13 12:39:42 +09:00
Paul Masurel
2104c0277c Updating uuid 2018-09-13 09:13:37 +09:00
Paul Masurel
dd37e109f2 Merge branch 'issue/368b' 2018-09-11 20:16:14 +09:00
Paul Masurel
cc23194c58 Editing document 2018-09-11 20:15:38 +09:00
Paul Masurel
63868733a3 Added SnippetGenerator 2018-09-11 09:45:27 +09:00
Paul Masurel
644d8a3a10 Added snippet generator 2018-09-10 16:39:45 +09:00
Paul Masurel
e32dba1a97 Phrase weight 2018-09-10 09:26:33 +09:00
Paul Masurel
a78aa4c259 updating doc 2018-09-09 17:23:30 +09:00
Paul Masurel
7e5f697d00 Closes #387 2018-09-09 16:23:56 +09:00
Paul Masurel
a78f4cca37 Merge branch 'issue/368' into issue/368b 2018-09-09 16:04:20 +09:00
Paul Masurel
2e44f0f099 blop 2018-09-09 14:23:24 +09:00
Vignesh Sarma K
9ccba9f864 Merge branch 'master' into issue/368 2018-09-07 20:27:38 +05:30
Paul Masurel
9101bf5753 Fragments 2018-09-07 09:57:12 +09:00
Paul Masurel
23e97da9f6 Merge branch 'master' of github.com:tantivy-search/tantivy 2018-09-07 08:44:14 +09:00
Paul Masurel
1d439e96f5 Using sort unstable by key. 2018-09-07 08:43:44 +09:00
Paul Masurel
934933582e Closes #402 (#403) 2018-09-06 10:12:26 +09:00
Paul Masurel
98c7fbdc6f Issue/378 (#392) 
* Added failing unit test

* Closes #378. Handling queries that end up empty after going through the analyzer.

* Fixed stop word example
 2018-09-06 10:11:54 +09:00
Paul Masurel
cec9956a01 Issue/389 (#405) 
* Setting up the dependency.

* Completed README
 2018-09-06 10:10:40 +09:00
Paul Masurel
c64972e039 Apply unicode lowercasing. (#408) 
Checks if the str is ASCII, and uses a fast track if it is the case.
If not, the std's definition of a lowercase character.

Closes #406
 2018-09-05 09:43:56 +09:00
Paul Masurel
b3b2421e8a Issue/367 (#404) 
* First stab

* Closes #367
 2018-09-04 09:17:00 +09:00
Paul Masurel
f570fe37d4 small changes 2018-08-31 09:03:44 +09:00
Paul Masurel
6704ab6987 Added methods to extract the matching terms. First stab 2018-08-30 09:47:19 +09:00
Paul Masurel
a12d211330 Extracting terms matching query in the document 2018-08-30 09:23:34 +09:00
Paul Masurel
ee681a4dd1 Added say thanks badge 2018-08-29 11:06:04 +09:00
petr-tik
d15efd6635 Closes #235 - adds a new error type (#398) 
error message suggests possible causes

Addressed code review 1 thread + smaller heap size
 2018-08-29 08:26:59 +09:00
Vignesh Sarma K (വിഘ്നേഷ് ശ൪മ കെ)
18814ba0c1 add a test for second fragment having higher score 2018-08-28 22:27:56 +05:30
Vignesh Sarma K (വിഘ്നേഷ് ശ൪മ കെ)
f247935bb9 Use HighlightSection::new rather than just directly creating the object 2018-08-28 22:16:22 +05:30
Vignesh Sarma K (വിഘ്നേഷ് ശ൪മ കെ)
6a197e023e ran rustfmt 2018-08-28 20:41:58 +05:30
Vignesh Sarma K (വിഘ്നേഷ് ശ൪മ കെ)
96a313c6dd add more tests 2018-08-28 20:41:58 +05:30
Vignesh Sarma K (വിഘ്നേഷ് ശ൪മ കെ)
fb9b1c1f41 add a test and fix the bug of not calculating first token 2018-08-28 20:41:58 +05:30
Vignesh Sarma K (വിഘ്നേഷ് ശ൪മ കെ)
e1bca6db9d update calculate_score to try_add_token 
`try_add_token` will now update the stop_offset as well.
`FragmentCandidate::new` now just takes `start_offset`,
it expects `try_add_token` to be called to add a token.
 2018-08-28 20:41:58 +05:30
Vignesh Sarma K (വിഘ്നേഷ് ശ൪മ കെ)
8438eda01a use while let instead of loop and if. 
as per CR comment
 2018-08-28 20:41:57 +05:30
Vignesh Sarma K (വിഘ്നേഷ് ശ൪മ കെ)
b373f00840 add htmlescape and update to_html fn to use it. 
tests and imports also updated.
 2018-08-28 20:41:57 +05:30
Vignesh Sarma K (വിഘ്നേഷ് ശ൪മ കെ)
46decdb0ea compare against accumulator rather than init value 2018-08-28 20:41:41 +05:30
Vignesh Sarma K (വിഘ്നേഷ് ശ൪മ കെ)
835cdc2fe8 Initial version of snippet 
refer #368
 2018-08-28 20:41:41 +05:30
Paul Masurel
19756bb7d6 Getting started on #368 2018-08-28 20:41:41 +05:30
CJP10
57e1f8ed28 Missed a closing bracket (#397) 2018-08-28 23:17:59 +09:00
Paul Masurel
2649c8a715 Issue/246 (#393) 
* Moving Range and All to Leaves

* Parsing OR/AND

* Simplify user input ast

* AND and OR supported. Returning an error when mixing syntax

Closes #246

* Added support for NOT

* Updated changelog
 2018-08-28 11:03:54 +09:00
177 changed files with 15565 additions and 5347 deletions
Expand all files Collapse all files

1
.gitignore vendored
View File

@@ -1,3 +1,4 @@
tantivy.iml
*.swp
target
target/debug

7
.travis.yml
View File

@@ -68,6 +68,11 @@ cache: cargo
before_cache:
# Travis can't cache files that are not readable by "others"
- chmod -R a+r $HOME/.cargo
- find ./target/debug -type f -maxdepth 1 -delete
- rm -f ./target/.rustc_info.json
- rm -fr ./target/debug/{deps,.fingerprint}/tantivy*
- rm -r target/debug/examples/
- ls -1 examples/ | sed -e 's/\.rs$//' | xargs -I "{}" find target/* -name "*{}*" -type f -delete
#branches:
# only:
@@ -77,4 +82,4 @@ before_cache:
notifications:
email:
on_success: never
on_success: never

111
CHANGELOG.md
View File

@@ -1,10 +1,119 @@
Tantivy 0.10.0
=====================
*Tantivy 0.10.0 index format is compatible with the index format in 0.9.0.*
- Added an ASCII folding filter (@drusellers)
- Bugfix in `query.count` in presence of deletes (@pmasurel)
Minor
---------
- Small simplification of the code.
Calling .freq() or .doc() when .advance() has never
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.
## How to update?
Your existing indexes are usable as is. Your may or may need some
trivial updates.
### 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
=========================

53
Cargo.toml
View File

@@ -1,6 +1,6 @@
[package]
name = "tantivy"
version = "0.7.0-dev"
version = "0.10.0-dev"
authors = ["Paul Masurel <paul.masurel@gmail.com>"]
license = "MIT"
categories = ["database-implementations", "data-structures"]
@@ -12,60 +12,73 @@ readme = "README.md"
keywords = ["search", "information", "retrieval"]
[dependencies]
base64 = "0.9.1"
base64 = "0.10.0"
byteorder = "1.0"
lazy_static = "1"
tinysegmenter = "0.1.0"
regex = "1.0"
fst = {version="0.3", default-features=false}
fst-regex = { version="0.2" }
tantivy-fst = "0.1"
memmap = {version = "0.7", optional=true}
lz4 = {version="1.20", optional=true}
snap = {version="0.2"}
atomicwrites = {version="0.2.2", optional=true}
tempfile = "3.0"
log = "0.4"
combine = "3"
combine = ">=3.6.0,<4.0.0"
tempdir = "0.3"
serde = "1.0"
serde_derive = "1.0"
serde_json = "1.0"
num_cpus = "1.2"
itertools = "0.7"
fs2={version="0.4", optional=true}
itertools = "0.8"
levenshtein_automata = {version="0.1", features=["fst_automaton"]}
notify = {version="4", optional=true}
bit-set = "0.5"
uuid = { version = "0.6", features = ["v4", "serde"] }
crossbeam = "0.4"
crossbeam-channel = "0.2"
uuid = { version = "0.7.2", features = ["v4", "serde"] }
crossbeam = "0.5"
futures = "0.1"
futures-cpupool = "0.1"
owning_ref = "0.4"
stable_deref_trait = "1.0.0"
rust-stemmers = "1"
downcast = { version="0.9" }
matches = "0.1"
bitpacking = "0.5"
census = "0.1"
rust-stemmers = "1.1"
downcast-rs = { version="1.0" }
bitpacking = "0.6"
census = "0.2"
fnv = "1.0.6"
owned-read = "0.4"
failure = "0.1"
htmlescape = "0.3.1"
fail = "0.2"
scoped-pool = "1.0"
murmurhash32 = "0.2"
chrono = "0.4"
[target.'cfg(windows)'.dependencies]
winapi = "0.2"
[dev-dependencies]
rand = "0.5"
rand = "0.6"
maplit = "1"
matches = "0.1.8"
time = "0.1.42"
[profile.release]
opt-level = 3
debug = false
lto = true
debug-assertions = false
[profile.test]
debug-assertions = true
overflow-checks = true
[features]
default = ["mmap"]
mmap = ["fst/mmap", "atomicwrites"]
# by default no-fail is disabled. We manually enable it when running test.
default = ["mmap", "no_fail"]
mmap = ["atomicwrites", "fs2", "memmap", "notify"]
lz4-compression = ["lz4"]
no_fail = ["fail/no_fail"]
unstable = [] # useful for benches.
wasm-bindgen = ["uuid/wasm-bindgen"]
[badges]
travis-ci = { repository = "tantivy-search/tantivy" }

42
README.md
View File

@@ -4,6 +4,7 @@
[![Join the chat at https://gitter.im/tantivy-search/tantivy](https://badges.gitter.im/tantivy-search/tantivy.svg)](https://gitter.im/tantivy-search/tantivy?utm_source=badge&utm_medium=badge&utm_campaign=pr-badge&utm_content=badge)
[![License: MIT](https://img.shields.io/badge/License-MIT-yellow.svg)](https://opensource.org/licenses/MIT)
[![Build status](https://ci.appveyor.com/api/projects/status/r7nb13kj23u8m9pj/branch/master?svg=true)](https://ci.appveyor.com/project/fulmicoton/tantivy/branch/master)
[![Say Thanks!](https://img.shields.io/badge/Say%20Thanks-!-1EAEDB.svg)](https://saythanks.io/to/fulmicoton)
![Tantivy](https://tantivy-search.github.io/logo/tantivy-logo.png)
@@ -16,39 +17,52 @@
[![](https://sourcerer.io/fame/fulmicoton/tantivy-search/tantivy/images/6)](https://sourcerer.io/fame/fulmicoton/tantivy-search/tantivy/links/6)
[![](https://sourcerer.io/fame/fulmicoton/tantivy-search/tantivy/images/7)](https://sourcerer.io/fame/fulmicoton/tantivy-search/tantivy/links/7)
[![Become a patron](https://c5.patreon.com/external/logo/become_a_patron_button.png)](https://www.patreon.com/fulmicoton)
**Tantivy** is a **full text search engine library** written in rust.
It is closer to Lucene than to Elastic Search and Solr in the sense it is not
It is closer to [Apache Lucene](https://lucene.apache.org/) than to [Elasticsearch](https://www.elastic.co/products/elasticsearch) and [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.
Tantivy is, in fact, strongly inspired by Lucene's design.
# Benchmark
Tantivy is typically faster than Lucene, but the results will depend on
the nature of the queries in your workload.
The following [benchmark](https://tantivy-search.github.io/bench/) break downs
performance for different type of queries / collection.
# Features
- Full-text search
- Configurable tokenizer. (stemming available for 17 latin languages. Third party support for Chinese ([tantivy-jieba](https://crates.io/crates/tantivy-jieba) and [cang-jie](https://crates.io/crates/cang-jie)) and [Japanese](https://crates.io/crates/tantivy-tokenizer-tiny-segmenter)
- 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)
- Basic query language (`+michael +jackson`)
- Phrase queries search (\"michael jackson\"`)
- Natural query language `(michael AND jackson) OR "king of pop"`
- Phrase queries search (`"michael jackson"`)
- Incremental indexing
- Multithreaded indexing (indexing English Wikipedia takes < 3 minutes on my desktop)
- Mmap directory
- SIMD integer compression when the platform/CPU includes the SSE2 instruction set.
- Single valued and multivalued u64 and i64 fast fields (equivalent of doc values in Lucene)
- `&[u8]` fast fields
- Text, i64, u64, dates and hierarchical facet fields
- LZ4 compressed document store
- Range queries
- Faceted search
- Configurable indexing (optional term frequency and position indexing
- Configurable indexing (optional term frequency and position indexing)
- Cheesy logo with a horse
# Non-features
- Distributed search and will not be in the scope of tantivy.
- Distributed search is out of the scope of tantivy. That being said, tantivy is meant as a
library upon which one could build a distributed search. Serializable/mergeable collector state for instance,
are within the scope of tantivy.
# Supported OS and compiler
@@ -73,11 +87,23 @@ It will walk you through getting a wikipedia search engine up and running in a f
Tantivy compiles on stable rust but requires `Rust >= 1.27`.
To check out and run tests, you can simply run :
git clone git@github.com:tantivy-search/tantivy.git
git clone https://github.com/tantivy-search/tantivy.git
cd tantivy
cargo build
## Running tests
# Contribute
Some tests will not run with just `cargo test` because of `fail-rs`.
To run the tests exhaustively, run `./run-tests.sh`.
Send me an email (paul.masurel at gmail.com) if you want to contribute to tantivy.
# How can I support this project ?
There are many ways to support this project.
- If you use tantivy, tell us about your experience on [gitter](https://gitter.im/tantivy-search/tantivy) or by email (paul.masurel@gmail.com)
- Report bugs
- Write a blog post
- Complete documentation
- Contribute code (you can join [our gitter](https://gitter.im/tantivy-search/tantivy) )
- Talk about tantivy around you
- Drop a word on on [![Say Thanks!](https://img.shields.io/badge/Say%20Thanks-!-1EAEDB.svg)](https://saythanks.io/to/fulmicoton) or even [![Become a patron](https://c5.patreon.com/external/logo/become_a_patron_button.png)](https://www.patreon.com/fulmicoton)

4
appveyor.yml
View File

@@ -18,5 +18,5 @@ install:
build: false
test_script:
- REM SET RUST_LOG=tantivy,test & cargo test --verbose
- REM SET RUST_BACKTRACE=1 & cargo build --examples
- REM SET RUST_LOG=tantivy,test & cargo test --verbose --no-default-features --features mmap -- --test-threads 1
- REM SET RUST_BACKTRACE=1 & cargo build --examples

3
ci/script.sh
View File

@@ -11,12 +11,11 @@ main() {
else
echo "Build"
cross build --target $TARGET
cross build --target $TARGET --release
if [ ! -z $DISABLE_TESTS ]; then
return
fi
echo "Test"
cross test --target $TARGET
cross test --target $TARGET --no-default-features --features mmap -- --test-threads 1
fi
for example in $(ls examples/*.rs)
do

1
doc/src/SUMMARY.md
View File

@@ -9,6 +9,7 @@
- [Facetting](./facetting.md)
- [Innerworkings](./innerworkings.md)
- [Inverted index](./inverted_index.md)
- [Best practise](./inverted_index.md)
[Frequently Asked Questions](./faq.md)
[Examples](./examples.md)

29
doc/src/avant-propos.md
View File

@@ -2,8 +2,8 @@
> Tantivy is a **search** engine **library** for Rust.
If you are familiar with Lucene, tantivy is heavily inspired by Lucene's design and
they both have the same scope and targetted users.
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 targetted use cases.
If you are not familiar with Lucene, let's break down our little tagline.
@@ -17,15 +17,18 @@ relevancy, collapsing, highlighting, spatial 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.
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.
- **Library** means that you will have to write code. tantivy is not an *all-in-one* server solution like elastic search for instance.
Most frequently this will mean writing your own `Collector`, your own `Scorer` or your own
`Tokenizer/TokenFilter`... But some of your requirement may also be related 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 its API to do all of these things.
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.

81
doc/src/basis.md
View File

@@ -2,47 +2,76 @@
## Straight from disk
By default, tantivy accesses its data using its `MMapDirectory`.
While this design has some downsides, this greatly simplifies the source code of tantivy,
and entirely delegates the caching to the OS.
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`.
`tantivy` works entirely (or almost) by directly reading the datastructures as they are layed 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 a query
can typically be done in a matter of milliseconds.
While this design has some downsides, this greatly simplifies the source code of
tantivy. Caching is also entirely delegated to the OS.
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.
`tantivy` works entirely (or almost) by directly reading the datastructures as they are layed 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.
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 the 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 a great property when working with `SSD` or `RAM`,
as it makes our read patterns very predictable for the CPU.
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 compact layout comes at one cost: it prevents our datastructures from being dynamic.
In fact, a trait called `Directory` is in charge of abstracting all of tantivy's data access
and its API does not even allow editing these file once they are written.
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*.
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 designed to be dynamic but
cannot be accessed for search. 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.
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>.*`.
> 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 to segment `D-7`.
> A tantivy `Index` is a collection of `Segments`.
# 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.

0
doc/src/best_practise.md.rs Normal file
View File

2
doc/src/examples.md
View File

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

104
examples/basic_search.rs
View File

@@ -10,17 +10,18 @@
// - search for the best document matchings "sea whale"
// - retrieve the best document original content.
extern crate tempdir;
// ---
// Importing tantivy...
#[macro_use]
extern crate tantivy;
use tantivy::collector::TopCollector;
use tantivy::collector::TopDocs;
use tantivy::query::QueryParser;
use tantivy::schema::*;
use tantivy::Index;
use tantivy::ReloadPolicy;
use tempdir::TempDir;
fn main() -> tantivy::Result<()> {
// Let's create a temporary directory for the
@@ -35,7 +36,7 @@ fn main() -> tantivy::Result<()> {
// be indexed".
// first we need to define a schema ...
let mut schema_builder = SchemaBuilder::default();
let mut schema_builder = Schema::builder();
// Our first field is title.
// We want full-text search for it, and we also want
@@ -106,37 +107,37 @@ fn main() -> tantivy::Result<()> {
// 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"
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 => "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"
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."
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
@@ -170,24 +171,33 @@ fn main() -> tantivy::Result<()> {
//
// ### Searcher
//
// Let's search our index. Start by reloading
// searchers in the index. This should be done
// after every `commit()`.
index.load_searchers()?;
// A reader is required to get search the 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.
// Some search experience might require more than
// one query.
//
// The searcher ensure that we get to work
// with a consistent version of the index.
// A searcher points to 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
// You should acquire a searcher every time you start processing a request and
// and release it right after your query is finished.
let searcher = index.searcher();
let searcher = reader.searcher();
// ### Query
@@ -213,15 +223,10 @@ fn main() -> tantivy::Result<()> {
//
// We are not interested in all of the documents but
// only in the top 10. Keeping track of our top 10 best documents
// is the role of the TopCollector.
let mut top_collector = TopCollector::with_limit(10);
// is the role of the TopDocs.
// 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();
let top_docs = searcher.search(&query, &TopDocs::with_limit(10))?;
// The actual documents still need to be
// retrieved from Tantivy's store.
@@ -229,15 +234,10 @@ fn main() -> tantivy::Result<()> {
// 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)?;
for (_score, doc_address) in top_docs {
let retrieved_doc = searcher.doc(doc_address)?;
println!("{}", schema.to_json(&retrieved_doc));
}
Ok(())
}
use tempdir::TempDir;

196
examples/custom_collector.rs Normal file
View File

@@ -0,0 +1,196 @@
// # 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.
extern crate tempdir;
// ---
// Importing tantivy...
#[macro_use]
extern crate tantivy;
use tantivy::collector::{Collector, SegmentCollector};
use tantivy::fastfield::FastFieldReader;
use tantivy::query::QueryParser;
use tantivy::schema::Field;
use tantivy::schema::{Schema, FAST, INDEXED, TEXT};
use tantivy::SegmentReader;
use tantivy::{Index, TantivyError};
#[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_or_else(|| {
let field_name = segment_reader.schema().get_field_name(self.field);
TantivyError::SchemaError(format!(
"Field {:?} is not a u64 fast field.",
field_name
))
})?;
Ok(StatsSegmentCollector {
fast_field_reader,
stats: Stats::default(),
})
}
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_opt in segment_stats {
if let Some(segment_stats) = segment_stats_opt {
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: FastFieldReader<u64>,
stats: Stats,
}
impl SegmentCollector for StatsSegmentCollector {
type Fruit = Option<Stats>;
fn collect(&mut self, doc: u32, _score: f32) {
let value = self.fast_field_reader.get(doc) 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.clone());
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(())
}

28
examples/custom_tokenizer.rs
View File

@@ -3,16 +3,14 @@
// In this example, we'll see how to define a tokenizer pipeline
// by aligning a bunch of `TokenFilter`.
#[macro_use]
extern crate tantivy;
use tantivy::collector::TopCollector;
use tantivy::collector::TopDocs;
use tantivy::query::QueryParser;
use tantivy::schema::*;
use tantivy::tokenizer::NgramTokenizer;
use tantivy::Index;
fn main() -> tantivy::Result<()> {
// # Defining the schema
//
@@ -22,7 +20,7 @@ fn main() -> tantivy::Result<()> {
// be indexed".
// first we need to define a schema ...
let mut schema_builder = SchemaBuilder::default();
let mut schema_builder = Schema::builder();
// Our first field is title.
// In this example we want to use NGram searching
@@ -70,12 +68,12 @@ fn main() -> tantivy::Result<()> {
// heap 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."
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",
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
@@ -86,16 +84,16 @@ fn main() -> tantivy::Result<()> {
limbs and branches that arch over the pool"#
));
index_writer.add_document(doc!(
title => "Frankenstein",
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()?;
index.load_searchers()?;
let searcher = index.searcher();
let reader = index.reader()?;
let searcher = reader.searcher();
// The query parser can interpret human queries.
// Here, if the user does not specify which
@@ -106,12 +104,10 @@ fn main() -> tantivy::Result<()> {
// here we want to get a hit on the 'ken' in Frankenstein
let query = query_parser.parse_query("ken")?;
let mut top_collector = TopCollector::with_limit(10);
searcher.search(&*query, &mut top_collector)?;
let top_docs = searcher.search(&query, &TopDocs::with_limit(10))?;
let doc_addresses = top_collector.docs();
for doc_address in doc_addresses {
let retrieved_doc = searcher.doc(&doc_address)?;
for (_, doc_address) in top_docs {
let retrieved_doc = searcher.doc(doc_address)?;
println!("{}", schema.to_json(&retrieved_doc));
}

34
examples/deleting_updating_documents.rs
View File

@@ -10,17 +10,20 @@
// Importing tantivy...
#[macro_use]
extern crate tantivy;
use tantivy::collector::TopCollector;
use tantivy::collector::TopDocs;
use tantivy::query::TermQuery;
use tantivy::schema::*;
use tantivy::Index;
use tantivy::query::TermQuery;
use tantivy::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(index: &Index, isbn_term: &Term) -> tantivy::Result<Option<Document>> {
let searcher = index.searcher();
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.
@@ -28,11 +31,10 @@ fn extract_doc_given_isbn(index: &Index, isbn_term: &Term) -> tantivy::Result<Op
// 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 mut top_collector = TopCollector::with_limit(1);
searcher.search(&term_query, &mut top_collector)?;
let top_docs = searcher.search(&term_query, &TopDocs::with_limit(1))?;
if let Some(doc_address) = top_collector.docs().first() {
let doc = searcher.doc(doc_address)?;
if let Some((_score, doc_address)) = top_docs.first() {
let doc = searcher.doc(*doc_address)?;
Ok(Some(doc))
} else {
// no doc matching this ID.
@@ -41,12 +43,11 @@ fn extract_doc_given_isbn(index: &Index, isbn_term: &Term) -> tantivy::Result<Op
}
fn main() -> tantivy::Result<()> {
// # Defining the schema
//
// Check out the *basic_search* example if this makes
// small sense to you.
let mut schema_builder = SchemaBuilder::default();
let mut schema_builder = Schema::builder();
// Tantivy does not really have a notion of primary id.
// This may change in the future.
@@ -88,12 +89,12 @@ fn main() -> tantivy::Result<()> {
isbn => "978-9176370711",
));
index_writer.commit()?;
index.load_searchers()?;
let reader = index.reader()?;
let frankenstein_isbn = Term::from_field_text(isbn, "978-9176370711");
// Oops our frankenstein doc seems mispelled
let frankenstein_doc_misspelled = extract_doc_given_isbn(&index, &frankenstein_isbn)?.unwrap();
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"]}"#,
@@ -126,21 +127,20 @@ fn main() -> tantivy::Result<()> {
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.
index.load_searchers()?;
reader.reload()?;
// No more typo!
let frankenstein_new_doc = extract_doc_given_isbn(&index, &frankenstein_isbn)?.unwrap();
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(())
}
}

65
examples/faceted_search.rs
View File

@@ -22,60 +22,59 @@ use tantivy::schema::*;
use tantivy::Index;
fn main() -> tantivy::Result<()> {
// Let's create a temporary directory for the
// sake of this example
let index_path = TempDir::new("tantivy_facet_example_dir")?;
let mut schema_builder = SchemaBuilder::default();
// Let's create a temporary directory for the
// sake of this example
let index_path = TempDir::new("tantivy_facet_example_dir")?;
let mut schema_builder = Schema::builder();
schema_builder.add_text_field("name", TEXT | STORED);
schema_builder.add_text_field("name", TEXT | STORED);
// this is our faceted field
schema_builder.add_facet_field("tags");
// this is our faceted field
schema_builder.add_facet_field("tags");
let schema = schema_builder.build();
let schema = schema_builder.build();
let index = Index::create_in_dir(&index_path, schema.clone())?;
let index = Index::create_in_dir(&index_path, schema.clone())?;
let mut index_writer = index.writer(50_000_000)?;
let mut index_writer = index.writer(50_000_000)?;
let name = schema.get_field("name").unwrap();
let tags = schema.get_field("tags").unwrap();
let name = schema.get_field("name").unwrap();
let tags = schema.get_field("tags").unwrap();
// For convenience, tantivy also comes with a macro to
// reduce the boilerplate above.
index_writer.add_document(doc!(
// For convenience, tantivy also comes with a macro to
// reduce the boilerplate above.
index_writer.add_document(doc!(
name => "the ditch",
tags => Facet::from("/pools/north")
));
index_writer.add_document(doc!(
index_writer.add_document(doc!(
name => "little stacey",
tags => Facet::from("/pools/south")
));
index_writer.commit()?;
index_writer.commit()?;
index.load_searchers()?;
let reader = index.reader()?;
let searcher = index.searcher();
let searcher = reader.searcher();
let mut facet_collector = FacetCollector::for_field(tags);
facet_collector.add_facet("/pools");
let mut facet_collector = FacetCollector::for_field(tags);
facet_collector.add_facet("/pools");
searcher.search(&AllQuery, &mut facet_collector).unwrap();
let facet_counts = searcher.search(&AllQuery, &facet_collector).unwrap();
let counts = facet_collector.harvest();
// This lists all of the facet counts
let facets: Vec<(&Facet, u64)> = counts.get("/pools").collect();
assert_eq!(
facets,
vec![
(&Facet::from("/pools/north"), 1),
(&Facet::from("/pools/south"), 1)
]
);
// This lists all of the facet counts
let facets: Vec<(&Facet, u64)> = facet_counts.get("/pools").collect();
assert_eq!(
facets,
vec![
(&Facet::from("/pools/north"), 1),
(&Facet::from("/pools/south"), 1),
]
);
Ok(())
Ok(())
}
use tempdir::TempDir;

43
examples/integer_range_search.rs Normal file
View File

@@ -0,0 +1,43 @@
// # 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.
#[macro_use]
extern crate tantivy;
use tantivy::collector::Count;
use tantivy::query::RangeQuery;
use tantivy::schema::{Schema, INDEXED};
use tantivy::Index;
use tantivy::Result;
fn run() -> 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(())
}
fn main() {
run().unwrap()
}

26
examples/iterating_docs_and_positions.rs
View File

@@ -7,21 +7,18 @@
// the list of documents containing a term, getting
// its term frequency, and accessing its positions.
// ---
// Importing tantivy...
#[macro_use]
extern crate tantivy;
use tantivy::schema::*;
use tantivy::Index;
use tantivy::{DocSet, DocId, Postings};
use tantivy::{DocId, DocSet, Postings};
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 = SchemaBuilder::default();
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.
@@ -36,9 +33,9 @@ fn main() -> tantivy::Result<()> {
index_writer.add_document(doc!(title => "The modern Promotheus"));
index_writer.commit()?;
index.load_searchers()?;
let reader = index.reader()?;
let searcher = index.searcher();
let searcher = reader.searcher();
// A tantivy index is actually a collection of segments.
// Similarly, a searcher just wraps a list `segment_reader`.
@@ -47,7 +44,6 @@ fn main() -> tantivy::Result<()> {
// 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
@@ -58,19 +54,18 @@ fn main() -> tantivy::Result<()> {
// 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) {
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);
while segment_postings.advance() {
// the number of time the term appears in the document.
let doc_id: DocId = segment_postings.doc(); //< do not try to access this before calling advance once.
@@ -98,7 +93,6 @@ fn main() -> tantivy::Result<()> {
}
}
// 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");
@@ -111,7 +105,6 @@ fn main() -> tantivy::Result<()> {
// 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
@@ -124,7 +117,9 @@ fn main() -> tantivy::Result<()> {
//
// If you don't need all this information, you may get better performance by decompressing less
// information.
if let Some(mut block_segment_postings) = inverted_index.read_block_postings(&term_the, IndexRecordOption::Basic) {
if let Some(mut block_segment_postings) =
inverted_index.read_block_postings(&term_the, IndexRecordOption::Basic)
{
while block_segment_postings.advance() {
// Once again these docs MAY contains deleted documents as well.
let docs = block_segment_postings.docs();
@@ -136,4 +131,3 @@ fn main() -> tantivy::Result<()> {
Ok(())
}

86
examples/snippet.rs Normal file
View File

@@ -0,0 +1,86 @@
// # 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.
extern crate tempdir;
// ---
// Importing tantivy...
#[macro_use]
extern crate tantivy;
use tantivy::collector::TopDocs;
use tantivy::query::QueryParser;
use tantivy::schema::*;
use tantivy::Index;
use tantivy::{Snippet, SnippetGenerator};
use tempdir::TempDir;
fn main() -> tantivy::Result<()> {
// Let's create a temporary directory for the
// sake of this example
let index_path = TempDir::new("tantivy_example_dir")?;
// # 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.clone())?;
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().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 (start, end) in snippet.highlighted().iter().map(|h| h.bounds()) {
result.push_str(&snippet.fragments()[start_from..start]);
result.push_str(" --> ");
result.push_str(&snippet.fragments()[start..end]);
result.push_str(" <-- ");
start_from = end;
}
result.push_str(&snippet.fragments()[start_from..]);
result
}

150
examples/stop_words.rs
View File

@@ -15,115 +15,103 @@ extern crate tempdir;
// Importing tantivy...
#[macro_use]
extern crate tantivy;
use tantivy::collector::TopCollector;
use tantivy::collector::TopDocs;
use tantivy::query::QueryParser;
use tantivy::schema::*;
use tantivy::tokenizer::*;
use tantivy::Index;
fn main() -> tantivy::Result<()> {
// this example assumes you understand the content in `basic_search`
let index_path = TempDir::new("tantivy_stopwords_example_dir")?;
let mut schema_builder = SchemaBuilder::default();
// 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();
// 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 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);
// 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 schema = schema_builder.build();
let index = Index::create_in_dir(&index_path, schema.clone())?;
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 = SimpleTokenizer
.filter(LowerCaser)
.filter(StopWordFilter::remove(vec![
"the".to_string(),
"and".to_string(),
]));
// This tokenizer lowers all of the text (to help with stop word matching)
// then removes all instances of `the` and `and` from the corpus
let tokenizer = SimpleTokenizer
.filter(LowerCaser)
.filter(StopWordFilter::remove(vec![
"the".to_string(),
"and".to_string(),
]));
index.tokenizers().register("stoppy", tokenizer);
index.tokenizers().register("stoppy", tokenizer);
let mut index_writer = index.writer(50_000_000)?;
let mut index_writer = index.writer(50_000_000)?;
let title = schema.get_field("title").unwrap();
let body = schema.get_field("body").unwrap();
let title = schema.get_field("title").unwrap();
let body = schema.get_field("body").unwrap();
index_writer.add_document(doc!(
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.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()?;
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.load_searchers()?;
index_writer.commit()?;
let searcher = index.searcher();
let reader = index.reader()?;
let query_parser = QueryParser::for_index(&index, vec![title, body]);
let searcher = reader.searcher();
// this will have NO hits because it was filtered out
// because the query is run through the analyzer you
// actually will get an error here because the query becomes
// empty
assert!(query_parser.parse_query("the").is_err());
let query_parser = QueryParser::for_index(&index, vec![title, body]);
// this will have hits
let query = query_parser.parse_query("is")?;
// 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))?;
let mut top_collector = TopCollector::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));
}
searcher.search(&*query, &mut top_collector)?;
let doc_addresses = top_collector.docs();
for doc_address in doc_addresses {
let retrieved_doc = searcher.doc(&doc_address)?;
println!("{}", schema.to_json(&retrieved_doc));
}
Ok(())
Ok(())
}
use tempdir::TempDir;

4
examples/working_with_json.rs
View File

@@ -9,10 +9,10 @@ fn main() -> tantivy::Result<()> {
// Check out the basic example if this is confusing to you.
//
// first we need to define a schema ...
let mut schema_builder = SchemaBuilder::default();
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", INT_INDEXED);
schema_builder.add_u64_field("year", INDEXED);
let schema = schema_builder.build();
// Let's assume we have a json-serialized document.

2
run-tests.sh Executable file
View File

@@ -0,0 +1,2 @@
#!/bin/bash
cargo test --no-default-features --features mmap -- --test-threads 1

142
src/collector/chained_collector.rs
View File

@@ -1,142 +0,0 @@
use collector::Collector;
use DocId;
use Result;
use Score;
use SegmentLocalId;
use SegmentReader;
/// Collector that does nothing.
/// This is used in the chain Collector and will hopefully
/// be optimized away by the compiler.
pub struct DoNothingCollector;
impl Collector for DoNothingCollector {
#[inline]
fn set_segment(&mut self, _: SegmentLocalId, _: &SegmentReader) -> Result<()> {
Ok(())
}
#[inline]
fn collect(&mut self, _doc: DocId, _score: Score) {}
#[inline]
fn requires_scoring(&self) -> bool {
false
}
}
/// Zero-cost abstraction used to collect on multiple collectors.
/// This contraption is only usable if the type of your collectors
/// are known at compile time.
///
/// ```rust
/// #[macro_use]
/// extern crate tantivy;
/// use tantivy::schema::{SchemaBuilder, TEXT};
/// use tantivy::{Index, Result};
/// use tantivy::collector::{CountCollector, TopCollector, chain};
/// use tantivy::query::QueryParser;
///
/// # fn main() { example().unwrap(); }
/// fn example() -> Result<()> {
/// let mut schema_builder = SchemaBuilder::new();
/// let title = schema_builder.add_text_field("title", TEXT);
/// let schema = schema_builder.build();
/// let index = Index::create_in_ram(schema);
/// {
/// let mut index_writer = index.writer(3_000_000)?;
/// index_writer.add_document(doc!(
/// title => "The Name of the Wind",
/// ));
/// index_writer.add_document(doc!(
/// title => "The Diary of Muadib",
/// ));
/// index_writer.add_document(doc!(
/// title => "A Dairy Cow",
/// ));
/// index_writer.add_document(doc!(
/// title => "The Diary of a Young Girl",
/// ));
/// index_writer.commit().unwrap();
/// }
///
/// index.load_searchers()?;
/// let searcher = index.searcher();
///
/// {
/// let mut top_collector = TopCollector::with_limit(2);
/// let mut count_collector = CountCollector::default();
/// {
/// let mut collectors = chain().push(&mut top_collector).push(&mut count_collector);
/// let query_parser = QueryParser::for_index(&index, vec![title]);
/// let query = query_parser.parse_query("diary")?;
/// searcher.search(&*query, &mut collectors).unwrap();
/// }
/// assert_eq!(count_collector.count(), 2);
/// assert!(top_collector.at_capacity());
/// }
///
/// Ok(())
/// }
/// ```
pub struct ChainedCollector<Left: Collector, Right: Collector> {
left: Left,
right: Right,
}
impl<Left: Collector, Right: Collector> ChainedCollector<Left, Right> {
/// Adds a collector
pub fn push<C: Collector>(self, new_collector: &mut C) -> ChainedCollector<Self, &mut C> {
ChainedCollector {
left: self,
right: new_collector,
}
}
}
impl<Left: Collector, Right: Collector> Collector for ChainedCollector<Left, Right> {
fn set_segment(
&mut self,
segment_local_id: SegmentLocalId,
segment: &SegmentReader,
) -> Result<()> {
self.left.set_segment(segment_local_id, segment)?;
self.right.set_segment(segment_local_id, segment)?;
Ok(())
}
fn collect(&mut self, doc: DocId, score: Score) {
self.left.collect(doc, score);
self.right.collect(doc, score);
}
fn requires_scoring(&self) -> bool {
self.left.requires_scoring() || self.right.requires_scoring()
}
}
/// Creates a `ChainedCollector`
pub fn chain() -> ChainedCollector<DoNothingCollector, DoNothingCollector> {
ChainedCollector {
left: DoNothingCollector,
right: DoNothingCollector,
}
}
#[cfg(test)]
mod tests {
use super::*;
use collector::{Collector, CountCollector, TopCollector};
#[test]
fn test_chained_collector() {
let mut top_collector = TopCollector::with_limit(2);
let mut count_collector = CountCollector::default();
{
let mut collectors = chain().push(&mut top_collector).push(&mut count_collector);
collectors.collect(1, 0.2);
collectors.collect(2, 0.1);
collectors.collect(3, 0.5);
}
assert_eq!(count_collector.count(), 3);
assert!(top_collector.at_capacity());
}
}

100
src/collector/count_collector.rs
View File

@@ -1,4 +1,5 @@
use super::Collector;
use collector::SegmentCollector;
use DocId;
use Result;
use Score;
@@ -11,14 +12,14 @@ use SegmentReader;
/// ```rust
/// #[macro_use]
/// extern crate tantivy;
/// use tantivy::schema::{SchemaBuilder, TEXT};
/// use tantivy::schema::{Schema, TEXT};
/// use tantivy::{Index, Result};
/// use tantivy::collector::CountCollector;
/// use tantivy::collector::Count;
/// use tantivy::query::QueryParser;
///
/// # fn main() { example().unwrap(); }
/// fn example() -> Result<()> {
/// let mut schema_builder = SchemaBuilder::new();
/// let mut schema_builder = Schema::builder();
/// let title = schema_builder.add_text_field("title", TEXT);
/// let schema = schema_builder.build();
/// let index = Index::create_in_ram(schema);
@@ -39,63 +40,90 @@ use SegmentReader;
/// index_writer.commit().unwrap();
/// }
///
/// index.load_searchers()?;
/// let searcher = index.searcher();
/// let reader = index.reader()?;
/// let searcher = reader.searcher();
///
/// {
/// let mut count_collector = CountCollector::default();
/// let query_parser = QueryParser::for_index(&index, vec![title]);
/// let query = query_parser.parse_query("diary")?;
/// searcher.search(&*query, &mut count_collector).unwrap();
/// let count = searcher.search(&query, &Count).unwrap();
///
/// assert_eq!(count_collector.count(), 2);
/// assert_eq!(count, 2);
/// }
///
/// Ok(())
/// }
/// ```
#[derive(Default)]
pub struct CountCollector {
count: usize,
}
pub struct Count;
impl CountCollector {
/// Returns the count of documents that were
/// collected.
pub fn count(&self) -> usize {
self.count
}
}
impl Collector for Count {
type Fruit = usize;
impl Collector for CountCollector {
fn set_segment(&mut self, _: SegmentLocalId, _: &SegmentReader) -> Result<()> {
Ok(())
}
type Child = SegmentCountCollector;
fn collect(&mut self, _: DocId, _: Score) {
self.count += 1;
fn for_segment(&self, _: SegmentLocalId, _: &SegmentReader) -> Result<SegmentCountCollector> {
Ok(SegmentCountCollector::default())
}
fn requires_scoring(&self) -> bool {
false
}
fn merge_fruits(&self, segment_counts: Vec<usize>) -> Result<usize> {
Ok(segment_counts.into_iter().sum())
}
}
#[derive(Default)]
pub struct SegmentCountCollector {
count: usize,
}
impl SegmentCollector for SegmentCountCollector {
type Fruit = usize;
fn collect(&mut self, _: DocId, _: Score) {
self.count += 1;
}
fn harvest(self) -> usize {
self.count
}
}
#[cfg(test)]
mod tests {
use collector::{Collector, CountCollector};
use super::{Count, SegmentCountCollector};
use collector::Collector;
use collector::SegmentCollector;
#[test]
fn test_count_collector() {
let mut count_collector = CountCollector::default();
assert_eq!(count_collector.count(), 0);
count_collector.collect(0u32, 1f32);
assert_eq!(count_collector.count(), 1);
assert_eq!(count_collector.count(), 1);
count_collector.collect(1u32, 1f32);
assert_eq!(count_collector.count(), 2);
assert!(!count_collector.requires_scoring());
fn test_count_collect_does_not_requires_scoring() {
assert!(!Count.requires_scoring());
}
#[test]
fn test_segment_count_collector() {
{
let count_collector = SegmentCountCollector::default();
assert_eq!(count_collector.harvest(), 0);
}
{
let mut count_collector = SegmentCountCollector::default();
count_collector.collect(0u32, 1f32);
assert_eq!(count_collector.harvest(), 1);
}
{
let mut count_collector = SegmentCountCollector::default();
count_collector.collect(0u32, 1f32);
assert_eq!(count_collector.harvest(), 1);
}
{
let mut count_collector = SegmentCountCollector::default();
count_collector.collect(0u32, 1f32);
count_collector.collect(1u32, 1f32);
assert_eq!(count_collector.harvest(), 2);
}
}
}

394
src/collector/facet_collector.rs
View File

@@ -1,25 +1,23 @@
use collector::Collector;
use collector::SegmentCollector;
use docset::SkipResult;
use fastfield::FacetReader;
use schema::Facet;
use schema::Field;
use std::cell::UnsafeCell;
use std::cmp::Ordering;
use std::collections::btree_map;
use std::collections::BTreeMap;
use std::collections::BTreeSet;
use std::collections::BinaryHeap;
use std::collections::Bound;
use std::iter::Peekable;
use std::mem;
use std::{u64, usize};
use termdict::TermMerger;
use std::cmp::Ordering;
use DocId;
use Result;
use Score;
use SegmentLocalId;
use SegmentReader;
use TantivyError;
struct Hit<'a> {
count: u64,
@@ -46,12 +44,6 @@ impl<'a> Ord for Hit<'a> {
}
}
struct SegmentFacetCounter {
pub facet_reader: FacetReader,
pub facet_ords: Vec<u64>,
pub facet_counts: Vec<u64>,
}
fn facet_depth(facet_bytes: &[u8]) -> usize {
if facet_bytes.is_empty() {
0
@@ -91,14 +83,14 @@ fn facet_depth(facet_bytes: &[u8]) -> usize {
/// ```rust
/// #[macro_use]
/// extern crate tantivy;
/// use tantivy::schema::{Facet, SchemaBuilder, TEXT};
/// use tantivy::schema::{Facet, Schema, TEXT};
/// use tantivy::{Index, Result};
/// use tantivy::collector::FacetCollector;
/// use tantivy::query::AllQuery;
///
/// # fn main() { example().unwrap(); }
/// fn example() -> Result<()> {
/// let mut schema_builder = SchemaBuilder::new();
/// let mut schema_builder = Schema::builder();
///
/// // Facet have their own specific type.
/// // It is not a bad practise to put all of your
@@ -131,23 +123,19 @@ fn facet_depth(facet_bytes: &[u8]) -> usize {
/// facet => Facet::from("/lang/en"),
/// facet => Facet::from("/category/biography")
/// ));
/// index_writer.commit().unwrap();
/// index_writer.commit()?;
/// }
///
/// index.load_searchers()?;
/// let searcher = index.searcher();
/// let reader = index.reader()?;
/// let searcher = reader.searcher();
///
/// {
/// let mut facet_collector = FacetCollector::for_field(facet);
/// facet_collector.add_facet("/lang");
/// facet_collector.add_facet("/category");
/// searcher.search(&AllQuery, &mut facet_collector).unwrap();
///
/// // this object contains count aggregate for all of the facets.
/// let counts = facet_collector.harvest();
/// let facet_counts = searcher.search(&AllQuery, &facet_collector)?;
///
/// // This lists all of the facet counts
/// let facets: Vec<(&Facet, u64)> = counts
/// let facets: Vec<(&Facet, u64)> = facet_counts
/// .get("/category")
/// .collect();
/// assert_eq!(facets, vec![
@@ -159,13 +147,10 @@ fn facet_depth(facet_bytes: &[u8]) -> usize {
/// {
/// let mut facet_collector = FacetCollector::for_field(facet);
/// facet_collector.add_facet("/category/fiction");
/// searcher.search(&AllQuery, &mut facet_collector).unwrap();
///
/// // this object contains count aggregate for all of the facets.
/// let counts = facet_collector.harvest();
/// let facet_counts = searcher.search(&AllQuery, &facet_collector)?;
///
/// // This lists all of the facet counts
/// let facets: Vec<(&Facet, u64)> = counts
/// let facets: Vec<(&Facet, u64)> = facet_counts
/// .get("/category/fiction")
/// .collect();
/// assert_eq!(facets, vec![
@@ -178,13 +163,10 @@ fn facet_depth(facet_bytes: &[u8]) -> usize {
/// {
/// let mut facet_collector = FacetCollector::for_field(facet);
/// facet_collector.add_facet("/category/fiction");
/// searcher.search(&AllQuery, &mut facet_collector).unwrap();
///
/// // this object contains count aggregate for all of the facets.
/// let counts = facet_collector.harvest();
/// let facet_counts = searcher.search(&AllQuery, &facet_collector)?;
///
/// // This lists all of the facet counts
/// let facets: Vec<(&Facet, u64)> = counts.top_k("/category/fiction", 1);
/// let facets: Vec<(&Facet, u64)> = facet_counts.top_k("/category/fiction", 1);
/// assert_eq!(facets, vec![
/// (&Facet::from("/category/fiction/fantasy"), 2)
/// ]);
@@ -194,28 +176,28 @@ fn facet_depth(facet_bytes: &[u8]) -> usize {
/// }
/// ```
pub struct FacetCollector {
facet_ords: Vec<u64>,
field: Field,
ff_reader: Option<UnsafeCell<FacetReader>>,
segment_counters: Vec<SegmentFacetCounter>,
// facet_ord -> collapse facet_id
current_segment_collapse_mapping: Vec<usize>,
// collapse facet_id -> count
current_segment_counts: Vec<u64>,
// collapse facet_id -> facet_ord
current_collapse_facet_ords: Vec<u64>,
facets: BTreeSet<Facet>,
}
pub struct FacetSegmentCollector {
reader: FacetReader,
facet_ords_buf: Vec<u64>,
// facet_ord -> collapse facet_id
collapse_mapping: Vec<usize>,
// collapse facet_id -> count
counts: Vec<u64>,
// collapse facet_id -> facet_ord
collapse_facet_ords: Vec<u64>,
}
fn skip<'a, I: Iterator<Item = &'a Facet>>(
target: &[u8],
collapse_it: &mut Peekable<I>,
) -> SkipResult {
loop {
match collapse_it.peek() {
Some(facet_bytes) => match facet_bytes.encoded_bytes().cmp(target) {
Some(facet_bytes) => match facet_bytes.encoded_str().as_bytes().cmp(target) {
Ordering::Less => {}
Ordering::Greater => {
return SkipResult::OverStep;
@@ -240,15 +222,8 @@ impl FacetCollector {
/// is of the proper type.
pub fn for_field(field: Field) -> FacetCollector {
FacetCollector {
facet_ords: Vec::with_capacity(255),
segment_counters: Vec::new(),
field,
ff_reader: None,
facets: BTreeSet::new(),
current_segment_collapse_mapping: Vec::new(),
current_collapse_facet_ords: Vec::new(),
current_segment_counts: Vec::new(),
facets: BTreeSet::default(),
}
}
@@ -278,141 +253,103 @@ impl FacetCollector {
}
self.facets.insert(facet);
}
fn set_collapse_mapping(&mut self, facet_reader: &FacetReader) {
self.current_segment_collapse_mapping.clear();
self.current_collapse_facet_ords.clear();
self.current_segment_counts.clear();
let mut collapse_facet_it = self.facets.iter().peekable();
self.current_collapse_facet_ords.push(0);
let mut facet_streamer = facet_reader.facet_dict().range().into_stream();
if !facet_streamer.advance() {
return;
}
'outer: loop {
// at the begining of this loop, facet_streamer
// is positionned on a term that has not been processed yet.
let skip_result = skip(facet_streamer.key(), &mut collapse_facet_it);
match skip_result {
SkipResult::Reached => {
// we reach a facet we decided to collapse.
let collapse_depth = facet_depth(facet_streamer.key());
let mut collapsed_id = 0;
self.current_segment_collapse_mapping.push(0);
while facet_streamer.advance() {
let depth = facet_depth(facet_streamer.key());
if depth <= collapse_depth {
continue 'outer;
}
if depth == collapse_depth + 1 {
collapsed_id = self.current_collapse_facet_ords.len();
self.current_collapse_facet_ords
.push(facet_streamer.term_ord());
self.current_segment_collapse_mapping.push(collapsed_id);
} else {
self.current_segment_collapse_mapping.push(collapsed_id);
}
}
break;
}
SkipResult::End | SkipResult::OverStep => {
self.current_segment_collapse_mapping.push(0);
if !facet_streamer.advance() {
break;
}
}
}
}
}
fn finalize_segment(&mut self) {
if self.ff_reader.is_some() {
self.segment_counters.push(SegmentFacetCounter {
facet_reader: self.ff_reader.take().unwrap().into_inner(),
facet_ords: mem::replace(&mut self.current_collapse_facet_ords, Vec::new()),
facet_counts: mem::replace(&mut self.current_segment_counts, Vec::new()),
});
}
}
/// Returns the results of the collection.
///
/// This method does not just return the counters,
/// it also translates the facet ordinals of the last segment.
pub fn harvest(mut self) -> FacetCounts {
self.finalize_segment();
let collapsed_facet_ords: Vec<&[u64]> = self.segment_counters
.iter()
.map(|segment_counter| &segment_counter.facet_ords[..])
.collect();
let collapsed_facet_counts: Vec<&[u64]> = self.segment_counters
.iter()
.map(|segment_counter| &segment_counter.facet_counts[..])
.collect();
let facet_streams = self.segment_counters
.iter()
.map(|seg_counts| seg_counts.facet_reader.facet_dict().range().into_stream())
.collect::<Vec<_>>();
let mut facet_merger = TermMerger::new(facet_streams);
let mut facet_counts = BTreeMap::new();
while facet_merger.advance() {
let count = facet_merger
.current_kvs()
.iter()
.map(|it| {
let seg_ord = it.segment_ord;
let term_ord = it.streamer.term_ord();
collapsed_facet_ords[seg_ord]
.binary_search(&term_ord)
.map(|collapsed_term_id| {
if collapsed_term_id == 0 {
0
} else {
collapsed_facet_counts[seg_ord][collapsed_term_id]
}
})
.unwrap_or(0)
})
.sum();
if count > 0u64 {
let bytes: Vec<u8> = facet_merger.key().to_owned();
// may create an corrupted facet if the term dicitonary is corrupted
let facet = unsafe { Facet::from_encoded(bytes) };
facet_counts.insert(facet, count);
}
}
FacetCounts { facet_counts }
}
}
impl Collector for FacetCollector {
fn set_segment(&mut self, _: SegmentLocalId, reader: &SegmentReader) -> Result<()> {
self.finalize_segment();
let facet_reader = reader.facet_reader(self.field)?;
self.set_collapse_mapping(&facet_reader);
self.current_segment_counts
.resize(self.current_collapse_facet_ords.len(), 0);
self.ff_reader = Some(UnsafeCell::new(facet_reader));
Ok(())
type Fruit = FacetCounts;
type Child = FacetSegmentCollector;
fn for_segment(
&self,
_: SegmentLocalId,
reader: &SegmentReader,
) -> Result<FacetSegmentCollector> {
let field_name = reader.schema().get_field_name(self.field);
let facet_reader = reader.facet_reader(self.field).ok_or_else(|| {
TantivyError::SchemaError(format!("Field {:?} is not a facet field.", field_name))
})?;
let mut collapse_mapping = Vec::new();
let mut counts = Vec::new();
let mut collapse_facet_ords = Vec::new();
let mut collapse_facet_it = self.facets.iter().peekable();
collapse_facet_ords.push(0);
{
let mut facet_streamer = facet_reader.facet_dict().range().into_stream();
if facet_streamer.advance() {
'outer: loop {
// at the begining of this loop, facet_streamer
// is positionned on a term that has not been processed yet.
let skip_result = skip(facet_streamer.key(), &mut collapse_facet_it);
match skip_result {
SkipResult::Reached => {
// we reach a facet we decided to collapse.
let collapse_depth = facet_depth(facet_streamer.key());
let mut collapsed_id = 0;
collapse_mapping.push(0);
while facet_streamer.advance() {
let depth = facet_depth(facet_streamer.key());
if depth <= collapse_depth {
continue 'outer;
}
if depth == collapse_depth + 1 {
collapsed_id = collapse_facet_ords.len();
collapse_facet_ords.push(facet_streamer.term_ord());
collapse_mapping.push(collapsed_id);
} else {
collapse_mapping.push(collapsed_id);
}
}
break;
}
SkipResult::End | SkipResult::OverStep => {
collapse_mapping.push(0);
if !facet_streamer.advance() {
break;
}
}
}
}
}
}
counts.resize(collapse_facet_ords.len(), 0);
Ok(FacetSegmentCollector {
reader: facet_reader,
facet_ords_buf: Vec::with_capacity(255),
collapse_mapping,
counts,
collapse_facet_ords,
})
}
fn requires_scoring(&self) -> bool {
false
}
fn merge_fruits(&self, segments_facet_counts: Vec<FacetCounts>) -> Result<FacetCounts> {
let mut facet_counts: BTreeMap<Facet, u64> = BTreeMap::new();
for segment_facet_counts in segments_facet_counts {
for (facet, count) in segment_facet_counts.facet_counts {
*(facet_counts.entry(facet).or_insert(0)) += count;
}
}
Ok(FacetCounts { facet_counts })
}
}
impl SegmentCollector for FacetSegmentCollector {
type Fruit = FacetCounts;
fn collect(&mut self, doc: DocId, _: Score) {
let facet_reader: &mut FacetReader = unsafe {
&mut *self.ff_reader
.as_ref()
.expect("collect() was called before set_segment. This should never happen.")
.get()
};
facet_reader.facet_ords(doc, &mut self.facet_ords);
self.reader.facet_ords(doc, &mut self.facet_ords_buf);
let mut previous_collapsed_ord: usize = usize::MAX;
for &facet_ord in &self.facet_ords {
let collapsed_ord = self.current_segment_collapse_mapping[facet_ord as usize];
self.current_segment_counts[collapsed_ord] += if collapsed_ord == previous_collapsed_ord
{
for &facet_ord in &self.facet_ords_buf {
let collapsed_ord = self.collapse_mapping[facet_ord as usize];
self.counts[collapsed_ord] += if collapsed_ord == previous_collapsed_ord {
0
} else {
1
@@ -421,8 +358,24 @@ impl Collector for FacetCollector {
}
}
fn requires_scoring(&self) -> bool {
false
/// Returns the results of the collection.
///
/// This method does not just return the counters,
/// it also translates the facet ordinals of the last segment.
fn harvest(self) -> FacetCounts {
let mut facet_counts = BTreeMap::new();
let facet_dict = self.reader.facet_dict();
for (collapsed_facet_ord, count) in self.counts.iter().cloned().enumerate() {
if count == 0 {
continue;
}
let mut facet = vec![];
let facet_ord = self.collapse_facet_ords[collapsed_facet_ord];
facet_dict.ord_to_term(facet_ord as u64, &mut facet);
// TODO
facet_counts.insert(Facet::from_encoded(facet).unwrap(), count);
}
FacetCounts { facet_counts }
}
}
@@ -454,9 +407,9 @@ impl FacetCounts {
let right_bound = if facet.is_root() {
Bound::Unbounded
} else {
let mut facet_after_bytes: Vec<u8> = facet.encoded_bytes().to_owned();
facet_after_bytes.push(1u8);
let facet_after = unsafe { Facet::from_encoded(facet_after_bytes) }; // ok logic
let mut facet_after_bytes: String = facet.encoded_str().to_owned();
facet_after_bytes.push('\u{1}');
let facet_after = Facet::from_encoded_string(facet_after_bytes);
Bound::Excluded(facet_after)
};
let underlying: btree_map::Range<_, _> = self.facet_counts.range((left_bound, right_bound));
@@ -476,9 +429,8 @@ impl FacetCounts {
heap.push(Hit { count, facet });
}
let mut lowest_count: u64 = heap.peek().map(|hit| hit.count)
.unwrap_or(u64::MIN); //< the `unwrap_or` case may be triggered but the value
// is never used in that case.
let mut lowest_count: u64 = heap.peek().map(|hit| hit.count).unwrap_or(u64::MIN); //< the `unwrap_or` case may be triggered but the value
// is never used in that case.
for (facet, count) in it {
if count > lowest_count {
@@ -504,14 +456,14 @@ mod tests {
use core::Index;
use query::AllQuery;
use rand::distributions::Uniform;
use rand::prelude::SliceRandom;
use rand::{thread_rng, Rng};
use schema::Field;
use schema::{Document, Facet, SchemaBuilder};
use schema::{Document, Facet, Field, Schema};
use std::iter;
#[test]
fn test_facet_collector_drilldown() {
let mut schema_builder = SchemaBuilder::new();
let mut schema_builder = Schema::builder();
let facet_field = schema_builder.add_facet_field("facet");
let schema = schema_builder.build();
let index = Index::create_in_ram(schema);
@@ -534,14 +486,12 @@ mod tests {
index_writer.add_document(doc);
}
index_writer.commit().unwrap();
index.load_searchers().unwrap();
let searcher = index.searcher();
let reader = index.reader().unwrap();
let searcher = reader.searcher();
let mut facet_collector = FacetCollector::for_field(facet_field);
facet_collector.add_facet(Facet::from("/top1"));
searcher.search(&AllQuery, &mut facet_collector).unwrap();
let counts = searcher.search(&AllQuery, &facet_collector).unwrap();
let counts: FacetCounts = facet_collector.harvest();
{
let facets: Vec<(String, u64)> = counts
.get("/top1")
@@ -554,18 +504,17 @@ mod tests {
("/top1/mid1", 50),
("/top1/mid2", 50),
("/top1/mid3", 50),
].iter()
.map(|&(facet_str, count)| (String::from(facet_str), count))
.collect::<Vec<_>>()
]
.iter()
.map(|&(facet_str, count)| (String::from(facet_str), count))
.collect::<Vec<_>>()
);
}
}
#[test]
#[should_panic(
expected = "Tried to add a facet which is a descendant of \
an already added facet."
)]
#[should_panic(expected = "Tried to add a facet which is a descendant of \
an already added facet.")]
fn test_misused_facet_collector() {
let mut facet_collector = FacetCollector::for_field(Field(0));
facet_collector.add_facet(Facet::from("/country"));
@@ -574,7 +523,7 @@ mod tests {
#[test]
fn test_doc_unsorted_multifacet() {
let mut schema_builder = SchemaBuilder::new();
let mut schema_builder = Schema::builder();
let facet_field = schema_builder.add_facet_field("facets");
let schema = schema_builder.build();
let index = Index::create_in_ram(schema);
@@ -586,13 +535,12 @@ mod tests {
facet_field => Facet::from_text(&"/subjects/B/b"),
));
index_writer.commit().unwrap();
index.load_searchers().unwrap();
let searcher = index.searcher();
let reader = index.reader().unwrap();
let searcher = reader.searcher();
assert_eq!(searcher.num_docs(), 1);
let mut facet_collector = FacetCollector::for_field(facet_field);
facet_collector.add_facet("/subjects");
searcher.search(&AllQuery, &mut facet_collector).unwrap();
let counts = facet_collector.harvest();
let counts = searcher.search(&AllQuery, &facet_collector).unwrap();
let facets: Vec<(&Facet, u64)> = counts.get("/subjects").collect();
assert_eq!(facets[0].1, 1);
}
@@ -606,7 +554,7 @@ mod tests {
#[test]
fn test_facet_collector_topk() {
let mut schema_builder = SchemaBuilder::new();
let mut schema_builder = Schema::builder();
let facet_field = schema_builder.add_facet_field("facet");
let schema = schema_builder.build();
let index = Index::create_in_ram(schema);
@@ -619,24 +567,27 @@ mod tests {
let doc = doc!(facet_field => facet);
iter::repeat(doc).take(count)
})
.map(|mut doc| { doc.add_facet(facet_field, &format!("/facet/{}", thread_rng().sample(&uniform) )); doc})
.map(|mut doc| {
doc.add_facet(
facet_field,
&format!("/facet/{}", thread_rng().sample(&uniform)),
);
doc
})
.collect();
thread_rng().shuffle(&mut docs[..]);
docs[..].shuffle(&mut thread_rng());
let mut index_writer = index.writer_with_num_threads(1, 3_000_000).unwrap();
for doc in docs {
index_writer.add_document(doc);
}
index_writer.commit().unwrap();
index.load_searchers().unwrap();
let searcher = index.searcher();
let searcher = index.reader().unwrap().searcher();
let mut facet_collector = FacetCollector::for_field(facet_field);
facet_collector.add_facet("/facet");
searcher.search(&AllQuery, &mut facet_collector).unwrap();
let counts: FacetCounts = searcher.search(&AllQuery, &facet_collector).unwrap();
let counts: FacetCounts = facet_collector.harvest();
{
let facets: Vec<(&Facet, u64)> = counts.top_k("/facet", 3);
assert_eq!(
@@ -659,13 +610,13 @@ mod bench {
use query::AllQuery;
use rand::{thread_rng, Rng};
use schema::Facet;
use schema::SchemaBuilder;
use schema::Schema;
use test::Bencher;
use Index;
#[bench]
fn bench_facet_collector(b: &mut Bencher) {
let mut schema_builder = SchemaBuilder::new();
let mut schema_builder = Schema::builder();
let facet_field = schema_builder.add_facet_field("facet");
let schema = schema_builder.build();
let index = Index::create_in_ram(schema);
@@ -685,12 +636,11 @@ mod bench {
index_writer.add_document(doc);
}
index_writer.commit().unwrap();
index.load_searchers().unwrap();
let reader = index.reader().unwrap();
b.iter(|| {
let searcher = index.searcher();
let mut facet_collector = FacetCollector::for_field(facet_field);
searcher.search(&AllQuery, &mut facet_collector).unwrap();
let facet_collector = FacetCollector::for_field(facet_field);
searcher.search(&AllQuery, &facet_collector).unwrap();
});
}
}

7
src/collector/int_facet_collector.rs
View File

@@ -79,7 +79,7 @@ mod tests {
// make sure we have facet counters correctly filled
fn test_facet_collector_results() {
let mut schema_builder = schema::SchemaBuilder::new();
let mut schema_builder = schema::Schema::builder();
let num_field_i64 = schema_builder.add_i64_field("num_i64", FAST);
let num_field_u64 = schema_builder.add_u64_field("num_u64", FAST);
let text_field = schema_builder.add_text_field("text", STRING);
@@ -88,7 +88,7 @@ mod tests {
let index = Index::create_in_ram(schema.clone());
{
let mut index_writer = index.writer_with_num_threads(1, 40_000_000).unwrap();
let mut index_writer = index.writer_with_num_threads(1, 3_000_000).unwrap();
{
for i in 0u64..10u64 {
index_writer.add_document(doc!(
@@ -101,8 +101,7 @@ mod tests {
assert_eq!(index_writer.commit().unwrap(), 10u64);
}
index.load_searchers().unwrap();
let searcher = index.searcher();
let searcher = index.reader().searcher();
let mut ffvf_i64: IntFacetCollector<I64FastFieldReader> = IntFacetCollector::new(num_field_i64);
let mut ffvf_u64: IntFacetCollector<U64FastFieldReader> = IntFacetCollector::new(num_field_u64);

512
src/collector/mod.rs
View File

@@ -1,7 +1,91 @@
/*!
Defines how the documents matching a search query should be processed.
# Collectors
Collectors define the information you want to extract from the documents matching the queries.
In tantivy jargon, we call this information your search "fruit".
Your fruit could for instance be :
- [the count of matching documents](./struct.Count.html)
- [the top 10 documents, by relevancy or by a fast field](./struct.TopDocs.html)
- [facet counts](./struct.FacetCollector.html)
At one point in your code, you will trigger the actual search operation by calling
[the `search(...)` method of your `Searcher` object](../struct.Searcher.html#method.search).
This call will look like this.
```verbatim
let fruit = searcher.search(&query, &collector)?;
```
Here the type of fruit is actually determined as an associated type of the collector (`Collector::Fruit`).
# Combining several collectors
A rich search experience often requires to run several collectors on your search query.
For instance,
- selecting the top-K products matching your query
- counting the matching documents
- computing several facets
- computing statistics about the matching product prices
A simple and efficient way to do that is to pass your collectors as one tuple.
The resulting `Fruit` will then be a typed tuple with each collector's original fruits
in their respective position.
```rust
# extern crate tantivy;
# use tantivy::schema::*;
# use tantivy::*;
# use tantivy::query::*;
use tantivy::collector::{Count, TopDocs};
#
# fn main() -> tantivy::Result<()> {
# let mut schema_builder = Schema::builder();
# let title = schema_builder.add_text_field("title", TEXT);
# let schema = schema_builder.build();
# let index = Index::create_in_ram(schema);
# let mut index_writer = index.writer(3_000_000)?;
# index_writer.add_document(doc!(
# title => "The Name of the Wind",
# ));
# index_writer.add_document(doc!(
# title => "The Diary of Muadib",
# ));
# index_writer.commit()?;
# let reader = index.reader()?;
# let searcher = reader.searcher();
# let query_parser = QueryParser::for_index(&index, vec![title]);
# let query = query_parser.parse_query("diary")?;
let (doc_count, top_docs): (usize, Vec<(Score, DocAddress)>) =
searcher.search(&query, &(Count, TopDocs::with_limit(2)))?;
# Ok(())
# }
```
The `Collector` trait is implemented for up to 4 collectors.
If you have more than 4 collectors, you can either group them into
tuples of tuples `(a,(b,(c,d)))`, or rely on `MultiCollector`'s.
# Combining several collectors dynamically
Combining collectors into a tuple is a zero-cost abstraction: everything
happens as if you had manually implemented a single collector
combining all of our features.
Unfortunately it requires you to know at compile time your collector types.
If on the other hand, the collectors depend on some query parameter,
you can rely on `MultiCollector`'s.
# Implementing your own collectors.
See the `custom_collector` example.
*/
use downcast_rs;
use DocId;
use Result;
use Score;
@@ -9,238 +93,272 @@ use SegmentLocalId;
use SegmentReader;
mod count_collector;
pub use self::count_collector::CountCollector;
pub use self::count_collector::Count;
mod multi_collector;
pub use self::multi_collector::MultiCollector;
mod top_collector;
pub use self::top_collector::TopCollector;
mod top_score_collector;
pub use self::top_score_collector::TopDocs;
mod top_field_collector;
pub use self::top_field_collector::TopDocsByField;
mod facet_collector;
pub use self::facet_collector::FacetCollector;
mod chained_collector;
pub use self::chained_collector::{chain, ChainedCollector};
/// `Fruit` is the type for the result of our collection.
/// e.g. `usize` for the `Count` collector.
pub trait Fruit: Send + downcast_rs::Downcast {}
impl<T> Fruit for T where T: Send + downcast_rs::Downcast {}
/// Collectors are in charge of collecting and retaining relevant
/// information from the document found and scored by the query.
///
///
/// For instance,
///
/// - keeping track of the top 10 best documents
/// - computing a breakdown over a fast field
/// - computing the number of documents matching the query
///
/// Queries are in charge of pushing the `DocSet` to the collector.
/// Our search index is in fact a collection of segments, so
/// a `Collector` trait is actually more of a factory to instance
/// `SegmentCollector`s for each segments.
///
/// As they work on multiple segments, they first inform
/// the collector of a change in a segment and then
/// call the `collect` method to push the document to the collector.
///
/// Temporally, our collector will receive calls
/// - `.set_segment(0, segment_reader_0)`
/// - `.collect(doc0_of_segment_0)`
/// - `.collect(...)`
/// - `.collect(last_doc_of_segment_0)`
/// - `.set_segment(1, segment_reader_1)`
/// - `.collect(doc0_of_segment_1)`
/// - `.collect(...)`
/// - `.collect(last_doc_of_segment_1)`
/// - `...`
/// - `.collect(last_doc_of_last_segment)`
/// The collection logic itself is in the `SegmentCollector`.
///
/// Segments are not guaranteed to be visited in any specific order.
pub trait Collector {
pub trait Collector: Sync {
/// `Fruit` is the type for the result of our collection.
/// e.g. `usize` for the `Count` collector.
type Fruit: Fruit;
/// Type of the `SegmentCollector` associated to this collector.
type Child: SegmentCollector<Fruit = Self::Fruit>;
/// `set_segment` is called before beginning to enumerate
/// on this segment.
fn set_segment(
&mut self,
fn for_segment(
&self,
segment_local_id: SegmentLocalId,
segment: &SegmentReader,
) -> Result<()>;
/// The query pushes the scored document to the collector via this method.
fn collect(&mut self, doc: DocId, score: Score);
) -> Result<Self::Child>;
/// Returns true iff the collector requires to compute scores for documents.
fn requires_scoring(&self) -> bool;
/// Combines the fruit associated to the collection of each segments
/// into one fruit.
fn merge_fruits(&self, segment_fruits: Vec<Self::Fruit>) -> Result<Self::Fruit>;
}
impl<'a, C: Collector> Collector for &'a mut C {
fn set_segment(
&mut self,
segment_local_id: SegmentLocalId,
segment: &SegmentReader,
) -> Result<()> {
(*self).set_segment(segment_local_id, segment)
}
/// The `SegmentCollector` is the trait in charge of defining the
/// collect operation at the scale of the segment.
///
/// `.collect(doc, score)` will be called for every documents
/// matching the query.
pub trait SegmentCollector: 'static {
/// `Fruit` is the type for the result of our collection.
/// e.g. `usize` for the `Count` collector.
type Fruit: Fruit;
/// The query pushes the scored document to the collector via this method.
fn collect(&mut self, doc: DocId, score: Score) {
C::collect(self, doc, score)
fn collect(&mut self, doc: DocId, score: Score);
/// Extract the fruit of the collection from the `SegmentCollector`.
fn harvest(self) -> Self::Fruit;
}
// -----------------------------------------------
// Tuple implementations.
impl<Left, Right> Collector for (Left, Right)
where
Left: Collector,
Right: Collector,
{
type Fruit = (Left::Fruit, Right::Fruit);
type Child = (Left::Child, Right::Child);
fn for_segment(&self, segment_local_id: u32, segment: &SegmentReader) -> Result<Self::Child> {
let left = self.0.for_segment(segment_local_id, segment)?;
let right = self.1.for_segment(segment_local_id, segment)?;
Ok((left, right))
}
fn requires_scoring(&self) -> bool {
C::requires_scoring(self)
self.0.requires_scoring() || self.1.requires_scoring()
}
fn merge_fruits(
&self,
children: Vec<(Left::Fruit, Right::Fruit)>,
) -> Result<(Left::Fruit, Right::Fruit)> {
let mut left_fruits = vec![];
let mut right_fruits = vec![];
for (left_fruit, right_fruit) in children {
left_fruits.push(left_fruit);
right_fruits.push(right_fruit);
}
Ok((
self.0.merge_fruits(left_fruits)?,
self.1.merge_fruits(right_fruits)?,
))
}
}
impl<Left, Right> SegmentCollector for (Left, Right)
where
Left: SegmentCollector,
Right: SegmentCollector,
{
type Fruit = (Left::Fruit, Right::Fruit);
fn collect(&mut self, doc: DocId, score: Score) {
self.0.collect(doc, score);
self.1.collect(doc, score);
}
fn harvest(self) -> <Self as SegmentCollector>::Fruit {
(self.0.harvest(), self.1.harvest())
}
}
// 3-Tuple
impl<One, Two, Three> Collector for (One, Two, Three)
where
One: Collector,
Two: Collector,
Three: Collector,
{
type Fruit = (One::Fruit, Two::Fruit, Three::Fruit);
type Child = (One::Child, Two::Child, Three::Child);
fn for_segment(&self, segment_local_id: u32, segment: &SegmentReader) -> Result<Self::Child> {
let one = self.0.for_segment(segment_local_id, segment)?;
let two = self.1.for_segment(segment_local_id, segment)?;
let three = self.2.for_segment(segment_local_id, segment)?;
Ok((one, two, three))
}
fn requires_scoring(&self) -> bool {
self.0.requires_scoring() || self.1.requires_scoring() || self.2.requires_scoring()
}
fn merge_fruits(&self, children: Vec<Self::Fruit>) -> Result<Self::Fruit> {
let mut one_fruits = vec![];
let mut two_fruits = vec![];
let mut three_fruits = vec![];
for (one_fruit, two_fruit, three_fruit) in children {
one_fruits.push(one_fruit);
two_fruits.push(two_fruit);
three_fruits.push(three_fruit);
}
Ok((
self.0.merge_fruits(one_fruits)?,
self.1.merge_fruits(two_fruits)?,
self.2.merge_fruits(three_fruits)?,
))
}
}
impl<One, Two, Three> SegmentCollector for (One, Two, Three)
where
One: SegmentCollector,
Two: SegmentCollector,
Three: SegmentCollector,
{
type Fruit = (One::Fruit, Two::Fruit, Three::Fruit);
fn collect(&mut self, doc: DocId, score: Score) {
self.0.collect(doc, score);
self.1.collect(doc, score);
self.2.collect(doc, score);
}
fn harvest(self) -> <Self as SegmentCollector>::Fruit {
(self.0.harvest(), self.1.harvest(), self.2.harvest())
}
}
// 4-Tuple
impl<One, Two, Three, Four> Collector for (One, Two, Three, Four)
where
One: Collector,
Two: Collector,
Three: Collector,
Four: Collector,
{
type Fruit = (One::Fruit, Two::Fruit, Three::Fruit, Four::Fruit);
type Child = (One::Child, Two::Child, Three::Child, Four::Child);
fn for_segment(&self, segment_local_id: u32, segment: &SegmentReader) -> Result<Self::Child> {
let one = self.0.for_segment(segment_local_id, segment)?;
let two = self.1.for_segment(segment_local_id, segment)?;
let three = self.2.for_segment(segment_local_id, segment)?;
let four = self.3.for_segment(segment_local_id, segment)?;
Ok((one, two, three, four))
}
fn requires_scoring(&self) -> bool {
self.0.requires_scoring()
|| self.1.requires_scoring()
|| self.2.requires_scoring()
|| self.3.requires_scoring()
}
fn merge_fruits(&self, children: Vec<Self::Fruit>) -> Result<Self::Fruit> {
let mut one_fruits = vec![];
let mut two_fruits = vec![];
let mut three_fruits = vec![];
let mut four_fruits = vec![];
for (one_fruit, two_fruit, three_fruit, four_fruit) in children {
one_fruits.push(one_fruit);
two_fruits.push(two_fruit);
three_fruits.push(three_fruit);
four_fruits.push(four_fruit);
}
Ok((
self.0.merge_fruits(one_fruits)?,
self.1.merge_fruits(two_fruits)?,
self.2.merge_fruits(three_fruits)?,
self.3.merge_fruits(four_fruits)?,
))
}
}
impl<One, Two, Three, Four> SegmentCollector for (One, Two, Three, Four)
where
One: SegmentCollector,
Two: SegmentCollector,
Three: SegmentCollector,
Four: SegmentCollector,
{
type Fruit = (One::Fruit, Two::Fruit, Three::Fruit, Four::Fruit);
fn collect(&mut self, doc: DocId, score: Score) {
self.0.collect(doc, score);
self.1.collect(doc, score);
self.2.collect(doc, score);
self.3.collect(doc, score);
}
fn harvest(self) -> <Self as SegmentCollector>::Fruit {
(
self.0.harvest(),
self.1.harvest(),
self.2.harvest(),
self.3.harvest(),
)
}
}
impl_downcast!(Fruit);
#[cfg(test)]
pub mod tests {
use super::*;
use core::SegmentReader;
use fastfield::BytesFastFieldReader;
use fastfield::FastFieldReader;
use schema::Field;
use DocId;
use Score;
use SegmentLocalId;
/// Stores all of the doc ids.
/// This collector is only used for tests.
/// It is unusable in practise, as it does not store
/// the segment ordinals
pub struct TestCollector {
offset: DocId,
segment_max_doc: DocId,
docs: Vec<DocId>,
scores: Vec<Score>,
}
impl TestCollector {
/// Return the exhalist of documents.
pub fn docs(self) -> Vec<DocId> {
self.docs
}
pub fn scores(self) -> Vec<Score> {
self.scores
}
}
impl Default for TestCollector {
fn default() -> TestCollector {
TestCollector {
offset: 0,
segment_max_doc: 0,
docs: Vec::new(),
scores: Vec::new(),
}
}
}
impl Collector for TestCollector {
fn set_segment(&mut self, _: SegmentLocalId, reader: &SegmentReader) -> Result<()> {
self.offset += self.segment_max_doc;
self.segment_max_doc = reader.max_doc();
Ok(())
}
fn collect(&mut self, doc: DocId, score: Score) {
self.docs.push(doc + self.offset);
self.scores.push(score);
}
fn requires_scoring(&self) -> bool {
true
}
}
/// Collects in order all of the fast fields for all of the
/// doc in the `DocSet`
///
/// This collector is mainly useful for tests.
pub struct FastFieldTestCollector {
vals: Vec<u64>,
field: Field,
ff_reader: Option<FastFieldReader<u64>>,
}
impl FastFieldTestCollector {
pub fn for_field(field: Field) -> FastFieldTestCollector {
FastFieldTestCollector {
vals: Vec::new(),
field,
ff_reader: None,
}
}
pub fn vals(self) -> Vec<u64> {
self.vals
}
}
impl Collector for FastFieldTestCollector {
fn set_segment(&mut self, _: SegmentLocalId, reader: &SegmentReader) -> Result<()> {
self.ff_reader = Some(reader.fast_field_reader(self.field)?);
Ok(())
}
fn collect(&mut self, doc: DocId, _score: Score) {
let val = self.ff_reader.as_ref().unwrap().get(doc);
self.vals.push(val);
}
fn requires_scoring(&self) -> bool {
false
}
}
/// Collects in order all of the fast field bytes for all of the
/// docs in the `DocSet`
///
/// This collector is mainly useful for tests.
pub struct BytesFastFieldTestCollector {
vals: Vec<u8>,
field: Field,
ff_reader: Option<BytesFastFieldReader>,
}
impl BytesFastFieldTestCollector {
pub fn for_field(field: Field) -> BytesFastFieldTestCollector {
BytesFastFieldTestCollector {
vals: Vec::new(),
field,
ff_reader: None,
}
}
pub fn vals(self) -> Vec<u8> {
self.vals
}
}
impl Collector for BytesFastFieldTestCollector {
fn set_segment(&mut self, _segment_local_id: u32, segment: &SegmentReader) -> Result<()> {
self.ff_reader = Some(segment.bytes_fast_field_reader(self.field)?);
Ok(())
}
fn collect(&mut self, doc: u32, _score: f32) {
let val = self.ff_reader.as_ref().unwrap().get_val(doc);
self.vals.extend(val);
}
fn requires_scoring(&self) -> bool {
false
}
}
}
#[cfg(all(test, feature = "unstable"))]
mod bench {
use collector::{Collector, CountCollector};
use test::Bencher;
#[bench]
fn build_collector(b: &mut Bencher) {
b.iter(|| {
let mut count_collector = CountCollector::default();
let docs: Vec<u32> = (0..1_000_000).collect();
for doc in docs {
count_collector.collect(doc, 1f32);
}
count_collector.count()
});
}
}
pub mod tests;

270
src/collector/multi_collector.rs
View File

@@ -1,26 +1,120 @@
use super::Collector;
use super::SegmentCollector;
use collector::Fruit;
use std::marker::PhantomData;
use std::ops::Deref;
use DocId;
use Result;
use Score;
use SegmentLocalId;
use SegmentReader;
use TantivyError;
pub struct MultiFruit {
sub_fruits: Vec<Option<Box<Fruit>>>,
}
pub struct CollectorWrapper<TCollector: Collector>(TCollector);
impl<TCollector: Collector> Collector for CollectorWrapper<TCollector> {
type Fruit = Box<Fruit>;
type Child = Box<BoxableSegmentCollector>;
fn for_segment(
&self,
segment_local_id: u32,
reader: &SegmentReader,
) -> Result<Box<BoxableSegmentCollector>> {
let child = self.0.for_segment(segment_local_id, reader)?;
Ok(Box::new(SegmentCollectorWrapper(child)))
}
fn requires_scoring(&self) -> bool {
self.0.requires_scoring()
}
fn merge_fruits(&self, children: Vec<<Self as Collector>::Fruit>) -> Result<Box<Fruit>> {
let typed_fruit: Vec<TCollector::Fruit> = children
.into_iter()
.map(|untyped_fruit| {
untyped_fruit
.downcast::<TCollector::Fruit>()
.map(|boxed_but_typed| *boxed_but_typed)
.map_err(|_| {
TantivyError::InvalidArgument("Failed to cast child fruit.".to_string())
})
})
.collect::<Result<_>>()?;
let merged_fruit = self.0.merge_fruits(typed_fruit)?;
Ok(Box::new(merged_fruit))
}
}
impl SegmentCollector for Box<BoxableSegmentCollector> {
type Fruit = Box<Fruit>;
fn collect(&mut self, doc: u32, score: f32) {
self.as_mut().collect(doc, score);
}
fn harvest(self) -> Box<Fruit> {
BoxableSegmentCollector::harvest_from_box(self)
}
}
pub trait BoxableSegmentCollector {
fn collect(&mut self, doc: u32, score: f32);
fn harvest_from_box(self: Box<Self>) -> Box<Fruit>;
}
pub struct SegmentCollectorWrapper<TSegmentCollector: SegmentCollector>(TSegmentCollector);
impl<TSegmentCollector: SegmentCollector> BoxableSegmentCollector
for SegmentCollectorWrapper<TSegmentCollector>
{
fn collect(&mut self, doc: u32, score: f32) {
self.0.collect(doc, score);
}
fn harvest_from_box(self: Box<Self>) -> Box<Fruit> {
Box::new(self.0.harvest())
}
}
pub struct FruitHandle<TFruit: Fruit> {
pos: usize,
_phantom: PhantomData<TFruit>,
}
impl<TFruit: Fruit> FruitHandle<TFruit> {
pub fn extract(self, fruits: &mut MultiFruit) -> TFruit {
let boxed_fruit = fruits.sub_fruits[self.pos].take().expect("");
*boxed_fruit
.downcast::<TFruit>()
.map_err(|_| ())
.expect("Failed to downcast collector fruit.")
}
}
/// Multicollector makes it possible to collect on more than one collector.
/// It should only be used for use cases where the Collector types is unknown
/// at compile time.
/// If the type of the collectors is known, you should prefer to use `ChainedCollector`.
///
/// If the type of the collectors is known, you can just group yours collectors
/// in a tuple. See the
/// [Combining several collectors section of the collector documentation](./index.html#combining-several-collectors).
///
/// ```rust
/// #[macro_use]
/// extern crate tantivy;
/// use tantivy::schema::{SchemaBuilder, TEXT};
/// use tantivy::schema::{Schema, TEXT};
/// use tantivy::{Index, Result};
/// use tantivy::collector::{CountCollector, TopCollector, MultiCollector};
/// use tantivy::collector::{Count, TopDocs, MultiCollector};
/// use tantivy::query::QueryParser;
///
/// # fn main() { example().unwrap(); }
/// fn example() -> Result<()> {
/// let mut schema_builder = SchemaBuilder::new();
/// let mut schema_builder = Schema::builder();
/// let title = schema_builder.add_text_field("title", TEXT);
/// let schema = schema_builder.build();
/// let index = Index::create_in_ram(schema);
@@ -41,58 +135,121 @@ use SegmentReader;
/// index_writer.commit().unwrap();
/// }
///
/// index.load_searchers()?;
/// let searcher = index.searcher();
/// let reader = index.reader()?;
/// let searcher = reader.searcher();
///
/// {
/// let mut top_collector = TopCollector::with_limit(2);
/// let mut count_collector = CountCollector::default();
/// {
/// let mut collectors =
/// MultiCollector::from(vec![&mut top_collector, &mut count_collector]);
/// let query_parser = QueryParser::for_index(&index, vec![title]);
/// let query = query_parser.parse_query("diary")?;
/// searcher.search(&*query, &mut collectors).unwrap();
/// }
/// assert_eq!(count_collector.count(), 2);
/// assert!(top_collector.at_capacity());
/// }
/// let mut collectors = MultiCollector::new();
/// let top_docs_handle = collectors.add_collector(TopDocs::with_limit(2));
/// let count_handle = collectors.add_collector(Count);
/// let query_parser = QueryParser::for_index(&index, vec![title]);
/// let query = query_parser.parse_query("diary")?;
/// let mut multi_fruit = searcher.search(&query, &collectors)?;
///
/// let count = count_handle.extract(&mut multi_fruit);
/// let top_docs = top_docs_handle.extract(&mut multi_fruit);
///
/// # assert_eq!(count, 2);
/// # assert_eq!(top_docs.len(), 2);
///
/// Ok(())
/// }
/// ```
#[allow(clippy::type_complexity)]
#[derive(Default)]
pub struct MultiCollector<'a> {
collectors: Vec<&'a mut Collector>,
collector_wrappers:
Vec<Box<Collector<Child = Box<BoxableSegmentCollector>, Fruit = Box<Fruit>> + 'a>>,
}
impl<'a> MultiCollector<'a> {
/// Constructor
pub fn from(collectors: Vec<&'a mut Collector>) -> MultiCollector {
MultiCollector { collectors }
/// Create a new `MultiCollector`
pub fn new() -> Self {
Default::default()
}
/// Add a new collector to our `MultiCollector`.
pub fn add_collector<'b: 'a, TCollector: Collector + 'b>(
&mut self,
collector: TCollector,
) -> FruitHandle<TCollector::Fruit> {
let pos = self.collector_wrappers.len();
self.collector_wrappers
.push(Box::new(CollectorWrapper(collector)));
FruitHandle {
pos,
_phantom: PhantomData,
}
}
}
impl<'a> Collector for MultiCollector<'a> {
fn set_segment(
&mut self,
type Fruit = MultiFruit;
type Child = MultiCollectorChild;
fn for_segment(
&self,
segment_local_id: SegmentLocalId,
segment: &SegmentReader,
) -> Result<()> {
for collector in &mut self.collectors {
collector.set_segment(segment_local_id, segment)?;
}
Ok(())
) -> Result<MultiCollectorChild> {
let children = self
.collector_wrappers
.iter()
.map(|collector_wrapper| collector_wrapper.for_segment(segment_local_id, segment))
.collect::<Result<Vec<_>>>()?;
Ok(MultiCollectorChild { children })
}
fn requires_scoring(&self) -> bool {
self.collector_wrappers
.iter()
.map(Deref::deref)
.any(Collector::requires_scoring)
}
fn merge_fruits(&self, segments_multifruits: Vec<MultiFruit>) -> Result<MultiFruit> {
let mut segment_fruits_list: Vec<Vec<Box<Fruit>>> = (0..self.collector_wrappers.len())
.map(|_| Vec::with_capacity(segments_multifruits.len()))
.collect::<Vec<_>>();
for segment_multifruit in segments_multifruits {
for (idx, segment_fruit_opt) in segment_multifruit.sub_fruits.into_iter().enumerate() {
if let Some(segment_fruit) = segment_fruit_opt {
segment_fruits_list[idx].push(segment_fruit);
}
}
}
let sub_fruits = self
.collector_wrappers
.iter()
.zip(segment_fruits_list)
.map(|(child_collector, segment_fruits)| {
Ok(Some(child_collector.merge_fruits(segment_fruits)?))
})
.collect::<Result<_>>()?;
Ok(MultiFruit { sub_fruits })
}
}
pub struct MultiCollectorChild {
children: Vec<Box<BoxableSegmentCollector>>,
}
impl SegmentCollector for MultiCollectorChild {
type Fruit = MultiFruit;
fn collect(&mut self, doc: DocId, score: Score) {
for collector in &mut self.collectors {
collector.collect(doc, score);
for child in &mut self.children {
child.collect(doc, score);
}
}
fn requires_scoring(&self) -> bool {
self.collectors
.iter()
.any(|collector| collector.requires_scoring())
fn harvest(self) -> MultiFruit {
MultiFruit {
sub_fruits: self
.children
.into_iter()
.map(|child| Some(child.harvest()))
.collect(),
}
}
}
@@ -100,20 +257,41 @@ impl<'a> Collector for MultiCollector<'a> {
mod tests {
use super::*;
use collector::{Collector, CountCollector, TopCollector};
use collector::{Count, TopDocs};
use query::TermQuery;
use schema::IndexRecordOption;
use schema::{Schema, TEXT};
use Index;
use Term;
#[test]
fn test_multi_collector() {
let mut top_collector = TopCollector::with_limit(2);
let mut count_collector = CountCollector::default();
let mut schema_builder = Schema::builder();
let text = schema_builder.add_text_field("text", TEXT);
let schema = schema_builder.build();
let index = Index::create_in_ram(schema);
{
let mut collectors =
MultiCollector::from(vec![&mut top_collector, &mut count_collector]);
collectors.collect(1, 0.2);
collectors.collect(2, 0.1);
collectors.collect(3, 0.5);
let mut index_writer = index.writer_with_num_threads(1, 3_000_000).unwrap();
index_writer.add_document(doc!(text=>"abc"));
index_writer.add_document(doc!(text=>"abc abc abc"));
index_writer.add_document(doc!(text=>"abc abc"));
index_writer.commit().unwrap();
index_writer.add_document(doc!(text=>""));
index_writer.add_document(doc!(text=>"abc abc abc abc"));
index_writer.add_document(doc!(text=>"abc"));
index_writer.commit().unwrap();
}
assert_eq!(count_collector.count(), 3);
assert!(top_collector.at_capacity());
let searcher = index.reader().unwrap().searcher();
let term = Term::from_field_text(text, "abc");
let query = TermQuery::new(term, IndexRecordOption::Basic);
let mut collectors = MultiCollector::new();
let topdocs_handler = collectors.add_collector(TopDocs::with_limit(2));
let count_handler = collectors.add_collector(Count);
let mut multifruits = searcher.search(&query, &mut collectors).unwrap();
assert_eq!(count_handler.extract(&mut multifruits), 5);
assert_eq!(topdocs_handler.extract(&mut multifruits).len(), 2);
}
}

208
src/collector/tests.rs Normal file
View File

@@ -0,0 +1,208 @@
use super::*;
use core::SegmentReader;
use fastfield::BytesFastFieldReader;
use fastfield::FastFieldReader;
use schema::Field;
use DocAddress;
use DocId;
use Score;
use SegmentLocalId;
/// Stores all of the doc ids.
/// This collector is only used for tests.
/// It is unusable in pr
///
/// actise, as it does not store
/// the segment ordinals
pub struct TestCollector;
pub struct TestSegmentCollector {
segment_id: SegmentLocalId,
fruit: TestFruit,
}
#[derive(Default)]
pub struct TestFruit {
docs: Vec<DocAddress>,
scores: Vec<Score>,
}
impl TestFruit {
/// Return the list of matching documents exhaustively.
pub fn docs(&self) -> &[DocAddress] {
&self.docs[..]
}
pub fn scores(&self) -> &[Score] {
&self.scores[..]
}
}
impl Collector for TestCollector {
type Fruit = TestFruit;
type Child = TestSegmentCollector;
fn for_segment(
&self,
segment_id: SegmentLocalId,
_reader: &SegmentReader,
) -> Result<TestSegmentCollector> {
Ok(TestSegmentCollector {
segment_id,
fruit: TestFruit::default(),
})
}
fn requires_scoring(&self) -> bool {
true
}
fn merge_fruits(&self, mut children: Vec<TestFruit>) -> Result<TestFruit> {
children.sort_by_key(|fruit| {
if fruit.docs().is_empty() {
0
} else {
fruit.docs()[0].segment_ord()
}
});
let mut docs = vec![];
let mut scores = vec![];
for child in children {
docs.extend(child.docs());
scores.extend(child.scores);
}
Ok(TestFruit { docs, scores })
}
}
impl SegmentCollector for TestSegmentCollector {
type Fruit = TestFruit;
fn collect(&mut self, doc: DocId, score: Score) {
self.fruit.docs.push(DocAddress(self.segment_id, doc));
self.fruit.scores.push(score);
}
fn harvest(self) -> <Self as SegmentCollector>::Fruit {
self.fruit
}
}
/// Collects in order all of the fast fields for all of the
/// doc in the `DocSet`
///
/// This collector is mainly useful for tests.
pub struct FastFieldTestCollector {
field: Field,
}
pub struct FastFieldSegmentCollector {
vals: Vec<u64>,
reader: FastFieldReader<u64>,
}
impl FastFieldTestCollector {
pub fn for_field(field: Field) -> FastFieldTestCollector {
FastFieldTestCollector { field }
}
}
impl Collector for FastFieldTestCollector {
type Fruit = Vec<u64>;
type Child = FastFieldSegmentCollector;
fn for_segment(
&self,
_: SegmentLocalId,
segment_reader: &SegmentReader,
) -> Result<FastFieldSegmentCollector> {
let reader = segment_reader
.fast_fields()
.u64(self.field)
.expect("Requested field is not a fast field.");
Ok(FastFieldSegmentCollector {
vals: Vec::new(),
reader,
})
}
fn requires_scoring(&self) -> bool {
false
}
fn merge_fruits(&self, children: Vec<Vec<u64>>) -> Result<Vec<u64>> {
Ok(children.into_iter().flat_map(|v| v.into_iter()).collect())
}
}
impl SegmentCollector for FastFieldSegmentCollector {
type Fruit = Vec<u64>;
fn collect(&mut self, doc: DocId, _score: Score) {
let val = self.reader.get(doc);
self.vals.push(val);
}
fn harvest(self) -> Vec<u64> {
self.vals
}
}
/// Collects in order all of the fast field bytes for all of the
/// docs in the `DocSet`
///
/// This collector is mainly useful for tests.
pub struct BytesFastFieldTestCollector {
field: Field,
}
pub struct BytesFastFieldSegmentCollector {
vals: Vec<u8>,
reader: BytesFastFieldReader,
}
impl BytesFastFieldTestCollector {
pub fn for_field(field: Field) -> BytesFastFieldTestCollector {
BytesFastFieldTestCollector { field }
}
}
impl Collector for BytesFastFieldTestCollector {
type Fruit = Vec<u8>;
type Child = BytesFastFieldSegmentCollector;
fn for_segment(
&self,
_segment_local_id: u32,
segment_reader: &SegmentReader,
) -> Result<BytesFastFieldSegmentCollector> {
Ok(BytesFastFieldSegmentCollector {
vals: Vec::new(),
reader: segment_reader
.fast_fields()
.bytes(self.field)
.expect("Field is not a bytes fast field."),
})
}
fn requires_scoring(&self) -> bool {
false
}
fn merge_fruits(&self, children: Vec<Vec<u8>>) -> Result<Vec<u8>> {
Ok(children.into_iter().flat_map(|c| c.into_iter()).collect())
}
}
impl SegmentCollector for BytesFastFieldSegmentCollector {
type Fruit = Vec<u8>;
fn collect(&mut self, doc: u32, _score: f32) {
let data = self.reader.get_bytes(doc);
self.vals.extend(data);
}
fn harvest(self) -> <Self as SegmentCollector>::Fruit {
self.vals
}
}

302
src/collector/top_collector.rs
View File

@@ -1,244 +1,224 @@
use super::Collector;
use serde::export::PhantomData;
use std::cmp::Ordering;
use std::collections::BinaryHeap;
use DocAddress;
use DocId;
use Result;
use Score;
use SegmentLocalId;
use SegmentReader;
// Rust heap is a max-heap and we need a min heap.
#[derive(Clone, Copy)]
struct GlobalScoredDoc {
score: Score,
doc_address: DocAddress,
/// Contains a feature (field, score, etc.) of a document along with the document address.
///
/// It has a custom implementation of `PartialOrd` that reverses the order. This is because the
/// default Rust heap is a max heap, whereas a min heap is needed.
///
/// WARNING: equality is not what you would expect here.
/// Two elements are equal if their feature is equal, and regardless of whether `doc`
/// is equal. This should be perfectly fine for this usage, but let's make sure this
/// struct is never public.
struct ComparableDoc<T, D> {
feature: T,
doc: D,
}
impl PartialOrd for GlobalScoredDoc {
fn partial_cmp(&self, other: &GlobalScoredDoc) -> Option<Ordering> {
impl<T: PartialOrd, D> PartialOrd for ComparableDoc<T, D> {
fn partial_cmp(&self, other: &Self) -> Option<Ordering> {
Some(self.cmp(other))
}
}
impl Ord for GlobalScoredDoc {
impl<T: PartialOrd, D> Ord for ComparableDoc<T, D> {
#[inline]
fn cmp(&self, other: &GlobalScoredDoc) -> Ordering {
fn cmp(&self, other: &Self) -> Ordering {
other
.score
.partial_cmp(&self.score)
.unwrap_or_else(|| other.doc_address.cmp(&self.doc_address))
.feature
.partial_cmp(&self.feature)
.unwrap_or_else(|| Ordering::Equal)
}
}
impl PartialEq for GlobalScoredDoc {
fn eq(&self, other: &GlobalScoredDoc) -> bool {
impl<T: PartialOrd, D> PartialEq for ComparableDoc<T, D> {
fn eq(&self, other: &Self) -> bool {
self.cmp(other) == Ordering::Equal
}
}
impl Eq for GlobalScoredDoc {}
impl<T: PartialOrd, D> Eq for ComparableDoc<T, D> {}
/// The Top Collector keeps track of the K documents
/// with the best scores.
///
/// The implementation is based on a `BinaryHeap`.
/// The theorical complexity for collecting the top `K` out of `n` documents
/// is `O(n log K)`.
///
/// ```rust
/// #[macro_use]
/// extern crate tantivy;
/// use tantivy::schema::{SchemaBuilder, TEXT};
/// use tantivy::{Index, Result, DocId, Score};
/// use tantivy::collector::TopCollector;
/// use tantivy::query::QueryParser;
///
/// # fn main() { example().unwrap(); }
/// fn example() -> Result<()> {
/// let mut schema_builder = SchemaBuilder::new();
/// let title = schema_builder.add_text_field("title", TEXT);
/// let schema = schema_builder.build();
/// let index = Index::create_in_ram(schema);
/// {
/// let mut index_writer = index.writer_with_num_threads(1, 3_000_000)?;
/// index_writer.add_document(doc!(
/// title => "The Name of the Wind",
/// ));
/// index_writer.add_document(doc!(
/// title => "The Diary of Muadib",
/// ));
/// index_writer.add_document(doc!(
/// title => "A Dairy Cow",
/// ));
/// index_writer.add_document(doc!(
/// title => "The Diary of a Young Girl",
/// ));
/// index_writer.commit().unwrap();
/// }
///
/// index.load_searchers()?;
/// let searcher = index.searcher();
///
/// {
/// let mut top_collector = TopCollector::with_limit(2);
/// let query_parser = QueryParser::for_index(&index, vec![title]);
/// let query = query_parser.parse_query("diary")?;
/// searcher.search(&*query, &mut top_collector).unwrap();
///
/// let score_docs: Vec<(Score, DocId)> = top_collector
/// .score_docs()
/// .into_iter()
/// .map(|(score, doc_address)| (score, doc_address.doc()))
/// .collect();
///
/// assert_eq!(score_docs, vec![(0.7261542, 1), (0.6099695, 3)]);
/// }
///
/// Ok(())
/// }
/// ```
pub struct TopCollector {
pub(crate) struct TopCollector<T> {
limit: usize,
heap: BinaryHeap<GlobalScoredDoc>,
segment_id: u32,
_marker: PhantomData<T>,
}
impl TopCollector {
impl<T> TopCollector<T>
where
T: PartialOrd + Clone,
{
/// Creates a top collector, with a number of documents equal to "limit".
///
/// # Panics
/// The method panics if limit is 0
pub fn with_limit(limit: usize) -> TopCollector {
pub fn with_limit(limit: usize) -> TopCollector<T> {
if limit < 1 {
panic!("Limit must be strictly greater than 0.");
}
TopCollector {
limit,
heap: BinaryHeap::with_capacity(limit),
segment_id: 0,
_marker: PhantomData,
}
}
/// Returns K best documents sorted in decreasing order.
///
/// Calling this method triggers the sort.
/// The result of the sort is not cached.
pub fn docs(&self) -> Vec<DocAddress> {
self.score_docs()
.into_iter()
.map(|score_doc| score_doc.1)
.collect()
pub fn limit(&self) -> usize {
self.limit
}
/// Returns K best ScoredDocument sorted in decreasing order.
///
/// Calling this method triggers the sort.
/// The result of the sort is not cached.
pub fn score_docs(&self) -> Vec<(Score, DocAddress)> {
let mut scored_docs: Vec<GlobalScoredDoc> = self.heap.iter().cloned().collect();
scored_docs.sort();
scored_docs
pub fn merge_fruits(
&self,
children: Vec<Vec<(T, DocAddress)>>,
) -> Result<Vec<(T, DocAddress)>> {
if self.limit == 0 {
return Ok(Vec::new());
}
let mut top_collector = BinaryHeap::new();
for child_fruit in children {
for (feature, doc) in child_fruit {
if top_collector.len() < self.limit {
top_collector.push(ComparableDoc { feature, doc });
} else if let Some(mut head) = top_collector.peek_mut() {
if head.feature < feature {
*head = ComparableDoc { feature, doc };
}
}
}
}
Ok(top_collector
.into_sorted_vec()
.into_iter()
.map(|GlobalScoredDoc { score, doc_address }| (score, doc_address))
.map(|cdoc| (cdoc.feature, cdoc.doc))
.collect())
}
pub(crate) fn for_segment<F: PartialOrd>(
&self,
segment_id: SegmentLocalId,
_: &SegmentReader,
) -> Result<TopSegmentCollector<F>> {
Ok(TopSegmentCollector::new(segment_id, self.limit))
}
}
/// The Top Collector keeps track of the K documents
/// sorted by type `T`.
///
/// The implementation is based on a `BinaryHeap`.
/// The theorical complexity for collecting the top `K` out of `n` documents
/// is `O(n log K)`.
pub(crate) struct TopSegmentCollector<T> {
limit: usize,
heap: BinaryHeap<ComparableDoc<T, DocId>>,
segment_id: u32,
}
impl<T: PartialOrd> TopSegmentCollector<T> {
fn new(segment_id: SegmentLocalId, limit: usize) -> TopSegmentCollector<T> {
TopSegmentCollector {
limit,
heap: BinaryHeap::with_capacity(limit),
segment_id,
}
}
}
impl<T: PartialOrd + Clone> TopSegmentCollector<T> {
pub fn harvest(self) -> Vec<(T, DocAddress)> {
let segment_id = self.segment_id;
self.heap
.into_sorted_vec()
.into_iter()
.map(|comparable_doc| {
(
comparable_doc.feature,
DocAddress(segment_id, comparable_doc.doc),
)
})
.collect()
}
/// Return true iff at least K documents have gone through
/// the collector.
#[inline]
pub fn at_capacity(&self) -> bool {
#[inline(always)]
pub(crate) fn at_capacity(&self) -> bool {
self.heap.len() >= self.limit
}
}
impl Collector for TopCollector {
fn set_segment(&mut self, segment_id: SegmentLocalId, _: &SegmentReader) -> Result<()> {
self.segment_id = segment_id;
Ok(())
}
fn collect(&mut self, doc: DocId, score: Score) {
/// Collects a document scored by the given feature
///
/// It collects documents until it has reached the max capacity. Once it reaches capacity, it
/// will compare the lowest scoring item with the given one and keep whichever is greater.
#[inline(always)]
pub fn collect(&mut self, doc: DocId, feature: T) {
if self.at_capacity() {
// It's ok to unwrap as long as a limit of 0 is forbidden.
let limit_doc: GlobalScoredDoc = *self.heap
.peek()
.expect("Top collector with size 0 is forbidden");
if limit_doc.score < score {
let mut mut_head = self.heap
.peek_mut()
.expect("Top collector with size 0 is forbidden");
mut_head.score = score;
mut_head.doc_address = DocAddress(self.segment_id, doc);
if let Some(limit_feature) = self.heap.peek().map(|head| head.feature.clone()) {
if limit_feature < feature {
if let Some(mut head) = self.heap.peek_mut() {
head.feature = feature;
head.doc = doc;
}
}
}
} else {
let wrapped_doc = GlobalScoredDoc {
score,
doc_address: DocAddress(self.segment_id, doc),
};
self.heap.push(wrapped_doc);
// we have not reached capacity yet, so we can just push the
// element.
self.heap.push(ComparableDoc { feature, doc });
}
}
fn requires_scoring(&self) -> bool {
true
}
}
#[cfg(test)]
mod tests {
use super::*;
use collector::Collector;
use DocId;
use super::{TopCollector, TopSegmentCollector};
use DocAddress;
use Score;
#[test]
fn test_top_collector_not_at_capacity() {
let mut top_collector = TopCollector::with_limit(4);
let mut top_collector = TopSegmentCollector::new(0, 4);
top_collector.collect(1, 0.8);
top_collector.collect(3, 0.2);
top_collector.collect(5, 0.3);
assert!(!top_collector.at_capacity());
let score_docs: Vec<(Score, DocId)> = top_collector
.score_docs()
.into_iter()
.map(|(score, doc_address)| (score, doc_address.doc()))
.collect();
assert_eq!(score_docs, vec![(0.8, 1), (0.3, 5), (0.2, 3)]);
assert_eq!(
top_collector.harvest(),
vec![
(0.8, DocAddress(0, 1)),
(0.3, DocAddress(0, 5)),
(0.2, DocAddress(0, 3))
]
);
}
#[test]
fn test_top_collector_at_capacity() {
let mut top_collector = TopCollector::with_limit(4);
let mut top_collector = TopSegmentCollector::new(0, 4);
top_collector.collect(1, 0.8);
top_collector.collect(3, 0.2);
top_collector.collect(5, 0.3);
top_collector.collect(7, 0.9);
top_collector.collect(9, -0.2);
assert!(top_collector.at_capacity());
{
let score_docs: Vec<(Score, DocId)> = top_collector
.score_docs()
.into_iter()
.map(|(score, doc_address)| (score, doc_address.doc()))
.collect();
assert_eq!(score_docs, vec![(0.9, 7), (0.8, 1), (0.3, 5), (0.2, 3)]);
}
{
let docs: Vec<DocId> = top_collector
.docs()
.into_iter()
.map(|doc_address| doc_address.doc())
.collect();
assert_eq!(docs, vec![7, 1, 5, 3]);
}
assert_eq!(
top_collector.harvest(),
vec![
(0.9, DocAddress(0, 7)),
(0.8, DocAddress(0, 1)),
(0.3, DocAddress(0, 5)),
(0.2, DocAddress(0, 3))
]
);
}
#[test]
#[should_panic]
fn test_top_0() {
TopCollector::with_limit(0);
let _collector: TopCollector<Score> = TopCollector::with_limit(0);
}
}

271
src/collector/top_field_collector.rs Normal file
View File

@@ -0,0 +1,271 @@
use super::Collector;
use collector::top_collector::TopCollector;
use collector::top_collector::TopSegmentCollector;
use collector::SegmentCollector;
use fastfield::FastFieldReader;
use fastfield::FastValue;
use schema::Field;
use std::marker::PhantomData;
use DocAddress;
use Result;
use SegmentLocalId;
use SegmentReader;
use TantivyError;
/// The Top Field Collector keeps track of the K documents
/// sorted by a fast field in the index
///
/// The implementation is based on a `BinaryHeap`.
/// The theorical complexity for collecting the top `K` out of `n` documents
/// is `O(n log K)`.
///
/// ```rust
/// #[macro_use]
/// extern crate tantivy;
/// # use tantivy::schema::{Schema, Field, FAST, TEXT};
/// # use tantivy::{Index, Result, DocAddress};
/// # use tantivy::query::{Query, QueryParser};
/// use tantivy::Searcher;
/// use tantivy::collector::TopDocs;
///
/// # fn main() -> tantivy::Result<()> {
/// # let mut schema_builder = Schema::builder();
/// # let title = schema_builder.add_text_field("title", TEXT);
/// # let rating = schema_builder.add_u64_field("rating", FAST);
/// # let schema = schema_builder.build();
/// # let index = Index::create_in_ram(schema);
/// # let mut index_writer = index.writer_with_num_threads(1, 3_000_000)?;
/// # index_writer.add_document(doc!(
/// # title => "The Name of the Wind",
/// # rating => 92u64,
/// # ));
/// # index_writer.add_document(doc!(title => "The Diary of Muadib", rating => 97u64));
/// # index_writer.add_document(doc!(title => "A Dairy Cow", rating => 63u64));
/// # index_writer.add_document(doc!(title => "The Diary of a Young Girl", rating => 80u64));
/// # index_writer.commit()?;
/// # let reader = index.reader()?;
/// # let query = QueryParser::for_index(&index, vec![title]).parse_query("diary")?;
/// # let top_docs = docs_sorted_by_rating(&reader.searcher(), &query, rating)?;
/// # assert_eq!(top_docs,
/// # vec![(97u64, DocAddress(0u32, 1)),
/// # (80u64, DocAddress(0u32, 3))]);
/// # Ok(())
/// # }
/// #
/// /// Searches the document matching the given query, and
/// /// collects the top 10 documents, order by the `field`
/// /// given in argument.
/// ///
/// /// `field` is required to be a FAST field.
/// fn docs_sorted_by_rating(searcher: &Searcher,
/// query: &Query,
/// sort_by_field: Field)
/// -> Result<Vec<(u64, DocAddress)>> {
///
/// // This is where we build our collector!
/// let top_docs_by_rating = TopDocs::with_limit(2).order_by_field(sort_by_field);
///
/// // ... and here is our documents. Not this is a simple vec.
/// // The `u64` in the pair is the value of our fast field for each documents.
/// searcher.search(query, &top_docs_by_rating)
/// }
/// ```
pub struct TopDocsByField<T> {
collector: TopCollector<T>,
field: Field,
}
impl<T: FastValue + PartialOrd + Clone> TopDocsByField<T> {
/// Creates a top field collector, with a number of documents equal to "limit".
///
/// The given field name must be a fast field, otherwise the collector have an error while
/// collecting results.
///
/// This constructor is crate-private. Client are supposed to call
/// build `TopDocsByField` object using the `TopDocs` API.
///
/// e.g.:
/// `TopDocs::with_limit(2).order_by_field(sort_by_field)`
///
/// # Panics
/// The method panics if limit is 0
pub(crate) fn new(field: Field, limit: usize) -> TopDocsByField<T> {
TopDocsByField {
collector: TopCollector::with_limit(limit),
field,
}
}
}
impl<T: FastValue + PartialOrd + Send + Sync + 'static> Collector for TopDocsByField<T> {
type Fruit = Vec<(T, DocAddress)>;
type Child = TopFieldSegmentCollector<T>;
fn for_segment(
&self,
segment_local_id: SegmentLocalId,
reader: &SegmentReader,
) -> Result<TopFieldSegmentCollector<T>> {
let collector = self.collector.for_segment(segment_local_id, reader)?;
let reader = reader.fast_fields().u64(self.field).ok_or_else(|| {
let field_name = reader.schema().get_field_name(self.field);
TantivyError::SchemaError(format!("Failed to find fast field reader {:?}", field_name))
})?;
Ok(TopFieldSegmentCollector {
collector,
reader,
_type: PhantomData,
})
}
fn requires_scoring(&self) -> bool {
false
}
fn merge_fruits(
&self,
segment_fruits: Vec<Vec<(T, DocAddress)>>,
) -> Result<Vec<(T, DocAddress)>> {
self.collector.merge_fruits(segment_fruits)
}
}
pub struct TopFieldSegmentCollector<T> {
collector: TopSegmentCollector<u64>,
reader: FastFieldReader<u64>,
_type: PhantomData<T>,
}
impl<T: FastValue + PartialOrd + Send + Sync + 'static> SegmentCollector
for TopFieldSegmentCollector<T>
{
type Fruit = Vec<(T, DocAddress)>;
fn collect(&mut self, doc: u32, _score: f32) {
let field_value = self.reader.get(doc);
self.collector.collect(doc, field_value);
}
fn harvest(self) -> Vec<(T, DocAddress)> {
self.collector
.harvest()
.into_iter()
.map(|(val, doc_address)| (T::from_u64(val), doc_address))
.collect()
}
}
#[cfg(test)]
mod tests {
use super::TopDocsByField;
use collector::Collector;
use collector::TopDocs;
use query::Query;
use query::QueryParser;
use schema::Field;
use schema::IntOptions;
use schema::{Schema, FAST, TEXT};
use DocAddress;
use Index;
use IndexWriter;
use TantivyError;
const TITLE: &str = "title";
const SIZE: &str = "size";
#[test]
fn test_top_collector_not_at_capacity() {
let mut schema_builder = Schema::builder();
let title = schema_builder.add_text_field(TITLE, TEXT);
let size = schema_builder.add_u64_field(SIZE, FAST);
let schema = schema_builder.build();
let (index, query) = index("beer", title, schema, |index_writer| {
index_writer.add_document(doc!(
title => "bottle of beer",
size => 12u64,
));
index_writer.add_document(doc!(
title => "growler of beer",
size => 64u64,
));
index_writer.add_document(doc!(
title => "pint of beer",
size => 16u64,
));
});
let searcher = index.reader().unwrap().searcher();
let top_collector = TopDocs::with_limit(4).order_by_field(size);
let top_docs: Vec<(u64, DocAddress)> = searcher.search(&query, &top_collector).unwrap();
assert_eq!(
top_docs,
vec![
(64, DocAddress(0, 1)),
(16, DocAddress(0, 2)),
(12, DocAddress(0, 0))
]
);
}
#[test]
#[should_panic]
fn test_field_does_not_exist() {
let mut schema_builder = Schema::builder();
let title = schema_builder.add_text_field(TITLE, TEXT);
let size = schema_builder.add_u64_field(SIZE, FAST);
let schema = schema_builder.build();
let (index, _) = index("beer", title, schema, |index_writer| {
index_writer.add_document(doc!(
title => "bottle of beer",
size => 12u64,
));
});
let searcher = index.reader().unwrap().searcher();
let top_collector: TopDocsByField<u64> = TopDocs::with_limit(4).order_by_field(Field(2));
let segment_reader = searcher.segment_reader(0u32);
top_collector
.for_segment(0, segment_reader)
.expect("should panic");
}
#[test]
fn test_field_not_fast_field() {
let mut schema_builder = Schema::builder();
let title = schema_builder.add_text_field(TITLE, TEXT);
let size = schema_builder.add_u64_field(SIZE, IntOptions::default());
let schema = schema_builder.build();
let (index, _) = index("beer", title, schema, |index_writer| {
index_writer.add_document(doc!(
title => "bottle of beer",
size => 12u64,
));
});
let searcher = index.reader().unwrap().searcher();
let segment = searcher.segment_reader(0);
let top_collector: TopDocsByField<u64> = TopDocs::with_limit(4).order_by_field(size);
assert_matches!(
top_collector
.for_segment(0, segment)
.map(|_| ())
.unwrap_err(),
TantivyError::SchemaError(_)
);
}
fn index(
query: &str,
query_field: Field,
schema: Schema,
mut doc_adder: impl FnMut(&mut IndexWriter) -> (),
) -> (Index, Box<Query>) {
let index = Index::create_in_ram(schema);
let mut index_writer = index.writer_with_num_threads(1, 3_000_000).unwrap();
doc_adder(&mut index_writer);
index_writer.commit().unwrap();
let query_parser = QueryParser::for_index(&index, vec![query_field]);
let query = query_parser.parse_query(query).unwrap();
(index, query)
}
}

203
src/collector/top_score_collector.rs Normal file
View File

@@ -0,0 +1,203 @@
use super::Collector;
use collector::top_collector::TopCollector;
use collector::top_collector::TopSegmentCollector;
use collector::SegmentCollector;
use collector::TopDocsByField;
use fastfield::FastValue;
use schema::Field;
use DocAddress;
use DocId;
use Result;
use Score;
use SegmentLocalId;
use SegmentReader;
/// The Top Score Collector keeps track of the K documents
/// sorted by their score.
///
/// The implementation is based on a `BinaryHeap`.
/// The theorical complexity for collecting the top `K` out of `n` documents
/// is `O(n log K)`.
///
/// ```rust
/// #[macro_use]
/// extern crate tantivy;
/// use tantivy::DocAddress;
/// use tantivy::schema::{Schema, TEXT};
/// use tantivy::{Index, Result};
/// use tantivy::collector::TopDocs;
/// use tantivy::query::QueryParser;
///
/// # fn main() { example().unwrap(); }
/// fn example() -> Result<()> {
/// let mut schema_builder = Schema::builder();
/// let title = schema_builder.add_text_field("title", TEXT);
/// let schema = schema_builder.build();
/// let index = Index::create_in_ram(schema);
/// {
/// let mut index_writer = index.writer_with_num_threads(1, 3_000_000)?;
/// index_writer.add_document(doc!(
/// title => "The Name of the Wind",
/// ));
/// index_writer.add_document(doc!(
/// title => "The Diary of Muadib",
/// ));
/// index_writer.add_document(doc!(
/// title => "A Dairy Cow",
/// ));
/// index_writer.add_document(doc!(
/// title => "The Diary of a Young Girl",
/// ));
/// index_writer.commit().unwrap();
/// }
///
/// let reader = index.reader()?;
/// let searcher = reader.searcher();
///
/// let query_parser = QueryParser::for_index(&index, vec![title]);
/// let query = query_parser.parse_query("diary")?;
/// let top_docs = searcher.search(&query, &TopDocs::with_limit(2))?;
///
/// assert_eq!(&top_docs[0], &(0.7261542, DocAddress(0, 1)));
/// assert_eq!(&top_docs[1], &(0.6099695, DocAddress(0, 3)));
///
/// Ok(())
/// }
/// ```
pub struct TopDocs(TopCollector<Score>);
impl TopDocs {
/// Creates a top score collector, with a number of documents equal to "limit".
///
/// # Panics
/// The method panics if limit is 0
pub fn with_limit(limit: usize) -> TopDocs {
TopDocs(TopCollector::with_limit(limit))
}
/// Set top-K to rank documents by a given fast field.
///
/// (By default, `TopDocs` collects the top-K documents sorted by
/// the similarity score.)
pub fn order_by_field<T: PartialOrd + FastValue + Clone>(
self,
field: Field,
) -> TopDocsByField<T> {
TopDocsByField::new(field, self.0.limit())
}
}
impl Collector for TopDocs {
type Fruit = Vec<(Score, DocAddress)>;
type Child = TopScoreSegmentCollector;
fn for_segment(
&self,
segment_local_id: SegmentLocalId,
reader: &SegmentReader,
) -> Result<Self::Child> {
let collector = self.0.for_segment(segment_local_id, reader)?;
Ok(TopScoreSegmentCollector(collector))
}
fn requires_scoring(&self) -> bool {
true
}
fn merge_fruits(&self, child_fruits: Vec<Vec<(Score, DocAddress)>>) -> Result<Self::Fruit> {
self.0.merge_fruits(child_fruits)
}
}
/// Segment Collector associated to `TopDocs`.
pub struct TopScoreSegmentCollector(TopSegmentCollector<Score>);
impl SegmentCollector for TopScoreSegmentCollector {
type Fruit = Vec<(Score, DocAddress)>;
fn collect(&mut self, doc: DocId, score: Score) {
self.0.collect(doc, score)
}
fn harvest(self) -> Vec<(Score, DocAddress)> {
self.0.harvest()
}
}
#[cfg(test)]
mod tests {
use super::TopDocs;
use query::QueryParser;
use schema::Schema;
use schema::TEXT;
use DocAddress;
use Index;
use Score;
fn make_index() -> Index {
let mut schema_builder = Schema::builder();
let text_field = schema_builder.add_text_field("text", TEXT);
let schema = schema_builder.build();
let index = Index::create_in_ram(schema);
{
// writing the segment
let mut index_writer = index.writer_with_num_threads(1, 3_000_000).unwrap();
index_writer.add_document(doc!(text_field=>"Hello happy tax payer."));
index_writer.add_document(doc!(text_field=>"Droopy says hello happy tax payer"));
index_writer.add_document(doc!(text_field=>"I like Droopy"));
assert!(index_writer.commit().is_ok());
}
index
}
#[test]
fn test_top_collector_not_at_capacity() {
let index = make_index();
let field = index.schema().get_field("text").unwrap();
let query_parser = QueryParser::for_index(&index, vec![field]);
let text_query = query_parser.parse_query("droopy tax").unwrap();
let score_docs: Vec<(Score, DocAddress)> = index
.reader()
.unwrap()
.searcher()
.search(&text_query, &TopDocs::with_limit(4))
.unwrap();
assert_eq!(
score_docs,
vec![
(0.81221175, DocAddress(0u32, 1)),
(0.5376842, DocAddress(0u32, 2)),
(0.48527452, DocAddress(0, 0))
]
);
}
#[test]
fn test_top_collector_at_capacity() {
let index = make_index();
let field = index.schema().get_field("text").unwrap();
let query_parser = QueryParser::for_index(&index, vec![field]);
let text_query = query_parser.parse_query("droopy tax").unwrap();
let score_docs: Vec<(Score, DocAddress)> = index
.reader()
.unwrap()
.searcher()
.search(&text_query, &TopDocs::with_limit(2))
.unwrap();
assert_eq!(
score_docs,
vec![
(0.81221175, DocAddress(0u32, 1)),
(0.5376842, DocAddress(0u32, 2)),
]
);
}
#[test]
#[should_panic]
fn test_top_0() {
TopDocs::with_limit(0);
}
}

70
src/common/bitpacker.rs
View File

@@ -1,9 +1,6 @@
use common::serialize::BinarySerializable;
use byteorder::{ByteOrder, LittleEndian, WriteBytesExt};
use std::io;
use std::io::Write;
use std::mem;
use std::ops::Deref;
use std::ptr;
pub(crate) struct BitPacker {
mini_buffer: u64,
@@ -18,7 +15,7 @@ impl BitPacker {
}
}
pub fn write<TWrite: Write>(
pub fn write<TWrite: io::Write>(
&mut self,
val: u64,
num_bits: u8,
@@ -28,14 +25,14 @@ impl BitPacker {
let num_bits = num_bits as usize;
if self.mini_buffer_written + num_bits > 64 {
self.mini_buffer |= val_u64.wrapping_shl(self.mini_buffer_written as u32);
self.mini_buffer.serialize(output)?;
output.write_u64::<LittleEndian>(self.mini_buffer)?;
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 {
self.mini_buffer.serialize(output)?;
output.write_u64::<LittleEndian>(self.mini_buffer)?;
self.mini_buffer_written = 0;
self.mini_buffer = 0u64;
}
@@ -43,17 +40,18 @@ impl BitPacker {
Ok(())
}
pub fn flush<TWrite: Write>(&mut self, output: &mut TWrite) -> io::Result<()> {
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 arr: [u8; 8] = unsafe { mem::transmute::<u64, [u8; 8]>(self.mini_buffer.to_le()) };
let mut arr: [u8; 8] = [0u8; 8];
LittleEndian::write_u64(&mut arr, self.mini_buffer);
output.write_all(&arr[..num_bytes])?;
self.mini_buffer_written = 0;
}
Ok(())
}
pub fn close<TWrite: Write>(&mut self, output: &mut TWrite) -> io::Result<()> {
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])?;
@@ -66,7 +64,7 @@ pub struct BitUnpacker<Data>
where
Data: Deref<Target = [u8]>,
{
num_bits: usize,
num_bits: u64,
mask: u64,
data: Data,
}
@@ -82,13 +80,13 @@ where
(1u64 << num_bits) - 1u64
};
BitUnpacker {
num_bits: num_bits as usize,
num_bits: u64::from(num_bits),
mask,
data,
}
}
pub fn get(&self, idx: usize) -> u64 {
pub fn get(&self, idx: u64) -> u64 {
if self.num_bits == 0 {
return 0u64;
}
@@ -99,40 +97,13 @@ where
let addr = addr_in_bits >> 3;
let bit_shift = addr_in_bits & 7;
debug_assert!(
addr + 8 <= data.len(),
addr + 8 <= data.len() as u64,
"The fast field field should have been padded with 7 bytes."
);
let val_unshifted_unmasked: u64 =
u64::from_le(unsafe { ptr::read_unaligned(data[addr..].as_ptr() as *const u64) });
let val_unshifted_unmasked: u64 = LittleEndian::read_u64(&data[(addr as usize)..]);
let val_shifted = (val_unshifted_unmasked >> bit_shift) as u64;
val_shifted & mask
}
/// Reads a range of values from the fast field.
///
/// The range of values read is from
/// `[start..start + output.len()[`
pub fn get_range(&self, start: u32, output: &mut [u64]) {
if self.num_bits == 0 {
for val in output.iter_mut() {
*val = 0u64;
}
} else {
let data: &[u8] = &*self.data;
let num_bits = self.num_bits;
let mask = self.mask;
let mut addr_in_bits = (start as usize) * num_bits;
for output_val in output.iter_mut() {
let addr = addr_in_bits >> 3;
let bit_shift = addr_in_bits & 7;
let val_unshifted_unmasked: u64 =
unsafe { ptr::read_unaligned(data[addr..].as_ptr() as *const u64) };
let val_shifted = (val_unshifted_unmasked >> bit_shift) as u64;
*output_val = val_shifted & mask;
addr_in_bits += num_bits;
}
}
}
}
#[cfg(test)]
@@ -158,7 +129,7 @@ mod test {
fn test_bitpacker_util(len: usize, num_bits: u8) {
let (bitunpacker, vals) = create_fastfield_bitpacker(len, num_bits);
for (i, val) in vals.iter().enumerate() {
assert_eq!(bitunpacker.get(i), *val);
assert_eq!(bitunpacker.get(i as u64), *val);
}
}
@@ -170,17 +141,4 @@ mod test {
test_bitpacker_util(6, 14);
test_bitpacker_util(1000, 14);
}
#[test]
fn test_bitpacker_range() {
let (bitunpacker, vals) = create_fastfield_bitpacker(100_000, 12);
let buffer_len = 100;
let mut buffer = vec![0u64; buffer_len];
for start in vec![0, 10, 20, 100, 1_000] {
bitunpacker.get_range(start as u32, &mut buffer[..]);
for i in 0..buffer_len {
assert_eq!(buffer[i], vals[start + i]);
}
}
}
}

16
src/common/bitset.rs
View File

@@ -34,17 +34,17 @@ impl TinySet {
}
/// Returns the complement of the set in `[0, 64[`.
fn complement(&self) -> TinySet {
fn complement(self) -> TinySet {
TinySet(!self.0)
}
/// Returns true iff the `TinySet` contains the element `el`.
pub fn contains(&self, el: u32) -> bool {
pub fn contains(self, el: u32) -> bool {
!self.intersect(TinySet::singleton(el)).is_empty()
}
/// Returns the intersection of `self` and `other`
pub fn intersect(&self, other: TinySet) -> TinySet {
pub fn intersect(self, other: TinySet) -> TinySet {
TinySet(self.0 & other.0)
}
@@ -77,7 +77,7 @@ impl TinySet {
/// Returns true iff the `TinySet` is empty.
#[inline(always)]
pub fn is_empty(&self) -> bool {
pub fn is_empty(self) -> bool {
self.0 == 0u64
}
@@ -114,7 +114,7 @@ impl TinySet {
self.0 = 0u64;
}
pub fn len(&self) -> u32 {
pub fn len(self) -> u32 {
self.0.count_ones()
}
}
@@ -266,14 +266,14 @@ mod tests {
#[test]
fn test_bitset_large() {
let arr = generate_nonunique_unsorted(1_000_000, 50_000);
let arr = generate_nonunique_unsorted(100_000, 5_000);
let mut btreeset: BTreeSet<u32> = BTreeSet::new();
let mut bitset = BitSet::with_max_value(1_000_000);
let mut bitset = BitSet::with_max_value(100_000);
for el in arr {
btreeset.insert(el);
bitset.insert(el);
}
for i in 0..1_000_000 {
for i in 0..100_000 {
assert_eq!(btreeset.contains(&i), bitset.contains(i));
}
assert_eq!(btreeset.len(), bitset.len());

18
src/common/composite_file.rs
View File

@@ -4,6 +4,8 @@ use common::VInt;
use directory::ReadOnlySource;
use directory::WritePtr;
use schema::Field;
use space_usage::FieldUsage;
use space_usage::PerFieldSpaceUsage;
use std::collections::HashMap;
use std::io::Write;
use std::io::{self, Read};
@@ -37,7 +39,7 @@ impl BinarySerializable for FileAddr {
/// A `CompositeWrite` is used to write a `CompositeFile`.
pub struct CompositeWrite<W = WritePtr> {
write: CountingWriter<W>,
offsets: HashMap<FileAddr, usize>,
offsets: HashMap<FileAddr, u64>,
}
impl<W: Write> CompositeWrite<W> {
@@ -72,7 +74,8 @@ impl<W: Write> CompositeWrite<W> {
let footer_offset = self.write.written_bytes();
VInt(self.offsets.len() as u64).serialize(&mut self.write)?;
let mut offset_fields: Vec<_> = self.offsets
let mut offset_fields: Vec<_> = self
.offsets
.iter()
.map(|(file_addr, offset)| (*offset, *file_addr))
.collect();
@@ -165,6 +168,17 @@ impl CompositeFile {
.get(&FileAddr { field, idx })
.map(|&(from, to)| self.data.slice(from, to))
}
pub fn space_usage(&self) -> PerFieldSpaceUsage {
let mut fields = HashMap::new();
for (&field_addr, &(start, end)) in self.offsets_index.iter() {
fields
.entry(field_addr.field)
.or_insert_with(|| FieldUsage::empty(field_addr.field))
.add_field_idx(field_addr.idx, end - start);
}
PerFieldSpaceUsage::new(fields)
}
}
#[cfg(test)]

18
src/common/counting_writer.rs
View File

@@ -3,7 +3,7 @@ use std::io::Write;
pub struct CountingWriter<W> {
underlying: W,
written_bytes: usize,
written_bytes: u64,
}
impl<W: Write> CountingWriter<W> {
@@ -14,11 +14,11 @@ impl<W: Write> CountingWriter<W> {
}
}
pub fn written_bytes(&self) -> usize {
pub fn written_bytes(&self) -> u64 {
self.written_bytes
}
pub fn finish(mut self) -> io::Result<(W, usize)> {
pub fn finish(mut self) -> io::Result<(W, u64)> {
self.flush()?;
Ok((self.underlying, self.written_bytes))
}
@@ -27,10 +27,16 @@ impl<W: Write> CountingWriter<W> {
impl<W: Write> Write for CountingWriter<W> {
fn write(&mut self, buf: &[u8]) -> io::Result<usize> {
let written_size = self.underlying.write(buf)?;
self.written_bytes += written_size;
self.written_bytes += written_size as u64;
Ok(written_size)
}
fn write_all(&mut self, buf: &[u8]) -> io::Result<()> {
self.underlying.write_all(buf)?;
self.written_bytes += buf.len() as u64;
Ok(())
}
fn flush(&mut self) -> io::Result<()> {
self.underlying.flush()
}
@@ -48,8 +54,8 @@ mod test {
let mut counting_writer = CountingWriter::wrap(buffer);
let bytes = (0u8..10u8).collect::<Vec<u8>>();
counting_writer.write_all(&bytes).unwrap();
let (w, len): (Vec<u8>, usize) = counting_writer.finish().unwrap();
assert_eq!(len, 10);
let (w, len): (Vec<u8>, u64) = counting_writer.finish().unwrap();
assert_eq!(len, 10u64);
assert_eq!(w.len(), 10);
}
}

19
src/common/mod.rs
View File

@@ -10,10 +10,13 @@ pub(crate) use self::bitset::TinySet;
pub(crate) use self::composite_file::{CompositeFile, CompositeWrite};
pub use self::counting_writer::CountingWriter;
pub use self::serialize::{BinarySerializable, FixedSize};
pub use self::vint::VInt;
pub use self::vint::{read_u32_vint, serialize_vint_u32, write_u32_vint, VInt};
pub use byteorder::LittleEndian as Endianness;
use std::io;
/// Segment's max doc must be `< MAX_DOC_LIMIT`.
///
/// We do not allow segments with more than
pub const MAX_DOC_LIMIT: u32 = 1 << 31;
/// Computes the number of bits that will be used for bitpacking.
///
@@ -52,11 +55,6 @@ pub(crate) fn is_power_of_2(n: usize) -> bool {
(n > 0) && (n & (n - 1) == 0)
}
/// Create a default io error given a string.
pub(crate) fn make_io_err(msg: String) -> io::Error {
io::Error::new(io::ErrorKind::Other, msg)
}
/// Has length trait
pub trait HasLen {
/// Return length
@@ -134,4 +132,11 @@ pub(crate) mod test {
assert_eq!(compute_num_bits(256), 9u8);
assert_eq!(compute_num_bits(5_000_000_000), 33u8);
}
#[test]
fn test_max_doc() {
// this is the first time I write a unit test for a constant.
assert!(((super::MAX_DOC_LIMIT - 1) as i32) >= 0);
assert!((super::MAX_DOC_LIMIT as i32) < 0);
}
}

141
src/common/vint.rs
View File

@@ -1,4 +1,5 @@
use super::BinarySerializable;
use byteorder::{ByteOrder, LittleEndian};
use std::io;
use std::io::Read;
use std::io::Write;
@@ -9,9 +10,101 @@ pub struct VInt(pub u64);
const STOP_BIT: u8 = 128;
pub fn serialize_vint_u32(val: u32) -> (u64, usize) {
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;
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,
),
}
}
/// 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 vlen = vint_len(*data);
let mut result = 0u32;
let mut shift = 0u64;
for &b in &data[..vlen] {
result |= u32::from(b & 127u8) << shift;
shift += 7;
}
*data = &data[vlen..];
result
}
/// Write a `u32` as a vint payload.
pub fn write_u32_vint<W: io::Write>(val: u32, writer: &mut W) -> io::Result<()> {
let (val, num_bytes) = serialize_vint_u32(val);
let mut buffer = [0u8; 8];
LittleEndian::write_u64(&mut buffer, val);
writer.write_all(&buffer[..num_bytes])
}
impl VInt {
pub fn val(&self) -> u64 {
self.0
}
@@ -20,14 +113,13 @@ impl VInt {
VInt::deserialize(reader).map(|vint| vint.0)
}
pub fn serialize_into_vec(&self, output: &mut Vec<u8>){
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]);
}
fn serialize_into(&self, buffer: &mut [u8; 10]) -> usize {
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;
@@ -67,18 +159,19 @@ impl BinarySerializable for VInt {
return Err(io::Error::new(
io::ErrorKind::InvalidData,
"Reach end of buffer while reading VInt",
))
));
}
}
}
}
}
#[cfg(test)]
mod tests {
use super::serialize_vint_u32;
use super::VInt;
use byteorder::{ByteOrder, LittleEndian};
use common::BinarySerializable;
fn aux_test_vint(val: u64) {
@@ -89,10 +182,10 @@ mod tests {
}
assert!(num_bytes > 0);
if num_bytes < 10 {
assert!(1u64 << (7*num_bytes) > val);
assert!(1u64 << (7 * num_bytes) > val);
}
if num_bytes > 1 {
assert!(1u64 << (7*(num_bytes-1)) <= val);
assert!(1u64 << (7 * (num_bytes - 1)) <= val);
}
let serdeser_val = VInt::deserialize(&mut &v[..]).unwrap();
assert_eq!(val, serdeser_val.0);
@@ -105,11 +198,35 @@ mod tests {
aux_test_vint(5);
aux_test_vint(u64::max_value());
for i in 1..9 {
let power_of_128 = 1u64 << (7*i);
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);
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; 10];
let len_vint = VInt(val as u64).serialize_into(&mut buffer);
let (vint, len) = serialize_vint_u32(val);
assert_eq!(len, len_vint, "len wrong for val {}", val);
LittleEndian::write_u64(&mut buffer2, vint);
assert_eq!(&buffer[..len], &buffer2[..len], "array wrong for {}", val);
}
#[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_value());
}
}

136
src/core/executor.rs Normal file
View File

@@ -0,0 +1,136 @@
use crossbeam::channel;
use scoped_pool::{Pool, ThreadConfig};
use Result;
/// Search executor whether search request are single thread or multithread.
///
/// We don't expose Rayon thread pool directly here for several reasons.
///
/// First dependency hell. It is not a good idea to expose the
/// API of a dependency, knowing it might conflict with a different version
/// used by the client. Second, we may stop using rayon in the future.
pub enum Executor {
SingleThread,
ThreadPool(Pool),
}
impl Executor {
/// Creates an Executor that performs all task in the caller thread.
pub fn single_thread() -> Executor {
Executor::SingleThread
}
// Creates an Executor that dispatches the tasks in a thread pool.
pub fn multi_thread(num_threads: usize, prefix: &'static str) -> Executor {
let thread_config = ThreadConfig::new().prefix(prefix);
let pool = Pool::with_thread_config(num_threads, thread_config);
Executor::ThreadPool(pool)
}
// Perform a map in the thread pool.
//
// Regardless of the executor (`SingleThread` or `ThreadPool`), panics in the task
// will propagate to the caller.
pub fn map<
A: Send,
R: Send,
AIterator: Iterator<Item = A>,
F: Sized + Sync + Fn(A) -> Result<R>,
>(
&self,
f: F,
args: AIterator,
) -> Result<Vec<R>> {
match self {
Executor::SingleThread => args.map(f).collect::<Result<_>>(),
Executor::ThreadPool(pool) => {
let args_with_indices: Vec<(usize, A)> = args.enumerate().collect();
let num_fruits = args_with_indices.len();
let fruit_receiver = {
let (fruit_sender, fruit_receiver) = channel::unbounded();
pool.scoped(|scope| {
for arg_with_idx in args_with_indices {
scope.execute(|| {
let (idx, arg) = arg_with_idx;
let fruit = f(arg);
if let Err(err) = fruit_sender.send((idx, fruit)) {
error!("Failed to send search task. It probably means all search threads have panicked. {:?}", err);
}
});
}
});
fruit_receiver
// This ends the scope of fruit_sender.
// This is important as it makes it possible for the fruit_receiver iteration to
// terminate.
};
// This is lame, but safe.
let mut results_with_position = Vec::with_capacity(num_fruits);
for (pos, fruit_res) in fruit_receiver {
let fruit = fruit_res?;
results_with_position.push((pos, fruit));
}
results_with_position.sort_by_key(|(pos, _)| *pos);
assert_eq!(results_with_position.len(), num_fruits);
Ok(results_with_position
.into_iter()
.map(|(_, fruit)| fruit)
.collect::<Vec<_>>())
}
}
}
}
#[cfg(test)]
mod tests {
use super::Executor;
#[test]
#[should_panic(expected = "panic should propagate")]
fn test_panic_propagates_single_thread() {
let _result: Vec<usize> = Executor::single_thread()
.map(
|_| {
panic!("panic should propagate");
},
vec![0].into_iter(),
)
.unwrap();
}
#[test]
#[should_panic] //< unfortunately the panic message is not propagated
fn test_panic_propagates_multi_thread() {
let _result: Vec<usize> = Executor::multi_thread(1, "search-test")
.map(
|_| {
panic!("panic should propagate");
},
vec![0].into_iter(),
)
.unwrap();
}
#[test]
fn test_map_singlethread() {
let result: Vec<usize> = Executor::single_thread()
.map(|i| Ok(i * 2), 0..1_000)
.unwrap();
assert_eq!(result.len(), 1_000);
for i in 0..1_000 {
assert_eq!(result[i], i * 2);
}
}
#[test]
fn test_map_multithread() {
let result: Vec<usize> = Executor::multi_thread(3, "search-test")
.map(|i| Ok(i * 2), 0..10)
.unwrap();
assert_eq!(result.len(), 10);
for i in 0..10 {
assert_eq!(result[i], i * 2);
}
}
}

415
src/core/index.rs
View File

@@ -1,52 +1,88 @@
use super::segment::create_segment;
use super::segment::Segment;
use core::Executor;
use core::IndexMeta;
use core::SegmentId;
use core::SegmentMeta;
use core::META_FILEPATH;
use directory::ManagedDirectory;
#[cfg(feature = "mmap")]
use directory::MmapDirectory;
use directory::INDEX_WRITER_LOCK;
use directory::{Directory, RAMDirectory};
use error::DataCorruption;
use error::TantivyError;
use indexer::index_writer::open_index_writer;
use indexer::index_writer::HEAP_SIZE_MIN;
use indexer::segment_updater::save_new_metas;
use num_cpus;
use reader::IndexReader;
use reader::IndexReaderBuilder;
use schema::Field;
use schema::FieldType;
use schema::Schema;
use serde_json;
use std::borrow::BorrowMut;
use std::fmt;
use std::sync::atomic::{AtomicUsize, Ordering};
use std::sync::Arc;
use Result;
use super::pool::LeasedItem;
use super::pool::Pool;
use super::segment::create_segment;
use super::segment::Segment;
use core::searcher::Searcher;
use core::IndexMeta;
use core::SegmentMeta;
use core::SegmentReader;
use core::META_FILEPATH;
#[cfg(feature = "mmap")]
use directory::MmapDirectory;
use directory::{Directory, RAMDirectory};
use directory::{DirectoryClone, ManagedDirectory};
use indexer::index_writer::open_index_writer;
use indexer::index_writer::HEAP_SIZE_MIN;
use indexer::segment_updater::save_new_metas;
use indexer::DirectoryLock;
use num_cpus;
use std::path::Path;
use std::sync::Arc;
use tokenizer::BoxedTokenizer;
use tokenizer::TokenizerManager;
use IndexWriter;
use Result;
fn load_metas(directory: &Directory) -> Result<IndexMeta> {
let meta_data = directory.atomic_read(&META_FILEPATH)?;
let meta_string = String::from_utf8_lossy(&meta_data);
serde_json::from_str(&meta_string)
.map_err(|_| TantivyError::CorruptedFile(META_FILEPATH.clone()))
.map_err(|e| {
DataCorruption::new(
META_FILEPATH.clone(),
format!("Meta file cannot be deserialized. {:?}.", e),
)
})
.map_err(From::from)
}
/// Search Index
#[derive(Clone)]
pub struct Index {
directory: ManagedDirectory,
schema: Schema,
num_searchers: Arc<AtomicUsize>,
searcher_pool: Arc<Pool<Searcher>>,
executor: Arc<Executor>,
tokenizers: TokenizerManager,
}
impl Index {
/// Examines the director to see if it contains an index
pub fn exists<Dir: Directory>(dir: &Dir) -> bool {
dir.exists(&META_FILEPATH)
}
/// Accessor to the search executor.
///
/// This pool is used by default when calling `searcher.search(...)`
/// to perform search on the individual segments.
///
/// By default the executor is single thread, and simply runs in the calling thread.
pub fn search_executor(&self) -> &Executor {
self.executor.as_ref()
}
/// Replace the default single thread search executor pool
/// by a thread pool with a given number of threads.
pub fn set_multithread_executor(&mut self, num_threads: usize) {
self.executor = Arc::new(Executor::multi_thread(num_threads, "thrd-tantivy-search-"));
}
/// Replace the default single thread search executor pool
/// by a thread pool with a given number of threads.
pub fn set_default_multithread_executor(&mut self) {
let default_num_threads = num_cpus::get();
self.set_multithread_executor(default_num_threads);
}
/// Creates a new index using the `RAMDirectory`.
///
/// The index will be allocated in anonymous memory.
@@ -63,9 +99,29 @@ impl Index {
#[cfg(feature = "mmap")]
pub fn create_in_dir<P: AsRef<Path>>(directory_path: P, schema: Schema) -> Result<Index> {
let mmap_directory = MmapDirectory::open(directory_path)?;
if Index::exists(&mmap_directory) {
return Err(TantivyError::IndexAlreadyExists);
}
Index::create(mmap_directory, schema)
}
/// Opens or creates a new index in the provided directory
pub fn open_or_create<Dir: Directory>(dir: Dir, schema: Schema) -> Result<Index> {
if Index::exists(&dir) {
let index = Index::open(dir)?;
if index.schema() == schema {
Ok(index)
} else {
Err(TantivyError::SchemaError(
"An index exists but the schema does not match.".to_string(),
))
}
} else {
Index::create(dir, schema)
}
}
/// Creates a new index in a temp directory.
///
/// The index will use the `MMapDirectory` in a newly created directory.
@@ -82,13 +138,15 @@ impl Index {
/// Creates a new index given an implementation of the trait `Directory`
pub fn create<Dir: Directory>(dir: Dir, schema: Schema) -> Result<Index> {
let directory = ManagedDirectory::new(dir)?;
let directory = ManagedDirectory::wrap(dir)?;
Index::from_directory(directory, schema)
}
/// Create a new index from a directory.
///
/// This will overwrite existing meta.json
fn from_directory(mut directory: ManagedDirectory, schema: Schema) -> Result<Index> {
save_new_metas(schema.clone(), 0, directory.borrow_mut())?;
save_new_metas(schema.clone(), directory.borrow_mut())?;
let metas = IndexMeta::with_schema(schema);
Index::create_from_metas(directory, &metas)
}
@@ -96,15 +154,12 @@ impl Index {
/// Creates a new index given a directory and an `IndexMeta`.
fn create_from_metas(directory: ManagedDirectory, metas: &IndexMeta) -> Result<Index> {
let schema = metas.schema.clone();
let n_cpus = num_cpus::get();
let index = Index {
directory,
schema,
num_searchers: Arc::new(AtomicUsize::new(n_cpus)),
searcher_pool: Arc::new(Pool::new()),
tokenizers: TokenizerManager::default(),
executor: Arc::new(Executor::single_thread()),
};
index.load_searchers()?;
Ok(index)
}
@@ -113,6 +168,43 @@ impl Index {
&self.tokenizers
}
/// Helper to access the tokenizer associated to a specific field.
pub fn tokenizer_for_field(&self, field: Field) -> Result<Box<BoxedTokenizer>> {
let field_entry = self.schema.get_field_entry(field);
let field_type = field_entry.field_type();
let tokenizer_manager: &TokenizerManager = self.tokenizers();
let tokenizer_name_opt: Option<Box<BoxedTokenizer>> = match field_type {
FieldType::Str(text_options) => text_options
.get_indexing_options()
.map(|text_indexing_options| text_indexing_options.tokenizer().to_string())
.and_then(|tokenizer_name| tokenizer_manager.get(&tokenizer_name)),
_ => None,
};
match tokenizer_name_opt {
Some(tokenizer) => Ok(tokenizer),
None => Err(TantivyError::SchemaError(format!(
"{:?} is not a text field.",
field_entry.name()
))),
}
}
/// Create a default `IndexReader` for the given index.
///
/// See [`Index.reader_builder()`](#method.reader_builder).
pub fn reader(&self) -> Result<IndexReader> {
self.reader_builder().try_into()
}
/// Create a `IndexReader` for the given index.
///
/// Most project should create at most one reader for a given index.
/// This method is typically called only once per `Index` instance,
/// over the lifetime of most problem.
pub fn reader_builder(&self) -> IndexReaderBuilder {
IndexReaderBuilder::new(self.clone())
}
/// Opens a new directory from an index path.
#[cfg(feature = "mmap")]
pub fn open_in_dir<P: AsRef<Path>>(directory_path: P) -> Result<Index> {
@@ -122,7 +214,7 @@ impl Index {
/// Open the index using the provided directory
pub fn open<D: Directory>(directory: D) -> Result<Index> {
let directory = ManagedDirectory::new(directory)?;
let directory = ManagedDirectory::wrap(directory)?;
let metas = load_metas(&directory)?;
Index::create_from_metas(directory, &metas)
}
@@ -148,7 +240,8 @@ impl Index {
/// Each thread will receive a budget of `overall_heap_size_in_bytes / num_threads`.
///
/// # Errors
/// If the lockfile already exists, returns `Error::FileAlreadyExists`.
/// If the lockfile already exists, returns `Error::DirectoryLockBusy` or an `Error::IOError`.
///
/// # Panics
/// If the heap size per thread is too small, panics.
pub fn writer_with_num_threads(
@@ -156,7 +249,21 @@ impl Index {
num_threads: usize,
overall_heap_size_in_bytes: usize,
) -> Result<IndexWriter> {
let directory_lock = DirectoryLock::lock(self.directory().box_clone())?;
let directory_lock = self
.directory
.acquire_lock(&INDEX_WRITER_LOCK)
.map_err(|err| {
TantivyError::LockFailure(
err,
Some(
"Failed to acquire index lock. If you are using\
a regular directory, this means there is already an \
`IndexWriter` working on this `Directory`, in this process \
or in a different process."
.to_string(),
),
)
})?;
let heap_size_in_bytes_per_thread = overall_heap_size_in_bytes / num_threads;
open_index_writer(
self,
@@ -194,7 +301,8 @@ impl Index {
/// Returns the list of segments that are searchable
pub fn searchable_segments(&self) -> Result<Vec<Segment>> {
Ok(self.searchable_segment_metas()?
Ok(self
.searchable_segment_metas()?
.into_iter()
.map(|segment_meta| self.segment(segment_meta))
.collect())
@@ -229,53 +337,12 @@ impl Index {
/// Returns the list of segment ids that are searchable.
pub fn searchable_segment_ids(&self) -> Result<Vec<SegmentId>> {
Ok(self.searchable_segment_metas()?
Ok(self
.searchable_segment_metas()?
.iter()
.map(|segment_meta| segment_meta.id())
.map(SegmentMeta::id)
.collect())
}
/// Sets the number of searchers to use
///
/// Only works after the next call to `load_searchers`
pub fn set_num_searchers(&mut self, num_searchers: usize) {
self.num_searchers.store(num_searchers, Ordering::Release);
}
/// Creates a new generation of searchers after
/// a change of the set of searchable indexes.
///
/// This needs to be called when a new segment has been
/// published or after a merge.
pub fn load_searchers(&self) -> Result<()> {
let searchable_segments = self.searchable_segments()?;
let segment_readers: Vec<SegmentReader> = searchable_segments
.iter()
.map(SegmentReader::open)
.collect::<Result<_>>()?;
let schema = self.schema();
let num_searchers: usize = self.num_searchers.load(Ordering::Acquire);
let searchers = (0..num_searchers)
.map(|_| Searcher::new(schema.clone(), segment_readers.clone()))
.collect();
self.searcher_pool.publish_new_generation(searchers);
Ok(())
}
/// Returns a searcher
///
/// This method should be called every single time a search
/// query is performed.
/// The searchers are taken from a pool of `num_searchers` searchers.
/// If no searcher is available
/// this may block.
///
/// The same searcher must be used for a given query, as it ensures
/// the use of a consistent segment set.
pub fn searcher(&self) -> LeasedItem<Searcher> {
self.searcher_pool.acquire()
}
}
impl fmt::Debug for Index {
@@ -284,14 +351,190 @@ impl fmt::Debug for Index {
}
}
impl Clone for Index {
fn clone(&self) -> Index {
Index {
directory: self.directory.clone(),
schema: self.schema.clone(),
num_searchers: Arc::clone(&self.num_searchers),
searcher_pool: Arc::clone(&self.searcher_pool),
tokenizers: self.tokenizers.clone(),
#[cfg(test)]
mod tests {
use directory::RAMDirectory;
use schema::Field;
use schema::{Schema, INDEXED, TEXT};
use std::thread;
use std::time::Duration;
use Index;
use IndexReader;
use IndexWriter;
use ReloadPolicy;
#[test]
fn test_indexer_for_field() {
let mut schema_builder = Schema::builder();
let num_likes_field = schema_builder.add_u64_field("num_likes", INDEXED);
let body_field = schema_builder.add_text_field("body", TEXT);
let schema = schema_builder.build();
let index = Index::create_in_ram(schema);
assert!(index.tokenizer_for_field(body_field).is_ok());
assert_eq!(
format!("{:?}", index.tokenizer_for_field(num_likes_field).err()),
"Some(SchemaError(\"\\\"num_likes\\\" is not a text field.\"))"
);
}
#[test]
fn test_index_exists() {
let directory = RAMDirectory::create();
assert!(!Index::exists(&directory));
assert!(Index::create(directory.clone(), throw_away_schema()).is_ok());
assert!(Index::exists(&directory));
}
#[test]
fn open_or_create_should_create() {
let directory = RAMDirectory::create();
assert!(!Index::exists(&directory));
assert!(Index::open_or_create(directory.clone(), throw_away_schema()).is_ok());
assert!(Index::exists(&directory));
}
#[test]
fn open_or_create_should_open() {
let directory = RAMDirectory::create();
assert!(Index::create(directory.clone(), throw_away_schema()).is_ok());
assert!(Index::exists(&directory));
assert!(Index::open_or_create(directory, throw_away_schema()).is_ok());
}
#[test]
fn create_should_wipeoff_existing() {
let directory = RAMDirectory::create();
assert!(Index::create(directory.clone(), throw_away_schema()).is_ok());
assert!(Index::exists(&directory));
assert!(Index::create(directory.clone(), Schema::builder().build()).is_ok());
}
#[test]
fn open_or_create_exists_but_schema_does_not_match() {
let directory = RAMDirectory::create();
assert!(Index::create(directory.clone(), throw_away_schema()).is_ok());
assert!(Index::exists(&directory));
assert!(Index::open_or_create(directory.clone(), throw_away_schema()).is_ok());
let err = Index::open_or_create(directory, Schema::builder().build());
assert_eq!(
format!("{:?}", err.unwrap_err()),
"SchemaError(\"An index exists but the schema does not match.\")"
);
}
fn throw_away_schema() -> Schema {
let mut schema_builder = Schema::builder();
let _ = schema_builder.add_u64_field("num_likes", INDEXED);
schema_builder.build()
}
#[test]
fn test_index_on_commit_reload_policy() {
let schema = throw_away_schema();
let field = schema.get_field("num_likes").unwrap();
let index = Index::create_in_ram(schema);
let reader = index
.reader_builder()
.reload_policy(ReloadPolicy::OnCommit)
.try_into()
.unwrap();
assert_eq!(reader.searcher().num_docs(), 0);
let mut writer = index.writer_with_num_threads(1, 3_000_000).unwrap();
test_index_on_commit_reload_policy_aux(field, &mut writer, &reader);
}
#[cfg(feature = "mmap")]
mod mmap_specific {
use super::*;
use std::path::PathBuf;
use tempdir::TempDir;
#[test]
fn test_index_on_commit_reload_policy_mmap() {
let schema = throw_away_schema();
let field = schema.get_field("num_likes").unwrap();
let tempdir = TempDir::new("index").unwrap();
let tempdir_path = PathBuf::from(tempdir.path());
let index = Index::create_in_dir(&tempdir_path, schema).unwrap();
let mut writer = index.writer_with_num_threads(1, 3_000_000).unwrap();
writer.commit().unwrap();
let reader = index
.reader_builder()
.reload_policy(ReloadPolicy::OnCommit)
.try_into()
.unwrap();
assert_eq!(reader.searcher().num_docs(), 0);
test_index_on_commit_reload_policy_aux(field, &mut writer, &reader);
}
#[test]
fn test_index_manual_policy_mmap() {
let schema = throw_away_schema();
let field = schema.get_field("num_likes").unwrap();
let index = Index::create_from_tempdir(schema).unwrap();
let mut writer = index.writer_with_num_threads(1, 3_000_000).unwrap();
writer.commit().unwrap();
let reader = index
.reader_builder()
.reload_policy(ReloadPolicy::Manual)
.try_into()
.unwrap();
assert_eq!(reader.searcher().num_docs(), 0);
writer.add_document(doc!(field=>1u64));
writer.commit().unwrap();
thread::sleep(Duration::from_millis(500));
assert_eq!(reader.searcher().num_docs(), 0);
reader.reload().unwrap();
assert_eq!(reader.searcher().num_docs(), 1);
}
#[test]
fn test_index_on_commit_reload_policy_different_directories() {
let schema = throw_away_schema();
let field = schema.get_field("num_likes").unwrap();
let tempdir = TempDir::new("index").unwrap();
let tempdir_path = PathBuf::from(tempdir.path());
let write_index = Index::create_in_dir(&tempdir_path, schema).unwrap();
let read_index = Index::open_in_dir(&tempdir_path).unwrap();
let reader = read_index
.reader_builder()
.reload_policy(ReloadPolicy::OnCommit)
.try_into()
.unwrap();
assert_eq!(reader.searcher().num_docs(), 0);
let mut writer = write_index.writer_with_num_threads(1, 3_000_000).unwrap();
test_index_on_commit_reload_policy_aux(field, &mut writer, &reader);
}
}
fn test_index_on_commit_reload_policy_aux(
field: Field,
writer: &mut IndexWriter,
reader: &IndexReader,
) {
assert_eq!(reader.searcher().num_docs(), 0);
writer.add_document(doc!(field=>1u64));
writer.commit().unwrap();
let mut count = 0;
for _ in 0..100 {
count = reader.searcher().num_docs();
if count > 0 {
break;
}
thread::sleep(Duration::from_millis(100));
}
assert_eq!(count, 1);
writer.add_document(doc!(field=>2u64));
writer.commit().unwrap();
let mut count = 0;
for _ in 0..10 {
count = reader.searcher().num_docs();
if count > 1 {
break;
}
thread::sleep(Duration::from_millis(100));
}
assert_eq!(count, 2);
}
}

6
src/core/index_meta.rs
View File

@@ -46,19 +46,19 @@ impl fmt::Debug for IndexMeta {
mod tests {
use super::IndexMeta;
use schema::{SchemaBuilder, TEXT};
use schema::{Schema, TEXT};
use serde_json;
#[test]
fn test_serialize_metas() {
let schema = {
let mut schema_builder = SchemaBuilder::new();
let mut schema_builder = Schema::builder();
schema_builder.add_text_field("text", TEXT);
schema_builder.build()
};
let index_metas = IndexMeta {
segments: Vec::new(),
schema: schema,
schema,
opstamp: 0u64,
payload: None,
};

15
src/core/inverted_index_reader.rs
View File

@@ -1,13 +1,13 @@
use common::BinarySerializable;
use directory::ReadOnlySource;
use owned_read::OwnedRead;
use positions::PositionReader;
use postings::TermInfo;
use postings::{BlockSegmentPostings, SegmentPostings};
use schema::FieldType;
use schema::IndexRecordOption;
use schema::Term;
use termdict::TermDictionary;
use owned_read::OwnedRead;
use positions::PositionReader;
/// The inverted index reader is in charge of accessing
/// the inverted index associated to a specific field.
@@ -32,6 +32,7 @@ pub struct InvertedIndexReader {
}
impl InvertedIndexReader {
#[cfg_attr(feature = "cargo-clippy", allow(clippy::needless_pass_by_value))] // for symetry
pub(crate) fn new(
termdict: TermDictionary,
postings_source: ReadOnlySource,
@@ -54,12 +55,12 @@ impl InvertedIndexReader {
/// Creates an empty `InvertedIndexReader` object, which
/// contains no terms at all.
pub fn empty(field_type: FieldType) -> InvertedIndexReader {
pub fn empty(field_type: &FieldType) -> InvertedIndexReader {
let record_option = field_type
.get_index_record_option()
.unwrap_or(IndexRecordOption::Basic);
InvertedIndexReader {
termdict: TermDictionary::empty(field_type),
termdict: TermDictionary::empty(&field_type),
postings_source: ReadOnlySource::empty(),
positions_source: ReadOnlySource::empty(),
positions_idx_source: ReadOnlySource::empty(),
@@ -100,7 +101,6 @@ impl InvertedIndexReader {
block_postings.reset(term_info.doc_freq, postings_reader);
}
/// Returns a block postings given a `Term`.
/// This method is for an advanced usage only.
///
@@ -111,7 +111,7 @@ impl InvertedIndexReader {
option: IndexRecordOption,
) -> Option<BlockSegmentPostings> {
self.get_term_info(term)
.map(move|term_info| self.read_block_postings_from_terminfo(&term_info, option))
.map(move |term_info| self.read_block_postings_from_terminfo(&term_info, option))
}
/// Returns a block postings given a `term_info`.
@@ -147,7 +147,8 @@ impl InvertedIndexReader {
if option.has_positions() {
let position_reader = self.positions_source.clone();
let skip_reader = self.positions_idx_source.clone();
let position_reader = PositionReader::new(position_reader, skip_reader, term_info.positions_idx);
let position_reader =
PositionReader::new(position_reader, skip_reader, term_info.positions_idx);
Some(position_reader)
} else {
None

10
src/core/mod.rs
View File

@@ -1,7 +1,7 @@
mod executor;
pub mod index;
mod index_meta;
mod inverted_index_reader;
mod pool;
pub mod searcher;
mod segment;
mod segment_component;
@@ -9,6 +9,7 @@ mod segment_id;
mod segment_meta;
mod segment_reader;
pub use self::executor::Executor;
pub use self::index::Index;
pub use self::index_meta::IndexMeta;
pub use self::inverted_index_reader::InvertedIndexReader;
@@ -23,6 +24,7 @@ pub use self::segment_reader::SegmentReader;
use std::path::PathBuf;
lazy_static! {
/// The meta file contains all the information about the list of segments and the schema
/// of the index.
pub static ref META_FILEPATH: PathBuf = PathBuf::from("meta.json");
@@ -33,10 +35,4 @@ lazy_static! {
/// Removing this file is safe, but will prevent the garbage collection of all of the file that
/// are currently in the directory
pub static ref MANAGED_FILEPATH: PathBuf = PathBuf::from(".managed.json");
/// Only one process should be able to write tantivy's index at a time.
/// This file, when present, is in charge of preventing other processes to open an IndexWriter.
///
/// If the process is killed and this file remains, it is safe to remove it manually.
pub static ref LOCKFILE_FILEPATH: PathBuf = PathBuf::from(".tantivy-indexer.lock");
}

134
src/core/pool.rs
View File

@@ -1,134 +0,0 @@
use crossbeam::queue::MsQueue;
use std::mem;
use std::ops::{Deref, DerefMut};
use std::sync::atomic::AtomicUsize;
use std::sync::atomic::Ordering;
use std::sync::Arc;
pub struct GenerationItem<T> {
generation: usize,
item: T,
}
pub struct Pool<T> {
queue: Arc<MsQueue<GenerationItem<T>>>,
freshest_generation: AtomicUsize,
next_generation: AtomicUsize,
}
impl<T> Pool<T> {
pub fn new() -> Pool<T> {
let queue = Arc::new(MsQueue::new());
Pool {
queue,
freshest_generation: AtomicUsize::default(),
next_generation: AtomicUsize::default(),
}
}
pub fn publish_new_generation(&self, items: Vec<T>) {
let next_generation = self.next_generation.fetch_add(1, Ordering::SeqCst) + 1;
for item in items {
let gen_item = GenerationItem {
item,
generation: next_generation,
};
self.queue.push(gen_item);
}
self.advertise_generation(next_generation);
}
/// At the exit of this method,
/// - freshest_generation has a value greater or equal than generation
/// - freshest_generation has a value that has been advertised
/// - freshest_generation has)
fn advertise_generation(&self, generation: usize) {
// not optimal at all but the easiest to read proof.
loop {
let former_generation = self.freshest_generation.load(Ordering::Acquire);
if former_generation >= generation {
break;
}
self.freshest_generation.compare_and_swap(
former_generation,
generation,
Ordering::SeqCst,
);
}
}
fn generation(&self) -> usize {
self.freshest_generation.load(Ordering::Acquire)
}
pub fn acquire(&self) -> LeasedItem<T> {
let generation = self.generation();
loop {
let gen_item = self.queue.pop();
if gen_item.generation >= generation {
return LeasedItem {
gen_item: Some(gen_item),
recycle_queue: Arc::clone(&self.queue),
};
} else {
// this searcher is obsolete,
// removing it from the pool.
}
}
}
}
pub struct LeasedItem<T> {
gen_item: Option<GenerationItem<T>>,
recycle_queue: Arc<MsQueue<GenerationItem<T>>>,
}
impl<T> Deref for LeasedItem<T> {
type Target = T;
fn deref(&self) -> &T {
&self.gen_item
.as_ref()
.expect("Unwrapping a leased item should never fail")
.item // unwrap is safe here
}
}
impl<T> DerefMut for LeasedItem<T> {
fn deref_mut(&mut self) -> &mut T {
&mut self.gen_item
.as_mut()
.expect("Unwrapping a mut leased item should never fail")
.item // unwrap is safe here
}
}
impl<T> Drop for LeasedItem<T> {
fn drop(&mut self) {
let gen_item: GenerationItem<T> = mem::replace(&mut self.gen_item, None)
.expect("Unwrapping a leased item should never fail");
self.recycle_queue.push(gen_item);
}
}
#[cfg(test)]
mod tests {
use super::Pool;
use std::iter;
#[test]
fn test_pool() {
let items10: Vec<usize> = iter::repeat(10).take(10).collect();
let pool = Pool::new();
pool.publish_new_generation(items10);
for _ in 0..20 {
assert_eq!(*pool.acquire(), 10);
}
let items11: Vec<usize> = iter::repeat(11).take(10).collect();
pool.publish_new_generation(items11);
for _ in 0..20 {
assert_eq!(*pool.acquire(), 11);
}
}
}

136
src/core/searcher.rs
View File

@@ -1,16 +1,43 @@
use collector::Collector;
use collector::SegmentCollector;
use core::Executor;
use core::InvertedIndexReader;
use core::SegmentReader;
use query::Query;
use query::Scorer;
use query::Weight;
use schema::Document;
use schema::Schema;
use schema::{Field, Term};
use space_usage::SearcherSpaceUsage;
use std::fmt;
use std::sync::Arc;
use store::StoreReader;
use termdict::TermMerger;
use DocAddress;
use Index;
use Result;
fn collect_segment<C: Collector>(
collector: &C,
weight: &Weight,
segment_ord: u32,
segment_reader: &SegmentReader,
) -> Result<C::Fruit> {
let mut scorer = weight.scorer(segment_reader)?;
let mut segment_collector = collector.for_segment(segment_ord as u32, segment_reader)?;
if let Some(delete_bitset) = segment_reader.delete_bitset() {
scorer.for_each(&mut |doc, score| {
if !delete_bitset.is_deleted(doc) {
segment_collector.collect(doc, score);
}
});
} else {
scorer.for_each(&mut |doc, score| segment_collector.collect(doc, score));
}
Ok(segment_collector.harvest())
}
/// Holds a list of `SegmentReader`s ready for search.
///
/// It guarantees that the `Segment` will not be removed before
@@ -18,25 +45,43 @@ use Result;
///
pub struct Searcher {
schema: Schema,
index: Index,
segment_readers: Vec<SegmentReader>,
store_readers: Vec<StoreReader>,
}
impl Searcher {
/// Creates a new `Searcher`
pub(crate) fn new(schema: Schema, segment_readers: Vec<SegmentReader>) -> Searcher {
pub(crate) fn new(
schema: Schema,
index: Index,
segment_readers: Vec<SegmentReader>,
) -> Searcher {
let store_readers = segment_readers
.iter()
.map(SegmentReader::get_store_reader)
.collect();
Searcher {
schema,
index,
segment_readers,
store_readers,
}
}
/// Returns the `Index` associated to the `Searcher`
pub fn index(&self) -> &Index {
&self.index
}
/// Fetches a document from tantivy's store given a `DocAddress`.
///
/// The searcher uses the segment ordinal to route the
/// the request to the right `Segment`.
pub fn doc(&self, doc_address: &DocAddress) -> Result<Document> {
let DocAddress(segment_local_id, doc_id) = *doc_address;
let segment_reader = &self.segment_readers[segment_local_id as usize];
segment_reader.doc(doc_id)
pub fn doc(&self, doc_address: DocAddress) -> Result<Document> {
let DocAddress(segment_local_id, doc_id) = doc_address;
let store_reader = &self.store_readers[segment_local_id as usize];
store_reader.get(doc_id)
}
/// Access the schema associated to the index of this searcher.
@@ -48,7 +93,7 @@ impl Searcher {
pub fn num_docs(&self) -> u64 {
self.segment_readers
.iter()
.map(|segment_reader| segment_reader.num_docs() as u64)
.map(|segment_reader| u64::from(segment_reader.num_docs()))
.sum::<u64>()
}
@@ -57,7 +102,9 @@ impl Searcher {
pub fn doc_freq(&self, term: &Term) -> u64 {
self.segment_readers
.iter()
.map(|segment_reader| segment_reader.inverted_index(term.field()).doc_freq(term) as u64)
.map(|segment_reader| {
u64::from(segment_reader.inverted_index(term.field()).doc_freq(term))
})
.sum::<u64>()
}
@@ -71,19 +118,78 @@ impl Searcher {
&self.segment_readers[segment_ord as usize]
}
/// Runs a query on the segment readers wrapped by the searcher
pub fn search<C: Collector>(&self, query: &Query, collector: &mut C) -> Result<()> {
query.search(self, collector)
/// Runs a query on the segment readers wrapped by the searcher.
///
/// Search works as follows :
///
/// First the weight object associated to the query is created.
///
/// Then, the query loops over the segments and for each segment :
/// - setup the collector and informs it that the segment being processed has changed.
/// - creates a SegmentCollector for collecting documents associated to the segment
/// - creates a `Scorer` object associated for this segment
/// - iterate through the matched documents and push them to the segment collector.
///
/// Finally, the Collector merges each of the child collectors into itself for result usability
/// by the caller.
pub fn search<C: Collector>(&self, query: &Query, collector: &C) -> Result<C::Fruit> {
let executor = self.index.search_executor();
self.search_with_executor(query, collector, executor)
}
/// Same as [`search(...)`](#method.search) but multithreaded.
///
/// The current implementation is rather naive :
/// multithreading is by splitting search into as many task
/// as there are segments.
///
/// It is powerless at making search faster if your index consists in
/// one large segment.
///
/// Also, keep in my multithreading a single query on several
/// threads will not improve your throughput. It can actually
/// hurt it. It will however, decrease the average response time.
pub fn search_with_executor<C: Collector>(
&self,
query: &Query,
collector: &C,
executor: &Executor,
) -> Result<C::Fruit> {
let scoring_enabled = collector.requires_scoring();
let weight = query.weight(self, scoring_enabled)?;
let segment_readers = self.segment_readers();
let fruits = executor.map(
|(segment_ord, segment_reader)| {
collect_segment(
collector,
weight.as_ref(),
segment_ord as u32,
segment_reader,
)
},
segment_readers.iter().enumerate(),
)?;
collector.merge_fruits(fruits)
}
/// Return the field searcher associated to a `Field`.
pub fn field(&self, field: Field) -> FieldSearcher {
let inv_index_readers = self.segment_readers
let inv_index_readers = self
.segment_readers
.iter()
.map(|segment_reader| segment_reader.inverted_index(field))
.collect::<Vec<_>>();
FieldSearcher::new(inv_index_readers)
}
/// Summarize total space usage of this searcher.
pub fn space_usage(&self) -> SearcherSpaceUsage {
let mut space_usage = SearcherSpaceUsage::new();
for segment_reader in self.segment_readers.iter() {
space_usage.add_segment(segment_reader.space_usage());
}
space_usage
}
}
pub struct FieldSearcher {
@@ -98,7 +204,8 @@ impl FieldSearcher {
/// Returns a Stream over all of the sorted unique terms of
/// for the given field.
pub fn terms(&self) -> TermMerger {
let term_streamers: Vec<_> = self.inv_index_readers
let term_streamers: Vec<_> = self
.inv_index_readers
.iter()
.map(|inverted_index| inverted_index.terms().stream())
.collect();
@@ -108,9 +215,10 @@ impl FieldSearcher {
impl fmt::Debug for Searcher {
fn fmt(&self, f: &mut fmt::Formatter) -> fmt::Result {
let segment_ids = self.segment_readers
let segment_ids = self
.segment_readers
.iter()
.map(|segment_reader| segment_reader.segment_id())
.map(SegmentReader::segment_id)
.collect::<Vec<_>>();
write!(f, "Searcher({:?})", segment_ids)
}

2
src/core/segment_component.rs
View File

@@ -41,6 +41,6 @@ impl SegmentComponent {
SegmentComponent::STORE,
SegmentComponent::DELETE,
];
SEGMENT_COMPONENTS.into_iter()
SEGMENT_COMPONENTS.iter()
}
}

8
src/core/segment_id.rs
View File

@@ -19,7 +19,7 @@ pub struct SegmentId(Uuid);
#[cfg(test)]
lazy_static! {
static ref AUTO_INC_COUNTER: atomic::AtomicUsize = atomic::AtomicUsize::default();
static ref EMPTY_ARR: [u8; 8] = [0u8; 8];
static ref ZERO_ARRAY: [u8; 8] = [0u8; 8];
}
// During tests, we generate the segment id in a autoincrement manner
@@ -30,7 +30,7 @@ lazy_static! {
#[cfg(test)]
fn create_uuid() -> Uuid {
let new_auto_inc_id = (*AUTO_INC_COUNTER).fetch_add(1, atomic::Ordering::SeqCst);
Uuid::from_fields(new_auto_inc_id as u32, 0, 0, &*EMPTY_ARR).unwrap()
Uuid::from_fields(new_auto_inc_id as u32, 0, 0, &*ZERO_ARRAY).unwrap()
}
#[cfg(not(test))]
@@ -52,12 +52,12 @@ impl SegmentId {
/// Picking the first 8 chars is ok to identify
/// segments in a display message.
pub fn short_uuid_string(&self) -> String {
(&self.0.simple().to_string()[..8]).to_string()
(&self.0.to_simple_ref().to_string()[..8]).to_string()
}
/// Returns a segment uuid string.
pub fn uuid_string(&self) -> String {
self.0.simple().to_string()
self.0.to_simple_ref().to_string()
}
}

2
src/core/segment_meta.rs
View File

@@ -50,7 +50,7 @@ impl<'a> serde::Deserialize<'a> for SegmentMeta {
{
let inner = InnerSegmentMeta::deserialize(deserializer)?;
let tracked = INVENTORY.track(inner);
Ok(SegmentMeta { tracked: tracked })
Ok(SegmentMeta { tracked })
}
}

181
src/core/segment_reader.rs
View File

@@ -4,19 +4,15 @@ use core::InvertedIndexReader;
use core::Segment;
use core::SegmentComponent;
use core::SegmentId;
use core::SegmentMeta;
use error::TantivyError;
use directory::ReadOnlySource;
use fastfield::DeleteBitSet;
use fastfield::FacetReader;
use fastfield::FastFieldReader;
use fastfield::{self, FastFieldNotAvailableError};
use fastfield::{BytesFastFieldReader, FastValue, MultiValueIntFastFieldReader};
use fastfield::FastFieldReaders;
use fieldnorm::FieldNormReader;
use schema::Cardinality;
use schema::Document;
use schema::Field;
use schema::FieldType;
use schema::Schema;
use space_usage::SegmentSpaceUsage;
use std::collections::HashMap;
use std::fmt;
use std::sync::Arc;
@@ -44,16 +40,17 @@ pub struct SegmentReader {
inv_idx_reader_cache: Arc<RwLock<HashMap<Field, Arc<InvertedIndexReader>>>>,
segment_id: SegmentId,
segment_meta: SegmentMeta,
max_doc: DocId,
num_docs: DocId,
termdict_composite: CompositeFile,
postings_composite: CompositeFile,
positions_composite: CompositeFile,
positions_idx_composite: CompositeFile,
fast_fields_composite: CompositeFile,
fast_fields_readers: Arc<FastFieldReaders>,
fieldnorms_composite: CompositeFile,
store_reader: StoreReader,
store_source: ReadOnlySource,
delete_bitset_opt: Option<DeleteBitSet>,
schema: Schema,
}
@@ -64,7 +61,7 @@ impl SegmentReader {
/// Today, `tantivy` does not handle deletes, so it happens
/// to also be the number of documents in the index.
pub fn max_doc(&self) -> DocId {
self.segment_meta.max_doc()
self.max_doc
}
/// Returns the number of documents.
@@ -73,7 +70,7 @@ impl SegmentReader {
/// Today, `tantivy` does not handle deletes so max doc and
/// num_docs are the same.
pub fn num_docs(&self) -> DocId {
self.segment_meta.num_docs()
self.num_docs
}
/// Returns the schema of the index this segment belongs to.
@@ -104,98 +101,27 @@ impl SegmentReader {
///
/// # Panics
/// May panic if the index is corrupted.
pub fn fast_field_reader<Item: FastValue>(
&self,
field: Field,
) -> fastfield::Result<FastFieldReader<Item>> {
let field_entry = self.schema.get_field_entry(field);
if Item::fast_field_cardinality(field_entry.field_type()) == Some(Cardinality::SingleValue)
{
self.fast_fields_composite
.open_read(field)
.ok_or_else(|| FastFieldNotAvailableError::new(field_entry))
.map(FastFieldReader::open)
} else {
Err(FastFieldNotAvailableError::new(field_entry))
}
}
pub(crate) fn fast_field_reader_with_idx<Item: FastValue>(
&self,
field: Field,
idx: usize,
) -> fastfield::Result<FastFieldReader<Item>> {
if let Some(ff_source) = self.fast_fields_composite.open_read_with_idx(field, idx) {
Ok(FastFieldReader::open(ff_source))
} else {
let field_entry = self.schema.get_field_entry(field);
Err(FastFieldNotAvailableError::new(field_entry))
}
}
/// Accessor to the `MultiValueIntFastFieldReader` associated to a given `Field`.
/// May panick if the field is not a multivalued fastfield of the type `Item`.
pub fn multi_fast_field_reader<Item: FastValue>(
&self,
field: Field,
) -> fastfield::Result<MultiValueIntFastFieldReader<Item>> {
let field_entry = self.schema.get_field_entry(field);
if Item::fast_field_cardinality(field_entry.field_type()) == Some(Cardinality::MultiValues)
{
let idx_reader = self.fast_field_reader_with_idx(field, 0)?;
let vals_reader = self.fast_field_reader_with_idx(field, 1)?;
Ok(MultiValueIntFastFieldReader::open(idx_reader, vals_reader))
} else {
Err(FastFieldNotAvailableError::new(field_entry))
}
}
/// Accessor to the `BytesFastFieldReader` associated to a given `Field`.
pub fn bytes_fast_field_reader(&self, field: Field) -> fastfield::Result<BytesFastFieldReader> {
let field_entry = self.schema.get_field_entry(field);
match field_entry.field_type() {
&FieldType::Bytes => {}
_ => return Err(FastFieldNotAvailableError::new(field_entry)),
}
let idx_reader = self.fast_fields_composite
.open_read_with_idx(field, 0)
.ok_or_else(|| FastFieldNotAvailableError::new(field_entry))
.map(FastFieldReader::open)?;
let values = self.fast_fields_composite
.open_read_with_idx(field, 1)
.ok_or_else(|| FastFieldNotAvailableError::new(field_entry))?;
Ok(BytesFastFieldReader::open(idx_reader, values))
pub fn fast_fields(&self) -> &FastFieldReaders {
&self.fast_fields_readers
}
/// Accessor to the `FacetReader` associated to a given `Field`.
pub fn facet_reader(&self, field: Field) -> Result<FacetReader> {
pub fn facet_reader(&self, field: Field) -> Option<FacetReader> {
let field_entry = self.schema.get_field_entry(field);
if field_entry.field_type() != &FieldType::HierarchicalFacet {
return Err(TantivyError::InvalidArgument(format!(
"The field {:?} is not a \
hierarchical facet.",
field_entry
)).into());
return None;
}
let term_ords_reader = self.multi_fast_field_reader(field)?;
let termdict_source = self.termdict_composite.open_read(field).ok_or_else(|| {
TantivyError::InvalidArgument(format!(
"The field \"{}\" is a hierarchical \
but this segment does not seem to have the field term \
dictionary.",
field_entry.name()
))
})?;
let termdict = TermDictionary::from_source(termdict_source);
let term_ords_reader = self.fast_fields().u64s(field)?;
let termdict_source = self.termdict_composite.open_read(field)?;
let termdict = TermDictionary::from_source(&termdict_source);
let facet_reader = FacetReader::new(term_ords_reader, termdict);
Ok(facet_reader)
Some(facet_reader)
}
/// Accessor to the segment's `Field norms`'s reader.
///
/// Field norms are the length (in tokens) of the fields.
/// It is used in the computation of the [TfIdf]
/// (https://fulmicoton.gitbooks.io/tantivy-doc/content/tfidf.html).
/// It is used in the computation of the [TfIdf](https://fulmicoton.gitbooks.io/tantivy-doc/content/tfidf.html).
///
/// They are simply stored as a fast field, serialized in
/// the `.fieldnorm` file of the segment.
@@ -213,8 +139,8 @@ impl SegmentReader {
}
/// Accessor to the segment's `StoreReader`.
pub fn get_store_reader(&self) -> &StoreReader {
&self.store_reader
pub fn get_store_reader(&self) -> StoreReader {
StoreReader::from_source(self.store_source.clone())
}
/// Open a new segment for reading.
@@ -223,7 +149,8 @@ impl SegmentReader {
let termdict_composite = CompositeFile::open(&termdict_source)?;
let store_source = segment.open_read(SegmentComponent::STORE)?;
let store_reader = StoreReader::from_source(store_source);
fail_point!("SegmentReader::open#middle");
let postings_source = segment.open_read(SegmentComponent::POSTINGS)?;
let postings_composite = CompositeFile::open(&postings_source)?;
@@ -244,8 +171,12 @@ impl SegmentReader {
}
};
let schema = segment.schema();
let fast_fields_data = segment.open_read(SegmentComponent::FASTFIELDS)?;
let fast_fields_composite = CompositeFile::open(&fast_fields_data)?;
let fast_field_readers =
Arc::new(FastFieldReaders::load_all(&schema, &fast_fields_composite)?);
let fieldnorms_data = segment.open_read(SegmentComponent::FIELDNORMS)?;
let fieldnorms_composite = CompositeFile::open(&fieldnorms_data)?;
@@ -257,16 +188,16 @@ impl SegmentReader {
None
};
let schema = segment.schema();
Ok(SegmentReader {
inv_idx_reader_cache: Arc::new(RwLock::new(HashMap::new())),
segment_meta: segment.meta().clone(),
max_doc: segment.meta().max_doc(),
num_docs: segment.meta().num_docs(),
termdict_composite,
postings_composite,
fast_fields_composite,
fast_fields_readers: fast_field_readers,
fieldnorms_composite,
segment_id: segment.id(),
store_reader,
store_source,
delete_bitset_opt,
positions_composite,
positions_idx_composite,
@@ -282,7 +213,8 @@ impl SegmentReader {
/// term dictionary associated to a specific field,
/// and opening the posting list associated to any term.
pub fn inverted_index(&self, field: Field) -> Arc<InvertedIndexReader> {
if let Some(inv_idx_reader) = self.inv_idx_reader_cache
if let Some(inv_idx_reader) = self
.inv_idx_reader_cache
.read()
.expect("Lock poisoned. This should never happen")
.get(&field)
@@ -306,25 +238,28 @@ impl SegmentReader {
// As a result, no data is associated to the inverted index.
//
// Returns an empty inverted index.
return Arc::new(InvertedIndexReader::empty(field_type.clone()));
return Arc::new(InvertedIndexReader::empty(field_type));
}
let postings_source = postings_source_opt.unwrap();
let termdict_source = self.termdict_composite
let termdict_source = self
.termdict_composite
.open_read(field)
.expect("Failed to open field term dictionary in composite file. Is the field indexed");
let positions_source = self.positions_composite
let positions_source = self
.positions_composite
.open_read(field)
.expect("Index corrupted. Failed to open field positions in composite file.");
let positions_idx_source = self.positions_idx_composite
let positions_idx_source = self
.positions_idx_composite
.open_read(field)
.expect("Index corrupted. Failed to open field positions in composite file.");
let inv_idx_reader = Arc::new(InvertedIndexReader::new(
TermDictionary::from_source(termdict_source),
TermDictionary::from_source(&termdict_source),
postings_source,
positions_source,
positions_idx_source,
@@ -341,14 +276,6 @@ impl SegmentReader {
inv_idx_reader
}
/// Returns the document (or to be accurate, its stored field)
/// bearing the given doc id.
/// This method is slow and should seldom be called from
/// within a collector.
pub fn doc(&self, doc_id: DocId) -> Result<Document> {
self.store_reader.get(doc_id)
}
/// Returns the segment id
pub fn segment_id(&self) -> SegmentId {
self.segment_id
@@ -372,6 +299,24 @@ impl SegmentReader {
pub fn doc_ids_alive(&self) -> SegmentReaderAliveDocsIterator {
SegmentReaderAliveDocsIterator::new(&self)
}
/// Summarize total space usage of this segment.
pub fn space_usage(&self) -> SegmentSpaceUsage {
SegmentSpaceUsage::new(
self.num_docs(),
self.termdict_composite.space_usage(),
self.postings_composite.space_usage(),
self.positions_composite.space_usage(),
self.positions_idx_composite.space_usage(),
self.fast_fields_readers.space_usage(),
self.fieldnorms_composite.space_usage(),
self.get_store_reader().space_usage(),
self.delete_bitset_opt
.as_ref()
.map(DeleteBitSet::space_usage)
.unwrap_or(0),
)
}
}
impl fmt::Debug for SegmentReader {
@@ -391,7 +336,7 @@ pub struct SegmentReaderAliveDocsIterator<'a> {
impl<'a> SegmentReaderAliveDocsIterator<'a> {
pub fn new(reader: &'a SegmentReader) -> SegmentReaderAliveDocsIterator<'a> {
SegmentReaderAliveDocsIterator {
reader: reader,
reader,
max_doc: reader.max_doc(),
current: 0,
}
@@ -429,12 +374,12 @@ impl<'a> Iterator for SegmentReaderAliveDocsIterator<'a> {
#[cfg(test)]
mod test {
use core::Index;
use schema::{SchemaBuilder, Term, STORED, TEXT};
use schema::{Schema, Term, STORED, TEXT};
use DocId;
#[test]
fn test_alive_docs_iterator() {
let mut schema_builder = SchemaBuilder::new();
let mut schema_builder = Schema::builder();
schema_builder.add_text_field("name", TEXT | STORED);
let schema = schema_builder.build();
let index = Index::create_in_ram(schema.clone());
@@ -459,9 +404,7 @@ mod test {
// ok, now we should have a deleted doc
index_writer2.commit().unwrap();
}
index.load_searchers().unwrap();
let searcher = index.searcher();
let searcher = index.reader().unwrap().searcher();
let docs: Vec<DocId> = searcher.segment_reader(0).doc_ids_alive().collect();
assert_eq!(vec![0u32, 2u32], docs);
}

144
src/directory/directory.rs
View File

@@ -1,11 +1,104 @@
use directory::directory_lock::Lock;
use directory::error::LockError;
use directory::error::{DeleteError, OpenReadError, OpenWriteError};
use directory::WatchCallback;
use directory::WatchHandle;
use directory::{ReadOnlySource, WritePtr};
use std::fmt;
use std::io;
use std::io::Write;
use std::marker::Send;
use std::marker::Sync;
use std::path::Path;
use std::path::PathBuf;
use std::result;
use std::thread;
use std::time::Duration;
/// Retry the logic of acquiring locks is pretty simple.
/// We just retry `n` times after a given `duratio`, both
/// depending on the type of lock.
struct RetryPolicy {
num_retries: usize,
wait_in_ms: u64,
}
impl RetryPolicy {
fn no_retry() -> RetryPolicy {
RetryPolicy {
num_retries: 0,
wait_in_ms: 0,
}
}
fn wait_and_retry(&mut self) -> bool {
if self.num_retries == 0 {
false
} else {
self.num_retries -= 1;
let wait_duration = Duration::from_millis(self.wait_in_ms);
thread::sleep(wait_duration);
true
}
}
}
/// The `DirectoryLock` is an object that represents a file lock.
/// See [`LockType`](struct.LockType.html)
///
/// It is transparently associated to a lock file, that gets deleted
/// on `Drop.` The lock is released automatically on `Drop`.
pub struct DirectoryLock(Box<Drop + Send + 'static>);
struct DirectoryLockGuard {
directory: Box<Directory>,
path: PathBuf,
}
impl<T: Drop + Send + 'static> From<Box<T>> for DirectoryLock {
fn from(underlying: Box<T>) -> Self {
DirectoryLock(underlying)
}
}
impl Drop for DirectoryLockGuard {
fn drop(&mut self) {
if let Err(e) = self.directory.delete(&*self.path) {
error!("Failed to remove the lock file. {:?}", e);
}
}
}
enum TryAcquireLockError {
FileExists,
IOError(io::Error),
}
fn try_acquire_lock(
filepath: &Path,
directory: &mut Directory,
) -> Result<DirectoryLock, TryAcquireLockError> {
let mut write = directory.open_write(filepath).map_err(|e| match e {
OpenWriteError::FileAlreadyExists(_) => TryAcquireLockError::FileExists,
OpenWriteError::IOError(io_error) => TryAcquireLockError::IOError(io_error.into()),
})?;
write.flush().map_err(TryAcquireLockError::IOError)?;
Ok(DirectoryLock::from(Box::new(DirectoryLockGuard {
directory: directory.box_clone(),
path: filepath.to_owned(),
})))
}
fn retry_policy(is_blocking: bool) -> RetryPolicy {
if is_blocking {
RetryPolicy {
num_retries: 100,
wait_in_ms: 100,
}
} else {
RetryPolicy::no_retry()
}
}
/// Write-once read many (WORM) abstraction for where
/// tantivy's data should be stored.
@@ -73,19 +166,58 @@ pub trait Directory: DirectoryClone + fmt::Debug + Send + Sync + 'static {
///
/// The file may or may not previously exist.
fn atomic_write(&mut self, path: &Path, data: &[u8]) -> io::Result<()>;
/// Acquire a lock in the given directory.
///
/// The method is blocking or not depending on the `Lock` object.
fn acquire_lock(&self, lock: &Lock) -> Result<DirectoryLock, LockError> {
let mut box_directory = self.box_clone();
let mut retry_policy = retry_policy(lock.is_blocking);
loop {
match try_acquire_lock(&lock.filepath, &mut *box_directory) {
Ok(result) => {
return Ok(result);
}
Err(TryAcquireLockError::FileExists) => {
if !retry_policy.wait_and_retry() {
return Err(LockError::LockBusy);
}
}
Err(TryAcquireLockError::IOError(io_error)) => {
return Err(LockError::IOError(io_error));
}
}
}
}
/// Registers a callback that will be called whenever a change on the `meta.json`
/// using the `atomic_write` API is detected.
///
/// The behavior when using `.watch()` on a file using `.open_write(...)` is, on the other
/// hand, undefined.
///
/// The file will be watched for the lifetime of the returned `WatchHandle`. The caller is
/// required to keep it.
/// It does not override previous callbacks. When the file is modified, all callback that are
/// registered (and whose `WatchHandle` is still alive) are triggered.
///
/// Internally, tantivy only uses this API to detect new commits to implement the
/// `OnCommit` `ReloadPolicy`. Not implementing watch in a `Directory` only prevents the
/// `OnCommit` `ReloadPolicy` to work properly.
fn watch(&self, watch_callback: WatchCallback) -> WatchHandle;
}
/// DirectoryClone
pub trait DirectoryClone {
/// Clones the directory and boxes the clone
fn box_clone(&self) -> Box<Directory>;
/// Clones the directory and boxes the clone
fn box_clone(&self) -> Box<Directory>;
}
impl<T> DirectoryClone for T
where
T: 'static + Directory + Clone,
T: 'static + Directory + Clone,
{
fn box_clone(&self) -> Box<Directory> {
Box::new(self.clone())
}
fn box_clone(&self) -> Box<Directory> {
Box::new(self.clone())
}
}

56
src/directory/directory_lock.rs Normal file
View File

@@ -0,0 +1,56 @@
use std::path::PathBuf;
/// A directory lock.
///
/// A lock is associated to a specific path and some
/// [`LockParams`](./enum.LockParams.html).
/// Tantivy itself uses only two locks but client application
/// can use the directory facility to define their own locks.
/// - [INDEX_WRITER_LOCK](./struct.INDEX_WRITER_LOCK.html)
/// - [META_LOCK](./struct.META_LOCK.html)
///
/// Check out these locks documentation for more information.
///
#[derive(Debug)]
pub struct Lock {
/// The lock needs to be associated with its own file `path`.
/// Depending on the platform, the lock might rely on the creation
/// and deletion of this filepath.
pub filepath: PathBuf,
/// `lock_params` describes whether acquiring the lock is meant
/// to be a blocking operation or a non-blocking.
///
/// Acquiring a blocking lock blocks until the lock is
/// available.
/// Acquiring a blocking lock returns rapidly, either successfully
/// or with an error signifying that someone is already holding
/// the lock.
pub is_blocking: bool,
}
lazy_static! {
/// Only one process should be able to write tantivy's index at a time.
/// This lock file, when present, is in charge of preventing other processes to open an IndexWriter.
///
/// If the process is killed and this file remains, it is safe to remove it manually.
///
/// Failing to acquire this lock usually means a misuse of tantivy's API,
/// (creating more than one instance of the `IndexWriter`), are a spurious
/// lock file remaining after a crash. In the latter case, removing the file after
/// checking no process running tantivy is running is safe.
pub static ref INDEX_WRITER_LOCK: Lock = Lock {
filepath: PathBuf::from(".tantivy-writer.lock"),
is_blocking: false
};
/// The meta lock file is here to protect the segment files being opened by
/// `IndexReader::reload()` from being garbage collected.
/// It makes it possible for another process to safely consume
/// our index in-writing. Ideally, we may have prefered `RWLock` semantics
/// here, but it is difficult to achieve on Windows.
///
/// Opening segment readers is a very fast process.
pub static ref META_LOCK: Lock = Lock {
filepath: PathBuf::from(".tantivy-meta.lock"),
is_blocking: true
};
}

35
src/directory/error.rs
View File

@@ -3,6 +3,22 @@ use std::fmt;
use std::io;
use std::path::PathBuf;
/// Error while trying to acquire a directory lock.
#[derive(Debug, Fail)]
pub enum LockError {
/// Failed to acquired a lock as it is already hold by another
/// client.
/// - In the context of a blocking lock, this means the lock was not released within some `timeout` period.
/// - In the context of a non-blocking lock, this means the lock was busy at the moment of the call.
#[fail(
display = "Could not acquire lock as it is already held, possibly by a different process."
)]
LockBusy,
/// Trying to acquire a lock failed with an `IOError`
#[fail(display = "Failed to acquire the lock due to an io:Error.")]
IOError(io::Error),
}
/// General IO error with an optional path to the offending file.
#[derive(Debug)]
pub struct IOError {
@@ -10,6 +26,12 @@ pub struct IOError {
err: io::Error,
}
impl Into<io::Error> for IOError {
fn into(self) -> io::Error {
self.err
}
}
impl fmt::Display for IOError {
fn fmt(&self, f: &mut fmt::Formatter) -> fmt::Result {
match self.path {
@@ -51,6 +73,14 @@ pub enum OpenDirectoryError {
DoesNotExist(PathBuf),
/// The path exists but is not a directory.
NotADirectory(PathBuf),
/// IoError
IoError(io::Error),
}
impl From<io::Error> for OpenDirectoryError {
fn from(io_err: io::Error) -> Self {
OpenDirectoryError::IoError(io_err)
}
}
impl fmt::Display for OpenDirectoryError {
@@ -62,6 +92,11 @@ impl fmt::Display for OpenDirectoryError {
OpenDirectoryError::NotADirectory(ref path) => {
write!(f, "the path '{:?}' exists but is not a directory", path)
}
OpenDirectoryError::IoError(ref err) => write!(
f,
"IOError while trying to open/create the directory. {:?}",
err
),
}
}
}

254
src/directory/managed_directory.rs
View File

@@ -1,7 +1,11 @@
use core::MANAGED_FILEPATH;
use directory::error::{DeleteError, IOError, OpenReadError, OpenWriteError};
use directory::error::{DeleteError, IOError, LockError, OpenReadError, OpenWriteError};
use directory::DirectoryLock;
use directory::Lock;
use directory::META_LOCK;
use directory::{ReadOnlySource, WritePtr};
use error::TantivyError;
use directory::{WatchCallback, WatchHandle};
use error::DataCorruption;
use serde_json;
use std::collections::HashSet;
use std::io;
@@ -13,6 +17,17 @@ use std::sync::{Arc, RwLock};
use Directory;
use Result;
/// Returns true iff the file is "managed".
/// Non-managed file are not subject to garbage collection.
///
/// Filenames that starts by a "." -typically locks-
/// are not managed.
fn is_managed(path: &Path) -> bool {
path.to_str()
.map(|p_str| !p_str.starts_with('.'))
.unwrap_or(true)
}
/// Wrapper of directories that keeps track of files created by Tantivy.
///
/// A managed directory is just a wrapper of a directory
@@ -40,19 +55,24 @@ fn save_managed_paths(
wlock: &RwLockWriteGuard<MetaInformation>,
) -> io::Result<()> {
let mut w = serde_json::to_vec(&wlock.managed_paths)?;
write!(&mut w, "\n")?;
writeln!(&mut w)?;
directory.atomic_write(&MANAGED_FILEPATH, &w[..])?;
Ok(())
}
impl ManagedDirectory {
/// Wraps a directory as managed directory.
pub fn new<Dir: Directory>(directory: Dir) -> Result<ManagedDirectory> {
pub fn wrap<Dir: Directory>(directory: Dir) -> Result<ManagedDirectory> {
match directory.atomic_read(&MANAGED_FILEPATH) {
Ok(data) => {
let managed_files_json = String::from_utf8_lossy(&data);
let managed_files: HashSet<PathBuf> = serde_json::from_str(&managed_files_json)
.map_err(|_| TantivyError::CorruptedFile(MANAGED_FILEPATH.clone()))?;
.map_err(|e| {
DataCorruption::new(
MANAGED_FILEPATH.clone(),
format!("Managed file cannot be deserialized: {:?}. ", e),
)
})?;
Ok(ManagedDirectory {
directory: Box::new(directory),
meta_informations: Arc::new(RwLock::new(MetaInformation {
@@ -75,6 +95,9 @@ impl ManagedDirectory {
///
/// * `living_files` - List of files that are still used by the index.
///
/// The use a callback ensures that the list of living_files is computed
/// while we hold the lock on meta.
///
/// This method does not panick nor returns errors.
/// If a file cannot be deleted (for permission reasons for instance)
/// an error is simply logged, and the file remains in the list of managed
@@ -82,25 +105,35 @@ impl ManagedDirectory {
pub fn garbage_collect<L: FnOnce() -> HashSet<PathBuf>>(&mut self, get_living_files: L) {
info!("Garbage collect");
let mut files_to_delete = vec![];
// It is crucial to get the living files after acquiring the
// read lock of meta informations. That way, we
// avoid the following scenario.
//
// 1) we get the list of living files.
// 2) someone creates a new file.
// 3) we start garbage collection and remove this file
// even though it is a living file.
//
// releasing the lock as .delete() will use it too.
{
// releasing the lock as .delete() will use it too.
let meta_informations_rlock = self.meta_informations
let meta_informations_rlock = self
.meta_informations
.read()
.expect("Managed directory rlock poisoned in garbage collect.");
// It is crucial to get the living files after acquiring the
// read lock of meta informations. That way, we
// avoid the following scenario.
//
// 1) we get the list of living files.
// 2) someone creates a new file.
// 3) we start garbage collection and remove this file
// even though it is a living file.
let living_files = get_living_files();
for managed_path in &meta_informations_rlock.managed_paths {
if !living_files.contains(managed_path) {
files_to_delete.push(managed_path.clone());
// The point of this second "file" lock is to enforce the following scenario
// 1) process B tries to load a new set of searcher.
// The list of segments is loaded
// 2) writer change meta.json (for instance after a merge or a commit)
// 3) gc kicks in.
// 4) gc removes a file that was useful for process B, before process B opened it.
if let Ok(_meta_lock) = self.acquire_lock(&META_LOCK) {
let living_files = get_living_files();
for managed_path in &meta_informations_rlock.managed_paths {
if !living_files.contains(managed_path) {
files_to_delete.push(managed_path.clone());
}
}
}
}
@@ -134,7 +167,8 @@ impl ManagedDirectory {
if !deleted_files.is_empty() {
// update the list of managed files by removing
// the file that were removed.
let mut meta_informations_wlock = self.meta_informations
let mut meta_informations_wlock = self
.meta_informations
.write()
.expect("Managed directory wlock poisoned (2).");
{
@@ -156,8 +190,17 @@ impl ManagedDirectory {
/// registering the filepath and creating the file
/// will not lead to garbage files that will
/// never get removed.
///
/// File starting by "." are reserved to locks.
/// They are not managed and cannot be subjected
/// to garbage collection.
fn register_file_as_managed(&mut self, filepath: &Path) -> io::Result<()> {
let mut meta_wlock = self.meta_informations
// Files starting by "." (e.g. lock files) are not managed.
if !is_managed(filepath) {
return Ok(());
}
let mut meta_wlock = self
.meta_informations
.write()
.expect("Managed file lock poisoned");
let has_changed = meta_wlock.managed_paths.insert(filepath.to_owned());
@@ -195,6 +238,14 @@ impl Directory for ManagedDirectory {
fn exists(&self, path: &Path) -> bool {
self.directory.exists(path)
}
fn acquire_lock(&self, lock: &Lock) -> result::Result<DirectoryLock, LockError> {
self.directory.acquire_lock(lock)
}
fn watch(&self, watch_callback: WatchCallback) -> WatchHandle {
self.directory.watch(watch_callback)
}
}
impl Clone for ManagedDirectory {
@@ -209,95 +260,98 @@ impl Clone for ManagedDirectory {
#[cfg(test)]
mod tests {
use super::*;
#[cfg(feature = "mmap")]
use directory::MmapDirectory;
use std::io::Write;
use std::path::Path;
use tempdir::TempDir;
mod mmap_specific {
lazy_static! {
static ref TEST_PATH1: &'static Path = Path::new("some_path_for_test");
static ref TEST_PATH2: &'static Path = Path::new("some_path_for_test2");
}
use super::super::*;
use std::path::Path;
use tempdir::TempDir;
#[test]
#[cfg(feature = "mmap")]
fn test_managed_directory() {
let tempdir = TempDir::new("index").unwrap();
let tempdir_path = PathBuf::from(tempdir.path());
{
let mmap_directory = MmapDirectory::open(&tempdir_path).unwrap();
let mut managed_directory = ManagedDirectory::new(mmap_directory).unwrap();
lazy_static! {
static ref TEST_PATH1: &'static Path = Path::new("some_path_for_test");
static ref TEST_PATH2: &'static Path = Path::new("some_path_for_test2");
}
use directory::MmapDirectory;
use std::io::Write;
#[test]
fn test_managed_directory() {
let tempdir = TempDir::new("index").unwrap();
let tempdir_path = PathBuf::from(tempdir.path());
{
let mut write_file = managed_directory.open_write(*TEST_PATH1).unwrap();
write_file.flush().unwrap();
let mmap_directory = MmapDirectory::open(&tempdir_path).unwrap();
let mut managed_directory = ManagedDirectory::wrap(mmap_directory).unwrap();
{
let mut write_file = managed_directory.open_write(*TEST_PATH1).unwrap();
write_file.flush().unwrap();
}
{
managed_directory
.atomic_write(*TEST_PATH2, &vec![0u8, 1u8])
.unwrap();
}
{
assert!(managed_directory.exists(*TEST_PATH1));
assert!(managed_directory.exists(*TEST_PATH2));
}
{
let living_files: HashSet<PathBuf> =
[TEST_PATH1.to_owned()].into_iter().cloned().collect();
managed_directory.garbage_collect(|| living_files);
}
{
assert!(managed_directory.exists(*TEST_PATH1));
assert!(!managed_directory.exists(*TEST_PATH2));
}
}
{
managed_directory
.atomic_write(*TEST_PATH2, &vec![0u8, 1u8])
.unwrap();
}
{
assert!(managed_directory.exists(*TEST_PATH1));
assert!(managed_directory.exists(*TEST_PATH2));
}
{
let living_files: HashSet<PathBuf> =
[TEST_PATH1.to_owned()].into_iter().cloned().collect();
managed_directory.garbage_collect(|| living_files);
}
{
assert!(managed_directory.exists(*TEST_PATH1));
assert!(!managed_directory.exists(*TEST_PATH2));
let mmap_directory = MmapDirectory::open(&tempdir_path).unwrap();
let mut managed_directory = ManagedDirectory::wrap(mmap_directory).unwrap();
{
assert!(managed_directory.exists(*TEST_PATH1));
assert!(!managed_directory.exists(*TEST_PATH2));
}
{
let living_files: HashSet<PathBuf> = HashSet::new();
managed_directory.garbage_collect(|| living_files);
}
{
assert!(!managed_directory.exists(*TEST_PATH1));
assert!(!managed_directory.exists(*TEST_PATH2));
}
}
}
{
#[test]
fn test_managed_directory_gc_while_mmapped() {
let tempdir = TempDir::new("index").unwrap();
let tempdir_path = PathBuf::from(tempdir.path());
let living_files = HashSet::new();
let mmap_directory = MmapDirectory::open(&tempdir_path).unwrap();
let mut managed_directory = ManagedDirectory::new(mmap_directory).unwrap();
{
assert!(managed_directory.exists(*TEST_PATH1));
assert!(!managed_directory.exists(*TEST_PATH2));
}
{
let living_files: HashSet<PathBuf> = HashSet::new();
managed_directory.garbage_collect(|| living_files);
}
{
assert!(!managed_directory.exists(*TEST_PATH1));
assert!(!managed_directory.exists(*TEST_PATH2));
}
}
}
#[test]
#[cfg(feature = "mmap ")]
fn test_managed_directory_gc_while_mmapped() {
let tempdir = TempDir::new("index").unwrap();
let tempdir_path = PathBuf::from(tempdir.path());
let living_files = HashSet::new();
let mmap_directory = MmapDirectory::open(&tempdir_path).unwrap();
let mut managed_directory = ManagedDirectory::new(mmap_directory).unwrap();
managed_directory
.atomic_write(*TEST_PATH1, &vec![0u8, 1u8])
.unwrap();
assert!(managed_directory.exists(*TEST_PATH1));
let _mmap_read = managed_directory.open_read(*TEST_PATH1).unwrap();
managed_directory.garbage_collect(|| living_files.clone());
if cfg!(target_os = "windows") {
// On Windows, gc should try and fail the file as it is mmapped.
let mut managed_directory = ManagedDirectory::wrap(mmap_directory).unwrap();
managed_directory
.atomic_write(*TEST_PATH1, &vec![0u8, 1u8])
.unwrap();
assert!(managed_directory.exists(*TEST_PATH1));
// unmap should happen here.
drop(_mmap_read);
// The file should still be in the list of managed file and
// eventually be deleted once mmap is released.
managed_directory.garbage_collect(|| living_files);
assert!(!managed_directory.exists(*TEST_PATH1));
} else {
assert!(!managed_directory.exists(*TEST_PATH1));
let _mmap_read = managed_directory.open_read(*TEST_PATH1).unwrap();
managed_directory.garbage_collect(|| living_files.clone());
if cfg!(target_os = "windows") {
// On Windows, gc should try and fail the file as it is mmapped.
assert!(managed_directory.exists(*TEST_PATH1));
// unmap should happen here.
drop(_mmap_read);
// The file should still be in the list of managed file and
// eventually be deleted once mmap is released.
managed_directory.garbage_collect(|| living_files);
assert!(!managed_directory.exists(*TEST_PATH1));
} else {
assert!(!managed_directory.exists(*TEST_PATH1));
}
}
}
}

470
src/directory/mmap_directory.rs
View File

@@ -1,12 +1,24 @@
extern crate fs2;
extern crate notify;
use self::fs2::FileExt;
use self::notify::RawEvent;
use self::notify::RecursiveMode;
use self::notify::Watcher;
use atomicwrites;
use common::make_io_err;
use core::META_FILEPATH;
use directory::error::LockError;
use directory::error::{DeleteError, IOError, OpenDirectoryError, OpenReadError, OpenWriteError};
use directory::shared_vec_slice::SharedVecSlice;
use directory::read_only_source::BoxedData;
use directory::Directory;
use directory::DirectoryLock;
use directory::Lock;
use directory::ReadOnlySource;
use directory::WatchCallback;
use directory::WatchCallbackList;
use directory::WatchHandle;
use directory::WritePtr;
use fst::raw::MmapReadOnly;
use std::collections::hash_map::Entry as HashMapEntry;
use memmap::Mmap;
use std::collections::HashMap;
use std::convert::From;
use std::fmt;
@@ -16,14 +28,22 @@ use std::io::{self, Seek, SeekFrom};
use std::io::{BufWriter, Read, Write};
use std::path::{Path, PathBuf};
use std::result;
use std::sync::mpsc::{channel, Receiver, Sender};
use std::sync::Arc;
use std::sync::Mutex;
use std::sync::RwLock;
use std::sync::Weak;
use std::thread;
use tempdir::TempDir;
/// Create a default io error given a string.
pub(crate) fn make_io_err(msg: String) -> io::Error {
io::Error::new(io::ErrorKind::Other, msg)
}
/// Returns None iff the file exists, can be read, but is empty (and hence
/// cannot be mmapped).
///
fn open_mmap(full_path: &Path) -> result::Result<Option<MmapReadOnly>, OpenReadError> {
/// cannot be mmapped)
fn open_mmap(full_path: &Path) -> result::Result<Option<Mmap>, OpenReadError> {
let file = File::open(full_path).map_err(|e| {
if e.kind() == io::ErrorKind::NotFound {
OpenReadError::FileDoesNotExist(full_path.to_owned())
@@ -32,7 +52,8 @@ fn open_mmap(full_path: &Path) -> result::Result<Option<MmapReadOnly>, OpenReadE
}
})?;
let meta_data = file.metadata()
let meta_data = file
.metadata()
.map_err(|e| IOError::with_path(full_path.to_owned(), e))?;
if meta_data.len() == 0 {
// if the file size is 0, it will not be possible
@@ -41,7 +62,7 @@ fn open_mmap(full_path: &Path) -> result::Result<Option<MmapReadOnly>, OpenReadE
return Ok(None);
}
unsafe {
MmapReadOnly::open(&file)
memmap::Mmap::map(&file)
.map(Some)
.map_err(|e| From::from(IOError::with_path(full_path.to_owned(), e)))
}
@@ -64,7 +85,7 @@ pub struct CacheInfo {
struct MmapCache {
counters: CacheCounters,
cache: HashMap<PathBuf, MmapReadOnly>,
cache: HashMap<PathBuf, Weak<BoxedData>>,
}
impl Default for MmapCache {
@@ -77,12 +98,7 @@ impl Default for MmapCache {
}
impl MmapCache {
/// Removes a `MmapReadOnly` entry from the mmap cache.
fn discard_from_cache(&mut self, full_path: &Path) -> bool {
self.cache.remove(full_path).is_some()
}
fn get_info(&mut self) -> CacheInfo {
fn get_info(&self) -> CacheInfo {
let paths: Vec<PathBuf> = self.cache.keys().cloned().collect();
CacheInfo {
counters: self.counters.clone(),
@@ -90,57 +106,178 @@ impl MmapCache {
}
}
fn get_mmap(&mut self, full_path: &Path) -> Result<Option<MmapReadOnly>, OpenReadError> {
Ok(match self.cache.entry(full_path.to_owned()) {
HashMapEntry::Occupied(occupied_entry) => {
let mmap = occupied_entry.get();
fn remove_weak_ref(&mut self) {
let keys_to_remove: Vec<PathBuf> = self
.cache
.iter()
.filter(|(_, mmap_weakref)| mmap_weakref.upgrade().is_none())
.map(|(key, _)| key.clone())
.collect();
for key in keys_to_remove {
self.cache.remove(&key);
}
}
// Returns None if the file exists but as a len of 0 (and hence is not mmappable).
fn get_mmap(&mut self, full_path: &Path) -> Result<Option<Arc<BoxedData>>, OpenReadError> {
if let Some(mmap_weak) = self.cache.get(full_path) {
if let Some(mmap_arc) = mmap_weak.upgrade() {
self.counters.hit += 1;
Some(mmap.clone())
}
HashMapEntry::Vacant(vacant_entry) => {
self.counters.miss += 1;
if let Some(mmap) = open_mmap(full_path)? {
vacant_entry.insert(mmap.clone());
Some(mmap)
} else {
None
}
return Ok(Some(mmap_arc));
}
}
self.cache.remove(full_path);
self.counters.miss += 1;
Ok(if let Some(mmap) = open_mmap(full_path)? {
let mmap_arc: Arc<BoxedData> = Arc::new(Box::new(mmap));
let mmap_weak = Arc::downgrade(&mmap_arc);
self.cache.insert(full_path.to_owned(), mmap_weak);
Some(mmap_arc)
} else {
None
})
}
}
struct InnerWatcherWrapper {
_watcher: Mutex<notify::RecommendedWatcher>,
watcher_router: WatchCallbackList,
}
impl InnerWatcherWrapper {
pub fn new(path: &Path) -> Result<(Self, Receiver<notify::RawEvent>), notify::Error> {
let (tx, watcher_recv): (Sender<RawEvent>, Receiver<RawEvent>) = channel();
// We need to initialize the
let mut watcher = notify::raw_watcher(tx)?;
watcher.watch(path, RecursiveMode::Recursive)?;
let inner = InnerWatcherWrapper {
_watcher: Mutex::new(watcher),
watcher_router: Default::default(),
};
Ok((inner, watcher_recv))
}
}
#[derive(Clone)]
pub(crate) struct WatcherWrapper {
inner: Arc<InnerWatcherWrapper>,
}
impl WatcherWrapper {
pub fn new(path: &Path) -> Result<Self, OpenDirectoryError> {
let (inner, watcher_recv) = InnerWatcherWrapper::new(path).map_err(|err| match err {
notify::Error::PathNotFound => OpenDirectoryError::DoesNotExist(path.to_owned()),
_ => {
panic!("Unknown error while starting watching directory {:?}", path);
}
})?;
let watcher_wrapper = WatcherWrapper {
inner: Arc::new(inner),
};
let watcher_wrapper_clone = watcher_wrapper.clone();
thread::Builder::new()
.name("meta-file-watch-thread".to_string())
.spawn(move || {
loop {
match watcher_recv.recv().map(|evt| evt.path) {
Ok(Some(changed_path)) => {
// ... Actually subject to false positive.
// We might want to be more accurate than this at one point.
if let Some(filename) = changed_path.file_name() {
if filename == *META_FILEPATH {
watcher_wrapper_clone.inner.watcher_router.broadcast();
}
}
}
Ok(None) => {
// not an event we are interested in.
}
Err(_e) => {
// the watch send channel was dropped
break;
}
}
}
})
.expect("Failed to spawn thread to watch meta.json");
Ok(watcher_wrapper)
}
pub fn watch(&mut self, watch_callback: WatchCallback) -> WatchHandle {
self.inner.watcher_router.subscribe(watch_callback)
}
}
/// Directory storing data in files, read via mmap.
///
/// The Mmap object are cached to limit the
/// system calls.
///
/// In the `MmapDirectory`, locks are implemented using the `fs2` crate definition of locks.
///
/// On MacOS & linux, it relies on `flock` (aka `BSD Lock`). These locks solve most of the
/// problems related to POSIX Locks, but may their contract may not be respected on `NFS`
/// depending on the implementation.
///
/// On Windows the semantics are again different.
#[derive(Clone)]
pub struct MmapDirectory {
inner: Arc<MmapDirectoryInner>,
}
struct MmapDirectoryInner {
root_path: PathBuf,
mmap_cache: Arc<RwLock<MmapCache>>,
_temp_directory: Arc<Option<TempDir>>,
mmap_cache: RwLock<MmapCache>,
_temp_directory: Option<TempDir>,
watcher: RwLock<WatcherWrapper>,
}
impl MmapDirectoryInner {
fn new(
root_path: PathBuf,
temp_directory: Option<TempDir>,
) -> Result<MmapDirectoryInner, OpenDirectoryError> {
let watch_wrapper = WatcherWrapper::new(&root_path)?;
let mmap_directory_inner = MmapDirectoryInner {
root_path,
mmap_cache: Default::default(),
_temp_directory: temp_directory,
watcher: RwLock::new(watch_wrapper),
};
Ok(mmap_directory_inner)
}
fn watch(&self, watch_callback: WatchCallback) -> WatchHandle {
let mut wlock = self.watcher.write().unwrap();
wlock.watch(watch_callback)
}
}
impl fmt::Debug for MmapDirectory {
fn fmt(&self, f: &mut fmt::Formatter) -> fmt::Result {
write!(f, "MmapDirectory({:?})", self.root_path)
write!(f, "MmapDirectory({:?})", self.inner.root_path)
}
}
impl MmapDirectory {
fn new(
root_path: PathBuf,
temp_directory: Option<TempDir>,
) -> Result<MmapDirectory, OpenDirectoryError> {
let inner = MmapDirectoryInner::new(root_path, temp_directory)?;
Ok(MmapDirectory {
inner: Arc::new(inner),
})
}
/// Creates a new MmapDirectory in a temporary directory.
///
/// This is mostly useful to test the MmapDirectory itself.
/// For your unit tests, prefer the RAMDirectory.
pub fn create_from_tempdir() -> io::Result<MmapDirectory> {
let tempdir = TempDir::new("index")?;
pub fn create_from_tempdir() -> Result<MmapDirectory, OpenDirectoryError> {
let tempdir = TempDir::new("index").map_err(OpenDirectoryError::IoError)?;
let tempdir_path = PathBuf::from(tempdir.path());
let directory = MmapDirectory {
root_path: tempdir_path,
mmap_cache: Arc::new(RwLock::new(MmapCache::default())),
_temp_directory: Arc::new(Some(tempdir)),
};
Ok(directory)
MmapDirectory::new(tempdir_path, Some(tempdir))
}
/// Opens a MmapDirectory in a directory.
@@ -158,18 +295,14 @@ impl MmapDirectory {
directory_path,
)))
} else {
Ok(MmapDirectory {
root_path: PathBuf::from(directory_path),
mmap_cache: Arc::new(RwLock::new(MmapCache::default())),
_temp_directory: Arc::new(None),
})
Ok(MmapDirectory::new(PathBuf::from(directory_path), None)?)
}
}
/// Joins a relative_path to the directory `root_path`
/// to create a proper complete `filepath`.
fn resolve_path(&self, relative_path: &Path) -> PathBuf {
self.root_path.join(relative_path)
self.inner.root_path.join(relative_path)
}
/// Sync the root directory.
@@ -194,7 +327,7 @@ impl MmapDirectory {
.custom_flags(winbase::FILE_FLAG_BACKUP_SEMANTICS);
}
let fd = open_opts.open(&self.root_path)?;
let fd = open_opts.open(&self.inner.root_path)?;
fd.sync_all()?;
Ok(())
}
@@ -204,14 +337,35 @@ impl MmapDirectory {
///
/// The `MmapDirectory` embeds a `MmapDirectory`
/// to avoid multiplying the `mmap` system calls.
pub fn get_cache_info(&mut self) -> CacheInfo {
self.mmap_cache
pub fn get_cache_info(&self) -> CacheInfo {
self.inner
.mmap_cache
.write()
.expect("mmap cache lock is poisoned")
.remove_weak_ref();
self.inner
.mmap_cache
.read()
.expect("Mmap cache lock is poisoned.")
.get_info()
}
}
/// We rely on fs2 for file locking. On Windows & MacOS this
/// uses BSD locks (`flock`). The lock is actually released when
/// the `File` object is dropped and its associated file descriptor
/// is closed.
struct ReleaseLockFile {
_file: File,
path: PathBuf,
}
impl Drop for ReleaseLockFile {
fn drop(&mut self) {
debug!("Releasing lock {:?}", self.path);
}
}
/// This Write wraps a File, but has the specificity of
/// call `sync_all` on flush.
struct SafeFileWriter(File);
@@ -244,7 +398,7 @@ impl Directory for MmapDirectory {
debug!("Open Read {:?}", path);
let full_path = self.resolve_path(path);
let mut mmap_cache = self.mmap_cache.write().map_err(|_| {
let mut mmap_cache = self.inner.mmap_cache.write().map_err(|_| {
let msg = format!(
"Failed to acquired write lock \
on mmap cache while reading {:?}",
@@ -252,11 +406,34 @@ impl Directory for MmapDirectory {
);
IOError::with_path(path.to_owned(), make_io_err(msg))
})?;
Ok(mmap_cache
.get_mmap(&full_path)?
.map(ReadOnlySource::Mmap)
.unwrap_or_else(|| ReadOnlySource::Anonymous(SharedVecSlice::empty())))
.map(ReadOnlySource::from)
.unwrap_or_else(ReadOnlySource::empty))
}
/// Any entry associated to the path in the mmap will be
/// removed before the file is deleted.
fn delete(&self, path: &Path) -> result::Result<(), DeleteError> {
debug!("Deleting file {:?}", path);
let full_path = self.resolve_path(path);
match fs::remove_file(&full_path) {
Ok(_) => self
.sync_directory()
.map_err(|e| IOError::with_path(path.to_owned(), e).into()),
Err(e) => {
if e.kind() == io::ErrorKind::NotFound {
Err(DeleteError::FileDoesNotExist(path.to_owned()))
} else {
Err(IOError::with_path(path.to_owned(), e).into())
}
}
}
}
fn exists(&self, path: &Path) -> bool {
let full_path = self.resolve_path(path);
full_path.exists()
}
fn open_write(&mut self, path: &Path) -> Result<WritePtr, OpenWriteError> {
@@ -289,43 +466,6 @@ impl Directory for MmapDirectory {
Ok(BufWriter::new(Box::new(writer)))
}
/// Any entry associated to the path in the mmap will be
/// removed before the file is deleted.
fn delete(&self, path: &Path) -> result::Result<(), DeleteError> {
debug!("Deleting file {:?}", path);
let full_path = self.resolve_path(path);
let mut mmap_cache = self.mmap_cache.write().map_err(|_| {
let msg = format!(
"Failed to acquired write lock \
on mmap cache while deleting {:?}",
path
);
IOError::with_path(path.to_owned(), make_io_err(msg))
})?;
mmap_cache.discard_from_cache(path);
// Removing the entry in the MMap cache.
// The munmap will appear on Drop,
// when the last reference is gone.
mmap_cache.cache.remove(&full_path);
match fs::remove_file(&full_path) {
Ok(_) => self.sync_directory()
.map_err(|e| IOError::with_path(path.to_owned(), e).into()),
Err(e) => {
if e.kind() == io::ErrorKind::NotFound {
Err(DeleteError::FileDoesNotExist(path.to_owned()))
} else {
Err(IOError::with_path(path.to_owned(), e).into())
}
}
}
}
fn exists(&self, path: &Path) -> bool {
let full_path = self.resolve_path(path);
full_path.exists()
}
fn atomic_read(&self, path: &Path) -> Result<Vec<u8>, OpenReadError> {
let full_path = self.resolve_path(path);
let mut buffer = Vec::new();
@@ -352,6 +492,30 @@ impl Directory for MmapDirectory {
meta_file.write(|f| f.write_all(data))?;
Ok(())
}
fn acquire_lock(&self, lock: &Lock) -> Result<DirectoryLock, LockError> {
let full_path = self.resolve_path(&lock.filepath);
// We make sure that the file exists.
let file: File = OpenOptions::new()
.write(true)
.create(true) //< if the file does not exist yet, create it.
.open(&full_path)
.map_err(LockError::IOError)?;
if lock.is_blocking {
file.lock_exclusive().map_err(LockError::IOError)?;
} else {
file.try_lock_exclusive().map_err(|_| LockError::LockBusy)?
}
// dropping the file handle will release the lock.
Ok(DirectoryLock::from(Box::new(ReleaseLockFile {
path: lock.filepath.clone(),
_file: file,
})))
}
fn watch(&self, watch_callback: WatchCallback) -> WatchHandle {
self.inner.watch(watch_callback)
}
}
#[cfg(test)]
@@ -361,6 +525,18 @@ mod tests {
// The following tests are specific to the MmapDirectory
use super::*;
use schema::{Schema, SchemaBuilder, TEXT};
use std::fs;
use std::sync::atomic::{AtomicUsize, Ordering};
use std::thread;
use std::time::Duration;
use Index;
use ReloadPolicy;
#[test]
fn test_open_non_existant_path() {
assert!(MmapDirectory::open(PathBuf::from("./nowhere")).is_err());
}
#[test]
fn test_open_empty() {
@@ -380,7 +556,7 @@ mod tests {
#[test]
fn test_cache() {
let content = "abc".as_bytes();
let content = b"abc";
// here we test if the cache releases
// mmaps correctly.
@@ -396,26 +572,104 @@ mod tests {
w.flush().unwrap();
}
}
{
for (i, path) in paths.iter().enumerate() {
let _r = mmap_directory.open_read(path).unwrap();
assert_eq!(mmap_directory.get_cache_info().mmapped.len(), i + 1);
}
for path in paths.iter() {
let _r = mmap_directory.open_read(path).unwrap();
assert_eq!(mmap_directory.get_cache_info().mmapped.len(), num_paths);
}
for (i, path) in paths.iter().enumerate() {
mmap_directory.delete(path).unwrap();
assert_eq!(
mmap_directory.get_cache_info().mmapped.len(),
num_paths - i - 1
);
}
let mut keep = vec![];
for (i, path) in paths.iter().enumerate() {
keep.push(mmap_directory.open_read(path).unwrap());
assert_eq!(mmap_directory.get_cache_info().mmapped.len(), i + 1);
}
assert_eq!(mmap_directory.get_cache_info().counters.hit, 0);
assert_eq!(mmap_directory.get_cache_info().counters.miss, 10);
assert_eq!(mmap_directory.get_cache_info().mmapped.len(), 10);
for path in paths.iter() {
let _r = mmap_directory.open_read(path).unwrap();
assert_eq!(mmap_directory.get_cache_info().mmapped.len(), num_paths);
}
assert_eq!(mmap_directory.get_cache_info().counters.hit, 10);
assert_eq!(mmap_directory.get_cache_info().counters.miss, 10);
assert_eq!(mmap_directory.get_cache_info().mmapped.len(), 10);
for path in paths.iter() {
let _r = mmap_directory.open_read(path).unwrap();
assert_eq!(mmap_directory.get_cache_info().mmapped.len(), 10);
}
assert_eq!(mmap_directory.get_cache_info().counters.hit, 20);
assert_eq!(mmap_directory.get_cache_info().counters.miss, 10);
assert_eq!(mmap_directory.get_cache_info().mmapped.len(), 10);
drop(keep);
for path in paths.iter() {
let _r = mmap_directory.open_read(path).unwrap();
assert_eq!(mmap_directory.get_cache_info().mmapped.len(), 1);
}
assert_eq!(mmap_directory.get_cache_info().counters.hit, 20);
assert_eq!(mmap_directory.get_cache_info().counters.miss, 20);
assert_eq!(mmap_directory.get_cache_info().mmapped.len(), 0);
for path in &paths {
mmap_directory.delete(path).unwrap();
}
assert_eq!(mmap_directory.get_cache_info().counters.hit, 20);
assert_eq!(mmap_directory.get_cache_info().counters.miss, 20);
assert_eq!(mmap_directory.get_cache_info().mmapped.len(), 0);
for path in paths.iter() {
assert!(mmap_directory.open_read(path).is_err());
}
assert_eq!(mmap_directory.get_cache_info().counters.hit, 20);
assert_eq!(mmap_directory.get_cache_info().counters.miss, 30);
assert_eq!(mmap_directory.get_cache_info().mmapped.len(), 0);
}
#[test]
fn test_watch_wrapper() {
let counter: Arc<AtomicUsize> = Default::default();
let counter_clone = counter.clone();
let tmp_dir: TempDir = tempdir::TempDir::new("test_watch_wrapper").unwrap();
let tmp_dirpath = tmp_dir.path().to_owned();
let mut watch_wrapper = WatcherWrapper::new(&tmp_dirpath).unwrap();
let tmp_file = tmp_dirpath.join("coucou");
let _handle = watch_wrapper.watch(Box::new(move || {
counter_clone.fetch_add(1, Ordering::SeqCst);
}));
assert_eq!(counter.load(Ordering::SeqCst), 0);
fs::write(&tmp_file, b"whateverwilldo").unwrap();
thread::sleep(Duration::new(0, 1_000u32));
}
#[test]
fn test_mmap_released() {
let mmap_directory = MmapDirectory::create_from_tempdir().unwrap();
let mut schema_builder: SchemaBuilder = Schema::builder();
let text_field = schema_builder.add_text_field("text", TEXT);
let schema = schema_builder.build();
{
let index = Index::create(mmap_directory.clone(), schema).unwrap();
let mut index_writer = index.writer_with_num_threads(1, 3_000_000).unwrap();
for _num_commits in 0..16 {
for _ in 0..10 {
index_writer.add_document(doc!(text_field=>"abc"));
}
index_writer.commit().unwrap();
}
let reader = index
.reader_builder()
.reload_policy(ReloadPolicy::Manual)
.try_into()
.unwrap();
for _ in 0..30 {
index_writer.add_document(doc!(text_field=>"abc"));
index_writer.commit().unwrap();
reader.reload().unwrap();
}
index_writer.wait_merging_threads().unwrap();
reader.reload().unwrap();
let num_segments = reader.searcher().segment_readers().len();
assert_eq!(num_segments, 4);
assert_eq!(
num_segments * 7,
mmap_directory.get_cache_info().mmapped.len()
);
}
assert_eq!(mmap_directory.get_cache_info().mmapped.len(), 0);
}
}

136
src/directory/mod.rs
View File

@@ -8,19 +8,23 @@ WORM directory abstraction.
mod mmap_directory;
mod directory;
mod directory_lock;
mod managed_directory;
mod ram_directory;
mod read_only_source;
mod shared_vec_slice;
mod watch_event_router;
/// Errors specific to the directory module.
pub mod error;
use std::io::{BufWriter, Seek, Write};
pub use self::directory::DirectoryLock;
pub use self::directory::{Directory, DirectoryClone};
pub use self::directory_lock::{Lock, INDEX_WRITER_LOCK, META_LOCK};
pub use self::ram_directory::RAMDirectory;
pub use self::read_only_source::ReadOnlySource;
pub(crate) use self::watch_event_router::WatchCallbackList;
pub use self::watch_event_router::{WatchCallback, WatchHandle};
use std::io::{BufWriter, Seek, Write};
#[cfg(feature = "mmap")]
pub use self::mmap_directory::MmapDirectory;
@@ -38,128 +42,4 @@ impl<T: Seek + Write> SeekableWrite for T {}
pub type WritePtr = BufWriter<Box<SeekableWrite>>;
#[cfg(test)]
mod tests {
use super::*;
use std::io::{Seek, SeekFrom, Write};
use std::path::Path;
lazy_static! {
static ref TEST_PATH: &'static Path = Path::new("some_path_for_test");
}
#[test]
fn test_ram_directory() {
let mut ram_directory = RAMDirectory::create();
test_directory(&mut ram_directory);
}
#[test]
#[cfg(feature = "mmap")]
fn test_mmap_directory() {
let mut mmap_directory = MmapDirectory::create_from_tempdir().unwrap();
test_directory(&mut mmap_directory);
}
#[test]
#[should_panic]
fn ram_directory_panics_if_flush_forgotten() {
let mut ram_directory = RAMDirectory::create();
let mut write_file = ram_directory.open_write(*TEST_PATH).unwrap();
assert!(write_file.write_all(&[4]).is_ok());
}
fn test_simple(directory: &mut Directory) {
{
{
let mut write_file = directory.open_write(*TEST_PATH).unwrap();
assert!(directory.exists(*TEST_PATH));
write_file.write_all(&[4]).unwrap();
write_file.write_all(&[3]).unwrap();
write_file.write_all(&[7, 3, 5]).unwrap();
write_file.flush().unwrap();
}
let read_file = directory.open_read(*TEST_PATH).unwrap();
let data: &[u8] = &*read_file;
assert_eq!(data, &[4u8, 3u8, 7u8, 3u8, 5u8]);
}
assert!(directory.delete(*TEST_PATH).is_ok());
assert!(!directory.exists(*TEST_PATH));
}
fn test_seek(directory: &mut Directory) {
{
{
let mut write_file = directory.open_write(*TEST_PATH).unwrap();
write_file.write_all(&[4, 3, 7, 3, 5]).unwrap();
write_file.seek(SeekFrom::Start(0)).unwrap();
write_file.write_all(&[3, 1]).unwrap();
write_file.flush().unwrap();
}
let read_file = directory.open_read(*TEST_PATH).unwrap();
let data: &[u8] = &*read_file;
assert_eq!(data, &[3u8, 1u8, 7u8, 3u8, 5u8]);
}
assert!(directory.delete(*TEST_PATH).is_ok());
}
fn test_rewrite_forbidden(directory: &mut Directory) {
{
directory.open_write(*TEST_PATH).unwrap();
assert!(directory.exists(*TEST_PATH));
}
{
assert!(directory.open_write(*TEST_PATH).is_err());
}
assert!(directory.delete(*TEST_PATH).is_ok());
}
fn test_write_create_the_file(directory: &mut Directory) {
{
assert!(directory.open_read(*TEST_PATH).is_err());
let _w = directory.open_write(*TEST_PATH).unwrap();
assert!(directory.exists(*TEST_PATH));
assert!(directory.open_read(*TEST_PATH).is_ok());
assert!(directory.delete(*TEST_PATH).is_ok());
}
}
fn test_directory_delete(directory: &mut Directory) {
assert!(directory.open_read(*TEST_PATH).is_err());
let mut write_file = directory.open_write(*TEST_PATH).unwrap();
write_file.write_all(&[1, 2, 3, 4]).unwrap();
write_file.flush().unwrap();
{
let read_handle = directory.open_read(*TEST_PATH).unwrap();
{
assert_eq!(&*read_handle, &[1u8, 2u8, 3u8, 4u8]);
// Mapped files can't be deleted on Windows
if !cfg!(windows) {
assert!(directory.delete(*TEST_PATH).is_ok());
assert_eq!(&*read_handle, &[1u8, 2u8, 3u8, 4u8]);
}
assert!(directory.delete(Path::new("SomeOtherPath")).is_err());
}
}
if cfg!(windows) {
assert!(directory.delete(*TEST_PATH).is_ok());
}
assert!(directory.open_read(*TEST_PATH).is_err());
assert!(directory.delete(*TEST_PATH).is_err());
}
fn test_directory(directory: &mut Directory) {
test_simple(directory);
test_seek(directory);
test_rewrite_forbidden(directory);
test_write_create_the_file(directory);
test_directory_delete(directory);
}
}
mod tests;

147
src/directory/ram_directory.rs
View File

@@ -1,8 +1,8 @@
use super::shared_vec_slice::SharedVecSlice;
use common::make_io_err;
use directory::error::{DeleteError, IOError, OpenReadError, OpenWriteError};
use core::META_FILEPATH;
use directory::error::{DeleteError, OpenReadError, OpenWriteError};
use directory::WatchCallbackList;
use directory::WritePtr;
use directory::{Directory, ReadOnlySource};
use directory::{Directory, ReadOnlySource, WatchCallback, WatchHandle};
use std::collections::HashMap;
use std::fmt;
use std::io::{self, BufWriter, Cursor, Seek, SeekFrom, Write};
@@ -22,13 +22,13 @@ use std::sync::{Arc, RwLock};
///
struct VecWriter {
path: PathBuf,
shared_directory: InnerDirectory,
shared_directory: RAMDirectory,
data: Cursor<Vec<u8>>,
is_flushed: bool,
}
impl VecWriter {
fn new(path_buf: PathBuf, shared_directory: InnerDirectory) -> VecWriter {
fn new(path_buf: PathBuf, shared_directory: RAMDirectory) -> VecWriter {
VecWriter {
path: path_buf,
data: Cursor::new(Vec::new()),
@@ -64,75 +64,44 @@ impl Write for VecWriter {
fn flush(&mut self) -> io::Result<()> {
self.is_flushed = true;
self.shared_directory
.write(self.path.clone(), self.data.get_ref())?;
let mut fs = self.shared_directory.fs.write().unwrap();
fs.write(self.path.clone(), self.data.get_ref());
Ok(())
}
}
#[derive(Clone)]
struct InnerDirectory(Arc<RwLock<HashMap<PathBuf, Arc<Vec<u8>>>>>);
#[derive(Default)]
struct InnerDirectory {
fs: HashMap<PathBuf, ReadOnlySource>,
watch_router: WatchCallbackList,
}
impl InnerDirectory {
fn new() -> InnerDirectory {
InnerDirectory(Arc::new(RwLock::new(HashMap::new())))
}
fn write(&self, path: PathBuf, data: &[u8]) -> io::Result<bool> {
let mut map = self.0.write().map_err(|_| {
make_io_err(format!(
"Failed to lock the directory, when trying to write {:?}",
path
))
})?;
let prev_value = map.insert(path, Arc::new(Vec::from(data)));
Ok(prev_value.is_some())
fn write(&mut self, path: PathBuf, data: &[u8]) -> bool {
let data = ReadOnlySource::new(Vec::from(data));
self.fs.insert(path, data).is_some()
}
fn open_read(&self, path: &Path) -> Result<ReadOnlySource, OpenReadError> {
self.0
.read()
.map_err(|_| {
let msg = format!(
"Failed to acquire read lock for the \
directory when trying to read {:?}",
path
);
let io_err = make_io_err(msg);
OpenReadError::IOError(IOError::with_path(path.to_owned(), io_err))
})
.and_then(|readable_map| {
readable_map
.get(path)
.ok_or_else(|| OpenReadError::FileDoesNotExist(PathBuf::from(path)))
.map(Arc::clone)
.map(|data| ReadOnlySource::Anonymous(SharedVecSlice::new(data)))
})
self.fs
.get(path)
.ok_or_else(|| OpenReadError::FileDoesNotExist(PathBuf::from(path)))
.map(Clone::clone)
}
fn delete(&self, path: &Path) -> result::Result<(), DeleteError> {
self.0
.write()
.map_err(|_| {
let msg = format!(
"Failed to acquire write lock for the \
directory when trying to delete {:?}",
path
);
let io_err = make_io_err(msg);
DeleteError::IOError(IOError::with_path(path.to_owned(), io_err))
})
.and_then(|mut writable_map| match writable_map.remove(path) {
Some(_) => Ok(()),
None => Err(DeleteError::FileDoesNotExist(PathBuf::from(path))),
})
fn delete(&mut self, path: &Path) -> result::Result<(), DeleteError> {
match self.fs.remove(path) {
Some(_) => Ok(()),
None => Err(DeleteError::FileDoesNotExist(PathBuf::from(path))),
}
}
fn exists(&self, path: &Path) -> bool {
self.0
.read()
.expect("Failed to get read lock directory.")
.contains_key(path)
self.fs.contains_key(path)
}
fn watch(&mut self, watch_handle: WatchCallback) -> WatchHandle {
self.watch_router.subscribe(watch_handle)
}
}
@@ -147,33 +116,36 @@ impl fmt::Debug for RAMDirectory {
/// It is mainly meant for unit testing.
/// Writes are only made visible upon flushing.
///
#[derive(Clone)]
#[derive(Clone, Default)]
pub struct RAMDirectory {
fs: InnerDirectory,
fs: Arc<RwLock<InnerDirectory>>,
}
impl RAMDirectory {
/// Constructor
pub fn create() -> RAMDirectory {
RAMDirectory {
fs: InnerDirectory::new(),
}
Self::default()
}
}
impl Directory for RAMDirectory {
fn open_read(&self, path: &Path) -> result::Result<ReadOnlySource, OpenReadError> {
self.fs.open_read(path)
self.fs.read().unwrap().open_read(path)
}
fn delete(&self, path: &Path) -> result::Result<(), DeleteError> {
self.fs.write().unwrap().delete(path)
}
fn exists(&self, path: &Path) -> bool {
self.fs.read().unwrap().exists(path)
}
fn open_write(&mut self, path: &Path) -> Result<WritePtr, OpenWriteError> {
let mut fs = self.fs.write().unwrap();
let path_buf = PathBuf::from(path);
let vec_writer = VecWriter::new(path_buf.clone(), self.fs.clone());
let exists = self.fs
.write(path_buf.clone(), &Vec::new())
.map_err(|err| IOError::with_path(path.to_owned(), err))?;
let vec_writer = VecWriter::new(path_buf.clone(), self.clone());
let exists = fs.write(path_buf.clone(), &[]);
// force the creation of the file to mimic the MMap directory.
if exists {
Err(OpenWriteError::FileAlreadyExists(path_buf))
@@ -182,25 +154,30 @@ impl Directory for RAMDirectory {
}
}
fn delete(&self, path: &Path) -> result::Result<(), DeleteError> {
self.fs.delete(path)
}
fn exists(&self, path: &Path) -> bool {
self.fs.exists(path)
}
fn atomic_read(&self, path: &Path) -> Result<Vec<u8>, OpenReadError> {
let read = self.open_read(path)?;
Ok(read.as_slice().to_owned())
Ok(self.open_read(path)?.as_slice().to_owned())
}
fn atomic_write(&mut self, path: &Path, data: &[u8]) -> io::Result<()> {
fail_point!("RAMDirectory::atomic_write", |msg| Err(io::Error::new(
io::ErrorKind::Other,
msg.unwrap_or("Undefined".to_string())
)));
let path_buf = PathBuf::from(path);
let mut vec_writer = VecWriter::new(path_buf.clone(), self.fs.clone());
self.fs.write(path_buf, &Vec::new())?;
// Reserve the path to prevent calls to .write() to succeed.
self.fs.write().unwrap().write(path_buf.clone(), &[]);
let mut vec_writer = VecWriter::new(path_buf.clone(), self.clone());
vec_writer.write_all(data)?;
vec_writer.flush()?;
if path == Path::new(&*META_FILEPATH) {
self.fs.write().unwrap().watch_router.broadcast();
}
Ok(())
}
fn watch(&self, watch_callback: WatchCallback) -> WatchHandle {
self.fs.write().unwrap().watch(watch_callback)
}
}

78
src/directory/read_only_source.rs
View File

@@ -1,10 +1,9 @@
use super::shared_vec_slice::SharedVecSlice;
use common::HasLen;
#[cfg(feature = "mmap")]
use fst::raw::MmapReadOnly;
use stable_deref_trait::{CloneStableDeref, StableDeref};
use std::ops::Deref;
use std::sync::Arc;
pub type BoxedData = Box<Deref<Target = [u8]> + Send + Sync + 'static>;
/// Read object that represents files in tantivy.
///
@@ -12,12 +11,10 @@ use std::ops::Deref;
/// the data in the form of a constant read-only `&[u8]`.
/// Whatever happens to the directory file, the data
/// hold by this object should never be altered or destroyed.
pub enum ReadOnlySource {
/// Mmap source of data
#[cfg(feature = "mmap")]
Mmap(MmapReadOnly),
/// Wrapping a `Vec<u8>`
Anonymous(SharedVecSlice),
pub struct ReadOnlySource {
data: Arc<BoxedData>,
start: usize,
stop: usize,
}
unsafe impl StableDeref for ReadOnlySource {}
@@ -31,19 +28,38 @@ impl Deref for ReadOnlySource {
}
}
impl From<Arc<BoxedData>> for ReadOnlySource {
fn from(data: Arc<BoxedData>) -> Self {
let len = data.len();
ReadOnlySource {
data,
start: 0,
stop: len,
}
}
}
impl ReadOnlySource {
pub(crate) fn new<D>(data: D) -> ReadOnlySource
where
D: Deref<Target = [u8]> + Send + Sync + 'static,
{
let len = data.len();
ReadOnlySource {
data: Arc::new(Box::new(data)),
start: 0,
stop: len,
}
}
/// Creates an empty ReadOnlySource
pub fn empty() -> ReadOnlySource {
ReadOnlySource::Anonymous(SharedVecSlice::empty())
ReadOnlySource::new(&[][..])
}
/// Returns the data underlying the ReadOnlySource object.
pub fn as_slice(&self) -> &[u8] {
match *self {
#[cfg(feature = "mmap")]
ReadOnlySource::Mmap(ref mmap_read_only) => mmap_read_only.as_slice(),
ReadOnlySource::Anonymous(ref shared_vec) => shared_vec.as_slice(),
}
&self.data[self.start..self.stop]
}
/// Splits into 2 `ReadOnlySource`, at the offset given
@@ -64,22 +80,18 @@ impl ReadOnlySource {
/// worth of data in anonymous memory, and only a
/// 1KB slice is remaining, the whole `500MBs`
/// are retained in memory.
pub fn slice(&self, from_offset: usize, to_offset: usize) -> ReadOnlySource {
pub fn slice(&self, start: usize, stop: usize) -> ReadOnlySource {
assert!(
from_offset <= to_offset,
start <= stop,
"Requested negative slice [{}..{}]",
from_offset,
to_offset
start,
stop
);
match *self {
#[cfg(feature = "mmap")]
ReadOnlySource::Mmap(ref mmap_read_only) => {
let sliced_mmap = mmap_read_only.range(from_offset, to_offset - from_offset);
ReadOnlySource::Mmap(sliced_mmap)
}
ReadOnlySource::Anonymous(ref shared_vec) => {
ReadOnlySource::Anonymous(shared_vec.slice(from_offset, to_offset))
}
assert!(stop <= self.len());
ReadOnlySource {
data: self.data.clone(),
start: self.start + start,
stop: self.start + stop,
}
}
@@ -88,8 +100,7 @@ impl ReadOnlySource {
///
/// Equivalent to `.slice(from_offset, self.len())`
pub fn slice_from(&self, from_offset: usize) -> ReadOnlySource {
let len = self.len();
self.slice(from_offset, len)
self.slice(from_offset, self.len())
}
/// Like `.slice(...)` but enforcing only the `to`
@@ -103,19 +114,18 @@ impl ReadOnlySource {
impl HasLen for ReadOnlySource {
fn len(&self) -> usize {
self.as_slice().len()
self.stop - self.start
}
}
impl Clone for ReadOnlySource {
fn clone(&self) -> Self {
self.slice(0, self.len())
self.slice_from(0)
}
}
impl From<Vec<u8>> for ReadOnlySource {
fn from(data: Vec<u8>) -> ReadOnlySource {
let shared_data = SharedVecSlice::from(data);
ReadOnlySource::Anonymous(shared_data)
ReadOnlySource::new(data)
}
}

41
src/directory/shared_vec_slice.rs
View File

@@ -1,41 +0,0 @@
use std::sync::Arc;
#[derive(Clone)]
pub struct SharedVecSlice {
pub data: Arc<Vec<u8>>,
pub start: usize,
pub len: usize,
}
impl SharedVecSlice {
pub fn empty() -> SharedVecSlice {
SharedVecSlice::new(Arc::new(Vec::new()))
}
pub fn new(data: Arc<Vec<u8>>) -> SharedVecSlice {
let data_len = data.len();
SharedVecSlice {
data,
start: 0,
len: data_len,
}
}
pub fn as_slice(&self) -> &[u8] {
&self.data[self.start..self.start + self.len]
}
pub fn slice(&self, from_offset: usize, to_offset: usize) -> SharedVecSlice {
SharedVecSlice {
data: Arc::clone(&self.data),
start: self.start + from_offset,
len: to_offset - from_offset,
}
}
}
impl From<Vec<u8>> for SharedVecSlice {
fn from(data: Vec<u8>) -> SharedVecSlice {
SharedVecSlice::new(Arc::new(data))
}
}

222
src/directory/tests.rs Normal file
View File

@@ -0,0 +1,222 @@
use super::*;
use std::io::{Seek, SeekFrom, Write};
use std::mem;
use std::path::{Path, PathBuf};
use std::sync::atomic::AtomicUsize;
use std::sync::atomic::Ordering;
use std::sync::Arc;
use std::thread;
use std::time;
use std::time::Duration;
lazy_static! {
static ref TEST_PATH: &'static Path = Path::new("some_path_for_test");
}
#[test]
fn test_ram_directory() {
let mut ram_directory = RAMDirectory::create();
test_directory(&mut ram_directory);
}
#[test]
#[cfg(feature = "mmap")]
fn test_mmap_directory() {
let mut mmap_directory = MmapDirectory::create_from_tempdir().unwrap();
test_directory(&mut mmap_directory);
}
#[test]
#[should_panic]
fn ram_directory_panics_if_flush_forgotten() {
let mut ram_directory = RAMDirectory::create();
let mut write_file = ram_directory.open_write(*TEST_PATH).unwrap();
assert!(write_file.write_all(&[4]).is_ok());
}
fn test_simple(directory: &mut Directory) {
{
let mut write_file = directory.open_write(*TEST_PATH).unwrap();
assert!(directory.exists(*TEST_PATH));
write_file.write_all(&[4]).unwrap();
write_file.write_all(&[3]).unwrap();
write_file.write_all(&[7, 3, 5]).unwrap();
write_file.flush().unwrap();
}
{
let read_file = directory.open_read(*TEST_PATH).unwrap();
let data: &[u8] = &*read_file;
assert_eq!(data, &[4u8, 3u8, 7u8, 3u8, 5u8]);
}
assert!(directory.delete(*TEST_PATH).is_ok());
assert!(!directory.exists(*TEST_PATH));
}
fn test_seek(directory: &mut Directory) {
{
{
let mut write_file = directory.open_write(*TEST_PATH).unwrap();
write_file.write_all(&[4, 3, 7, 3, 5]).unwrap();
write_file.seek(SeekFrom::Start(0)).unwrap();
write_file.write_all(&[3, 1]).unwrap();
write_file.flush().unwrap();
}
let read_file = directory.open_read(*TEST_PATH).unwrap();
let data: &[u8] = &*read_file;
assert_eq!(data, &[3u8, 1u8, 7u8, 3u8, 5u8]);
}
assert!(directory.delete(*TEST_PATH).is_ok());
}
fn test_rewrite_forbidden(directory: &mut Directory) {
{
directory.open_write(*TEST_PATH).unwrap();
assert!(directory.exists(*TEST_PATH));
}
{
assert!(directory.open_write(*TEST_PATH).is_err());
}
assert!(directory.delete(*TEST_PATH).is_ok());
}
fn test_write_create_the_file(directory: &mut Directory) {
{
assert!(directory.open_read(*TEST_PATH).is_err());
let _w = directory.open_write(*TEST_PATH).unwrap();
assert!(directory.exists(*TEST_PATH));
assert!(directory.open_read(*TEST_PATH).is_ok());
assert!(directory.delete(*TEST_PATH).is_ok());
}
}
fn test_directory_delete(directory: &mut Directory) {
assert!(directory.open_read(*TEST_PATH).is_err());
let mut write_file = directory.open_write(*TEST_PATH).unwrap();
write_file.write_all(&[1, 2, 3, 4]).unwrap();
write_file.flush().unwrap();
{
let read_handle = directory.open_read(*TEST_PATH).unwrap();
{
assert_eq!(&*read_handle, &[1u8, 2u8, 3u8, 4u8]);
// Mapped files can't be deleted on Windows
if !cfg!(windows) {
assert!(directory.delete(*TEST_PATH).is_ok());
assert_eq!(&*read_handle, &[1u8, 2u8, 3u8, 4u8]);
}
assert!(directory.delete(Path::new("SomeOtherPath")).is_err());
}
}
if cfg!(windows) {
assert!(directory.delete(*TEST_PATH).is_ok());
}
assert!(directory.open_read(*TEST_PATH).is_err());
assert!(directory.delete(*TEST_PATH).is_err());
}
fn test_directory(directory: &mut Directory) {
test_simple(directory);
test_seek(directory);
test_rewrite_forbidden(directory);
test_write_create_the_file(directory);
test_directory_delete(directory);
test_lock_non_blocking(directory);
test_lock_blocking(directory);
test_watch(directory);
}
fn test_watch(directory: &mut Directory) {
let counter: Arc<AtomicUsize> = Default::default();
let counter_clone = counter.clone();
let watch_callback = Box::new(move || {
counter_clone.fetch_add(1, Ordering::SeqCst);
});
assert!(directory
.atomic_write(Path::new("meta.json"), b"random_test_data")
.is_ok());
thread::sleep(Duration::new(0, 10_000));
assert_eq!(0, counter.load(Ordering::SeqCst));
let watch_handle = directory.watch(watch_callback);
for i in 0..10 {
assert_eq!(i, counter.load(Ordering::SeqCst));
assert!(directory
.atomic_write(Path::new("meta.json"), b"random_test_data_2")
.is_ok());
for _ in 0..100 {
if counter.load(Ordering::SeqCst) > i {
break;
}
thread::sleep(Duration::from_millis(10));
}
assert_eq!(i + 1, counter.load(Ordering::SeqCst));
}
mem::drop(watch_handle);
assert!(directory
.atomic_write(Path::new("meta.json"), b"random_test_data")
.is_ok());
thread::sleep(Duration::from_millis(200));
assert_eq!(10, counter.load(Ordering::SeqCst));
}
fn test_lock_non_blocking(directory: &mut Directory) {
{
let lock_a_res = directory.acquire_lock(&Lock {
filepath: PathBuf::from("a.lock"),
is_blocking: false,
});
assert!(lock_a_res.is_ok());
let lock_b_res = directory.acquire_lock(&Lock {
filepath: PathBuf::from("b.lock"),
is_blocking: false,
});
assert!(lock_b_res.is_ok());
let lock_a_res2 = directory.acquire_lock(&Lock {
filepath: PathBuf::from("a.lock"),
is_blocking: false,
});
assert!(lock_a_res2.is_err());
}
let lock_a_res = directory.acquire_lock(&Lock {
filepath: PathBuf::from("a.lock"),
is_blocking: false,
});
assert!(lock_a_res.is_ok());
}
fn test_lock_blocking(directory: &mut Directory) {
let lock_a_res = directory.acquire_lock(&Lock {
filepath: PathBuf::from("a.lock"),
is_blocking: true,
});
assert!(lock_a_res.is_ok());
std::thread::spawn(move || {
//< lock_a_res is sent to the thread.
std::thread::sleep(time::Duration::from_millis(10));
// explicitely droping lock_a_res. It would have been sufficient to just force it
// to be part of the move, but the intent seems clearer that way.
drop(lock_a_res);
});
{
// A non-blocking call should fail, as the thread is running and holding the lock.
let lock_a_res = directory.acquire_lock(&Lock {
filepath: PathBuf::from("a.lock"),
is_blocking: false,
});
assert!(lock_a_res.is_err());
}
{
// the blocking call should wait for at least 10ms.
let start = time::Instant::now();
let lock_a_res = directory.acquire_lock(&Lock {
filepath: PathBuf::from("a.lock"),
is_blocking: true,
});
assert!(lock_a_res.is_ok());
assert!(start.elapsed().subsec_millis() >= 10);
}
}

156
src/directory/watch_event_router.rs Normal file
View File

@@ -0,0 +1,156 @@
use std::sync::Arc;
use std::sync::RwLock;
use std::sync::Weak;
/// Type alias for callbacks registered when watching files of a `Directory`.
pub type WatchCallback = Box<Fn() -> () + Sync + Send>;
/// Helper struct to implement the watch method in `Directory` implementations.
///
/// It registers callbacks (See `.subscribe(...)`) and
/// calls them upon calls to `.broadcast(...)`.
#[derive(Default)]
pub struct WatchCallbackList {
router: RwLock<Vec<Weak<WatchCallback>>>,
}
/// Controls how long a directory should watch for a file change.
///
/// After all the clones of `WatchHandle` are dropped, the associated will not be called when a
/// file change is detected.
#[must_use = "This `WatchHandle` controls the lifetime of the watch and should therefore be used."]
#[derive(Clone)]
pub struct WatchHandle(Arc<WatchCallback>);
impl WatchCallbackList {
/// Suscribes a new callback and returns a handle that controls the lifetime of the callback.
pub fn subscribe(&self, watch_callback: WatchCallback) -> WatchHandle {
let watch_callback_arc = Arc::new(watch_callback);
let watch_callback_weak = Arc::downgrade(&watch_callback_arc);
self.router.write().unwrap().push(watch_callback_weak);
WatchHandle(watch_callback_arc)
}
fn list_callback(&self) -> Vec<Arc<WatchCallback>> {
let mut callbacks = vec![];
let mut router_wlock = self.router.write().unwrap();
let mut i = 0;
while i < router_wlock.len() {
if let Some(watch) = router_wlock[i].upgrade() {
callbacks.push(watch);
i += 1;
} else {
router_wlock.swap_remove(i);
}
}
callbacks
}
/// Triggers all callbacks
pub fn broadcast(&self) {
let callbacks = self.list_callback();
let spawn_res = std::thread::Builder::new()
.name("watch-callbacks".to_string())
.spawn(move || {
for callback in callbacks {
callback();
}
});
if let Err(err) = spawn_res {
error!(
"Failed to spawn thread to call watch callbacks. Cause: {:?}",
err
);
}
}
}
#[cfg(test)]
mod tests {
use directory::WatchCallbackList;
use std::mem;
use std::sync::atomic::{AtomicUsize, Ordering};
use std::sync::Arc;
use std::thread;
use std::time::Duration;
const WAIT_TIME: u64 = 20;
#[test]
fn test_watch_event_router_simple() {
let watch_event_router = WatchCallbackList::default();
let counter: Arc<AtomicUsize> = Default::default();
let counter_clone = counter.clone();
let inc_callback = Box::new(move || {
counter_clone.fetch_add(1, Ordering::SeqCst);
});
watch_event_router.broadcast();
assert_eq!(0, counter.load(Ordering::SeqCst));
let handle_a = watch_event_router.subscribe(inc_callback);
thread::sleep(Duration::from_millis(WAIT_TIME));
assert_eq!(0, counter.load(Ordering::SeqCst));
watch_event_router.broadcast();
thread::sleep(Duration::from_millis(WAIT_TIME));
assert_eq!(1, counter.load(Ordering::SeqCst));
watch_event_router.broadcast();
watch_event_router.broadcast();
watch_event_router.broadcast();
thread::sleep(Duration::from_millis(WAIT_TIME));
assert_eq!(4, counter.load(Ordering::SeqCst));
mem::drop(handle_a);
watch_event_router.broadcast();
thread::sleep(Duration::from_millis(WAIT_TIME));
assert_eq!(4, counter.load(Ordering::SeqCst));
}
#[test]
fn test_watch_event_router_multiple_callback_same_key() {
let watch_event_router = WatchCallbackList::default();
let counter: Arc<AtomicUsize> = Default::default();
let inc_callback = |inc: usize| {
let counter_clone = counter.clone();
Box::new(move || {
counter_clone.fetch_add(inc, Ordering::SeqCst);
})
};
let handle_a = watch_event_router.subscribe(inc_callback(1));
let handle_a2 = watch_event_router.subscribe(inc_callback(10));
thread::sleep(Duration::from_millis(WAIT_TIME));
assert_eq!(0, counter.load(Ordering::SeqCst));
watch_event_router.broadcast();
watch_event_router.broadcast();
thread::sleep(Duration::from_millis(WAIT_TIME));
assert_eq!(22, counter.load(Ordering::SeqCst));
mem::drop(handle_a);
watch_event_router.broadcast();
thread::sleep(Duration::from_millis(WAIT_TIME));
assert_eq!(32, counter.load(Ordering::SeqCst));
mem::drop(handle_a2);
watch_event_router.broadcast();
watch_event_router.broadcast();
thread::sleep(Duration::from_millis(WAIT_TIME));
assert_eq!(32, counter.load(Ordering::SeqCst));
}
#[test]
fn test_watch_event_router_multiple_callback_different_key() {
let watch_event_router = WatchCallbackList::default();
let counter: Arc<AtomicUsize> = Default::default();
let counter_clone = counter.clone();
let inc_callback = Box::new(move || {
counter_clone.fetch_add(1, Ordering::SeqCst);
});
let handle_a = watch_event_router.subscribe(inc_callback);
assert_eq!(0, counter.load(Ordering::SeqCst));
watch_event_router.broadcast();
watch_event_router.broadcast();
thread::sleep(Duration::from_millis(WAIT_TIME));
assert_eq!(2, counter.load(Ordering::SeqCst));
thread::sleep(Duration::from_millis(WAIT_TIME));
mem::drop(handle_a);
watch_event_router.broadcast();
thread::sleep(Duration::from_millis(WAIT_TIME));
assert_eq!(2, counter.load(Ordering::SeqCst));
}
}

34
src/docset.rs
View File

@@ -1,4 +1,5 @@
use common::BitSet;
use fastfield::DeleteBitSet;
use std::borrow::Borrow;
use std::borrow::BorrowMut;
use std::cmp::Ordering;
@@ -95,9 +96,23 @@ pub trait DocSet {
}
/// Returns the number documents matching.
///
/// Calling this method consumes the `DocSet`.
fn count(&mut self) -> u32 {
fn count(&mut self, delete_bitset: &DeleteBitSet) -> u32 {
let mut count = 0u32;
while self.advance() {
if !delete_bitset.is_deleted(self.doc()) {
count += 1u32;
}
}
count
}
/// Returns the count of documents, deleted or not.
/// Calling this method consumes the `DocSet`.
///
/// Of course, the result is an upper bound of the result
/// given by `count()`.
fn count_including_deleted(&mut self) -> u32 {
let mut count = 0u32;
while self.advance() {
count += 1u32;
@@ -127,13 +142,18 @@ impl<TDocSet: DocSet + ?Sized> DocSet for Box<TDocSet> {
unboxed.size_hint()
}
fn count(&mut self) -> u32 {
let unboxed: &mut TDocSet = self.borrow_mut();
unboxed.count()
}
fn append_to_bitset(&mut self, bitset: &mut BitSet) {
let unboxed: &mut TDocSet = self.borrow_mut();
unboxed.append_to_bitset(bitset);
}
fn count(&mut self, delete_bitset: &DeleteBitSet) -> u32 {
let unboxed: &mut TDocSet = self.borrow_mut();
unboxed.count(delete_bitset)
}
fn count_including_deleted(&mut self) -> u32 {
let unboxed: &mut TDocSet = self.borrow_mut();
unboxed.count_including_deleted()
}
}

108
src/error.rs
View File

@@ -2,92 +2,143 @@
use std::io;
use directory::error::LockError;
use directory::error::{IOError, OpenDirectoryError, OpenReadError, OpenWriteError};
use fastfield::FastFieldNotAvailableError;
use query;
use schema;
use serde_json;
use std::fmt;
use std::path::PathBuf;
use std::sync::PoisonError;
pub struct DataCorruption {
filepath: Option<PathBuf>,
comment: String,
}
impl DataCorruption {
pub fn new(filepath: PathBuf, comment: String) -> DataCorruption {
DataCorruption {
filepath: Some(filepath),
comment,
}
}
pub fn comment_only(comment: String) -> DataCorruption {
DataCorruption {
filepath: None,
comment,
}
}
}
impl fmt::Debug for DataCorruption {
fn fmt(&self, f: &mut fmt::Formatter) -> Result<(), fmt::Error> {
write!(f, "Data corruption: ")?;
if let Some(ref filepath) = &self.filepath {
write!(f, "(in file `{:?}`)", filepath)?;
}
write!(f, ": {}.", self.comment)?;
Ok(())
}
}
/// The library's failure based error enum
#[derive(Debug, Fail)]
pub enum TantivyError {
/// Path does not exist.
#[fail(display = "path does not exist: '{:?}'", _0)]
#[fail(display = "Path does not exist: '{:?}'", _0)]
PathDoesNotExist(PathBuf),
/// File already exists, this is a problem when we try to write into a new file.
#[fail(display = "file already exists: '{:?}'", _0)]
#[fail(display = "File already exists: '{:?}'", _0)]
FileAlreadyExists(PathBuf),
/// Index already exists in this directory
#[fail(display = "Index already exists")]
IndexAlreadyExists,
/// Failed to acquire file lock
#[fail(display = "Failed to acquire Lockfile: {:?}. {:?}", _0, _1)]
LockFailure(LockError, Option<String>),
/// IO Error.
#[fail(display = "an IO error occurred: '{}'", _0)]
#[fail(display = "An IO error occurred: '{}'", _0)]
IOError(#[cause] IOError),
/// The data within is corrupted.
///
/// For instance, it contains invalid JSON.
#[fail(display = "file contains corrupted data: '{:?}'", _0)]
CorruptedFile(PathBuf),
/// Data corruption.
#[fail(display = "{:?}", _0)]
DataCorruption(DataCorruption),
/// A thread holding the locked panicked and poisoned the lock.
#[fail(display = "a thread holding the locked panicked and poisoned the lock")]
#[fail(display = "A thread holding the locked panicked and poisoned the lock")]
Poisoned,
/// Invalid argument was passed by the user.
#[fail(display = "an invalid argument was passed: '{}'", _0)]
#[fail(display = "An invalid argument was passed: '{}'", _0)]
InvalidArgument(String),
/// An Error happened in one of the thread.
#[fail(display = "an error occurred in a thread: '{}'", _0)]
#[fail(display = "An error occurred in a thread: '{}'", _0)]
ErrorInThread(String),
/// An Error appeared related to the schema.
#[fail(display = "Schema error: '{}'", _0)]
SchemaError(String),
/// Tried to access a fastfield reader for a field not configured accordingly.
#[fail(display = "fast field not available: '{:?}'", _0)]
#[fail(display = "Fast field not available: '{:?}'", _0)]
FastFieldError(#[cause] FastFieldNotAvailableError),
/// System error. (e.g.: We failed spawning a new thread)
#[fail(display = "System error.'{}'", _0)]
SystemError(String),
}
impl From<DataCorruption> for TantivyError {
fn from(data_corruption: DataCorruption) -> TantivyError {
TantivyError::DataCorruption(data_corruption)
}
}
impl From<FastFieldNotAvailableError> for TantivyError {
fn from(fastfield_error: FastFieldNotAvailableError) -> TantivyError {
TantivyError::FastFieldError(fastfield_error).into()
TantivyError::FastFieldError(fastfield_error)
}
}
impl From<LockError> for TantivyError {
fn from(lock_error: LockError) -> TantivyError {
TantivyError::LockFailure(lock_error, None)
}
}
impl From<IOError> for TantivyError {
fn from(io_error: IOError) -> TantivyError {
TantivyError::IOError(io_error).into()
TantivyError::IOError(io_error)
}
}
impl From<io::Error> for TantivyError {
fn from(io_error: io::Error) -> TantivyError {
TantivyError::IOError(io_error.into()).into()
TantivyError::IOError(io_error.into())
}
}
impl From<query::QueryParserError> for TantivyError {
fn from(parsing_error: query::QueryParserError) -> TantivyError {
TantivyError::InvalidArgument(format!("Query is invalid. {:?}", parsing_error)).into()
TantivyError::InvalidArgument(format!("Query is invalid. {:?}", parsing_error))
}
}
impl<Guard> From<PoisonError<Guard>> for TantivyError {
fn from(_: PoisonError<Guard>) -> TantivyError {
TantivyError::Poisoned.into()
TantivyError::Poisoned
}
}
impl From<OpenReadError> for TantivyError {
fn from(error: OpenReadError) -> TantivyError {
match error {
OpenReadError::FileDoesNotExist(filepath) => {
TantivyError::PathDoesNotExist(filepath).into()
}
OpenReadError::IOError(io_error) => TantivyError::IOError(io_error).into(),
OpenReadError::FileDoesNotExist(filepath) => TantivyError::PathDoesNotExist(filepath),
OpenReadError::IOError(io_error) => TantivyError::IOError(io_error),
}
}
}
impl From<schema::DocParsingError> for TantivyError {
fn from(error: schema::DocParsingError) -> TantivyError {
TantivyError::InvalidArgument(format!("Failed to parse document {:?}", error)).into()
TantivyError::InvalidArgument(format!("Failed to parse document {:?}", error))
}
}
@@ -98,7 +149,7 @@ impl From<OpenWriteError> for TantivyError {
TantivyError::FileAlreadyExists(filepath)
}
OpenWriteError::IOError(io_error) => TantivyError::IOError(io_error),
}.into()
}
}
}
@@ -106,11 +157,12 @@ impl From<OpenDirectoryError> for TantivyError {
fn from(error: OpenDirectoryError) -> TantivyError {
match error {
OpenDirectoryError::DoesNotExist(directory_path) => {
TantivyError::PathDoesNotExist(directory_path).into()
TantivyError::PathDoesNotExist(directory_path)
}
OpenDirectoryError::NotADirectory(directory_path) => TantivyError::InvalidArgument(
format!("{:?} is not a directory", directory_path),
).into(),
OpenDirectoryError::NotADirectory(directory_path) => {
TantivyError::InvalidArgument(format!("{:?} is not a directory", directory_path))
}
OpenDirectoryError::IoError(err) => TantivyError::IOError(IOError::from(err)),
}
}
}
@@ -118,6 +170,6 @@ impl From<OpenDirectoryError> for TantivyError {
impl From<serde_json::Error> for TantivyError {
fn from(error: serde_json::Error) -> TantivyError {
let io_err = io::Error::from(error);
TantivyError::IOError(io_err.into()).into()
TantivyError::IOError(io_err.into())
}
}

22
src/fastfield/bytes/mod.rs
View File

@@ -6,12 +6,12 @@ pub use self::writer::BytesFastFieldWriter;
#[cfg(test)]
mod tests {
use schema::SchemaBuilder;
use schema::Schema;
use Index;
#[test]
fn test_bytes() {
let mut schema_builder = SchemaBuilder::default();
let mut schema_builder = Schema::builder();
let field = schema_builder.add_bytes_field("bytesfield");
let schema = schema_builder.build();
let index = Index::create_in_ram(schema);
@@ -22,17 +22,15 @@ mod tests {
index_writer.add_document(doc!(field=>vec![1u8, 3, 5, 7, 9]));
index_writer.add_document(doc!(field=>vec![0u8; 1000]));
assert!(index_writer.commit().is_ok());
let searcher = index.reader().unwrap().searcher();
let segment_reader = searcher.segment_reader(0);
let bytes_reader = segment_reader.fast_fields().bytes(field).unwrap();
index.load_searchers().unwrap();
let searcher = index.searcher();
let reader = searcher.segment_reader(0);
let bytes_reader = reader.bytes_fast_field_reader(field).unwrap();
assert_eq!(bytes_reader.get_val(0), &[0u8, 1, 2, 3]);
assert!(bytes_reader.get_val(1).is_empty());
assert_eq!(bytes_reader.get_val(2), &[255u8]);
assert_eq!(bytes_reader.get_val(3), &[1u8, 3, 5, 7, 9]);
assert_eq!(bytes_reader.get_bytes(0), &[0u8, 1, 2, 3]);
assert!(bytes_reader.get_bytes(1).is_empty());
assert_eq!(bytes_reader.get_bytes(2), &[255u8]);
assert_eq!(bytes_reader.get_bytes(3), &[1u8, 3, 5, 7, 9]);
let long = vec![0u8; 1000];
assert_eq!(bytes_reader.get_val(4), long.as_slice());
assert_eq!(bytes_reader.get_bytes(4), long.as_slice());
}
}

15
src/fastfield/bytes/reader.rs
View File

@@ -14,6 +14,7 @@ use DocId;
///
/// Reading the value for a document is done by reading the start index for it,
/// and the start index for the next document, and keeping the bytes in between.
#[derive(Clone)]
pub struct BytesFastFieldReader {
idx_reader: FastFieldReader<u64>,
values: OwningRef<ReadOnlySource, [u8]>,
@@ -28,10 +29,20 @@ impl BytesFastFieldReader {
BytesFastFieldReader { idx_reader, values }
}
/// Returns the bytes associated to the given `doc`
pub fn get_val(&self, doc: DocId) -> &[u8] {
fn range(&self, doc: DocId) -> (usize, usize) {
let start = self.idx_reader.get(doc) as usize;
let stop = self.idx_reader.get(doc + 1) as usize;
(start, stop)
}
/// Returns the bytes associated to the given `doc`
pub fn get_bytes(&self, doc: DocId) -> &[u8] {
let (start, stop) = self.range(doc);
&self.values[start..stop]
}
/// Returns the overall number of bytes in this bytes fast field.
pub fn total_num_bytes(&self) -> usize {
self.values.len()
}
}

2
src/fastfield/bytes/writer.rs
View File

@@ -51,7 +51,7 @@ impl BytesFastFieldWriter {
self.next_doc();
for field_value in doc.field_values() {
if field_value.field() == self.field {
if let &Value::Bytes(ref bytes) = field_value.value() {
if let Value::Bytes(ref bytes) = *field_value.value() {
self.vals.extend_from_slice(bytes);
} else {
panic!(

29
src/fastfield/delete.rs
View File

@@ -2,6 +2,7 @@ use bit_set::BitSet;
use common::HasLen;
use directory::ReadOnlySource;
use directory::WritePtr;
use space_usage::ByteCount;
use std::io;
use std::io::Write;
use DocId;
@@ -41,7 +42,8 @@ pub struct DeleteBitSet {
impl DeleteBitSet {
/// Opens a delete bitset given its data source.
pub fn open(data: ReadOnlySource) -> DeleteBitSet {
let num_deleted: usize = data.as_slice()
let num_deleted: usize = data
.as_slice()
.iter()
.map(|b| b.count_ones() as usize)
.sum();
@@ -51,16 +53,23 @@ impl DeleteBitSet {
}
}
/// Returns whether the document has been marked as deleted.
/// Returns true iff the document is still "alive". In other words, if it has not been deleted.
pub fn is_alive(&self, doc: DocId) -> bool {
!self.is_deleted(doc)
}
/// Returns true iff the document has been marked as deleted.
#[inline(always)]
pub fn is_deleted(&self, doc: DocId) -> bool {
if self.len == 0 {
false
} else {
let byte_offset = doc / 8u32;
let b: u8 = (*self.data)[byte_offset as usize];
let shift = (doc & 7u32) as u8;
b & (1u8 << shift) != 0
}
let byte_offset = doc / 8u32;
let b: u8 = (*self.data)[byte_offset as usize];
let shift = (doc & 7u32) as u8;
b & (1u8 << shift) != 0
}
/// Summarize total space usage of this bitset.
pub fn space_usage(&self) -> ByteCount {
self.data.len()
}
}

17
src/fastfield/facet_reader.rs
View File

@@ -1,5 +1,6 @@
use super::MultiValueIntFastFieldReader;
use schema::Facet;
use std::str;
use termdict::TermDictionary;
use termdict::TermOrdinal;
use DocId;
@@ -20,6 +21,7 @@ use DocId;
pub struct FacetReader {
term_ords: MultiValueIntFastFieldReader<u64>,
term_dict: TermDictionary,
buffer: Vec<u8>,
}
impl FacetReader {
@@ -37,6 +39,7 @@ impl FacetReader {
FacetReader {
term_ords,
term_dict,
buffer: vec![],
}
}
@@ -55,10 +58,18 @@ impl FacetReader {
}
/// Given a term ordinal returns the term associated to it.
pub fn facet_from_ord(&self, facet_ord: TermOrdinal, output: &mut Facet) {
let found_term = self.term_dict
.ord_to_term(facet_ord as u64, output.inner_buffer_mut());
pub fn facet_from_ord(
&mut self,
facet_ord: TermOrdinal,
output: &mut Facet,
) -> Result<(), str::Utf8Error> {
let found_term = self
.term_dict
.ord_to_term(facet_ord as u64, &mut self.buffer);
assert!(found_term, "Term ordinal {} no found.", facet_ord);
let facet_str = str::from_utf8(&self.buffer[..])?;
output.set_facet_str(facet_str);
Ok(())
}
/// Return the list of facet ordinals associated to a document.

29
src/fastfield/mod.rs
View File

@@ -30,6 +30,7 @@ pub use self::error::{FastFieldNotAvailableError, Result};
pub use self::facet_reader::FacetReader;
pub use self::multivalued::{MultiValueIntFastFieldReader, MultiValueIntFastFieldWriter};
pub use self::reader::FastFieldReader;
pub use self::readers::FastFieldReaders;
pub use self::serializer::FastFieldSerializer;
pub use self::writer::{FastFieldsWriter, IntFastFieldWriter};
use common;
@@ -43,6 +44,7 @@ mod error;
mod facet_reader;
mod multivalued;
mod reader;
mod readers;
mod serializer;
mod writer;
@@ -78,10 +80,6 @@ impl FastValue for u64 {
*self
}
fn as_u64(&self) -> u64 {
*self
}
fn fast_field_cardinality(field_type: &FieldType) -> Option<Cardinality> {
match *field_type {
FieldType::U64(ref integer_options) => integer_options.get_fastfield_cardinality(),
@@ -89,6 +87,10 @@ impl FastValue for u64 {
_ => None,
}
}
fn as_u64(&self) -> u64 {
*self
}
}
impl FastValue for i64 {
@@ -127,19 +129,19 @@ mod tests {
use common::CompositeFile;
use directory::{Directory, RAMDirectory, WritePtr};
use fastfield::FastFieldReader;
use rand::Rng;
use rand::prelude::SliceRandom;
use rand::rngs::StdRng;
use rand::SeedableRng;
use rand::XorShiftRng;
use schema::Document;
use schema::Field;
use schema::Schema;
use schema::FAST;
use schema::{Schema, SchemaBuilder};
use std::collections::HashMap;
use std::path::Path;
lazy_static! {
pub static ref SCHEMA: Schema = {
let mut schema_builder = SchemaBuilder::default();
let mut schema_builder = Schema::builder();
schema_builder.add_u64_field("field", FAST);
schema_builder.build()
};
@@ -298,7 +300,7 @@ mod tests {
fn test_signed_intfastfield() {
let path = Path::new("test");
let mut directory: RAMDirectory = RAMDirectory::create();
let mut schema_builder = SchemaBuilder::new();
let mut schema_builder = Schema::builder();
let i64_field = schema_builder.add_i64_field("field", FAST);
let schema = schema_builder.build();
@@ -342,7 +344,7 @@ mod tests {
fn test_signed_intfastfield_default_val() {
let path = Path::new("test");
let mut directory: RAMDirectory = RAMDirectory::create();
let mut schema_builder = SchemaBuilder::new();
let mut schema_builder = Schema::builder();
let i64_field = schema_builder.add_i64_field("field", FAST);
let schema = schema_builder.build();
@@ -367,11 +369,10 @@ mod tests {
}
}
// Warning: this generates the same permutation at each call
pub fn generate_permutation() -> Vec<u64> {
let seed: [u8; 16] = [1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16];
let mut rng = XorShiftRng::from_seed(seed);
let mut permutation: Vec<u64> = (0u64..1_000_000u64).collect();
rng.shuffle(&mut permutation);
let mut permutation: Vec<u64> = (0u64..100_000u64).collect();
permutation.shuffle(&mut StdRng::from_seed([1u8; 32]));
permutation
}

166
src/fastfield/multivalued/mod.rs
View File

@@ -7,14 +7,20 @@ pub use self::writer::MultiValueIntFastFieldWriter;
#[cfg(test)]
mod tests {
extern crate time;
use self::time::Duration;
use collector::TopDocs;
use query::QueryParser;
use schema::Cardinality;
use schema::Facet;
use schema::IntOptions;
use schema::SchemaBuilder;
use schema::Schema;
use Index;
#[test]
fn test_multivalued_u64() {
let mut schema_builder = SchemaBuilder::default();
let mut schema_builder = Schema::builder();
let field = schema_builder.add_u64_field(
"multifield",
IntOptions::default().set_fast(Cardinality::MultiValues),
@@ -28,11 +34,10 @@ mod tests {
index_writer.add_document(doc!(field=>5u64, field=>20u64,field=>1u64));
assert!(index_writer.commit().is_ok());
index.load_searchers().unwrap();
let searcher = index.searcher();
let reader = searcher.segment_reader(0);
let searcher = index.reader().unwrap().searcher();
let segment_reader = searcher.segment_reader(0);
let mut vals = Vec::new();
let multi_value_reader = reader.multi_fast_field_reader::<u64>(field).unwrap();
let multi_value_reader = segment_reader.fast_fields().u64s(field).unwrap();
{
multi_value_reader.get_vals(2, &mut vals);
assert_eq!(&vals, &[4u64]);
@@ -47,9 +52,136 @@ mod tests {
}
}
#[test]
fn test_multivalued_date() {
let mut schema_builder = Schema::builder();
let date_field = schema_builder.add_date_field(
"multi_date_field",
IntOptions::default()
.set_fast(Cardinality::MultiValues)
.set_indexed()
.set_stored(),
);
let time_i =
schema_builder.add_i64_field("time_stamp_i", IntOptions::default().set_stored());
let schema = schema_builder.build();
let index = Index::create_in_ram(schema);
let mut index_writer = index.writer_with_num_threads(1, 3_000_000).unwrap();
let first_time_stamp = chrono::Utc::now();
index_writer.add_document(
doc!(date_field=>first_time_stamp, date_field=>first_time_stamp, time_i=>1i64),
);
index_writer.add_document(doc!(time_i=>0i64));
// add one second
index_writer
.add_document(doc!(date_field=>first_time_stamp + Duration::seconds(1), time_i=>2i64));
// add another second
let two_secs_ahead = first_time_stamp + Duration::seconds(2);
index_writer.add_document(doc!(date_field=>two_secs_ahead, date_field=>two_secs_ahead,date_field=>two_secs_ahead, time_i=>3i64));
assert!(index_writer.commit().is_ok());
let reader = index.reader().unwrap();
let searcher = reader.searcher();
let reader = searcher.segment_reader(0);
assert_eq!(reader.num_docs(), 4);
{
let parser = QueryParser::for_index(&index, vec![date_field]);
let query = parser
.parse_query(&format!("\"{}\"", first_time_stamp.to_rfc3339()).to_string())
.expect("could not parse query");
let results = searcher
.search(&query, &TopDocs::with_limit(5))
.expect("could not query index");
assert_eq!(results.len(), 1);
for (_score, doc_address) in results {
let retrieved_doc = searcher.doc(doc_address).expect("cannot fetch doc");
assert_eq!(
retrieved_doc
.get_first(date_field)
.expect("cannot find value")
.date_value()
.timestamp(),
first_time_stamp.timestamp()
);
assert_eq!(
retrieved_doc
.get_first(time_i)
.expect("cannot find value")
.i64_value(),
1i64
);
}
}
{
let parser = QueryParser::for_index(&index, vec![date_field]);
let query = parser
.parse_query(&format!("\"{}\"", two_secs_ahead.to_rfc3339()).to_string())
.expect("could not parse query");
let results = searcher
.search(&query, &TopDocs::with_limit(5))
.expect("could not query index");
assert_eq!(results.len(), 1);
for (_score, doc_address) in results {
let retrieved_doc = searcher.doc(doc_address).expect("cannot fetch doc");
assert_eq!(
retrieved_doc
.get_first(date_field)
.expect("cannot find value")
.date_value()
.timestamp(),
two_secs_ahead.timestamp()
);
assert_eq!(
retrieved_doc
.get_first(time_i)
.expect("cannot find value")
.i64_value(),
3i64
);
}
}
// TODO: support Date range queries
// {
// let parser = QueryParser::for_index(&index, vec![date_field]);
// let range_q = format!("\"{}\"..\"{}\"",
// (first_time_stamp + Duration::seconds(1)).to_rfc3339(),
// (first_time_stamp + Duration::seconds(3)).to_rfc3339()
// );
// let query = parser.parse_query(&range_q)
// .expect("could not parse query");
// let results = searcher.search(&query, &TopDocs::with_limit(5))
// .expect("could not query index");
//
//
// assert_eq!(results.len(), 2);
// for (i, doc_pair) in results.iter().enumerate() {
// let retrieved_doc = searcher.doc(doc_pair.1).expect("cannot fetch doc");
// let offset_sec = match i {
// 0 => 1,
// 1 => 3,
// _ => panic!("should not have more than 2 docs")
// };
// let time_i_val = match i {
// 0 => 2,
// 1 => 3,
// _ => panic!("should not have more than 2 docs")
// };
// assert_eq!(retrieved_doc.get_first(date_field).expect("cannot find value").date_value().timestamp(),
// (first_time_stamp + Duration::seconds(offset_sec)).timestamp());
// assert_eq!(retrieved_doc.get_first(time_i).expect("cannot find value").i64_value(), time_i_val);
// }
// }
}
#[test]
fn test_multivalued_i64() {
let mut schema_builder = SchemaBuilder::default();
let mut schema_builder = Schema::builder();
let field = schema_builder.add_i64_field(
"multifield",
IntOptions::default().set_fast(Cardinality::MultiValues),
@@ -63,11 +195,10 @@ mod tests {
index_writer.add_document(doc!(field=> -5i64, field => -20i64, field=>1i64));
assert!(index_writer.commit().is_ok());
index.load_searchers().unwrap();
let searcher = index.searcher();
let reader = searcher.segment_reader(0);
let searcher = index.reader().unwrap().searcher();
let segment_reader = searcher.segment_reader(0);
let mut vals = Vec::new();
let multi_value_reader = reader.multi_fast_field_reader::<i64>(field).unwrap();
let multi_value_reader = segment_reader.fast_fields().i64s(field).unwrap();
{
multi_value_reader.get_vals(2, &mut vals);
assert_eq!(&vals, &[-4i64]);
@@ -85,4 +216,17 @@ mod tests {
assert_eq!(&vals, &[-5i64, -20i64, 1i64]);
}
}
#[test]
#[ignore]
fn test_many_facets() {
let mut schema_builder = Schema::builder();
let field = schema_builder.add_facet_field("facetfield");
let schema = schema_builder.build();
let index = Index::create_in_ram(schema);
let mut index_writer = index.writer_with_num_threads(1, 3_000_000).unwrap();
for i in 0..100_000 {
index_writer.add_document(doc!(field=> Facet::from(format!("/lang/{}", i).as_str())));
}
assert!(index_writer.commit().is_ok());
}
}

57
src/fastfield/multivalued/reader.rs
View File

@@ -26,6 +26,13 @@ impl<Item: FastValue> MultiValueIntFastFieldReader<Item> {
}
}
pub(crate) fn into_u64s_reader(self) -> MultiValueIntFastFieldReader<u64> {
MultiValueIntFastFieldReader {
idx_reader: self.idx_reader,
vals_reader: self.vals_reader.into_u64_reader(),
}
}
/// Returns `(start, stop)`, such that the values associated
/// to the given document are `start..stop`.
fn range(&self, doc: DocId) -> (u64, u64) {
@@ -39,7 +46,18 @@ impl<Item: FastValue> MultiValueIntFastFieldReader<Item> {
let (start, stop) = self.range(doc);
let len = (stop - start) as usize;
vals.resize(len, Item::default());
self.vals_reader.get_range(start as u32, &mut vals[..]);
self.vals_reader.get_range_u64(start, &mut vals[..]);
}
/// Returns the number of values associated with the document `DocId`.
pub fn num_vals(&self, doc: DocId) -> usize {
let (start, stop) = self.range(doc);
(stop - start) as usize
}
/// Returns the overall number of values in this field .
pub fn total_num_vals(&self) -> u64 {
self.idx_reader.max_value()
}
}
@@ -47,55 +65,44 @@ impl<Item: FastValue> MultiValueIntFastFieldReader<Item> {
mod tests {
use core::Index;
use schema::{Document, Facet, SchemaBuilder};
use schema::{Facet, Schema};
#[test]
fn test_multifastfield_reader() {
let mut schema_builder = SchemaBuilder::new();
let mut schema_builder = Schema::builder();
let facet_field = schema_builder.add_facet_field("facets");
let schema = schema_builder.build();
let index = Index::create_in_ram(schema);
let mut index_writer = index
.writer_with_num_threads(1, 30_000_000)
.expect("Failed to create index writer.");
{
let mut doc = Document::new();
doc.add_facet(facet_field, "/category/cat2");
doc.add_facet(facet_field, "/category/cat1");
index_writer.add_document(doc);
}
{
let mut doc = Document::new();
doc.add_facet(facet_field, "/category/cat2");
index_writer.add_document(doc);
}
{
let mut doc = Document::new();
doc.add_facet(facet_field, "/category/cat3");
index_writer.add_document(doc);
}
index_writer.add_document(doc!(
facet_field => Facet::from("/category/cat2"),
facet_field => Facet::from("/category/cat1"),
));
index_writer.add_document(doc!(facet_field => Facet::from("/category/cat2")));
index_writer.add_document(doc!(facet_field => Facet::from("/category/cat3")));
index_writer.commit().expect("Commit failed");
index.load_searchers().expect("Reloading searchers");
let searcher = index.searcher();
let searcher = index.reader().unwrap().searcher();
let segment_reader = searcher.segment_reader(0);
let mut facet_reader = segment_reader.facet_reader(facet_field).unwrap();
let mut facet = Facet::root();
{
facet_reader.facet_from_ord(1, &mut facet);
facet_reader.facet_from_ord(1, &mut facet).unwrap();
assert_eq!(facet, Facet::from("/category"));
}
{
facet_reader.facet_from_ord(2, &mut facet);
facet_reader.facet_from_ord(2, &mut facet).unwrap();
assert_eq!(facet, Facet::from("/category/cat1"));
}
{
facet_reader.facet_from_ord(3, &mut facet);
facet_reader.facet_from_ord(3, &mut facet).unwrap();
assert_eq!(format!("{}", facet), "/category/cat2");
assert_eq!(facet, Facet::from("/category/cat2"));
}
{
facet_reader.facet_from_ord(4, &mut facet);
facet_reader.facet_from_ord(4, &mut facet).unwrap();
assert_eq!(facet, Facet::from("/category/cat3"));
}

6
src/fastfield/multivalued/writer.rs
View File

@@ -32,7 +32,7 @@ use DocId;
/// term ids when the segment is getting serialized.
pub struct MultiValueIntFastFieldWriter {
field: Field,
vals: Vec<u64>,
vals: Vec<UnorderedTermId>,
doc_index: Vec<u64>,
is_facet: bool,
}
@@ -132,7 +132,8 @@ impl MultiValueIntFastFieldWriter {
);
let mut doc_vals: Vec<u64> = Vec::with_capacity(100);
for (start, stop) in self.doc_index
for (start, stop) in self
.doc_index
.windows(2)
.map(|interval| (interval[0], interval[1]))
.chain(Some(last_interval).into_iter())
@@ -148,7 +149,6 @@ impl MultiValueIntFastFieldWriter {
value_serializer.add_val(val)?;
}
}
}
None => {
let val_min_max = self.vals.iter().cloned().minmax();

49
src/fastfield/reader.rs
View File

@@ -7,11 +7,10 @@ use directory::ReadOnlySource;
use directory::{Directory, RAMDirectory, WritePtr};
use fastfield::{FastFieldSerializer, FastFieldsWriter};
use owning_ref::OwningRef;
use schema::SchemaBuilder;
use schema::Schema;
use schema::FAST;
use std::collections::HashMap;
use std::marker::PhantomData;
use std::mem;
use std::path::Path;
use DocId;
@@ -51,6 +50,15 @@ impl<Item: FastValue> FastFieldReader<Item> {
}
}
pub(crate) fn into_u64_reader(self) -> FastFieldReader<u64> {
FastFieldReader {
bit_unpacker: self.bit_unpacker,
min_value_u64: self.min_value_u64,
max_value_u64: self.max_value_u64,
_phantom: PhantomData,
}
}
/// Return the value associated to the given document.
///
/// This accessor should return as fast as possible.
@@ -60,7 +68,29 @@ impl<Item: FastValue> FastFieldReader<Item> {
/// May panic if `doc` is greater than the segment
// `maxdoc`.
pub fn get(&self, doc: DocId) -> Item {
Item::from_u64(self.min_value_u64 + self.bit_unpacker.get(doc as usize))
self.get_u64(u64::from(doc))
}
pub(crate) fn get_u64(&self, doc: u64) -> Item {
Item::from_u64(self.min_value_u64 + self.bit_unpacker.get(doc))
}
/// Internally `multivalued` also use SingleValue Fast fields.
/// It works as follows... A first column contains the list of start index
/// for each document, a second column contains the actual values.
///
/// The values associated to a given doc, are then
/// `second_column[first_column.get(doc)..first_column.get(doc+1)]`.
///
/// Which means single value fast field reader can be indexed internally with
/// something different from a `DocId`. For this use case, we want to use `u64`
/// values.
///
/// See `get_range` for an actual documentation about this method.
pub(crate) fn get_range_u64(&self, start: u64, output: &mut [Item]) {
for (i, out) in output.iter_mut().enumerate() {
*out = self.get_u64(start + (i as u64));
}
}
/// Fills an output buffer with the fast field values
@@ -76,15 +106,8 @@ impl<Item: FastValue> FastFieldReader<Item> {
///
/// May panic if `start + output.len()` is greater than
/// the segment's `maxdoc`.
///
// TODO change start to `u64`.
// For multifastfield, start is an index in a second fastfield, not a `DocId`
pub fn get_range(&self, start: u32, output: &mut [Item]) {
let output_u64: &mut [u64] = unsafe { mem::transmute(output) }; // ok: Item is either `u64` or `i64`
self.bit_unpacker.get_range(start, output_u64);
for out in output_u64.iter_mut() {
*out = Item::from_u64(*out + self.min_value_u64).as_u64();
}
pub fn get_range(&self, start: DocId, output: &mut [Item]) {
self.get_range_u64(u64::from(start), output);
}
/// Returns the minimum value for this fast field.
@@ -108,7 +131,7 @@ impl<Item: FastValue> FastFieldReader<Item> {
impl<Item: FastValue> From<Vec<Item>> for FastFieldReader<Item> {
fn from(vals: Vec<Item>) -> FastFieldReader<Item> {
let mut schema_builder = SchemaBuilder::default();
let mut schema_builder = Schema::builder();
let field = schema_builder.add_u64_field("field", FAST);
let schema = schema_builder.build();
let path = Path::new("__dummy__");

191
src/fastfield/readers.rs Normal file
View File

@@ -0,0 +1,191 @@
use common::CompositeFile;
use fastfield::BytesFastFieldReader;
use fastfield::MultiValueIntFastFieldReader;
use fastfield::{FastFieldNotAvailableError, FastFieldReader};
use schema::{Cardinality, Field, FieldType, Schema};
use space_usage::PerFieldSpaceUsage;
use std::collections::HashMap;
use Result;
/// Provides access to all of the FastFieldReader.
///
/// Internally, `FastFieldReaders` have preloaded fast field readers,
/// and just wraps several `HashMap`.
pub struct FastFieldReaders {
fast_field_i64: HashMap<Field, FastFieldReader<i64>>,
fast_field_u64: HashMap<Field, FastFieldReader<u64>>,
fast_field_i64s: HashMap<Field, MultiValueIntFastFieldReader<i64>>,
fast_field_u64s: HashMap<Field, MultiValueIntFastFieldReader<u64>>,
fast_bytes: HashMap<Field, BytesFastFieldReader>,
fast_fields_composite: CompositeFile,
}
enum FastType {
I64,
U64,
}
fn type_and_cardinality(field_type: &FieldType) -> Option<(FastType, Cardinality)> {
match field_type {
FieldType::U64(options) => options
.get_fastfield_cardinality()
.map(|cardinality| (FastType::U64, cardinality)),
FieldType::I64(options) => options
.get_fastfield_cardinality()
.map(|cardinality| (FastType::I64, cardinality)),
FieldType::HierarchicalFacet => Some((FastType::U64, Cardinality::MultiValues)),
_ => None,
}
}
impl FastFieldReaders {
pub(crate) fn load_all(
schema: &Schema,
fast_fields_composite: &CompositeFile,
) -> Result<FastFieldReaders> {
let mut fast_field_readers = FastFieldReaders {
fast_field_i64: Default::default(),
fast_field_u64: Default::default(),
fast_field_i64s: Default::default(),
fast_field_u64s: Default::default(),
fast_bytes: Default::default(),
fast_fields_composite: fast_fields_composite.clone(),
};
for (field_id, field_entry) in schema.fields().iter().enumerate() {
let field = Field(field_id as u32);
let field_type = field_entry.field_type();
if field_type == &FieldType::Bytes {
let idx_reader = fast_fields_composite
.open_read_with_idx(field, 0)
.ok_or_else(|| FastFieldNotAvailableError::new(field_entry))
.map(FastFieldReader::open)?;
let data = fast_fields_composite
.open_read_with_idx(field, 1)
.ok_or_else(|| FastFieldNotAvailableError::new(field_entry))?;
fast_field_readers
.fast_bytes
.insert(field, BytesFastFieldReader::open(idx_reader, data));
} else if let Some((fast_type, cardinality)) = type_and_cardinality(field_type) {
match cardinality {
Cardinality::SingleValue => {
if let Some(fast_field_data) = fast_fields_composite.open_read(field) {
match fast_type {
FastType::U64 => {
let fast_field_reader = FastFieldReader::open(fast_field_data);
fast_field_readers
.fast_field_u64
.insert(field, fast_field_reader);
}
FastType::I64 => {
fast_field_readers.fast_field_i64.insert(
field,
FastFieldReader::open(fast_field_data.clone()),
);
}
}
} else {
return Err(From::from(FastFieldNotAvailableError::new(field_entry)));
}
}
Cardinality::MultiValues => {
let idx_opt = fast_fields_composite.open_read_with_idx(field, 0);
let data_opt = fast_fields_composite.open_read_with_idx(field, 1);
if let (Some(fast_field_idx), Some(fast_field_data)) = (idx_opt, data_opt) {
let idx_reader = FastFieldReader::open(fast_field_idx);
match fast_type {
FastType::I64 => {
let vals_reader = FastFieldReader::open(fast_field_data);
let multivalued_int_fast_field =
MultiValueIntFastFieldReader::open(idx_reader, vals_reader);
fast_field_readers
.fast_field_i64s
.insert(field, multivalued_int_fast_field);
}
FastType::U64 => {
let vals_reader = FastFieldReader::open(fast_field_data);
let multivalued_int_fast_field =
MultiValueIntFastFieldReader::open(idx_reader, vals_reader);
fast_field_readers
.fast_field_u64s
.insert(field, multivalued_int_fast_field);
}
}
} else {
return Err(From::from(FastFieldNotAvailableError::new(field_entry)));
}
}
}
}
}
Ok(fast_field_readers)
}
pub(crate) fn space_usage(&self) -> PerFieldSpaceUsage {
self.fast_fields_composite.space_usage()
}
/// Returns the `u64` fast field reader reader associated to `field`.
///
/// If `field` is not a u64 fast field, this method returns `None`.
pub fn u64(&self, field: Field) -> Option<FastFieldReader<u64>> {
self.fast_field_u64.get(&field).cloned()
}
/// If the field is a u64-fast field return the associated reader.
/// If the field is a i64-fast field, return the associated u64 reader. Values are
/// mapped from i64 to u64 using a (well the, it is unique) monotonic mapping. ///
///
/// This method is useful when merging segment reader.
pub(crate) fn u64_lenient(&self, field: Field) -> Option<FastFieldReader<u64>> {
if let Some(u64_ff_reader) = self.u64(field) {
return Some(u64_ff_reader);
}
if let Some(i64_ff_reader) = self.i64(field) {
return Some(i64_ff_reader.into_u64_reader());
}
None
}
/// Returns the `i64` fast field reader reader associated to `field`.
///
/// If `field` is not a i64 fast field, this method returns `None`.
pub fn i64(&self, field: Field) -> Option<FastFieldReader<i64>> {
self.fast_field_i64.get(&field).cloned()
}
/// Returns a `u64s` multi-valued fast field reader reader associated to `field`.
///
/// If `field` is not a u64 multi-valued fast field, this method returns `None`.
pub fn u64s(&self, field: Field) -> Option<MultiValueIntFastFieldReader<u64>> {
self.fast_field_u64s.get(&field).cloned()
}
/// If the field is a u64s-fast field return the associated reader.
/// If the field is a i64s-fast field, return the associated u64s reader. Values are
/// mapped from i64 to u64 using a (well the, it is unique) monotonic mapping.
///
/// This method is useful when merging segment reader.
pub(crate) fn u64s_lenient(&self, field: Field) -> Option<MultiValueIntFastFieldReader<u64>> {
if let Some(u64s_ff_reader) = self.u64s(field) {
return Some(u64s_ff_reader);
}
if let Some(i64s_ff_reader) = self.i64s(field) {
return Some(i64s_ff_reader.into_u64s_reader());
}
None
}
/// Returns a `i64s` multi-valued fast field reader reader associated to `field`.
///
/// If `field` is not a i64 multi-valued fast field, this method returns `None`.
pub fn i64s(&self, field: Field) -> Option<MultiValueIntFastFieldReader<i64>> {
self.fast_field_i64s.get(&field).cloned()
}
/// Returns the `bytes` fast field reader associated to `field`.
///
/// If `field` is not a bytes fast field, returns `None`.
pub fn bytes(&self, field: Field) -> Option<BytesFastFieldReader> {
self.fast_bytes.get(&field).cloned()
}
}

35
src/fieldnorm/code.rs
View File

@@ -10,27 +10,28 @@ pub fn fieldnorm_to_id(fieldnorm: u32) -> u8 {
.unwrap_or_else(|idx| idx - 1) as u8
}
#[cfg_attr(feature = "cargo-clippy", allow(clippy::unreadable_literal))]
pub const FIELD_NORMS_TABLE: [u32; 256] = [
0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25,
26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, 42, 44, 46, 48, 50, 52, 54, 56, 60,
64, 68, 72, 76, 80, 84, 88, 96, 104, 112, 120, 128, 136, 144, 152, 168, 184, 200, 216, 232,
248, 264, 280, 312, 344, 376, 408, 440, 472, 504, 536, 600, 664, 728, 792, 856, 920, 984, 1048,
1176, 1304, 1432, 1560, 1688, 1816, 1944, 2072, 2328, 2584, 2840, 3096, 3352, 3608, 3864, 4120,
4632, 5144, 5656, 6168, 6680, 7192, 7704, 8216, 9240, 10264, 11288, 12312, 13336, 14360, 15384,
16408, 18456, 20504, 22552, 24600, 26648, 28696, 30744, 32792, 36888, 40984, 45080, 49176,
53272, 57368, 61464, 65560, 73752, 81944, 90136, 98328, 106520, 114712, 122904, 131096, 147480,
163864, 180248, 196632, 213016, 229400, 245784, 262168, 294936, 327704, 360472, 393240, 426008,
458776, 491544, 524312, 589848, 655384, 720920, 786456, 851992, 917528, 983064, 1048600,
1179672, 1310744, 1441816, 1572888, 1703960, 1835032, 1966104, 2097176, 2359320, 2621464,
2883608, 3145752, 3407896, 3670040, 3932184, 4194328, 4718616, 5242904, 5767192, 6291480,
6815768, 7340056, 7864344, 8388632, 9437208, 10485784, 11534360, 12582936, 13631512, 14680088,
15728664, 16777240, 18874392, 20971544, 23068696, 25165848, 27263000, 29360152, 31457304,
33554456, 37748760, 41943064, 46137368, 50331672, 54525976, 58720280, 62914584, 67108888,
75497496, 83886104, 92274712, 100663320, 109051928, 117440536, 125829144, 134217752, 150994968,
167772184, 184549400, 201326616, 218103832, 234881048, 251658264, 268435480, 301989912,
335544344, 369098776, 402653208, 436207640, 469762072, 503316504, 536870936, 603979800,
671088664, 738197528, 805306392, 872415256, 939524120, 1006632984, 1073741848, 1207959576,
1342177304, 1476395032, 1610612760, 1744830488, 1879048216, 2013265944,
248, 264, 280, 312, 344, 376, 408, 440, 472, 504, 536, 600, 664, 728, 792, 856, 920, 984,
1_048, 1176, 1304, 1432, 1560, 1688, 1816, 1944, 2072, 2328, 2584, 2840, 3096, 3352, 3608,
3864, 4120, 4632, 5144, 5656, 6168, 6680, 7192, 7704, 8216, 9240, 10264, 11288, 12312, 13336,
14360, 15384, 16408, 18456, 20504, 22552, 24600, 26648, 28696, 30744, 32792, 36888, 40984,
45080, 49176, 53272, 57368, 61464, 65560, 73752, 81944, 90136, 98328, 106520, 114712, 122904,
131096, 147480, 163864, 180248, 196632, 213016, 229400, 245784, 262168, 294936, 327704, 360472,
393240, 426008, 458776, 491544, 524312, 589848, 655384, 720920, 786456, 851992, 917528, 983064,
1048600, 1179672, 1310744, 1441816, 1572888, 1703960, 1835032, 1966104, 2097176, 2359320,
2621464, 2883608, 3145752, 3407896, 3670040, 3932184, 4194328, 4718616, 5242904, 5767192,
6291480, 6815768, 7340056, 7864344, 8388632, 9437208, 10485784, 11534360, 12582936, 13631512,
14680088, 15728664, 16777240, 18874392, 20971544, 23068696, 25165848, 27263000, 29360152,
31457304, 33554456, 37748760, 41943064, 46137368, 50331672, 54525976, 58720280, 62914584,
67108888, 75497496, 83886104, 92274712, 100663320, 109051928, 117440536, 125829144, 134217752,
150994968, 167772184, 184549400, 201326616, 218103832, 234881048, 251658264, 268435480,
301989912, 335544344, 369098776, 402653208, 436207640, 469762072, 503316504, 536870936,
603979800, 671088664, 738197528, 805306392, 872415256, 939524120, 1006632984, 1073741848,
1207959576, 1342177304, 1476395032, 1610612760, 1744830488, 1879048216, 2013265944,
];
#[cfg(test)]

2
src/fieldnorm/mod.rs
View File

@@ -15,7 +15,7 @@
//! precompute computationally expensive functions of the fieldnorm
//! in a very short array.
//!
//! This trick is used by the [BM25 similarity]().
//! This trick is used by the BM25 similarity.
mod code;
mod reader;
mod serializer;

16
src/functional_test.rs
View File

@@ -2,7 +2,6 @@ use rand::thread_rng;
use std::collections::HashSet;
use rand::Rng;
use rand::distributions::Range;
use schema::*;
use Index;
use Searcher;
@@ -14,17 +13,16 @@ fn check_index_content(searcher: &Searcher, vals: &HashSet<u64>) {
#[test]
#[ignore]
#[cfg(feature = "mmap")]
fn test_indexing() {
let mut schema_builder = SchemaBuilder::default();
let mut schema_builder = Schema::builder();
let id_field = schema_builder.add_u64_field("id", INT_INDEXED);
let multiples_field = schema_builder.add_u64_field("multiples", INT_INDEXED);
let id_field = schema_builder.add_u64_field("id", INDEXED);
let multiples_field = schema_builder.add_u64_field("multiples", INDEXED);
let schema = schema_builder.build();
let index = Index::create_from_tempdir(schema).unwrap();
let reader = index.reader().unwrap();
let universe = Range::new(0u64, 20u64);
let mut rng = thread_rng();
let mut index_writer = index.writer_with_num_threads(3, 120_000_000).unwrap();
@@ -33,13 +31,13 @@ fn test_indexing() {
let mut uncommitted_docs: HashSet<u64> = HashSet::new();
for _ in 0..200 {
let random_val = rng.sample(&universe);
let random_val = rng.gen_range(0, 20);
if random_val == 0 {
index_writer.commit().expect("Commit failed");
committed_docs.extend(&uncommitted_docs);
uncommitted_docs.clear();
index.load_searchers().unwrap();
let searcher = index.searcher();
reader.reload().unwrap();
let searcher = reader.searcher();
// check that everything is correct.
check_index_content(&searcher, &committed_docs);
} else {

29
src/indexer/delete_queue.rs
View File

@@ -52,7 +52,8 @@ impl DeleteQueue {
//
// Past delete operations are not accessible.
pub fn cursor(&self) -> DeleteCursor {
let last_block = self.inner
let last_block = self
.inner
.read()
.expect("Read lock poisoned when opening delete queue cursor")
.last_block
@@ -92,7 +93,8 @@ impl DeleteQueue {
// be some unflushed operations.
//
fn flush(&self) -> Option<Arc<Block>> {
let mut self_wlock = self.inner
let mut self_wlock = self
.inner
.write()
.expect("Failed to acquire write lock on delete queue writer");
@@ -132,7 +134,8 @@ impl From<DeleteQueue> for NextBlock {
impl NextBlock {
fn next_block(&self) -> Option<Arc<Block>> {
{
let next_read_lock = self.0
let next_read_lock = self
.0
.read()
.expect("Failed to acquire write lock in delete queue");
if let InnerNextBlock::Closed(ref block) = *next_read_lock {
@@ -141,7 +144,8 @@ impl NextBlock {
}
let next_block;
{
let mut next_write_lock = self.0
let mut next_write_lock = self
.0
.write()
.expect("Failed to acquire write lock in delete queue");
match *next_write_lock {
@@ -182,19 +186,18 @@ impl DeleteCursor {
/// `opstamp >= target_opstamp`.
pub fn skip_to(&mut self, target_opstamp: u64) {
// TODO Can be optimize as we work with block.
#[cfg_attr(feature = "cargo-clippy", allow(while_let_loop))]
loop {
if let Some(operation) = self.get() {
if operation.opstamp >= target_opstamp {
break;
}
} else {
break;
}
while self.is_behind_opstamp(target_opstamp) {
self.advance();
}
}
#[cfg_attr(feature = "cargo-clippy", allow(clippy::wrong_self_convention))]
fn is_behind_opstamp(&mut self, target_opstamp: u64) -> bool {
self.get()
.map(|operation| operation.opstamp < target_opstamp)
.unwrap_or(false)
}
/// If the current block has been entirely
/// consumed, try to load the next one.
///

27
src/indexer/directory_lock.rs
View File

@@ -1,27 +0,0 @@
use core::LOCKFILE_FILEPATH;
use directory::error::OpenWriteError;
use Directory;
/// The directory lock is a mechanism used to
/// prevent the creation of two [`IndexWriter`](struct.IndexWriter.html)
///
/// Only one lock can exist at a time for a given directory.
/// The lock is release automatically on `Drop`.
pub struct DirectoryLock {
directory: Box<Directory>,
}
impl DirectoryLock {
pub fn lock(mut directory: Box<Directory>) -> Result<DirectoryLock, OpenWriteError> {
directory.open_write(&*LOCKFILE_FILEPATH)?;
Ok(DirectoryLock { directory })
}
}
impl Drop for DirectoryLock {
fn drop(&mut self) {
if let Err(e) = self.directory.delete(&*LOCKFILE_FILEPATH) {
error!("Failed to remove the lock file. {:?}", e);
}
}
}

423
src/indexer/index_writer.rs
View File

@@ -1,4 +1,4 @@
use super::operation::AddOperation;
use super::operation::{AddOperation, UserOperation};
use super::segment_updater::SegmentUpdater;
use super::PreparedCommit;
use bit_set::BitSet;
@@ -8,16 +8,16 @@ use core::SegmentComponent;
use core::SegmentId;
use core::SegmentMeta;
use core::SegmentReader;
use crossbeam_channel as channel;
use crossbeam::channel;
use directory::DirectoryLock;
use docset::DocSet;
use error::TantivyError;
use fastfield::write_delete_bitset;
use futures::sync::oneshot::Receiver;
use futures::{Canceled, Future};
use indexer::delete_queue::{DeleteCursor, DeleteQueue};
use indexer::doc_opstamp_mapping::DocToOpstampMapping;
use indexer::operation::DeleteOperation;
use indexer::stamper::Stamper;
use indexer::DirectoryLock;
use indexer::MergePolicy;
use indexer::SegmentEntry;
use indexer::SegmentWriter;
@@ -26,7 +26,8 @@ use schema::Document;
use schema::IndexRecordOption;
use schema::Term;
use std::mem;
use std::mem::swap;
use std::ops::Range;
use std::sync::Arc;
use std::thread;
use std::thread::JoinHandle;
use Result;
@@ -43,8 +44,8 @@ pub const HEAP_SIZE_MAX: usize = u32::max_value() as usize - MARGIN_IN_BYTES;
// reaches `PIPELINE_MAX_SIZE_IN_DOCS`
const PIPELINE_MAX_SIZE_IN_DOCS: usize = 10_000;
type DocumentSender = channel::Sender<AddOperation>;
type DocumentReceiver = channel::Receiver<AddOperation>;
type OperationSender = channel::Sender<Vec<AddOperation>>;
type OperationReceiver = channel::Receiver<Vec<AddOperation>>;
/// Split the thread memory budget into
/// - the heap size
@@ -52,16 +53,19 @@ type DocumentReceiver = channel::Receiver<AddOperation>;
///
/// Returns (the heap size in bytes, the hash table size in number of bits)
fn initial_table_size(per_thread_memory_budget: usize) -> usize {
assert!(per_thread_memory_budget > 1_000);
let table_size_limit: usize = per_thread_memory_budget / 3;
(1..)
.into_iter()
if let Some(limit) = (1..)
.take_while(|num_bits: &usize| compute_table_size(*num_bits) < table_size_limit)
.last()
.expect(&format!(
{
limit.min(19) // we cap it at 2^19 = 512K.
} else {
unreachable!(
"Per thread memory is too small: {}",
per_thread_memory_budget
))
.min(19) // we cap it at 512K
);
}
}
/// `IndexWriter` is the user entry-point to add document to an index.
@@ -81,8 +85,8 @@ pub struct IndexWriter {
workers_join_handle: Vec<JoinHandle<Result<()>>>,
document_receiver: DocumentReceiver,
document_sender: DocumentSender,
operation_receiver: OperationReceiver,
operation_sender: OperationSender,
segment_updater: SegmentUpdater,
@@ -129,7 +133,7 @@ pub fn open_index_writer(
let err_msg = format!("The heap size per thread cannot exceed {}", HEAP_SIZE_MAX);
return Err(TantivyError::InvalidArgument(err_msg));
}
let (document_sender, document_receiver): (DocumentSender, DocumentReceiver) =
let (document_sender, document_receiver): (OperationSender, OperationReceiver) =
channel::bounded(PIPELINE_MAX_SIZE_IN_DOCS);
let delete_queue = DeleteQueue::new();
@@ -139,7 +143,7 @@ pub fn open_index_writer(
let stamper = Stamper::new(current_opstamp);
let segment_updater =
SegmentUpdater::new(index.clone(), stamper.clone(), &delete_queue.cursor())?;
SegmentUpdater::create(index.clone(), stamper.clone(), &delete_queue.cursor())?;
let mut index_writer = IndexWriter {
_directory_lock: Some(directory_lock),
@@ -147,8 +151,8 @@ pub fn open_index_writer(
heap_size_in_bytes_per_thread,
index: index.clone(),
document_receiver,
document_sender,
operation_receiver: document_receiver,
operation_sender: document_sender,
segment_updater,
@@ -177,7 +181,7 @@ pub fn compute_deleted_bitset(
) -> Result<bool> {
let mut might_have_changed = false;
#[cfg_attr(feature = "cargo-clippy", allow(while_let_loop))]
#[cfg_attr(feature = "cargo-clippy", allow(clippy::while_let_loop))]
loop {
if let Some(delete_op) = delete_cursor.get() {
if delete_op.opstamp > target_opstamp {
@@ -255,7 +259,7 @@ pub fn advance_deletes(
write_delete_bitset(&delete_bitset, &mut delete_file)?;
}
}
segment_entry.set_meta((*segment.meta()).clone());
segment_entry.set_meta(segment.meta().clone());
Ok(())
}
@@ -263,7 +267,7 @@ fn index_documents(
memory_budget: usize,
segment: &Segment,
generation: usize,
document_iterator: &mut Iterator<Item = AddOperation>,
document_iterator: &mut Iterator<Item = Vec<AddOperation>>,
segment_updater: &mut SegmentUpdater,
mut delete_cursor: DeleteCursor,
) -> Result<bool> {
@@ -271,11 +275,11 @@ fn index_documents(
let segment_id = segment.id();
let table_size = initial_table_size(memory_budget);
let mut segment_writer = SegmentWriter::for_segment(table_size, segment.clone(), &schema)?;
for doc in document_iterator {
segment_writer.add_document(doc, &schema)?;
for documents in document_iterator {
for doc in documents {
segment_writer.add_document(doc, &schema)?;
}
let mem_usage = segment_writer.mem_usage();
if mem_usage >= memory_budget - MARGIN_IN_BYTES {
info!(
"Buffer limit reached, flushing segment with maxdoc={}.",
@@ -301,25 +305,28 @@ fn index_documents(
let last_docstamp: u64 = *(doc_opstamps.last().unwrap());
let doc_to_opstamps = DocToOpstampMapping::from(doc_opstamps);
let segment_reader = SegmentReader::open(segment)?;
let mut deleted_bitset = BitSet::with_capacity(num_docs as usize);
let may_have_deletes = compute_deleted_bitset(
&mut deleted_bitset,
&segment_reader,
&mut delete_cursor,
&doc_to_opstamps,
last_docstamp,
)?;
let segment_entry = SegmentEntry::new(segment_meta, delete_cursor, {
let delete_bitset_opt = if delete_cursor.get().is_some() {
let doc_to_opstamps = DocToOpstampMapping::from(doc_opstamps);
let segment_reader = SegmentReader::open(segment)?;
let mut deleted_bitset = BitSet::with_capacity(num_docs as usize);
let may_have_deletes = compute_deleted_bitset(
&mut deleted_bitset,
&segment_reader,
&mut delete_cursor,
&doc_to_opstamps,
last_docstamp,
)?;
if may_have_deletes {
Some(deleted_bitset)
} else {
None
}
});
} else {
// if there are no delete operation in the queue, no need
// to even open the segment.
None
};
let segment_entry = SegmentEntry::new(segment_meta, delete_cursor, delete_bitset_opt);
Ok(segment_updater.add_segment(generation, segment_entry))
}
@@ -328,7 +335,7 @@ impl IndexWriter {
pub fn wait_merging_threads(mut self) -> Result<()> {
// this will stop the indexing thread,
// dropping the last reference to the segment_updater.
drop(self.document_sender);
drop(self.operation_sender);
let former_workers_handles = mem::replace(&mut self.workers_join_handle, vec![]);
for join_handle in former_workers_handles {
@@ -341,7 +348,8 @@ impl IndexWriter {
}
drop(self.workers_join_handle);
let result = self.segment_updater
let result = self
.segment_updater
.wait_merging_thread()
.map_err(|_| TantivyError::ErrorInThread("Failed to join merging thread.".into()));
@@ -360,20 +368,23 @@ impl IndexWriter {
.add_segment(self.generation, segment_entry);
}
/// *Experimental & Advanced API* Creates a new segment.
/// and marks it as currently in write.
/// Creates a new segment.
///
/// This method is useful only for users trying to do complex
/// operations, like converting an index format to another.
///
/// It is safe to start writing file associated to the new `Segment`.
/// These will not be garbage collected as long as an instance object of
/// `SegmentMeta` object associated to the new `Segment` is "alive".
pub fn new_segment(&self) -> Segment {
self.segment_updater.new_segment()
self.index.new_segment()
}
/// Spawns a new worker thread for indexing.
/// The thread consumes documents from the pipeline.
///
fn add_indexing_worker(&mut self) -> Result<()> {
let document_receiver_clone = self.document_receiver.clone();
let document_receiver_clone = self.operation_receiver.clone();
let mut segment_updater = self.segment_updater.clone();
let generation = self.generation;
@@ -381,9 +392,10 @@ impl IndexWriter {
let mut delete_cursor = self.delete_queue.cursor();
let mem_budget = self.heap_size_in_bytes_per_thread;
let index = self.index.clone();
let join_handle: JoinHandle<Result<()>> = thread::Builder::new()
.name(format!(
"indexing thread {} for gen {}",
"thrd-tantivy-index{}-gen{}",
self.worker_id, generation
))
.spawn(move || {
@@ -398,15 +410,19 @@ impl IndexWriter {
// this is a valid guarantee as the
// peeked document now belongs to
// our local iterator.
if let Some(operation) = document_iterator.peek() {
delete_cursor.skip_to(operation.opstamp);
if let Some(operations) = document_iterator.peek() {
if let Some(first) = operations.first() {
delete_cursor.skip_to(first.opstamp);
} else {
return Ok(());
}
} else {
// No more documents.
// Happens when there is a commit, or if the `IndexWriter`
// was dropped.
return Ok(());
}
let segment = segment_updater.new_segment();
let segment = index.new_segment();
index_documents(
mem_budget,
&segment,
@@ -423,7 +439,7 @@ impl IndexWriter {
}
/// Accessor to the merge policy.
pub fn get_merge_policy(&self) -> Box<MergePolicy> {
pub fn get_merge_policy(&self) -> Arc<Box<MergePolicy>> {
self.segment_updater.get_merge_policy()
}
@@ -448,7 +464,10 @@ impl IndexWriter {
/// Merges a given list of segments
///
/// `segment_ids` is required to be non-empty.
pub fn merge(&mut self, segment_ids: &[SegmentId]) -> Result<Receiver<SegmentMeta>> {
pub fn merge(
&mut self,
segment_ids: &[SegmentId],
) -> Result<impl Future<Item = SegmentMeta, Error = Canceled>> {
self.segment_updater.start_merge(segment_ids)
}
@@ -460,14 +479,11 @@ impl IndexWriter {
/// when no documents are remaining.
///
/// Returns the former segment_ready channel.
fn recreate_document_channel(&mut self) -> DocumentReceiver {
let (mut document_sender, mut document_receiver): (
DocumentSender,
DocumentReceiver,
) = channel::bounded(PIPELINE_MAX_SIZE_IN_DOCS);
swap(&mut self.document_sender, &mut document_sender);
swap(&mut self.document_receiver, &mut document_receiver);
document_receiver
fn recreate_document_channel(&mut self) -> OperationReceiver {
let (document_sender, document_receiver): (OperationSender, OperationReceiver) =
channel::bounded(PIPELINE_MAX_SIZE_IN_DOCS);
mem::replace(&mut self.operation_sender, document_sender);
mem::replace(&mut self.operation_receiver, document_receiver)
}
/// Rollback to the last commit
@@ -485,10 +501,11 @@ impl IndexWriter {
// segment updates will be ignored.
self.segment_updater.kill();
let document_receiver = self.document_receiver.clone();
let document_receiver = self.operation_receiver.clone();
// take the directory lock to create a new index_writer.
let directory_lock = self._directory_lock
let directory_lock = self
._directory_lock
.take()
.expect("The IndexWriter does not have any lock. This is a bug, please report.");
@@ -553,17 +570,12 @@ impl IndexWriter {
// and recreate a new one channels.
self.recreate_document_channel();
let mut former_workers_join_handle = Vec::new();
swap(
&mut former_workers_join_handle,
&mut self.workers_join_handle,
);
let former_workers_join_handle = mem::replace(&mut self.workers_join_handle, Vec::new());
for worker_handle in former_workers_join_handle {
let indexing_worker_result = worker_handle
.join()
.map_err(|e| TantivyError::ErrorInThread(format!("{:?}", e)))?;
indexing_worker_result?;
// add a new worker for the next generation.
self.add_indexing_worker()?;
@@ -636,37 +648,198 @@ impl IndexWriter {
pub fn add_document(&mut self, document: Document) -> u64 {
let opstamp = self.stamper.stamp();
let add_operation = AddOperation { opstamp, document };
self.document_sender.send(add_operation);
let send_result = self.operation_sender.send(vec![add_operation]);
if let Err(e) = send_result {
panic!("Failed to index document. Sending to indexing channel failed. This probably means all of the indexing threads have panicked. {:?}", e);
}
opstamp
}
/// Gets a range of stamps from the stamper and "pops" the last stamp
/// from the range returning a tuple of the last optstamp and the popped
/// range.
///
/// The total number of stamps generated by this method is `count + 1`;
/// each operation gets a stamp from the `stamps` iterator and `last_opstamp`
/// is for the batch itself.
fn get_batch_opstamps(&mut self, count: u64) -> (u64, Range<u64>) {
let Range { start, end } = self.stamper.stamps(count + 1u64);
let last_opstamp = end - 1;
let stamps = Range {
start,
end: last_opstamp,
};
(last_opstamp, stamps)
}
/// Runs a group of document operations ensuring that the operations are
/// assigned contigous u64 opstamps and that add operations of the same
/// group are flushed into the same segment.
///
/// If the indexing pipeline is full, this call may block.
///
/// Each operation of the given `user_operations` will receive an in-order,
/// contiguous u64 opstamp. The entire batch itself is also given an
/// opstamp that is 1 greater than the last given operation. This
/// `batch_opstamp` is the return value of `run`. An empty group of
/// `user_operations`, an empty `Vec<UserOperation>`, still receives
/// a valid opstamp even though no changes were _actually_ made to the index.
///
/// Like adds and deletes (see `IndexWriter.add_document` and
/// `IndexWriter.delete_term`), the changes made by calling `run` will be
/// visible to readers only after calling `commit()`.
pub fn run(&mut self, user_operations: Vec<UserOperation>) -> u64 {
let count = user_operations.len() as u64;
if count == 0 {
return self.stamper.stamp();
}
let (batch_opstamp, stamps) = self.get_batch_opstamps(count);
let mut adds: Vec<AddOperation> = Vec::new();
for (user_op, opstamp) in user_operations.into_iter().zip(stamps) {
match user_op {
UserOperation::Delete(term) => {
let delete_operation = DeleteOperation { opstamp, term };
self.delete_queue.push(delete_operation);
}
UserOperation::Add(document) => {
let add_operation = AddOperation { opstamp, document };
adds.push(add_operation);
}
}
}
let send_result = self.operation_sender.send(adds);
if let Err(e) = send_result {
panic!("Failed to index document. Sending to indexing channel failed. This probably means all of the indexing threads have panicked. {:?}", e);
};
batch_opstamp
}
}
#[cfg(test)]
mod tests {
use super::super::operation::UserOperation;
use super::initial_table_size;
use collector::TopDocs;
use directory::error::LockError;
use error::*;
use indexer::NoMergePolicy;
use schema::{self, Document};
use query::TermQuery;
use schema::{self, IndexRecordOption};
use Index;
use ReloadPolicy;
use Term;
#[test]
fn test_lockfile_stops_duplicates() {
let schema_builder = schema::SchemaBuilder::default();
fn test_operations_group() {
// an operations group with 2 items should cause 3 opstamps 0, 1, and 2.
let mut schema_builder = schema::Schema::builder();
let text_field = schema_builder.add_text_field("text", schema::TEXT);
let index = Index::create_in_ram(schema_builder.build());
let _index_writer = index.writer(40_000_000).unwrap();
match index.writer(40_000_000) {
Err(TantivyError::FileAlreadyExists(_)) => {}
_ => panic!("Expected FileAlreadyExists error"),
let mut index_writer = index.writer_with_num_threads(1, 3_000_000).unwrap();
let operations = vec![
UserOperation::Add(doc!(text_field=>"a")),
UserOperation::Add(doc!(text_field=>"b")),
];
let batch_opstamp1 = index_writer.run(operations);
assert_eq!(batch_opstamp1, 2u64);
}
#[test]
fn test_ordered_batched_operations() {
// * one delete for `doc!(field=>"a")`
// * one add for `doc!(field=>"a")`
// * one add for `doc!(field=>"b")`
// * one delete for `doc!(field=>"b")`
// after commit there is one doc with "a" and 0 doc with "b"
let mut schema_builder = schema::Schema::builder();
let text_field = schema_builder.add_text_field("text", schema::TEXT);
let index = Index::create_in_ram(schema_builder.build());
let reader = index
.reader_builder()
.reload_policy(ReloadPolicy::Manual)
.try_into()
.unwrap();
let mut index_writer = index.writer_with_num_threads(1, 3_000_000).unwrap();
let a_term = Term::from_field_text(text_field, "a");
let b_term = Term::from_field_text(text_field, "b");
let operations = vec![
UserOperation::Delete(a_term),
UserOperation::Add(doc!(text_field=>"a")),
UserOperation::Add(doc!(text_field=>"b")),
UserOperation::Delete(b_term),
];
index_writer.run(operations);
index_writer.commit().expect("failed to commit");
reader.reload().expect("failed to load searchers");
let a_term = Term::from_field_text(text_field, "a");
let b_term = Term::from_field_text(text_field, "b");
let a_query = TermQuery::new(a_term, IndexRecordOption::Basic);
let b_query = TermQuery::new(b_term, IndexRecordOption::Basic);
let searcher = reader.searcher();
let a_docs = searcher
.search(&a_query, &TopDocs::with_limit(1))
.expect("search for a failed");
let b_docs = searcher
.search(&b_query, &TopDocs::with_limit(1))
.expect("search for b failed");
assert_eq!(a_docs.len(), 1);
assert_eq!(b_docs.len(), 0);
}
#[test]
fn test_empty_operations_group() {
let schema_builder = schema::Schema::builder();
let index = Index::create_in_ram(schema_builder.build());
let mut index_writer = index.writer(3_000_000).unwrap();
let operations1 = vec![];
let batch_opstamp1 = index_writer.run(operations1);
assert_eq!(batch_opstamp1, 0u64);
let operations2 = vec![];
let batch_opstamp2 = index_writer.run(operations2);
assert_eq!(batch_opstamp2, 1u64);
}
#[test]
fn test_lockfile_stops_duplicates() {
let schema_builder = schema::Schema::builder();
let index = Index::create_in_ram(schema_builder.build());
let _index_writer = index.writer(3_000_000).unwrap();
match index.writer(3_000_000) {
Err(TantivyError::LockFailure(LockError::LockBusy, _)) => {}
_ => panic!("Expected a `LockFailure` error"),
}
}
#[test]
fn test_lockfile_already_exists_error_msg() {
let schema_builder = schema::Schema::builder();
let index = Index::create_in_ram(schema_builder.build());
let _index_writer = index.writer_with_num_threads(1, 3_000_000).unwrap();
match index.writer_with_num_threads(1, 3_000_000) {
Err(err) => {
let err_msg = err.to_string();
assert!(err_msg.contains("already an `IndexWriter`"));
}
_ => panic!("Expected LockfileAlreadyExists error"),
}
}
#[test]
fn test_set_merge_policy() {
let schema_builder = schema::SchemaBuilder::default();
let schema_builder = schema::Schema::builder();
let index = Index::create_in_ram(schema_builder.build());
let index_writer = index.writer(40_000_000).unwrap();
let index_writer = index.writer(3_000_000).unwrap();
assert_eq!(
format!("{:?}", index_writer.get_merge_policy()),
"LogMergePolicy { min_merge_size: 8, min_layer_size: 10000, \
@@ -682,24 +855,28 @@ mod tests {
#[test]
fn test_lockfile_released_on_drop() {
let schema_builder = schema::SchemaBuilder::default();
let schema_builder = schema::Schema::builder();
let index = Index::create_in_ram(schema_builder.build());
{
let _index_writer = index.writer(40_000_000).unwrap();
let _index_writer = index.writer(3_000_000).unwrap();
// the lock should be released when the
// index_writer leaves the scope.
}
let _index_writer_two = index.writer(40_000_000).unwrap();
let _index_writer_two = index.writer(3_000_000).unwrap();
}
#[test]
fn test_commit_and_rollback() {
let mut schema_builder = schema::SchemaBuilder::default();
let mut schema_builder = schema::Schema::builder();
let text_field = schema_builder.add_text_field("text", schema::TEXT);
let index = Index::create_in_ram(schema_builder.build());
let reader = index
.reader_builder()
.reload_policy(ReloadPolicy::Manual)
.try_into()
.unwrap();
let num_docs_containing = |s: &str| {
let searcher = index.searcher();
let searcher = reader.searcher();
let term = Term::from_field_text(text_field, s);
searcher.doc_freq(&term)
};
@@ -709,54 +886,54 @@ mod tests {
let mut index_writer = index.writer(3_000_000).unwrap();
index_writer.add_document(doc!(text_field=>"a"));
index_writer.rollback().unwrap();
assert_eq!(index_writer.commit_opstamp(), 0u64);
assert_eq!(num_docs_containing("a"), 0);
{
index_writer.add_document(doc!(text_field=>"b"));
index_writer.add_document(doc!(text_field=>"c"));
}
assert_eq!(index_writer.commit().unwrap(), 2u64);
index.load_searchers().unwrap();
assert!(index_writer.commit().is_ok());
reader.reload().unwrap();
assert_eq!(num_docs_containing("a"), 0);
assert_eq!(num_docs_containing("b"), 1);
assert_eq!(num_docs_containing("c"), 1);
}
index.load_searchers().unwrap();
index.searcher();
reader.reload().unwrap();
reader.searcher();
}
#[test]
fn test_with_merges() {
let mut schema_builder = schema::SchemaBuilder::default();
let mut schema_builder = schema::Schema::builder();
let text_field = schema_builder.add_text_field("text", schema::TEXT);
let index = Index::create_in_ram(schema_builder.build());
let reader = index
.reader_builder()
.reload_policy(ReloadPolicy::Manual)
.try_into()
.unwrap();
let num_docs_containing = |s: &str| {
let searcher = index.searcher();
let term_a = Term::from_field_text(text_field, s);
searcher.doc_freq(&term_a)
reader.searcher().doc_freq(&term_a)
};
{
// writing the segment
let mut index_writer = index.writer(12_000_000).unwrap();
// create 8 segments with 100 tiny docs
for _doc in 0..100 {
let mut doc = Document::default();
doc.add_text(text_field, "a");
index_writer.add_document(doc);
index_writer.add_document(doc!(text_field=>"a"));
}
index_writer.commit().expect("commit failed");
for _doc in 0..100 {
let mut doc = Document::default();
doc.add_text(text_field, "a");
index_writer.add_document(doc);
index_writer.add_document(doc!(text_field=>"a"));
}
// this should create 8 segments and trigger a merge.
// this should create 8 segments and trigger a merge.
index_writer.commit().expect("commit failed");
index_writer
.wait_merging_threads()
.expect("waiting merging thread failed");
index.load_searchers().unwrap();
reader.reload().unwrap();
assert_eq!(num_docs_containing("a"), 200);
assert!(index.searchable_segments().unwrap().len() < 8);
@@ -765,7 +942,7 @@ mod tests {
#[test]
fn test_prepare_with_commit_message() {
let mut schema_builder = schema::SchemaBuilder::default();
let mut schema_builder = schema::Schema::builder();
let text_field = schema_builder.add_text_field("text", schema::TEXT);
let index = Index::create_in_ram(schema_builder.build());
@@ -779,7 +956,6 @@ mod tests {
{
let mut prepared_commit = index_writer.prepare_commit().expect("commit failed");
prepared_commit.set_payload("first commit");
assert_eq!(prepared_commit.opstamp(), 100);
prepared_commit.commit().expect("commit failed");
}
{
@@ -799,7 +975,7 @@ mod tests {
#[test]
fn test_prepare_but_rollback() {
let mut schema_builder = schema::SchemaBuilder::default();
let mut schema_builder = schema::Schema::builder();
let text_field = schema_builder.add_text_field("text", schema::TEXT);
let index = Index::create_in_ram(schema_builder.build());
@@ -813,7 +989,6 @@ mod tests {
{
let mut prepared_commit = index_writer.prepare_commit().expect("commit failed");
prepared_commit.set_payload("first commit");
assert_eq!(prepared_commit.opstamp(), 100);
prepared_commit.abort().expect("commit failed");
}
{
@@ -825,11 +1000,15 @@ mod tests {
}
index_writer.commit().unwrap();
}
index.load_searchers().unwrap();
let num_docs_containing = |s: &str| {
let searcher = index.searcher();
let term_a = Term::from_field_text(text_field, s);
searcher.doc_freq(&term_a)
index
.reader_builder()
.reload_policy(ReloadPolicy::Manual)
.try_into()
.unwrap()
.searcher()
.doc_freq(&term_a)
};
assert_eq!(num_docs_containing("a"), 0);
assert_eq!(num_docs_containing("b"), 100);
@@ -837,10 +1016,36 @@ mod tests {
#[test]
fn test_hashmap_size() {
assert_eq!(initial_table_size(100_000), 12);
assert_eq!(initial_table_size(1_000_000), 15);
assert_eq!(initial_table_size(10_000_000), 18);
assert_eq!(initial_table_size(100_000), 11);
assert_eq!(initial_table_size(1_000_000), 14);
assert_eq!(initial_table_size(10_000_000), 17);
assert_eq!(initial_table_size(1_000_000_000), 19);
}
#[cfg(not(feature = "no_fail"))]
#[test]
fn test_write_commit_fails() {
use fail;
let mut schema_builder = schema::Schema::builder();
let text_field = schema_builder.add_text_field("text", schema::TEXT);
let index = Index::create_in_ram(schema_builder.build());
let mut index_writer = index.writer_with_num_threads(1, 3_000_000).unwrap();
for _ in 0..100 {
index_writer.add_document(doc!(text_field => "a"));
}
index_writer.commit().unwrap();
fail::cfg("RAMDirectory::atomic_write", "return(error_write_failed)").unwrap();
for _ in 0..100 {
index_writer.add_document(doc!(text_field => "b"));
}
assert!(index_writer.commit().is_err());
let num_docs_containing = |s: &str| {
let term_a = Term::from_field_text(text_field, s);
index.reader().unwrap().searcher().doc_freq(&term_a)
};
assert_eq!(num_docs_containing("a"), 100);
assert_eq!(num_docs_containing("b"), 0);
fail::cfg("RAMDirectory::atomic_write", "off").unwrap();
}
}

2
src/indexer/log_merge_policy.rs
View File

@@ -52,7 +52,7 @@ impl MergePolicy for LogMergePolicy {
let mut size_sorted_tuples = segments
.iter()
.map(|x| x.num_docs())
.map(SegmentMeta::num_docs)
.enumerate()
.collect::<Vec<(usize, u32)>>();

64
src/indexer/merge_operation.rs Normal file
View File

@@ -0,0 +1,64 @@
use census::{Inventory, TrackedObject};
use std::collections::HashSet;
use SegmentId;
#[derive(Default)]
pub struct MergeOperationInventory(Inventory<InnerMergeOperation>);
impl MergeOperationInventory {
pub fn segment_in_merge(&self) -> HashSet<SegmentId> {
let mut segment_in_merge = HashSet::default();
for merge_op in self.0.list() {
for &segment_id in &merge_op.segment_ids {
segment_in_merge.insert(segment_id);
}
}
segment_in_merge
}
}
/// A `MergeOperation` has two role.
/// It carries all of the information required to describe a merge :
/// - `target_opstamp` is the opstamp up to which we want to consume the
/// delete queue and reflect their deletes.
/// - `segment_ids` is the list of segment to be merged.
///
/// The second role is to ensure keep track of the fact that these
/// segments are in merge and avoid starting a merge operation that
/// may conflict with this one.
///
/// This works by tracking merge operations. When considering computing
/// merge candidates, we simply list tracked merge operations and remove
/// their segments from possible merge candidates.
pub struct MergeOperation {
inner: TrackedObject<InnerMergeOperation>,
}
struct InnerMergeOperation {
target_opstamp: u64,
segment_ids: Vec<SegmentId>,
}
impl MergeOperation {
pub fn new(
inventory: &MergeOperationInventory,
target_opstamp: u64,
segment_ids: Vec<SegmentId>,
) -> MergeOperation {
let inner_merge_operation = InnerMergeOperation {
target_opstamp,
segment_ids,
};
MergeOperation {
inner: inventory.0.track(inner_merge_operation),
}
}
pub fn target_opstamp(&self) -> u64 {
self.inner.target_opstamp
}
pub fn segment_ids(&self) -> &[SegmentId] {
&self.inner.segment_ids[..]
}
}

17
src/indexer/merge_policy.rs
View File

@@ -11,7 +11,7 @@ pub struct MergeCandidate(pub Vec<SegmentId>);
///
/// Every time a the list of segments changes, the segment updater
/// asks the merge policy if some segments should be merged.
pub trait MergePolicy: MergePolicyClone + marker::Send + marker::Sync + Debug {
pub trait MergePolicy: marker::Send + marker::Sync + Debug {
/// Given the list of segment metas, returns the list of merge candidates.
///
/// This call happens on the segment updater thread, and will block
@@ -19,21 +19,6 @@ pub trait MergePolicy: MergePolicyClone + marker::Send + marker::Sync + Debug {
fn compute_merge_candidates(&self, segments: &[SegmentMeta]) -> Vec<MergeCandidate>;
}
/// MergePolicyClone
pub trait MergePolicyClone {
/// Returns a boxed clone of the MergePolicy.
fn box_clone(&self) -> Box<MergePolicy>;
}
impl<T> MergePolicyClone for T
where
T: 'static + MergePolicy + Clone,
{
fn box_clone(&self) -> Box<MergePolicy> {
Box::new(self.clone())
}
}
/// Never merge segments.
#[derive(Debug, Clone)]
pub struct NoMergePolicy;

531
src/indexer/merger.rs
View File

@@ -1,7 +1,9 @@
use common::MAX_DOC_LIMIT;
use core::Segment;
use core::SegmentReader;
use core::SerializableSegment;
use docset::DocSet;
use fastfield::BytesFastFieldReader;
use fastfield::DeleteBitSet;
use fastfield::FastFieldReader;
use fastfield::FastFieldSerializer;
@@ -23,6 +25,7 @@ use termdict::TermMerger;
use termdict::TermOrdinal;
use DocId;
use Result;
use TantivyError;
fn compute_total_num_tokens(readers: &[SegmentReader], field: Field) -> u64 {
let mut total_tokens = 0u64;
@@ -46,7 +49,7 @@ fn compute_total_num_tokens(readers: &[SegmentReader], field: Field) -> u64 {
.cloned()
.enumerate()
.map(|(fieldnorm_ord, count)| {
count as u64 * FieldNormReader::id_to_fieldnorm(fieldnorm_ord as u8) as u64
count as u64 * u64::from(FieldNormReader::id_to_fieldnorm(fieldnorm_ord as u8))
})
.sum::<u64>()
}
@@ -70,7 +73,7 @@ fn compute_min_max_val(
// some deleted documents,
// we need to recompute the max / min
(0..max_doc)
.filter(|doc_id| !delete_bitset.is_deleted(*doc_id))
.filter(|doc_id| delete_bitset.is_alive(*doc_id))
.map(|doc_id| u64_reader.get(doc_id))
.minmax()
.into_option()
@@ -111,7 +114,7 @@ impl TermOrdinalMapping {
.iter()
.flat_map(|term_ordinals| term_ordinals.iter().cloned().max())
.max()
.unwrap_or(TermOrdinal::default())
.unwrap_or_else(TermOrdinal::default)
}
}
@@ -150,6 +153,14 @@ impl IndexMerger {
readers.push(reader);
}
}
if max_doc >= MAX_DOC_LIMIT {
let err_msg = format!(
"The segment resulting from this merge would have {} docs,\
which exceeds the limit {}.",
max_doc, MAX_DOC_LIMIT
);
return Err(TantivyError::InvalidArgument(err_msg));
}
Ok(IndexMerger {
schema,
readers,
@@ -190,21 +201,21 @@ impl IndexMerger {
`term_ordinal_mapping`.");
self.write_hierarchical_facet_field(
field,
term_ordinal_mapping,
&term_ordinal_mapping,
fast_field_serializer,
)?;
}
FieldType::U64(ref options) | FieldType::I64(ref options) => {
match options.get_fastfield_cardinality() {
Some(Cardinality::SingleValue) => {
self.write_single_fast_field(field, fast_field_serializer)?;
}
Some(Cardinality::MultiValues) => {
self.write_multi_fast_field(field, fast_field_serializer)?;
}
None => {}
FieldType::U64(ref options)
| FieldType::I64(ref options)
| FieldType::Date(ref options) => match options.get_fastfield_cardinality() {
Some(Cardinality::SingleValue) => {
self.write_single_fast_field(field, fast_field_serializer)?;
}
}
Some(Cardinality::MultiValues) => {
self.write_multi_fast_field(field, fast_field_serializer)?;
}
None => {}
},
FieldType::Str(_) => {
// We don't handle str fast field for the moment
// They can be implemented using what is done
@@ -229,7 +240,10 @@ impl IndexMerger {
let mut max_value = u64::min_value();
for reader in &self.readers {
let u64_reader: FastFieldReader<u64> = reader.fast_field_reader(field)?;
let u64_reader: FastFieldReader<u64> = reader
.fast_fields()
.u64_lenient(field)
.expect("Failed to find a reader for single fast field. This is a tantivy bug and it should never happen.");
if let Some((seg_min_val, seg_max_val)) =
compute_min_max_val(&u64_reader, reader.max_doc(), reader.delete_bitset())
{
@@ -272,24 +286,28 @@ impl IndexMerger {
fast_field_serializer: &mut FastFieldSerializer,
) -> Result<()> {
let mut total_num_vals = 0u64;
let mut u64s_readers: Vec<MultiValueIntFastFieldReader<u64>> = Vec::new();
// In the first pass, we compute the total number of vals.
//
// This is required by the bitpacker, as it needs to know
// what should be the bit length use for bitpacking.
for reader in &self.readers {
let idx_reader = reader.fast_field_reader_with_idx::<u64>(field, 0)?;
let u64s_reader = reader.fast_fields()
.u64s_lenient(field)
.expect("Failed to find index for multivalued field. This is a bug in tantivy, please report.");
if let Some(delete_bitset) = reader.delete_bitset() {
for doc in 0u32..reader.max_doc() {
if !delete_bitset.is_deleted(doc) {
let start = idx_reader.get(doc);
let end = idx_reader.get(doc + 1);
total_num_vals += end - start;
if delete_bitset.is_alive(doc) {
let num_vals = u64s_reader.num_vals(doc) as u64;
total_num_vals += num_vals;
}
}
} else {
total_num_vals += idx_reader.max_value();
total_num_vals += u64s_reader.total_num_vals();
}
u64s_readers.push(u64s_reader);
}
// We can now create our `idx` serializer, and in a second pass,
@@ -297,13 +315,10 @@ impl IndexMerger {
let mut serialize_idx =
fast_field_serializer.new_u64_fast_field_with_idx(field, 0, total_num_vals, 0)?;
let mut idx = 0;
for reader in &self.readers {
let idx_reader = reader.fast_field_reader_with_idx::<u64>(field, 0)?;
for doc in reader.doc_ids_alive() {
for (segment_reader, u64s_reader) in self.readers.iter().zip(&u64s_readers) {
for doc in segment_reader.doc_ids_alive() {
serialize_idx.add_val(idx)?;
let start = idx_reader.get(doc);
let end = idx_reader.get(doc + 1);
idx += end - start;
idx += u64s_reader.num_vals(doc) as u64;
}
}
serialize_idx.add_val(idx)?;
@@ -314,7 +329,7 @@ impl IndexMerger {
fn write_hierarchical_facet_field(
&self,
field: Field,
term_ordinal_mappings: TermOrdinalMapping,
term_ordinal_mappings: &TermOrdinalMapping,
fast_field_serializer: &mut FastFieldSerializer,
) -> Result<()> {
// Multifastfield consists in 2 fastfields.
@@ -334,8 +349,10 @@ impl IndexMerger {
for (segment_ord, segment_reader) in self.readers.iter().enumerate() {
let term_ordinal_mapping: &[TermOrdinal] =
term_ordinal_mappings.get_segment(segment_ord);
let ff_reader: MultiValueIntFastFieldReader<u64> =
segment_reader.multi_fast_field_reader(field)?;
let ff_reader: MultiValueIntFastFieldReader<u64> = segment_reader
.fast_fields()
.u64s(field)
.expect("Could not find multivalued u64 fast value reader.");
// TODO optimize if no deletes
for doc in segment_reader.doc_ids_alive() {
ff_reader.get_vals(doc, &mut vals);
@@ -367,6 +384,8 @@ impl IndexMerger {
let mut vals = Vec::with_capacity(100);
let mut ff_readers = Vec::new();
// Our values are bitpacked and we need to know what should be
// our bitwidth and our minimum value before serializing any values.
//
@@ -375,7 +394,10 @@ impl IndexMerger {
// maximum value and initialize our Serializer.
for reader in &self.readers {
let ff_reader: MultiValueIntFastFieldReader<u64> =
reader.multi_fast_field_reader(field)?;
reader.fast_fields().u64s_lenient(field).expect(
"Failed to find multivalued fast field reader. This is a bug in \
tantivy. Please report.",
);
for doc in reader.doc_ids_alive() {
ff_reader.get_vals(doc, &mut vals);
for &val in &vals {
@@ -383,6 +405,7 @@ impl IndexMerger {
max_value = cmp::max(val, max_value);
}
}
ff_readers.push(ff_reader);
// TODO optimize when no deletes
}
@@ -393,11 +416,9 @@ impl IndexMerger {
// We can now initialize our serializer, and push it the different values
{
let mut serialize_vals =
fast_field_serializer.new_u64_fast_field_with_idx(field, min_value, max_value, 1)?;
for reader in &self.readers {
let ff_reader: MultiValueIntFastFieldReader<u64> =
reader.multi_fast_field_reader(field)?;
let mut serialize_vals = fast_field_serializer
.new_u64_fast_field_with_idx(field, min_value, max_value, 1)?;
for (reader, ff_reader) in self.readers.iter().zip(ff_readers) {
// TODO optimize if no deletes
for doc in reader.doc_ids_alive() {
ff_reader.get_vals(doc, &mut vals);
@@ -416,19 +437,53 @@ impl IndexMerger {
field: Field,
fast_field_serializer: &mut FastFieldSerializer,
) -> Result<()> {
self.write_fast_field_idx(field, fast_field_serializer)?;
let mut total_num_vals = 0u64;
let mut bytes_readers: Vec<BytesFastFieldReader> = Vec::new();
for reader in &self.readers {
let bytes_reader = reader.fast_fields().bytes(field).expect(
"Failed to find bytes fast field reader. This is a bug in tantivy, please report.",
);
if let Some(delete_bitset) = reader.delete_bitset() {
for doc in 0u32..reader.max_doc() {
if delete_bitset.is_alive(doc) {
let num_vals = bytes_reader.get_bytes(doc).len() as u64;
total_num_vals += num_vals;
}
}
} else {
total_num_vals += bytes_reader.total_num_bytes() as u64;
}
bytes_readers.push(bytes_reader);
}
{
// We can now create our `idx` serializer, and in a second pass,
// can effectively push the different indexes.
let mut serialize_idx =
fast_field_serializer.new_u64_fast_field_with_idx(field, 0, total_num_vals, 0)?;
let mut idx = 0;
for (segment_reader, bytes_reader) in self.readers.iter().zip(&bytes_readers) {
for doc in segment_reader.doc_ids_alive() {
serialize_idx.add_val(idx)?;
idx += bytes_reader.get_bytes(doc).len() as u64;
}
}
serialize_idx.add_val(idx)?;
serialize_idx.close_field()?;
}
let mut serialize_vals = fast_field_serializer.new_bytes_fast_field_with_idx(field, 1)?;
for reader in &self.readers {
let bytes_reader = reader.bytes_fast_field_reader(field)?;
for segment_reader in &self.readers {
let bytes_reader = segment_reader.fast_fields().bytes(field)
.expect("Failed to find bytes field in fast field reader. This is a bug in tantivy. Please report.");
// TODO: optimize if no deletes
for doc in reader.doc_ids_alive() {
let val = bytes_reader.get_val(doc);
for doc in segment_reader.doc_ids_alive() {
let val = bytes_reader.get_bytes(doc);
serialize_vals.write_all(val)?;
}
}
serialize_vals.flush()?;
Ok(())
}
@@ -440,7 +495,8 @@ impl IndexMerger {
) -> Result<Option<TermOrdinalMapping>> {
let mut positions_buffer: Vec<u32> = Vec::with_capacity(1_000);
let mut delta_computer = DeltaComputer::new();
let field_readers = self.readers
let field_readers = self
.readers
.iter()
.map(|reader| reader.inverted_index(indexed_field))
.collect::<Vec<_>>();
@@ -616,7 +672,7 @@ impl IndexMerger {
store_writer.store(&doc)?;
}
} else {
store_writer.stack(store_reader)?;
store_writer.stack(&store_reader)?;
}
}
Ok(())
@@ -637,10 +693,9 @@ impl SerializableSegment for IndexMerger {
#[cfg(test)]
mod tests {
use byteorder::{BigEndian, ReadBytesExt, WriteBytesExt};
use collector::chain;
use collector::tests::TestCollector;
use collector::tests::{BytesFastFieldTestCollector, FastFieldTestCollector};
use collector::FacetCollector;
use collector::{Count, FacetCollector};
use core::Index;
use futures::Future;
use query::AllQuery;
@@ -649,10 +704,12 @@ mod tests {
use schema;
use schema::Cardinality;
use schema::Document;
use schema::Facet;
use schema::IndexRecordOption;
use schema::IntOptions;
use schema::Term;
use schema::TextFieldIndexing;
use schema::INDEXED;
use std::io::Cursor;
use DocAddress;
use IndexWriter;
@@ -660,7 +717,7 @@ mod tests {
#[test]
fn test_index_merger_no_deletes() {
let mut schema_builder = schema::SchemaBuilder::default();
let mut schema_builder = schema::Schema::builder();
let text_fieldtype = schema::TextOptions::default()
.set_indexing_options(
TextFieldIndexing::default()
@@ -669,11 +726,13 @@ mod tests {
)
.set_stored();
let text_field = schema_builder.add_text_field("text", text_fieldtype);
let date_field = schema_builder.add_date_field("date", INDEXED);
let score_fieldtype = schema::IntOptions::default().set_fast(Cardinality::SingleValue);
let score_field = schema_builder.add_u64_field("score", score_fieldtype);
let bytes_score_field = schema_builder.add_bytes_field("score_bytes");
let index = Index::create_in_ram(schema_builder.build());
let reader = index.reader().unwrap();
let curr_time = chrono::Utc::now();
let add_score_bytes = |doc: &mut Document, score: u32| {
let mut bytes = Vec::new();
bytes
@@ -690,6 +749,7 @@ mod tests {
let mut doc = Document::default();
doc.add_text(text_field, "af b");
doc.add_u64(score_field, 3);
doc.add_date(date_field, &curr_time);
add_score_bytes(&mut doc, 3);
index_writer.add_document(doc);
}
@@ -715,6 +775,7 @@ mod tests {
{
let mut doc = Document::default();
doc.add_text(text_field, "af b");
doc.add_date(date_field, &curr_time);
doc.add_u64(score_field, 11);
add_score_bytes(&mut doc, 11);
index_writer.add_document(doc);
@@ -742,66 +803,76 @@ mod tests {
index_writer.wait_merging_threads().unwrap();
}
{
index.load_searchers().unwrap();
let searcher = index.searcher();
reader.reload().unwrap();
let searcher = reader.searcher();
let get_doc_ids = |terms: Vec<Term>| {
let mut collector = TestCollector::default();
let query = BooleanQuery::new_multiterms_query(terms);
assert!(searcher.search(&query, &mut collector).is_ok());
collector.docs()
let top_docs = searcher.search(&query, &TestCollector).unwrap();
top_docs.docs().to_vec()
};
{
assert_eq!(
get_doc_ids(vec![Term::from_field_text(text_field, "a")]),
vec![1, 2, 4]
vec![DocAddress(0, 1), DocAddress(0, 2), DocAddress(0, 4)]
);
assert_eq!(
get_doc_ids(vec![Term::from_field_text(text_field, "af")]),
vec![0, 3]
vec![DocAddress(0, 0), DocAddress(0, 3)]
);
assert_eq!(
get_doc_ids(vec![Term::from_field_text(text_field, "g")]),
vec![4]
vec![DocAddress(0, 4)]
);
assert_eq!(
get_doc_ids(vec![Term::from_field_text(text_field, "b")]),
vec![0, 1, 2, 3, 4]
vec![
DocAddress(0, 0),
DocAddress(0, 1),
DocAddress(0, 2),
DocAddress(0, 3),
DocAddress(0, 4)
]
);
assert_eq!(
get_doc_ids(vec![Term::from_field_date(date_field, &curr_time)]),
vec![DocAddress(0, 0), DocAddress(0, 3)]
);
}
{
let doc = searcher.doc(&DocAddress(0, 0)).unwrap();
assert_eq!(doc.get_first(text_field).unwrap().text(), "af b");
let doc = searcher.doc(DocAddress(0, 0)).unwrap();
assert_eq!(doc.get_first(text_field).unwrap().text(), Some("af b"));
}
{
let doc = searcher.doc(&DocAddress(0, 1)).unwrap();
assert_eq!(doc.get_first(text_field).unwrap().text(), "a b c");
let doc = searcher.doc(DocAddress(0, 1)).unwrap();
assert_eq!(doc.get_first(text_field).unwrap().text(), Some("a b c"));
}
{
let doc = searcher.doc(&DocAddress(0, 2)).unwrap();
assert_eq!(doc.get_first(text_field).unwrap().text(), "a b c d");
let doc = searcher.doc(DocAddress(0, 2)).unwrap();
assert_eq!(doc.get_first(text_field).unwrap().text(), Some("a b c d"));
}
{
let doc = searcher.doc(&DocAddress(0, 3)).unwrap();
assert_eq!(doc.get_first(text_field).unwrap().text(), "af b");
let doc = searcher.doc(DocAddress(0, 3)).unwrap();
assert_eq!(doc.get_first(text_field).unwrap().text(), Some("af b"));
}
{
let doc = searcher.doc(&DocAddress(0, 4)).unwrap();
assert_eq!(doc.get_first(text_field).unwrap().text(), "a b c g");
let doc = searcher.doc(DocAddress(0, 4)).unwrap();
assert_eq!(doc.get_first(text_field).unwrap().text(), Some("a b c g"));
}
{
let get_fast_vals = |terms: Vec<Term>| {
let query = BooleanQuery::new_multiterms_query(terms);
let mut collector = FastFieldTestCollector::for_field(score_field);
assert!(searcher.search(&query, &mut collector).is_ok());
collector.vals()
searcher
.search(&query, &FastFieldTestCollector::for_field(score_field))
.unwrap()
};
let get_fast_vals_bytes = |terms: Vec<Term>| {
let query = BooleanQuery::new_multiterms_query(terms);
let mut collector = BytesFastFieldTestCollector::for_field(bytes_score_field);
searcher
.search(&query, &mut collector)
.expect("failed to search");
collector.vals()
.search(
&query,
&BytesFastFieldTestCollector::for_field(bytes_score_field),
)
.expect("failed to search")
};
assert_eq!(
get_fast_vals(vec![Term::from_field_text(text_field, "a")]),
@@ -817,7 +888,7 @@ mod tests {
#[test]
fn test_index_merger_with_deletes() {
let mut schema_builder = schema::SchemaBuilder::default();
let mut schema_builder = schema::Schema::builder();
let text_fieldtype = schema::TextOptions::default()
.set_indexing_options(
TextFieldIndexing::default().set_index_option(IndexRecordOption::WithFreqs),
@@ -828,24 +899,16 @@ mod tests {
let score_field = schema_builder.add_u64_field("score", score_fieldtype);
let bytes_score_field = schema_builder.add_bytes_field("score_bytes");
let index = Index::create_in_ram(schema_builder.build());
let mut index_writer = index.writer_with_num_threads(1, 40_000_000).unwrap();
let mut index_writer = index.writer_with_num_threads(1, 3_000_000).unwrap();
let reader = index.reader().unwrap();
let search_term = |searcher: &Searcher, term: Term| {
let mut collector = FastFieldTestCollector::for_field(score_field);
let mut bytes_collector = BytesFastFieldTestCollector::for_field(bytes_score_field);
let collector = FastFieldTestCollector::for_field(score_field);
let bytes_collector = BytesFastFieldTestCollector::for_field(bytes_score_field);
let term_query = TermQuery::new(term, IndexRecordOption::Basic);
{
let mut combined_collector =
chain().push(&mut collector).push(&mut bytes_collector);
searcher
.search(&term_query, &mut combined_collector)
.unwrap();
}
let scores = collector.vals();
let mut score_bytes = Cursor::new(bytes_collector.vals());
let (scores, bytes) = searcher
.search(&term_query, &(collector, bytes_collector))
.unwrap();
let mut score_bytes = Cursor::new(bytes);
for &score in &scores {
assert_eq!(score as u32, score_bytes.read_u32::<BigEndian>().unwrap());
}
@@ -858,24 +921,24 @@ mod tests {
{
// a first commit
index_writer.add_document(doc!(
text_field => "a b d",
score_field => 1u64,
bytes_score_field => vec![0u8, 0, 0, 1],
));
text_field => "a b d",
score_field => 1u64,
bytes_score_field => vec![0u8, 0, 0, 1],
));
index_writer.add_document(doc!(
text_field => "b c",
score_field => 2u64,
bytes_score_field => vec![0u8, 0, 0, 2],
));
text_field => "b c",
score_field => 2u64,
bytes_score_field => vec![0u8, 0, 0, 2],
));
index_writer.delete_term(Term::from_field_text(text_field, "c"));
index_writer.add_document(doc!(
text_field => "c d",
score_field => 3u64,
bytes_score_field => vec![0u8, 0, 0, 3],
));
text_field => "c d",
score_field => 3u64,
bytes_score_field => vec![0u8, 0, 0, 3],
));
index_writer.commit().expect("committed");
index.load_searchers().unwrap();
let ref searcher = *index.searcher();
reader.reload().unwrap();
let searcher = reader.searcher();
assert_eq!(searcher.num_docs(), 2);
assert_eq!(searcher.segment_readers()[0].num_docs(), 2);
assert_eq!(searcher.segment_readers()[0].max_doc(), 3);
@@ -899,37 +962,37 @@ mod tests {
{
// a second commit
index_writer.add_document(doc!(
text_field => "a d e",
score_field => 4_000u64,
bytes_score_field => vec![0u8, 0, 0, 4],
));
text_field => "a d e",
score_field => 4_000u64,
bytes_score_field => vec![0u8, 0, 0, 4],
));
index_writer.add_document(doc!(
text_field => "e f",
score_field => 5_000u64,
bytes_score_field => vec![0u8, 0, 0, 5],
));
text_field => "e f",
score_field => 5_000u64,
bytes_score_field => vec![0u8, 0, 0, 5],
));
index_writer.delete_term(Term::from_field_text(text_field, "a"));
index_writer.delete_term(Term::from_field_text(text_field, "f"));
index_writer.add_document(doc!(
text_field => "f g",
score_field => 6_000u64,
bytes_score_field => vec![0u8, 0, 23, 112],
));
text_field => "f g",
score_field => 6_000u64,
bytes_score_field => vec![0u8, 0, 23, 112],
));
index_writer.add_document(doc!(
text_field => "g h",
score_field => 7_000u64,
bytes_score_field => vec![0u8, 0, 27, 88],
));
text_field => "g h",
score_field => 7_000u64,
bytes_score_field => vec![0u8, 0, 27, 88],
));
index_writer.commit().expect("committed");
index.load_searchers().unwrap();
let searcher = index.searcher();
reader.reload().unwrap();
let searcher = reader.searcher();
assert_eq!(searcher.segment_readers().len(), 2);
assert_eq!(searcher.num_docs(), 3);
assert_eq!(searcher.segment_readers()[0].num_docs(), 1);
assert_eq!(searcher.segment_readers()[0].max_doc(), 3);
assert_eq!(searcher.segment_readers()[1].num_docs(), 2);
assert_eq!(searcher.segment_readers()[1].max_doc(), 4);
assert_eq!(searcher.segment_readers()[0].num_docs(), 2);
assert_eq!(searcher.segment_readers()[0].max_doc(), 4);
assert_eq!(searcher.segment_readers()[1].num_docs(), 1);
assert_eq!(searcher.segment_readers()[1].max_doc(), 3);
assert_eq!(
search_term(&searcher, Term::from_field_text(text_field, "a")),
empty_vec
@@ -961,17 +1024,19 @@ mod tests {
let score_field_reader = searcher
.segment_reader(0)
.fast_field_reader::<u64>(score_field)
.unwrap();
assert_eq!(score_field_reader.min_value(), 1);
assert_eq!(score_field_reader.max_value(), 3);
let score_field_reader = searcher
.segment_reader(1)
.fast_field_reader::<u64>(score_field)
.fast_fields()
.u64(score_field)
.unwrap();
assert_eq!(score_field_reader.min_value(), 4000);
assert_eq!(score_field_reader.max_value(), 7000);
let score_field_reader = searcher
.segment_reader(1)
.fast_fields()
.u64(score_field)
.unwrap();
assert_eq!(score_field_reader.min_value(), 1);
assert_eq!(score_field_reader.max_value(), 3);
}
{
// merging the segments
@@ -983,8 +1048,8 @@ mod tests {
.expect("Failed to initiate merge")
.wait()
.expect("Merging failed");
index.load_searchers().unwrap();
let ref searcher = *index.searcher();
reader.reload().unwrap();
let searcher = reader.searcher();
assert_eq!(searcher.segment_readers().len(), 1);
assert_eq!(searcher.num_docs(), 3);
assert_eq!(searcher.segment_readers()[0].num_docs(), 3);
@@ -1019,7 +1084,8 @@ mod tests {
);
let score_field_reader = searcher
.segment_reader(0)
.fast_field_reader::<u64>(score_field)
.fast_fields()
.u64(score_field)
.unwrap();
assert_eq!(score_field_reader.min_value(), 3);
assert_eq!(score_field_reader.max_value(), 7000);
@@ -1029,8 +1095,8 @@ mod tests {
index_writer.delete_term(Term::from_field_text(text_field, "c"));
index_writer.commit().unwrap();
index.load_searchers().unwrap();
let ref searcher = *index.searcher();
reader.reload().unwrap();
let searcher = reader.searcher();
assert_eq!(searcher.segment_readers().len(), 1);
assert_eq!(searcher.num_docs(), 2);
assert_eq!(searcher.segment_readers()[0].num_docs(), 2);
@@ -1065,7 +1131,8 @@ mod tests {
);
let score_field_reader = searcher
.segment_reader(0)
.fast_field_reader::<u64>(score_field)
.fast_fields()
.u64(score_field)
.unwrap();
assert_eq!(score_field_reader.min_value(), 3);
assert_eq!(score_field_reader.max_value(), 7000);
@@ -1080,9 +1147,9 @@ mod tests {
.expect("Failed to initiate merge")
.wait()
.expect("Merging failed");
index.load_searchers().unwrap();
reader.reload().unwrap();
let ref searcher = *index.searcher();
let searcher = reader.searcher();
assert_eq!(searcher.segment_readers().len(), 1);
assert_eq!(searcher.num_docs(), 2);
assert_eq!(searcher.segment_readers()[0].num_docs(), 2);
@@ -1117,7 +1184,8 @@ mod tests {
);
let score_field_reader = searcher
.segment_reader(0)
.fast_field_reader::<u64>(score_field)
.fast_fields()
.u64(score_field)
.unwrap();
assert_eq!(score_field_reader.min_value(), 6000);
assert_eq!(score_field_reader.max_value(), 7000);
@@ -1126,30 +1194,27 @@ mod tests {
{
// Test removing all docs
index_writer.delete_term(Term::from_field_text(text_field, "g"));
index_writer.commit().unwrap();
let segment_ids = index
.searchable_segment_ids()
.expect("Searchable segments failed.");
index_writer
.merge(&segment_ids)
.expect("Failed to initiate merge")
.wait()
.expect("Merging failed");
index.load_searchers().unwrap();
reader.reload().unwrap();
let ref searcher = *index.searcher();
assert_eq!(searcher.segment_readers().len(), 1);
let searcher = reader.searcher();
assert!(segment_ids.is_empty());
assert!(searcher.segment_readers().is_empty());
assert_eq!(searcher.num_docs(), 0);
}
}
#[test]
fn test_merge_facets() {
let mut schema_builder = schema::SchemaBuilder::default();
let mut schema_builder = schema::Schema::builder();
let facet_field = schema_builder.add_facet_field("facet");
let index = Index::create_in_ram(schema_builder.build());
use schema::Facet;
let reader = index.reader().unwrap();
{
let mut index_writer = index.writer_with_num_threads(1, 40_000_000).unwrap();
let mut index_writer = index.writer_with_num_threads(1, 3_000_000).unwrap();
let index_doc = |index_writer: &mut IndexWriter, doc_facets: &[&str]| {
let mut doc = Document::default();
for facet in doc_facets {
@@ -1176,20 +1241,16 @@ mod tests {
index_doc(&mut index_writer, &["/top/e", "/top/f"]);
index_writer.commit().expect("committed");
}
index.load_searchers().unwrap();
reader.reload().unwrap();
let test_searcher = |expected_num_docs: usize, expected: &[(&str, u64)]| {
let searcher = index.searcher();
let searcher = reader.searcher();
let mut facet_collector = FacetCollector::for_field(facet_field);
facet_collector.add_facet(Facet::from("/top"));
use collector::{CountCollector, MultiCollector};
let mut count_collector = CountCollector::default();
{
let mut multi_collectors =
MultiCollector::from(vec![&mut count_collector, &mut facet_collector]);
searcher.search(&AllQuery, &mut multi_collectors).unwrap();
}
assert_eq!(count_collector.count(), expected_num_docs);
let facet_counts = facet_collector.harvest();
let (count, facet_counts) = searcher
.search(&AllQuery, &(Count, facet_collector))
.unwrap();
assert_eq!(count, expected_num_docs);
let facets: Vec<(String, u64)> = facet_counts
.get("/top")
.map(|(facet, count)| (facet.to_string(), count))
@@ -1213,21 +1274,19 @@ mod tests {
("/top/f", 1),
],
);
// Merging the segments
{
let segment_ids = index
.searchable_segment_ids()
.expect("Searchable segments failed.");
let mut index_writer = index.writer_with_num_threads(1, 40_000_000).unwrap();
let mut index_writer = index.writer_with_num_threads(1, 3_000_000).unwrap();
index_writer
.merge(&segment_ids)
.expect("Failed to initiate merge")
.wait()
.expect("Merging failed");
index_writer.wait_merging_threads().unwrap();
index.load_searchers().unwrap();
reader.reload().unwrap();
test_searcher(
11,
&[
@@ -1243,12 +1302,12 @@ mod tests {
// Deleting one term
{
let mut index_writer = index.writer_with_num_threads(1, 40_000_000).unwrap();
let mut index_writer = index.writer_with_num_threads(1, 3_000_000).unwrap();
let facet = Facet::from_path(vec!["top", "a", "firstdoc"]);
let facet_term = Term::from_facet(facet_field, &facet);
index_writer.delete_term(facet_term);
index_writer.commit().unwrap();
index.load_searchers().unwrap();
reader.reload().unwrap();
test_searcher(
9,
&[
@@ -1263,17 +1322,45 @@ mod tests {
}
}
#[test]
fn test_bug_merge() {
let mut schema_builder = schema::Schema::builder();
let int_field = schema_builder.add_u64_field("intvals", INDEXED);
let index = Index::create_in_ram(schema_builder.build());
let mut index_writer = index.writer_with_num_threads(1, 3_000_000).unwrap();
index_writer.add_document(doc!(int_field => 1u64));
index_writer.commit().expect("commit failed");
index_writer.add_document(doc!(int_field => 1u64));
index_writer.commit().expect("commit failed");
let reader = index.reader().unwrap();
let searcher = reader.searcher();
assert_eq!(searcher.num_docs(), 2);
index_writer.delete_term(Term::from_field_u64(int_field, 1));
let segment_ids = index
.searchable_segment_ids()
.expect("Searchable segments failed.");
index_writer
.merge(&segment_ids)
.expect("Failed to initiate merge")
.wait()
.expect("Merging failed");
reader.reload().unwrap();
// commit has not been called yet. The document should still be
// there.
assert_eq!(reader.searcher().num_docs(), 2);
}
#[test]
fn test_merge_multivalued_int_fields_all_deleted() {
let mut schema_builder = schema::SchemaBuilder::default();
let mut schema_builder = schema::Schema::builder();
let int_options = IntOptions::default()
.set_fast(Cardinality::MultiValues)
.set_indexed();
let int_field = schema_builder.add_u64_field("intvals", int_options);
let index = Index::create_in_ram(schema_builder.build());
let reader = index.reader().unwrap();
{
let mut index_writer = index.writer_with_num_threads(1, 40_000_000).unwrap();
let mut index_writer = index.writer_with_num_threads(1, 3_000_000).unwrap();
let mut doc = Document::default();
doc.add_u64(int_field, 1);
index_writer.add_document(doc.clone());
@@ -1281,32 +1368,34 @@ mod tests {
index_writer.add_document(doc);
index_writer.commit().expect("commit failed");
index_writer.delete_term(Term::from_field_u64(int_field, 1));
index_writer.commit().expect("commit failed");
}
index.load_searchers().unwrap();
let searcher = index.searcher();
assert_eq!(searcher.num_docs(), 0);
// Merging the segments
{
let segment_ids = index
.searchable_segment_ids()
.expect("Searchable segments failed.");
let mut index_writer = index.writer_with_num_threads(1, 40_000_000).unwrap();
index_writer
.merge(&segment_ids)
.expect("Failed to initiate merge")
.wait()
.expect("Merging failed");
// assert delete has not been committed
reader.reload().expect("failed to load searcher 1");
let searcher = reader.searcher();
assert_eq!(searcher.num_docs(), 2);
index_writer.commit().unwrap();
index_writer.wait_merging_threads().unwrap();
}
index.load_searchers().unwrap();
let searcher = index.searcher();
reader.reload().unwrap();
let searcher = reader.searcher();
assert_eq!(searcher.num_docs(), 0);
}
#[test]
fn test_merge_multivalued_int_fields() {
let mut schema_builder = schema::SchemaBuilder::default();
fn test_merge_multivalued_int_fields_simple() {
let mut schema_builder = schema::Schema::builder();
let int_options = IntOptions::default()
.set_fast(Cardinality::MultiValues)
.set_indexed();
@@ -1314,7 +1403,7 @@ mod tests {
let index = Index::create_in_ram(schema_builder.build());
{
let mut index_writer = index.writer_with_num_threads(1, 40_000_000).unwrap();
let mut index_writer = index.writer_with_num_threads(1, 3_000_000).unwrap();
let index_doc = |index_writer: &mut IndexWriter, int_vals: &[u64]| {
let mut doc = Document::default();
for &val in int_vals {
@@ -1322,7 +1411,6 @@ mod tests {
}
index_writer.add_document(doc);
};
index_doc(&mut index_writer, &[1, 2]);
index_doc(&mut index_writer, &[1, 2, 3]);
index_doc(&mut index_writer, &[4, 5]);
@@ -1331,24 +1419,19 @@ mod tests {
index_doc(&mut index_writer, &[3]);
index_doc(&mut index_writer, &[17]);
index_writer.commit().expect("committed");
index_doc(&mut index_writer, &[20]);
index_writer.commit().expect("committed");
index_doc(&mut index_writer, &[28, 27]);
index_doc(&mut index_writer, &[1_000]);
index_writer.commit().expect("committed");
}
index.load_searchers().unwrap();
let searcher = index.searcher();
let reader = index.reader().unwrap();
let searcher = reader.searcher();
let mut vals: Vec<u64> = Vec::new();
{
let segment = searcher.segment_reader(0u32);
let ff_reader = segment.multi_fast_field_reader(int_field).unwrap();
let ff_reader = segment.fast_fields().u64s(int_field).unwrap();
ff_reader.get_vals(0, &mut vals);
assert_eq!(&vals, &[1, 2]);
@@ -1372,16 +1455,18 @@ mod tests {
assert_eq!(&vals, &[17]);
}
{
let segment = searcher.segment_reader(1u32);
let ff_reader = segment.multi_fast_field_reader(int_field).unwrap();
ff_reader.get_vals(0, &mut vals);
assert_eq!(&vals, &[20]);
}
println!(
"{:?}",
searcher
.segment_readers()
.iter()
.map(|reader| reader.max_doc())
.collect::<Vec<_>>()
);
{
let segment = searcher.segment_reader(2u32);
let ff_reader = segment.multi_fast_field_reader(int_field).unwrap();
let segment = searcher.segment_reader(1u32);
let ff_reader = segment.fast_fields().u64s(int_field).unwrap();
ff_reader.get_vals(0, &mut vals);
assert_eq!(&vals, &[28, 27]);
@@ -1389,26 +1474,42 @@ mod tests {
assert_eq!(&vals, &[1_000]);
}
{
let segment = searcher.segment_reader(2u32);
let ff_reader = segment.fast_fields().u64s(int_field).unwrap();
ff_reader.get_vals(0, &mut vals);
assert_eq!(&vals, &[20]);
}
// Merging the segments
{
let segment_ids = index
.searchable_segment_ids()
.expect("Searchable segments failed.");
let mut index_writer = index.writer_with_num_threads(1, 40_000_000).unwrap();
let mut index_writer = index.writer_with_num_threads(1, 3_000_000).unwrap();
index_writer
.merge(&segment_ids)
.expect("Failed to initiate merge")
.wait()
.expect("Merging failed");
index_writer.wait_merging_threads().unwrap();
index_writer
.wait_merging_threads()
.expect("Wait for merging threads");
}
index.load_searchers().unwrap();
reader.reload().expect("Load searcher");
{
let searcher = index.searcher();
let searcher = reader.searcher();
println!(
"{:?}",
searcher
.segment_readers()
.iter()
.map(|reader| reader.max_doc())
.collect::<Vec<_>>()
);
let segment = searcher.segment_reader(0u32);
let ff_reader = segment.multi_fast_field_reader(int_field).unwrap();
let ff_reader = segment.fast_fields().u64s(int_field).unwrap();
ff_reader.get_vals(0, &mut vals);
assert_eq!(&vals, &[1, 2]);
@@ -1432,13 +1533,13 @@ mod tests {
assert_eq!(&vals, &[17]);
ff_reader.get_vals(7, &mut vals);
assert_eq!(&vals, &[20]);
ff_reader.get_vals(8, &mut vals);
assert_eq!(&vals, &[28, 27]);
ff_reader.get_vals(9, &mut vals);
ff_reader.get_vals(8, &mut vals);
assert_eq!(&vals, &[1_000]);
ff_reader.get_vals(9, &mut vals);
assert_eq!(&vals, &[20]);
}
}
}

7
src/indexer/mod.rs
View File

@@ -1,8 +1,9 @@
pub mod delete_queue;
mod directory_lock;
mod doc_opstamp_mapping;
pub mod index_writer;
mod log_merge_policy;
mod merge_operation;
pub mod merge_policy;
pub mod merger;
pub mod operation;
@@ -15,12 +16,12 @@ pub mod segment_updater;
mod segment_writer;
mod stamper;
pub(crate) use self::directory_lock::DirectoryLock;
pub use self::index_writer::IndexWriter;
pub use self::log_merge_policy::LogMergePolicy;
pub use self::merge_operation::{MergeOperation, MergeOperationInventory};
pub use self::merge_policy::{MergeCandidate, MergePolicy, NoMergePolicy};
pub use self::prepared_commit::PreparedCommit;
pub use self::segment_entry::{SegmentEntry, SegmentState};
pub use self::segment_entry::SegmentEntry;
pub use self::segment_manager::SegmentManager;
pub use self::segment_serializer::SegmentSerializer;
pub use self::segment_writer::SegmentWriter;

7
src/indexer/operation.rs
View File

@@ -14,3 +14,10 @@ pub struct AddOperation {
pub opstamp: u64,
pub document: Document,
}
/// UserOperation is an enum type that encapsulates other operation types.
#[derive(Eq, PartialEq, Debug)]
pub enum UserOperation {
Add(Document),
Delete(Term),
}

50
src/indexer/segment_entry.rs
View File

@@ -4,21 +4,6 @@ use core::SegmentMeta;
use indexer::delete_queue::DeleteCursor;
use std::fmt;
#[derive(Clone, Copy, PartialEq, Eq, Debug)]
pub enum SegmentState {
Ready,
InMerge,
}
impl SegmentState {
pub fn letter_code(&self) -> char {
match *self {
SegmentState::InMerge => 'M',
SegmentState::Ready => 'R',
}
}
}
/// A segment entry describes the state of
/// a given segment, at a given instant.
///
@@ -35,7 +20,6 @@ impl SegmentState {
#[derive(Clone)]
pub struct SegmentEntry {
meta: SegmentMeta,
state: SegmentState,
delete_bitset: Option<BitSet>,
delete_cursor: DeleteCursor,
}
@@ -49,7 +33,6 @@ impl SegmentEntry {
) -> SegmentEntry {
SegmentEntry {
meta: segment_meta,
state: SegmentState::Ready,
delete_bitset,
delete_cursor,
}
@@ -72,14 +55,6 @@ impl SegmentEntry {
&mut self.delete_cursor
}
/// Return the `SegmentEntry`.
///
/// The state describes whether the segment is available for
/// a merge or not.
pub fn state(&self) -> SegmentState {
self.state
}
/// Returns the segment id.
pub fn segment_id(&self) -> SegmentId {
self.meta.id()
@@ -89,33 +64,10 @@ impl SegmentEntry {
pub fn meta(&self) -> &SegmentMeta {
&self.meta
}
/// Mark the `SegmentEntry` as in merge.
///
/// Only segments that are not already
/// in a merge are elligible for future merge.
pub fn start_merge(&mut self) {
self.state = SegmentState::InMerge;
}
/// Cancel a merge
///
/// If a merge fails, it is important to switch
/// the segment back to a idle state, so that it
/// may be elligible for future merges.
pub fn cancel_merge(&mut self) {
self.state = SegmentState::Ready;
}
/// Returns true iff a segment should
/// be considered for a merge.
pub fn is_ready(&self) -> bool {
self.state == SegmentState::Ready
}
}
impl fmt::Debug for SegmentEntry {
fn fmt(&self, formatter: &mut fmt::Formatter) -> fmt::Result {
write!(formatter, "SegmentEntry({:?}, {:?})", self.meta, self.state)
write!(formatter, "SegmentEntry({:?})", self.meta)
}
}

95
src/indexer/segment_manager.rs
View File

@@ -1,7 +1,7 @@
use super::segment_register::SegmentRegister;
use core::SegmentId;
use core::SegmentMeta;
use core::{LOCKFILE_FILEPATH, META_FILEPATH};
use core::META_FILEPATH;
use error::TantivyError;
use indexer::delete_queue::DeleteCursor;
use indexer::SegmentEntry;
@@ -16,7 +16,6 @@ use Result as TantivyResult;
struct SegmentRegisters {
uncommitted: SegmentRegister,
committed: SegmentRegister,
writing: HashSet<SegmentId>,
}
/// The segment manager stores the list of segments
@@ -41,12 +40,17 @@ impl Debug for SegmentManager {
}
pub fn get_mergeable_segments(
in_merge_segment_ids: &HashSet<SegmentId>,
segment_manager: &SegmentManager,
) -> (Vec<SegmentMeta>, Vec<SegmentMeta>) {
let registers_lock = segment_manager.read();
(
registers_lock.committed.get_mergeable_segments(),
registers_lock.uncommitted.get_mergeable_segments(),
registers_lock
.committed
.get_mergeable_segments(in_merge_segment_ids),
registers_lock
.uncommitted
.get_mergeable_segments(in_merge_segment_ids),
)
}
@@ -59,7 +63,6 @@ impl SegmentManager {
registers: RwLock::new(SegmentRegisters {
uncommitted: SegmentRegister::default(),
committed: SegmentRegister::new(segment_metas, delete_cursor),
writing: HashSet::new(),
}),
}
}
@@ -72,16 +75,13 @@ impl SegmentManager {
segment_entries
}
/// Returns the overall number of segments in the `SegmentManager`
pub fn num_segments(&self) -> usize {
let registers_lock = self.read();
registers_lock.committed.len() + registers_lock.uncommitted.len()
}
/// List the files that are useful to the index.
///
/// This does not include lock files, or files that are obsolete
/// but have not yet been deleted by the garbage collector.
pub fn list_files(&self) -> HashSet<PathBuf> {
let mut files = HashSet::new();
files.insert(META_FILEPATH.clone());
files.insert(LOCKFILE_FILEPATH.clone());
for segment_meta in SegmentMeta::all() {
files.extend(segment_meta.list_files());
}
@@ -103,6 +103,21 @@ impl SegmentManager {
.expect("Failed to acquire write lock on SegmentManager.")
}
/// Deletes all empty segments
fn remove_empty_segments(&self) {
let mut registers_lock = self.write();
registers_lock
.committed
.segment_entries()
.iter()
.filter(|segment| segment.meta().num_docs() == 0)
.for_each(|segment| {
registers_lock
.committed
.remove_segment(&segment.segment_id())
});
}
pub fn commit(&self, segment_entries: Vec<SegmentEntry>) {
let mut registers_lock = self.write();
registers_lock.committed.clear();
@@ -118,25 +133,22 @@ impl SegmentManager {
/// the `segment_ids` are not either all committed or all
/// uncommitted.
pub fn start_merge(&self, segment_ids: &[SegmentId]) -> TantivyResult<Vec<SegmentEntry>> {
let mut registers_lock = self.write();
let registers_lock = self.read();
let mut segment_entries = vec![];
if registers_lock.uncommitted.contains_all(segment_ids) {
for segment_id in segment_ids {
let segment_entry = registers_lock.uncommitted
.start_merge(segment_id)
.get(segment_id)
.expect("Segment id not found {}. Should never happen because of the contains all if-block.");
segment_entries.push(segment_entry);
}
} else if registers_lock.committed.contains_all(segment_ids) {
for segment_id in segment_ids {
let segment_entry = registers_lock.committed
.start_merge(segment_id)
.get(segment_id)
.expect("Segment id not found {}. Should never happen because of the contains all if-block.");
segment_entries.push(segment_entry);
}
for segment_id in segment_ids {
registers_lock.committed.start_merge(segment_id);
}
} else {
let error_msg = "Merge operation sent for segments that are not \
all uncommited or commited."
@@ -146,50 +158,8 @@ impl SegmentManager {
Ok(segment_entries)
}
pub fn cancel_merge(
&self,
before_merge_segment_ids: &[SegmentId],
after_merge_segment_id: SegmentId,
) {
let mut registers_lock = self.write();
// we mark all segments are ready for merge.
{
let target_segment_register: &mut SegmentRegister;
target_segment_register = {
if registers_lock
.uncommitted
.contains_all(before_merge_segment_ids)
{
&mut registers_lock.uncommitted
} else if registers_lock
.committed
.contains_all(before_merge_segment_ids)
{
&mut registers_lock.committed
} else {
warn!("couldn't find segment in SegmentManager");
return;
}
};
for segment_id in before_merge_segment_ids {
target_segment_register.cancel_merge(segment_id);
}
}
// ... and we make sure the target segment entry
// can be garbage collected.
registers_lock.writing.remove(&after_merge_segment_id);
}
pub fn write_segment(&self, segment_id: SegmentId) {
let mut registers_lock = self.write();
registers_lock.writing.insert(segment_id);
}
pub fn add_segment(&self, segment_entry: SegmentEntry) {
let mut registers_lock = self.write();
registers_lock.writing.remove(&segment_entry.segment_id());
registers_lock.uncommitted.add_segment_entry(segment_entry);
}
@@ -199,10 +169,6 @@ impl SegmentManager {
after_merge_segment_entry: SegmentEntry,
) {
let mut registers_lock = self.write();
registers_lock
.writing
.remove(&after_merge_segment_entry.segment_id());
let target_register: &mut SegmentRegister = {
if registers_lock
.uncommitted
@@ -226,6 +192,7 @@ impl SegmentManager {
}
pub fn committed_segment_metas(&self) -> Vec<SegmentMeta> {
self.remove_empty_segments();
let registers_lock = self.read();
registers_lock.committed.segment_metas()
}

75
src/indexer/segment_register.rs
View File

@@ -3,6 +3,7 @@ use core::SegmentMeta;
use indexer::delete_queue::DeleteCursor;
use indexer::segment_entry::SegmentEntry;
use std::collections::HashMap;
use std::collections::HashSet;
use std::fmt::{self, Debug, Formatter};
/// The segment register keeps track
@@ -21,8 +22,8 @@ pub struct SegmentRegister {
impl Debug for SegmentRegister {
fn fmt(&self, f: &mut Formatter) -> Result<(), fmt::Error> {
write!(f, "SegmentRegister(")?;
for (k, v) in &self.segment_states {
write!(f, "{}:{}, ", k.short_uuid_string(), v.state().letter_code())?;
for k in self.segment_states.keys() {
write!(f, "{}, ", k.short_uuid_string())?;
}
write!(f, ")")?;
Ok(())
@@ -34,14 +35,13 @@ impl SegmentRegister {
self.segment_states.clear();
}
pub fn len(&self) -> usize {
self.segment_states.len()
}
pub fn get_mergeable_segments(&self) -> Vec<SegmentMeta> {
pub fn get_mergeable_segments(
&self,
in_merge_segment_ids: &HashSet<SegmentId>,
) -> Vec<SegmentMeta> {
self.segment_states
.values()
.filter(|segment_entry| segment_entry.is_ready())
.filter(|segment_entry| !in_merge_segment_ids.contains(&segment_entry.segment_id()))
.map(|segment_entry| segment_entry.meta().clone())
.collect()
}
@@ -51,15 +51,16 @@ impl SegmentRegister {
}
pub fn segment_metas(&self) -> Vec<SegmentMeta> {
let mut segment_ids: Vec<SegmentMeta> = self.segment_states
let mut segment_ids: Vec<SegmentMeta> = self
.segment_states
.values()
.map(|segment_entry| segment_entry.meta().clone())
.collect();
segment_ids.sort_by_key(|meta| meta.id());
segment_ids.sort_by_key(SegmentMeta::id);
segment_ids
}
pub fn contains_all(&mut self, segment_ids: &[SegmentId]) -> bool {
pub fn contains_all(&self, segment_ids: &[SegmentId]) -> bool {
segment_ids
.iter()
.all(|segment_id| self.segment_states.contains_key(segment_id))
@@ -74,20 +75,8 @@ impl SegmentRegister {
self.segment_states.remove(segment_id);
}
pub fn cancel_merge(&mut self, segment_id: &SegmentId) {
self.segment_states
.get_mut(segment_id)
.expect("Received a merge notification for a segment that is not registered")
.cancel_merge();
}
pub fn start_merge(&mut self, segment_id: &SegmentId) -> Option<SegmentEntry> {
if let Some(segment_entry) = self.segment_states.get_mut(segment_id) {
segment_entry.start_merge();
Some(segment_entry.clone())
} else {
None
}
pub fn get(&self, segment_id: &SegmentId) -> Option<SegmentEntry> {
self.segment_states.get(segment_id).cloned()
}
pub fn new(segment_metas: Vec<SegmentMeta>, delete_cursor: &DeleteCursor) -> SegmentRegister {
@@ -99,11 +88,6 @@ impl SegmentRegister {
}
SegmentRegister { segment_states }
}
#[cfg(test)]
pub fn segment_entry(&self, segment_id: &SegmentId) -> Option<SegmentEntry> {
self.segment_states.get(segment_id).cloned()
}
}
#[cfg(test)]
@@ -112,7 +96,6 @@ mod tests {
use core::SegmentId;
use core::SegmentMeta;
use indexer::delete_queue::*;
use indexer::SegmentState;
fn segment_ids(segment_register: &SegmentRegister) -> Vec<SegmentId> {
segment_register
@@ -136,42 +119,12 @@ mod tests {
let segment_entry = SegmentEntry::new(segment_meta, delete_queue.cursor(), None);
segment_register.add_segment_entry(segment_entry);
}
assert_eq!(
segment_register
.segment_entry(&segment_id_a)
.unwrap()
.state(),
SegmentState::Ready
);
assert_eq!(segment_ids(&segment_register), vec![segment_id_a]);
{
let segment_meta = SegmentMeta::new(segment_id_b, 0u32);
let segment_entry = SegmentEntry::new(segment_meta, delete_queue.cursor(), None);
segment_register.add_segment_entry(segment_entry);
}
assert_eq!(
segment_register
.segment_entry(&segment_id_b)
.unwrap()
.state(),
SegmentState::Ready
);
segment_register.start_merge(&segment_id_a);
segment_register.start_merge(&segment_id_b);
assert_eq!(
segment_register
.segment_entry(&segment_id_a)
.unwrap()
.state(),
SegmentState::InMerge
);
assert_eq!(
segment_register
.segment_entry(&segment_id_b)
.unwrap()
.state(),
SegmentState::InMerge
);
segment_register.remove_segment(&segment_id_a);
segment_register.remove_segment(&segment_id_b);
{

354
src/indexer/segment_updater.rs
View File

@@ -16,9 +16,10 @@ use futures_cpupool::CpuFuture;
use futures_cpupool::CpuPool;
use indexer::delete_queue::DeleteCursor;
use indexer::index_writer::advance_deletes;
use indexer::merge_operation::MergeOperationInventory;
use indexer::merger::IndexMerger;
use indexer::stamper::Stamper;
use indexer::MergeCandidate;
use indexer::MergeOperation;
use indexer::SegmentEntry;
use indexer::SegmentSerializer;
use indexer::{DefaultMergePolicy, MergePolicy};
@@ -26,6 +27,7 @@ use schema::Schema;
use serde_json;
use std::borrow::BorrowMut;
use std::collections::HashMap;
use std::collections::HashSet;
use std::io::Write;
use std::mem;
use std::ops::DerefMut;
@@ -45,34 +47,31 @@ use Result;
/// and flushed.
///
/// This method is not part of tantivy's public API
pub fn save_new_metas(schema: Schema, opstamp: u64, directory: &mut Directory) -> Result<()> {
save_metas(vec![], schema, opstamp, None, directory)
pub fn save_new_metas(schema: Schema, directory: &mut Directory) -> Result<()> {
save_metas(
&IndexMeta {
segments: Vec::new(),
schema,
opstamp: 0u64,
payload: None,
},
directory,
)
}
/// Save the index meta file.
/// This operation is atomic:
/// Either
// - it fails, in which case an error is returned,
/// - it fails, in which case an error is returned,
/// and the `meta.json` remains untouched,
/// - it success, and `meta.json` is written
/// and flushed.
///
/// This method is not part of tantivy's public API
pub fn save_metas(
segment_metas: Vec<SegmentMeta>,
schema: Schema,
opstamp: u64,
payload: Option<String>,
directory: &mut Directory,
) -> Result<()> {
let metas = IndexMeta {
segments: segment_metas,
schema,
opstamp,
payload,
};
let mut buffer = serde_json::to_vec_pretty(&metas)?;
write!(&mut buffer, "\n")?;
fn save_metas(metas: &IndexMeta, directory: &mut Directory) -> Result<()> {
let mut buffer = serde_json::to_vec_pretty(metas)?;
// Just adding a new line at the end of the buffer.
writeln!(&mut buffer)?;
directory.atomic_write(&META_FILEPATH, &buffer[..])?;
debug!("Saved metas {:?}", serde_json::to_string_pretty(&metas));
Ok(())
@@ -83,16 +82,21 @@ pub fn save_metas(
//
// All this processing happens on a single thread
// consuming a common queue.
//
// We voluntarily pass a merge_operation ref to guarantee that
// the merge_operation is alive during the process
#[derive(Clone)]
pub struct SegmentUpdater(Arc<InnerSegmentUpdater>);
fn perform_merge(
merge_operation: &MergeOperation,
index: &Index,
mut segment_entries: Vec<SegmentEntry>,
mut merged_segment: Segment,
target_opstamp: u64,
) -> Result<SegmentEntry> {
let target_opstamp = merge_operation.target_opstamp();
// first we need to apply deletes to our segment.
let mut merged_segment = index.new_segment();
// TODO add logging
let schema = index.schema();
@@ -126,19 +130,27 @@ fn perform_merge(
}
struct InnerSegmentUpdater {
// we keep a copy of the current active IndexMeta to
// avoid loading the file everytime we need it in the
// `SegmentUpdater`.
//
// This should be up to date as all update happen through
// the unique active `SegmentUpdater`.
active_metas: RwLock<Arc<IndexMeta>>,
pool: CpuPool,
index: Index,
segment_manager: SegmentManager,
merge_policy: RwLock<Box<MergePolicy>>,
merge_policy: RwLock<Arc<Box<MergePolicy>>>,
merging_thread_id: AtomicUsize,
merging_threads: RwLock<HashMap<usize, JoinHandle<Result<()>>>>,
generation: AtomicUsize,
killed: AtomicBool,
stamper: Stamper,
merge_operations: MergeOperationInventory,
}
impl SegmentUpdater {
pub fn new(
pub fn create(
index: Index,
stamper: Stamper,
delete_cursor: &DeleteCursor,
@@ -149,32 +161,29 @@ impl SegmentUpdater {
.name_prefix("segment_updater")
.pool_size(1)
.create();
let index_meta = index.load_metas()?;
Ok(SegmentUpdater(Arc::new(InnerSegmentUpdater {
active_metas: RwLock::new(Arc::new(index_meta)),
pool,
index,
segment_manager,
merge_policy: RwLock::new(Box::new(DefaultMergePolicy::default())),
merge_policy: RwLock::new(Arc::new(Box::new(DefaultMergePolicy::default()))),
merging_thread_id: AtomicUsize::default(),
merging_threads: RwLock::new(HashMap::new()),
generation: AtomicUsize::default(),
killed: AtomicBool::new(false),
stamper,
merge_operations: Default::default(),
})))
}
pub fn new_segment(&self) -> Segment {
let new_segment = self.0.index.new_segment();
let segment_id = new_segment.id();
self.0.segment_manager.write_segment(segment_id);
new_segment
}
pub fn get_merge_policy(&self) -> Box<MergePolicy> {
self.0.merge_policy.read().unwrap().box_clone()
pub fn get_merge_policy(&self) -> Arc<Box<MergePolicy>> {
self.0.merge_policy.read().unwrap().clone()
}
pub fn set_merge_policy(&self, merge_policy: Box<MergePolicy>) {
*self.0.merge_policy.write().unwrap() = merge_policy;
let arc_merge_policy = Arc::new(merge_policy);
*self.0.merge_policy.write().unwrap() = arc_merge_policy;
}
fn get_merging_thread_id(&self) -> usize {
@@ -195,7 +204,8 @@ impl SegmentUpdater {
segment_updater.0.segment_manager.add_segment(segment_entry);
segment_updater.consider_merge_options();
true
}).forget();
})
.forget();
true
} else {
false
@@ -227,20 +237,39 @@ impl SegmentUpdater {
if self.is_alive() {
let index = &self.0.index;
let directory = index.directory();
save_metas(
self.0.segment_manager.committed_segment_metas(),
index.schema(),
let mut commited_segment_metas = self.0.segment_manager.committed_segment_metas();
// We sort segment_readers by number of documents.
// This is an heuristic to make multithreading more efficient.
//
// This is not done at the searcher level because I had a strange
// use case in which I was dealing with a large static index,
// dispatched over 5 SSD drives.
//
// A `UnionDirectory` makes it possible to read from these
// 5 different drives and creates a meta.json on the fly.
// In order to optimize the throughput, it creates a lasagna of segments
// from the different drives.
//
// Segment 1 from disk 1, Segment 1 from disk 2, etc.
commited_segment_metas.sort_by_key(|segment_meta| -(segment_meta.max_doc() as i32));
let index_meta = IndexMeta {
segments: commited_segment_metas,
schema: index.schema(),
opstamp,
commit_message,
directory.box_clone().borrow_mut(),
).expect("Could not save metas.");
payload: commit_message,
};
save_metas(&index_meta, directory.box_clone().borrow_mut())
.expect("Could not save metas.");
self.store_meta(&index_meta);
}
}
pub fn garbage_collect_files(&self) -> Result<()> {
self.run_async(move |segment_updater| {
segment_updater.garbage_collect_files_exec();
}).wait()
})
.wait()
}
fn garbage_collect_files_exec(&self) {
@@ -262,54 +291,67 @@ impl SegmentUpdater {
segment_updater.garbage_collect_files_exec();
segment_updater.consider_merge_options();
}
}).wait()
})
.wait()
}
pub fn start_merge(&self, segment_ids: &[SegmentId]) -> Result<Receiver<SegmentMeta>> {
//let future_merged_segment = */
let segment_ids_vec = segment_ids.to_vec();
self.run_async(move |segment_updater| {
segment_updater.start_merge_impl(&segment_ids_vec[..])
}).wait()?
let commit_opstamp = self.load_metas().opstamp;
let merge_operation = MergeOperation::new(
&self.0.merge_operations,
commit_opstamp,
segment_ids.to_vec(),
);
self.run_async(move |segment_updater| segment_updater.start_merge_impl(merge_operation))
.wait()?
}
fn store_meta(&self, index_meta: &IndexMeta) {
*self.0.active_metas.write().unwrap() = Arc::new(index_meta.clone());
}
fn load_metas(&self) -> Arc<IndexMeta> {
self.0.active_metas.read().unwrap().clone()
}
// `segment_ids` is required to be non-empty.
fn start_merge_impl(&self, segment_ids: &[SegmentId]) -> Result<Receiver<SegmentMeta>> {
assert!(!segment_ids.is_empty(), "Segment_ids cannot be empty.");
fn start_merge_impl(&self, merge_operation: MergeOperation) -> Result<Receiver<SegmentMeta>> {
assert!(
!merge_operation.segment_ids().is_empty(),
"Segment_ids cannot be empty."
);
let segment_updater_clone = self.clone();
let segment_entries: Vec<SegmentEntry> = self.0.segment_manager.start_merge(segment_ids)?;
let segment_entries: Vec<SegmentEntry> = self
.0
.segment_manager
.start_merge(merge_operation.segment_ids())?;
let segment_ids_vec = segment_ids.to_vec();
// let segment_ids_vec = merge_operation.segment_ids.to_vec();
let merging_thread_id = self.get_merging_thread_id();
info!(
"Starting merge thread #{} - {:?}",
merging_thread_id, segment_ids
merging_thread_id,
merge_operation.segment_ids()
);
let (merging_future_send, merging_future_recv) = oneshot();
let target_opstamp = self.0.stamper.stamp();
// first we need to apply deletes to our segment.
let merging_join_handle = thread::Builder::new()
.name(format!("mergingthread-{}", merging_thread_id))
.spawn(move || {
// first we need to apply deletes to our segment.
let merged_segment = segment_updater_clone.new_segment();
let merged_segment_id = merged_segment.id();
let merge_result = perform_merge(
&merge_operation,
&segment_updater_clone.0.index,
segment_entries,
merged_segment,
target_opstamp,
);
match merge_result {
Ok(after_merge_segment_entry) => {
let merged_segment_meta = after_merge_segment_entry.meta().clone();
segment_updater_clone
.end_merge(segment_ids_vec, after_merge_segment_entry)
.end_merge(merge_operation, after_merge_segment_entry)
.expect("Segment updater thread is corrupted.");
// the future may fail if the listener of the oneshot future
@@ -320,13 +362,18 @@ impl SegmentUpdater {
let _merging_future_res = merging_future_send.send(merged_segment_meta);
}
Err(e) => {
warn!("Merge of {:?} was cancelled: {:?}", segment_ids_vec, e);
warn!(
"Merge of {:?} was cancelled: {:?}",
merge_operation.segment_ids(),
e
);
// ... cancel merge
if cfg!(test) {
panic!("Merge failed.");
}
segment_updater_clone.cancel_merge(&segment_ids_vec, merged_segment_id);
// merging_future_send will be dropped, sending an error to the future.
// As `merge_operation` will be dropped, the segment in merge state will
// be available for merge again.
// `merging_future_send` will be dropped, sending an error to the future.
}
}
segment_updater_clone
@@ -347,16 +394,35 @@ impl SegmentUpdater {
}
fn consider_merge_options(&self) {
let merge_segment_ids: HashSet<SegmentId> = self.0.merge_operations.segment_in_merge();
let (committed_segments, uncommitted_segments) =
get_mergeable_segments(&self.0.segment_manager);
get_mergeable_segments(&merge_segment_ids, &self.0.segment_manager);
// Committed segments cannot be merged with uncommitted_segments.
// We therefore consider merges using these two sets of segments independently.
let merge_policy = self.get_merge_policy();
let mut merge_candidates = merge_policy.compute_merge_candidates(&uncommitted_segments);
let committed_merge_candidates = merge_policy.compute_merge_candidates(&committed_segments);
merge_candidates.extend_from_slice(&committed_merge_candidates[..]);
for MergeCandidate(segment_metas) in merge_candidates {
match self.start_merge_impl(&segment_metas) {
let current_opstamp = self.0.stamper.stamp();
let mut merge_candidates: Vec<MergeOperation> = merge_policy
.compute_merge_candidates(&uncommitted_segments)
.into_iter()
.map(|merge_candidate| {
MergeOperation::new(&self.0.merge_operations, current_opstamp, merge_candidate.0)
})
.collect();
let commit_opstamp = self.load_metas().opstamp;
let committed_merge_candidates = merge_policy
.compute_merge_candidates(&committed_segments)
.into_iter()
.map(|merge_candidate| {
MergeOperation::new(&self.0.merge_operations, commit_opstamp, merge_candidate.0)
})
.collect::<Vec<_>>();
merge_candidates.extend(committed_merge_candidates.into_iter());
for merge_operation in merge_candidates {
match self.start_merge_impl(merge_operation) {
Ok(merge_future) => {
if let Err(e) = merge_future.fuse().poll() {
error!("The merge task failed quickly after starting: {:?}", e);
@@ -372,31 +438,16 @@ impl SegmentUpdater {
}
}
fn cancel_merge(
&self,
before_merge_segment_ids: &[SegmentId],
after_merge_segment_entry: SegmentId,
) {
self.0
.segment_manager
.cancel_merge(before_merge_segment_ids, after_merge_segment_entry);
}
fn end_merge(
&self,
before_merge_segment_ids: Vec<SegmentId>,
merge_operation: MergeOperation,
mut after_merge_segment_entry: SegmentEntry,
) -> Result<()> {
self.run_async(move |segment_updater| {
info!("End merge {:?}", after_merge_segment_entry.meta());
let mut delete_cursor = after_merge_segment_entry.delete_cursor().clone();
if let Some(delete_operation) = delete_cursor.get() {
let committed_opstamp = segment_updater
.0
.index
.load_metas()
.expect("Failed to read opstamp")
.opstamp;
let committed_opstamp = segment_updater.load_metas().opstamp;
if delete_operation.opstamp < committed_opstamp {
let index = &segment_updater.0.index;
let segment = index.segment(after_merge_segment_entry.meta().clone());
@@ -405,16 +456,15 @@ impl SegmentUpdater {
{
error!(
"Merge of {:?} was cancelled (advancing deletes failed): {:?}",
before_merge_segment_ids, e
merge_operation.segment_ids(),
e
);
// ... cancel merge
if cfg!(test) {
panic!("Merge failed.");
}
segment_updater.cancel_merge(
&before_merge_segment_ids,
after_merge_segment_entry.segment_id(),
);
// ... cancel merge
// `merge_operations` are tracked. As it is dropped, the
// the segment_ids will be available again for merge.
return;
}
}
@@ -422,13 +472,14 @@ impl SegmentUpdater {
segment_updater
.0
.segment_manager
.end_merge(&before_merge_segment_ids, after_merge_segment_entry);
.end_merge(merge_operation.segment_ids(), after_merge_segment_entry);
segment_updater.consider_merge_options();
info!("save metas");
let previous_metas = segment_updater.0.index.load_metas().unwrap();
segment_updater.save_metas(previous_metas.opstamp, previous_metas.payload);
let previous_metas = segment_updater.load_metas();
segment_updater.save_metas(previous_metas.opstamp, previous_metas.payload.clone());
segment_updater.garbage_collect_files_exec();
}).wait()
})
.wait()
}
/// Wait for current merging threads.
@@ -447,32 +498,25 @@ impl SegmentUpdater {
/// Obsolete files will eventually be cleaned up
/// by the directory garbage collector.
pub fn wait_merging_thread(&self) -> Result<()> {
let mut num_segments: usize;
loop {
num_segments = self.0.segment_manager.num_segments();
let mut new_merging_threads = HashMap::new();
{
let merging_threads: HashMap<usize, JoinHandle<Result<()>>> = {
let mut merging_threads = self.0.merging_threads.write().unwrap();
mem::swap(&mut new_merging_threads, merging_threads.deref_mut());
mem::replace(merging_threads.deref_mut(), HashMap::new())
};
if merging_threads.is_empty() {
return Ok(());
}
debug!("wait merging thread {}", new_merging_threads.len());
for (_, merging_thread_handle) in new_merging_threads {
debug!("wait merging thread {}", merging_threads.len());
for (_, merging_thread_handle) in merging_threads {
merging_thread_handle
.join()
.map(|_| ())
.map_err(|_| TantivyError::ErrorInThread("Merging thread failed.".into()))?;
}
// Our merging thread may have queued their completed
// Our merging thread may have queued their completed merged segment.
// Let's wait for that too.
self.run_async(move |_| {}).wait()?;
let new_num_segments = self.0.segment_manager.num_segments();
if new_num_segments >= num_segments {
break;
}
}
Ok(())
}
}
@@ -485,14 +529,14 @@ mod tests {
#[test]
fn test_delete_during_merge() {
let mut schema_builder = SchemaBuilder::default();
let mut schema_builder = Schema::builder();
let text_field = schema_builder.add_text_field("text", TEXT);
let schema = schema_builder.build();
let index = Index::create_in_ram(schema);
// writing the segment
let mut index_writer = index.writer_with_num_threads(1, 40_000_000).unwrap();
let mut index_writer = index.writer_with_num_threads(1, 3_000_000).unwrap();
index_writer.set_merge_policy(Box::new(MergeWheneverPossible));
{
@@ -522,9 +566,8 @@ mod tests {
index_writer.delete_term(term);
assert!(index_writer.commit().is_ok());
}
index.load_searchers().unwrap();
assert_eq!(index.searcher().num_docs(), 302);
let reader = index.reader().unwrap();
assert_eq!(reader.searcher().num_docs(), 302);
{
index_writer
@@ -532,8 +575,79 @@ mod tests {
.expect("waiting for merging threads");
}
index.load_searchers().unwrap();
assert_eq!(index.searcher().segment_readers().len(), 1);
assert_eq!(index.searcher().num_docs(), 302);
reader.reload().unwrap();
assert_eq!(reader.searcher().segment_readers().len(), 1);
assert_eq!(reader.searcher().num_docs(), 302);
}
#[test]
fn delete_all_docs() {
let mut schema_builder = Schema::builder();
let text_field = schema_builder.add_text_field("text", TEXT);
let schema = schema_builder.build();
let index = Index::create_in_ram(schema);
// writing the segment
let mut index_writer = index.writer_with_num_threads(1, 3_000_000).unwrap();
{
for _ in 0..100 {
index_writer.add_document(doc!(text_field=>"a"));
index_writer.add_document(doc!(text_field=>"b"));
}
assert!(index_writer.commit().is_ok());
}
{
for _ in 0..100 {
index_writer.add_document(doc!(text_field=>"c"));
index_writer.add_document(doc!(text_field=>"d"));
}
assert!(index_writer.commit().is_ok());
}
{
index_writer.add_document(doc!(text_field=>"e"));
index_writer.add_document(doc!(text_field=>"f"));
assert!(index_writer.commit().is_ok());
}
{
let seg_ids = index
.searchable_segment_ids()
.expect("Searchable segments failed.");
// docs exist, should have at least 1 segment
assert!(seg_ids.len() > 0);
}
{
let term_vals = vec!["a", "b", "c", "d", "e", "f"];
for term_val in term_vals {
let term = Term::from_field_text(text_field, term_val);
index_writer.delete_term(term);
assert!(index_writer.commit().is_ok());
}
}
{
index_writer
.wait_merging_threads()
.expect("waiting for merging threads");
}
let reader = index.reader().unwrap();
assert_eq!(reader.searcher().num_docs(), 0);
let seg_ids = index
.searchable_segment_ids()
.expect("Searchable segments failed.");
assert!(seg_ids.is_empty());
reader.reload().unwrap();
assert_eq!(reader.searcher().num_docs(), 0);
// empty segments should be erased
assert!(index.searchable_segment_metas().unwrap().is_empty());
assert!(reader.searcher().segment_readers().is_empty());
}
}

48
src/indexer/segment_writer.rs
View File

@@ -5,6 +5,7 @@ use fastfield::FastFieldsWriter;
use fieldnorm::FieldNormsWriter;
use indexer::segment_serializer::SegmentSerializer;
use postings::MultiFieldPostingsWriter;
use schema::FieldEntry;
use schema::FieldType;
use schema::Schema;
use schema::Term;
@@ -49,20 +50,21 @@ impl SegmentWriter {
) -> Result<SegmentWriter> {
let segment_serializer = SegmentSerializer::for_segment(&mut segment)?;
let multifield_postings = MultiFieldPostingsWriter::new(schema, table_bits);
let tokenizers = schema
.fields()
.iter()
.map(|field_entry| field_entry.field_type())
.map(|field_type| match *field_type {
FieldType::Str(ref text_options) => text_options.get_indexing_options().and_then(
|text_index_option| {
let tokenizer_name = &text_index_option.tokenizer();
segment.index().tokenizers().get(tokenizer_name)
},
),
_ => None,
})
.collect();
let tokenizers =
schema
.fields()
.iter()
.map(FieldEntry::field_type)
.map(|field_type| match *field_type {
FieldType::Str(ref text_options) => text_options
.get_indexing_options()
.and_then(|text_index_option| {
let tokenizer_name = &text_index_option.tokenizer();
segment.index().tokenizers().get(tokenizer_name)
}),
_ => None,
})
.collect();
Ok(SegmentWriter {
max_doc: 0,
multifield_postings,
@@ -110,19 +112,18 @@ impl SegmentWriter {
}
match *field_options.field_type() {
FieldType::HierarchicalFacet => {
let facets: Vec<&[u8]> = field_values
let facets: Vec<&str> = field_values
.iter()
.flat_map(|field_value| match *field_value.value() {
Value::Facet(ref facet) => Some(facet.encoded_bytes()),
Value::Facet(ref facet) => Some(facet.encoded_str()),
_ => {
panic!("Expected hierarchical facet");
}
})
.collect();
let mut term = Term::for_field(field); // we set the Term
for facet_bytes in facets {
for fake_str in facets {
let mut unordered_term_id_opt = None;
let fake_str = unsafe { str::from_utf8_unchecked(facet_bytes) };
FacetTokenizer.token_stream(fake_str).process(&mut |token| {
term.set_text(&token.text);
let unordered_term_id =
@@ -171,6 +172,17 @@ impl SegmentWriter {
}
}
}
FieldType::Date(ref int_option) => {
if int_option.is_indexed() {
for field_value in field_values {
let term = Term::from_field_i64(
field_value.field(),
field_value.value().date_value().timestamp(),
);
self.multifield_postings.subscribe(doc_id, &term);
}
}
}
FieldType::I64(ref int_option) => {
if int_option.is_indexed() {
for field_value in field_values {

73
src/indexer/stamper.rs
View File

@@ -1,50 +1,77 @@
use std::ops::Range;
use std::sync::atomic::Ordering;
use std::sync::Arc;
// AtomicU64 have not landed in stable.
// For the moment let's just use AtomicUsize on
// x86/64 bit platform, and a mutex on other platform.
#[cfg(target = "x86_64")]
#[cfg(target_arch = "x86_64")]
mod archicture_impl {
use std::sync::atomic::{AtomicUsize, Ordering};
use std::sync::Arc;
#[derive(Clone, Default)]
pub struct Stamper(Arc<AtomicU64>);
#[derive(Default)]
pub struct AtomicU64Ersatz(AtomicUsize);
impl Stamper {
pub fn new(first_opstamp: u64) -> Stamper {
Stamper(Arc::new(AtomicU64::new(first_opstamp)))
impl AtomicU64Ersatz {
pub fn new(first_opstamp: u64) -> AtomicU64Ersatz {
AtomicU64Ersatz(AtomicUsize::new(first_opstamp as usize))
}
pub fn stamp(&self) -> u64 {
self.0.fetch_add(1u64, Ordering::SeqCst) as u64
pub fn fetch_add(&self, val: u64, order: Ordering) -> u64 {
self.0.fetch_add(val as usize, order) as u64
}
}
}
#[cfg(not(target = "x86_64"))]
#[cfg(not(target_arch = "x86_64"))]
mod archicture_impl {
use std::sync::{Arc, Mutex};
use std::sync::atomic::Ordering;
/// Under other architecture, we rely on a mutex.
use std::sync::RwLock;
#[derive(Clone, Default)]
pub struct Stamper(Arc<Mutex<u64>>);
#[derive(Default)]
pub struct AtomicU64Ersatz(RwLock<u64>);
impl Stamper {
pub fn new(first_opstamp: u64) -> Stamper {
Stamper(Arc::new(Mutex::new(first_opstamp)))
impl AtomicU64Ersatz {
pub fn new(first_opstamp: u64) -> AtomicU64Ersatz {
AtomicU64Ersatz(RwLock::new(first_opstamp))
}
pub fn stamp(&self) -> u64 {
let mut guard = self.0.lock().expect("Failed to lock the stamper");
let previous_val = *guard;
*guard = previous_val + 1;
pub fn fetch_add(&self, incr: u64, _order: Ordering) -> u64 {
let mut lock = self.0.write().unwrap();
let previous_val = *lock;
*lock = previous_val + incr;
previous_val
}
}
}
pub use self::archicture_impl::Stamper;
use self::archicture_impl::AtomicU64Ersatz;
#[derive(Clone, Default)]
pub struct Stamper(Arc<AtomicU64Ersatz>);
impl Stamper {
pub fn new(first_opstamp: u64) -> Stamper {
Stamper(Arc::new(AtomicU64Ersatz::new(first_opstamp)))
}
pub fn stamp(&self) -> u64 {
self.0.fetch_add(1u64, Ordering::SeqCst) as u64
}
/// Given a desired count `n`, `stamps` returns an iterator that
/// will supply `n` number of u64 stamps.
pub fn stamps(&self, n: u64) -> Range<u64> {
let start = self.0.fetch_add(n, Ordering::SeqCst);
Range {
start,
end: start + n,
}
}
}
#[cfg(test)]
mod test {
@@ -62,5 +89,7 @@ mod test {
assert_eq!(stamper.stamp(), 10u64);
assert_eq!(stamper_clone.stamp(), 11u64);
assert_eq!(stamper.stamps(3u64), (12..15));
assert_eq!(stamper.stamp(), 15u64);
}
}

506
src/lib.rs Normal file → Executable file
View File

@@ -1,11 +1,7 @@
#![doc(html_logo_url = "http://fulmicoton.com/tantivy-logo/tantivy-logo.png")]
#![cfg_attr(feature = "cargo-clippy", allow(module_inception))]
#![cfg_attr(feature = "cargo-clippy", allow(inline_always))]
#![cfg_attr(all(feature = "unstable", test), feature(test))]
#![cfg_attr(feature = "cargo-clippy", allow(clippy::module_inception))]
#![doc(test(attr(allow(unused_variables), deny(warnings))))]
#![allow(unknown_lints)]
#![allow(new_without_default)]
#![allow(decimal_literal_representation)]
#![warn(missing_docs)]
#![recursion_limit = "80"]
@@ -27,7 +23,8 @@
//! # use tempdir::TempDir;
//! # use tantivy::Index;
//! # use tantivy::schema::*;
//! # use tantivy::collector::TopCollector;
//! # use tantivy::{Score, DocAddress};
//! # use tantivy::collector::TopDocs;
//! # use tantivy::query::QueryParser;
//! #
//! # fn main() {
@@ -49,7 +46,7 @@
//! // in a compressed, row-oriented key-value store.
//! // This store is useful to reconstruct the
//! // documents that were selected during the search phase.
//! let mut schema_builder = SchemaBuilder::default();
//! 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();
@@ -78,9 +75,9 @@
//!
//! // # Searching
//!
//! index.load_searchers()?;
//! let reader = index.reader()?;
//!
//! let searcher = index.searcher();
//! let searcher = reader.searcher();
//!
//! let query_parser = QueryParser::for_index(&index, vec![title, body]);
//!
@@ -89,14 +86,14 @@
//! // A ticket has been opened regarding this problem.
//! let query = query_parser.parse_query("sea whale")?;
//!
//! let mut top_collector = TopCollector::with_limit(10);
//! searcher.search(&*query, &mut top_collector)?;
//! // Perform search.
//! // `topdocs` contains the 10 most relevant doc ids, sorted by decreasing scores...
//! let top_docs: Vec<(Score, DocAddress)> =
//! searcher.search(&query, &TopDocs::with_limit(10))?;
//!
//! // Our top collector now contains the 10
//! // most relevant doc ids...
//! let doc_addresses = top_collector.docs();
//! for doc_address in doc_addresses {
//! let retrieved_doc = searcher.doc(&doc_address)?;
//! for (_score, doc_address) in top_docs {
//! // Retrieve the actual content of documents given its `doc_address`.
//! let retrieved_doc = searcher.doc(doc_address)?;
//! println!("{}", schema.to_json(&retrieved_doc));
//! }
//!
@@ -132,25 +129,24 @@ extern crate base64;
extern crate bit_set;
extern crate bitpacking;
extern crate byteorder;
#[macro_use]
extern crate combine;
extern crate crossbeam;
extern crate crossbeam_channel;
extern crate fnv;
extern crate fst;
extern crate fst_regex;
extern crate futures;
extern crate futures_cpupool;
extern crate htmlescape;
extern crate itertools;
extern crate levenshtein_automata;
#[cfg(feature = "mmap")]
extern crate memmap;
extern crate num_cpus;
extern crate owning_ref;
extern crate regex;
extern crate rust_stemmers;
extern crate scoped_pool;
extern crate serde;
extern crate stable_deref_trait;
extern crate tantivy_fst;
extern crate tempdir;
extern crate tempfile;
extern crate uuid;
@@ -165,14 +161,20 @@ extern crate winapi;
#[cfg(test)]
extern crate rand;
#[cfg(test)]
#[macro_use]
extern crate maplit;
#[cfg(all(test, feature = "unstable"))]
extern crate test;
extern crate tinysegmenter;
#[macro_use]
extern crate downcast_rs;
#[macro_use]
extern crate downcast;
extern crate fail;
#[cfg(feature = "mmap")]
#[cfg(test)]
mod functional_test;
@@ -181,15 +183,19 @@ mod macros;
pub use error::TantivyError;
#[deprecated(since="0.7.0", note="please use `tantivy::TantivyError` instead")]
#[deprecated(since = "0.7.0", note = "please use `tantivy::TantivyError` instead")]
pub use error::TantivyError as Error;
extern crate census;
pub extern crate chrono;
extern crate owned_read;
/// Tantivy result.
pub type Result<T> = std::result::Result<T, error::TantivyError>;
/// Tantivy DateTime
pub type DateTime = chrono::DateTime<chrono::Utc>;
mod common;
mod core;
mod indexer;
@@ -206,9 +212,16 @@ pub(crate) mod positions;
pub mod postings;
pub mod query;
pub mod schema;
pub mod space_usage;
pub mod store;
pub mod termdict;
mod reader;
pub use self::reader::{IndexReader, IndexReaderBuilder, ReloadPolicy};
mod snippet;
pub use self::snippet::{Snippet, SnippetGenerator};
mod docset;
pub use self::docset::{DocSet, SkipResult};
@@ -225,11 +238,7 @@ pub use common::{i64_to_u64, u64_to_i64};
/// Expose the current version of tantivy, as well
/// whether it was compiled with the simd compression.
pub fn version() -> &'static str {
if cfg!(feature = "simdcompression") {
concat!(env!("CARGO_PKG_VERSION"), "-simd")
} else {
concat!(env!("CARGO_PKG_VERSION"), "-nosimd")
}
env!("CARGO_PKG_VERSION")
}
/// Defines tantivy's merging strategy
@@ -261,12 +270,12 @@ impl DocAddress {
/// The segment ordinal is an id identifying the segment
/// hosting the document. It is only meaningful, in the context
/// of a searcher.
pub fn segment_ord(&self) -> SegmentLocalId {
pub fn segment_ord(self) -> SegmentLocalId {
self.0
}
/// Return the segment local `DocId`
pub fn doc(&self) -> DocId {
pub fn doc(self) -> DocId {
self.1
}
}
@@ -290,12 +299,15 @@ mod tests {
use docset::DocSet;
use query::BooleanQuery;
use rand::distributions::Bernoulli;
use rand::distributions::Range;
use rand::{Rng, SeedableRng, XorShiftRng};
use rand::distributions::Uniform;
use rand::rngs::StdRng;
use rand::{Rng, SeedableRng};
use schema::*;
use DocAddress;
use Index;
use IndexWriter;
use Postings;
use ReloadPolicy;
pub fn assert_nearly_equals(expected: f32, val: f32) {
assert!(
@@ -311,16 +323,15 @@ mod tests {
}
pub fn generate_nonunique_unsorted(max_value: u32, n_elems: usize) -> Vec<u32> {
let seed: [u8; 16] = [0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15];
XorShiftRng::from_seed(seed)
.sample_iter(&Range::new(0u32, max_value))
let seed: [u8; 32] = [1; 32];
StdRng::from_seed(seed)
.sample_iter(&Uniform::new(0u32, max_value))
.take(n_elems)
.collect::<Vec<u32>>()
}
pub fn sample_with_seed(n: u32, ratio: f64, seed_val: u8) -> Vec<u32> {
let seed: [u8; 16] = [0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, seed_val];
XorShiftRng::from_seed(seed)
StdRng::from_seed([seed_val; 32])
.sample_iter(&Bernoulli::new(ratio))
.take(n as usize)
.enumerate()
@@ -335,13 +346,13 @@ mod tests {
#[test]
#[cfg(feature = "mmap")]
fn test_indexing() {
let mut schema_builder = SchemaBuilder::default();
let mut schema_builder = Schema::builder();
let text_field = schema_builder.add_text_field("text", TEXT);
let schema = schema_builder.build();
let index = Index::create_from_tempdir(schema).unwrap();
{
// writing the segment
let mut index_writer = index.writer_with_num_threads(1, 40_000_000).unwrap();
let mut index_writer = index.writer_with_num_threads(1, 3_000_000).unwrap();
{
let doc = doc!(text_field=>"af b");
index_writer.add_document(doc);
@@ -360,10 +371,10 @@ mod tests {
#[test]
fn test_docfreq1() {
let mut schema_builder = SchemaBuilder::default();
let mut schema_builder = Schema::builder();
let text_field = schema_builder.add_text_field("text", TEXT);
let index = Index::create_in_ram(schema_builder.build());
let mut index_writer = index.writer_with_num_threads(1, 40_000_000).unwrap();
let mut index_writer = index.writer_with_num_threads(1, 3_000_000).unwrap();
{
index_writer.add_document(doc!(text_field=>"a b c"));
index_writer.commit().unwrap();
@@ -385,8 +396,8 @@ mod tests {
index_writer.commit().unwrap();
}
{
index.load_searchers().unwrap();
let searcher = index.searcher();
let reader = index.reader().unwrap();
let searcher = reader.searcher();
let term_a = Term::from_field_text(text_field, "a");
assert_eq!(searcher.doc_freq(&term_a), 3);
let term_b = Term::from_field_text(text_field, "b");
@@ -400,12 +411,12 @@ mod tests {
#[test]
fn test_fieldnorm_no_docs_with_field() {
let mut schema_builder = SchemaBuilder::default();
let mut schema_builder = Schema::builder();
let title_field = schema_builder.add_text_field("title", TEXT);
let text_field = schema_builder.add_text_field("text", TEXT);
let index = Index::create_in_ram(schema_builder.build());
{
let mut index_writer = index.writer_with_num_threads(1, 40_000_000).unwrap();
let mut index_writer = index.writer_with_num_threads(1, 3_000_000).unwrap();
{
let doc = doc!(text_field=>"a b c");
index_writer.add_document(doc);
@@ -413,8 +424,8 @@ mod tests {
index_writer.commit().unwrap();
}
{
index.load_searchers().unwrap();
let searcher = index.searcher();
let index_reader = index.reader().unwrap();
let searcher = index_reader.searcher();
let reader = searcher.segment_reader(0);
{
let fieldnorm_reader = reader.get_fieldnorms_reader(text_field);
@@ -429,11 +440,11 @@ mod tests {
#[test]
fn test_fieldnorm() {
let mut schema_builder = SchemaBuilder::default();
let mut schema_builder = Schema::builder();
let text_field = schema_builder.add_text_field("text", TEXT);
let index = Index::create_in_ram(schema_builder.build());
{
let mut index_writer = index.writer_with_num_threads(1, 40_000_000).unwrap();
let mut index_writer = index.writer_with_num_threads(1, 3_000_000).unwrap();
{
let doc = doc!(text_field=>"a b c");
index_writer.add_document(doc);
@@ -449,8 +460,8 @@ mod tests {
index_writer.commit().unwrap();
}
{
index.load_searchers().unwrap();
let searcher = index.searcher();
let reader = index.reader().unwrap();
let searcher = reader.searcher();
let segment_reader: &SegmentReader = searcher.segment_reader(0);
let fieldnorms_reader = segment_reader.get_fieldnorms_reader(text_field);
assert_eq!(fieldnorms_reader.fieldnorm(0), 3);
@@ -470,7 +481,7 @@ mod tests {
#[test]
fn test_delete_postings1() {
let mut schema_builder = SchemaBuilder::default();
let mut schema_builder = Schema::builder();
let text_field = schema_builder.add_text_field("text", TEXT);
let term_abcd = Term::from_field_text(text_field, "abcd");
let term_a = Term::from_field_text(text_field, "a");
@@ -478,183 +489,151 @@ mod tests {
let term_c = Term::from_field_text(text_field, "c");
let schema = schema_builder.build();
let index = Index::create_in_ram(schema);
let reader = index
.reader_builder()
.reload_policy(ReloadPolicy::Manual)
.try_into()
.unwrap();
{
// writing the segment
let mut index_writer = index.writer_with_num_threads(1, 40_000_000).unwrap();
{
// 0
let doc = doc!(text_field=>"a b");
index_writer.add_document(doc);
}
{
// 1
let doc = doc!(text_field=>" a c");
index_writer.add_document(doc);
}
{
// 2
let doc = doc!(text_field=>" b c");
index_writer.add_document(doc);
}
{
// 3
let doc = doc!(text_field=>" b d");
index_writer.add_document(doc);
}
{
index_writer.delete_term(Term::from_field_text(text_field, "c"));
}
{
index_writer.delete_term(Term::from_field_text(text_field, "a"));
}
{
// 4
let doc = doc!(text_field=>" b c");
index_writer.add_document(doc);
}
{
// 5
let doc = doc!(text_field=>" a");
index_writer.add_document(doc);
}
let mut index_writer = index.writer_with_num_threads(1, 3_000_000).unwrap();
// 0
index_writer.add_document(doc!(text_field=>"a b"));
// 1
index_writer.add_document(doc!(text_field=>" a c"));
// 2
index_writer.add_document(doc!(text_field=>" b c"));
// 3
index_writer.add_document(doc!(text_field=>" b d"));
index_writer.delete_term(Term::from_field_text(text_field, "c"));
index_writer.delete_term(Term::from_field_text(text_field, "a"));
// 4
index_writer.add_document(doc!(text_field=>" b c"));
// 5
index_writer.add_document(doc!(text_field=>" a"));
index_writer.commit().unwrap();
}
{
index.load_searchers().unwrap();
let searcher = index.searcher();
let reader = searcher.segment_reader(0);
let inverted_index = reader.inverted_index(text_field);
assert!(
inverted_index
.read_postings(&term_abcd, IndexRecordOption::WithFreqsAndPositions)
.is_none()
);
reader.reload().unwrap();
let searcher = reader.searcher();
let segment_reader = searcher.segment_reader(0);
let inverted_index = segment_reader.inverted_index(text_field);
assert!(inverted_index
.read_postings(&term_abcd, IndexRecordOption::WithFreqsAndPositions)
.is_none());
{
let mut postings = inverted_index
.read_postings(&term_a, IndexRecordOption::WithFreqsAndPositions)
.unwrap();
assert!(advance_undeleted(&mut postings, reader));
assert!(advance_undeleted(&mut postings, segment_reader));
assert_eq!(postings.doc(), 5);
assert!(!advance_undeleted(&mut postings, reader));
assert!(!advance_undeleted(&mut postings, segment_reader));
}
{
let mut postings = inverted_index
.read_postings(&term_b, IndexRecordOption::WithFreqsAndPositions)
.unwrap();
assert!(advance_undeleted(&mut postings, reader));
assert!(advance_undeleted(&mut postings, segment_reader));
assert_eq!(postings.doc(), 3);
assert!(advance_undeleted(&mut postings, reader));
assert!(advance_undeleted(&mut postings, segment_reader));
assert_eq!(postings.doc(), 4);
assert!(!advance_undeleted(&mut postings, reader));
assert!(!advance_undeleted(&mut postings, segment_reader));
}
}
{
// writing the segment
let mut index_writer = index.writer_with_num_threads(1, 40_000_000).unwrap();
{
// 0
let doc = doc!(text_field=>"a b");
index_writer.add_document(doc);
}
{
// 1
index_writer.delete_term(Term::from_field_text(text_field, "c"));
}
let mut index_writer = index.writer_with_num_threads(1, 3_000_000).unwrap();
// 0
index_writer.add_document(doc!(text_field=>"a b"));
// 1
index_writer.delete_term(Term::from_field_text(text_field, "c"));
index_writer.rollback().unwrap();
}
{
index.load_searchers().unwrap();
let searcher = index.searcher();
let reader = searcher.segment_reader(0);
let inverted_index = reader.inverted_index(term_abcd.field());
reader.reload().unwrap();
let searcher = reader.searcher();
let seg_reader = searcher.segment_reader(0);
let inverted_index = seg_reader.inverted_index(term_abcd.field());
assert!(
inverted_index
.read_postings(&term_abcd, IndexRecordOption::WithFreqsAndPositions)
.is_none()
);
assert!(inverted_index
.read_postings(&term_abcd, IndexRecordOption::WithFreqsAndPositions)
.is_none());
{
let mut postings = inverted_index
.read_postings(&term_a, IndexRecordOption::WithFreqsAndPositions)
.unwrap();
assert!(advance_undeleted(&mut postings, reader));
assert!(advance_undeleted(&mut postings, seg_reader));
assert_eq!(postings.doc(), 5);
assert!(!advance_undeleted(&mut postings, reader));
assert!(!advance_undeleted(&mut postings, seg_reader));
}
{
let mut postings = inverted_index
.read_postings(&term_b, IndexRecordOption::WithFreqsAndPositions)
.unwrap();
assert!(advance_undeleted(&mut postings, reader));
assert!(advance_undeleted(&mut postings, seg_reader));
assert_eq!(postings.doc(), 3);
assert!(advance_undeleted(&mut postings, reader));
assert!(advance_undeleted(&mut postings, seg_reader));
assert_eq!(postings.doc(), 4);
assert!(!advance_undeleted(&mut postings, reader));
assert!(!advance_undeleted(&mut postings, seg_reader));
}
}
{
// writing the segment
let mut index_writer = index.writer_with_num_threads(1, 40_000_000).unwrap();
{
let doc = doc!(text_field=>"a b");
index_writer.add_document(doc);
}
{
index_writer.delete_term(Term::from_field_text(text_field, "c"));
}
let mut index_writer = index.writer_with_num_threads(1, 3_000_000).unwrap();
index_writer.add_document(doc!(text_field=>"a b"));
index_writer.delete_term(Term::from_field_text(text_field, "c"));
index_writer.rollback().unwrap();
index_writer.delete_term(Term::from_field_text(text_field, "a"));
index_writer.commit().unwrap();
}
{
index.load_searchers().unwrap();
let searcher = index.searcher();
let reader = searcher.segment_reader(0);
let inverted_index = reader.inverted_index(term_abcd.field());
assert!(
inverted_index
.read_postings(&term_abcd, IndexRecordOption::WithFreqsAndPositions)
.is_none()
);
reader.reload().unwrap();
let searcher = reader.searcher();
let segment_reader = searcher.segment_reader(0);
let inverted_index = segment_reader.inverted_index(term_abcd.field());
assert!(inverted_index
.read_postings(&term_abcd, IndexRecordOption::WithFreqsAndPositions)
.is_none());
{
let mut postings = inverted_index
.read_postings(&term_a, IndexRecordOption::WithFreqsAndPositions)
.unwrap();
assert!(!advance_undeleted(&mut postings, reader));
assert!(!advance_undeleted(&mut postings, segment_reader));
}
{
let mut postings = inverted_index
.read_postings(&term_b, IndexRecordOption::WithFreqsAndPositions)
.unwrap();
assert!(advance_undeleted(&mut postings, reader));
assert!(advance_undeleted(&mut postings, segment_reader));
assert_eq!(postings.doc(), 3);
assert!(advance_undeleted(&mut postings, reader));
assert!(advance_undeleted(&mut postings, segment_reader));
assert_eq!(postings.doc(), 4);
assert!(!advance_undeleted(&mut postings, reader));
assert!(!advance_undeleted(&mut postings, segment_reader));
}
{
let mut postings = inverted_index
.read_postings(&term_c, IndexRecordOption::WithFreqsAndPositions)
.unwrap();
assert!(advance_undeleted(&mut postings, reader));
assert!(advance_undeleted(&mut postings, segment_reader));
assert_eq!(postings.doc(), 4);
assert!(!advance_undeleted(&mut postings, reader));
assert!(!advance_undeleted(&mut postings, segment_reader));
}
}
}
#[test]
fn test_indexed_u64() {
let mut schema_builder = SchemaBuilder::default();
let field = schema_builder.add_u64_field("value", INT_INDEXED);
let mut schema_builder = Schema::builder();
let field = schema_builder.add_u64_field("value", INDEXED);
let schema = schema_builder.build();
let index = Index::create_in_ram(schema);
let mut index_writer = index.writer_with_num_threads(1, 40_000_000).unwrap();
let mut index_writer = index.writer_with_num_threads(1, 3_000_000).unwrap();
index_writer.add_document(doc!(field=>1u64));
index_writer.commit().unwrap();
index.load_searchers().unwrap();
let searcher = index.searcher();
let reader = index.reader().unwrap();
let searcher = reader.searcher();
let term = Term::from_field_u64(field, 1u64);
let mut postings = searcher
.segment_reader(0)
@@ -668,17 +647,17 @@ mod tests {
#[test]
fn test_indexed_i64() {
let mut schema_builder = SchemaBuilder::default();
let value_field = schema_builder.add_i64_field("value", INT_INDEXED);
let mut schema_builder = Schema::builder();
let value_field = schema_builder.add_i64_field("value", INDEXED);
let schema = schema_builder.build();
let index = Index::create_in_ram(schema);
let mut index_writer = index.writer_with_num_threads(1, 40_000_000).unwrap();
let mut index_writer = index.writer_with_num_threads(1, 3_000_000).unwrap();
let negative_val = -1i64;
index_writer.add_document(doc!(value_field => negative_val));
index_writer.commit().unwrap();
index.load_searchers().unwrap();
let searcher = index.searcher();
let reader = index.reader().unwrap();
let searcher = reader.searcher();
let term = Term::from_field_i64(value_field, negative_val);
let mut postings = searcher
.segment_reader(0)
@@ -692,29 +671,34 @@ mod tests {
#[test]
fn test_indexedfield_not_in_documents() {
let mut schema_builder = SchemaBuilder::default();
let mut schema_builder = Schema::builder();
let text_field = schema_builder.add_text_field("text", TEXT);
let absent_field = schema_builder.add_text_field("text", TEXT);
let schema = schema_builder.build();
let index = Index::create_in_ram(schema);
let mut index_writer = index.writer_with_num_threads(2, 40_000_000).unwrap();
let mut index_writer = index.writer_with_num_threads(2, 6_000_000).unwrap();
index_writer.add_document(doc!(text_field=>"a"));
assert!(index_writer.commit().is_ok());
assert!(index.load_searchers().is_ok());
let searcher = index.searcher();
let reader = index.reader().unwrap();
let searcher = reader.searcher();
let segment_reader = searcher.segment_reader(0);
segment_reader.inverted_index(absent_field); //< should not panic
}
#[test]
fn test_delete_postings2() {
let mut schema_builder = SchemaBuilder::default();
let mut schema_builder = Schema::builder();
let text_field = schema_builder.add_text_field("text", TEXT);
let schema = schema_builder.build();
let index = Index::create_in_ram(schema);
let reader = index
.reader_builder()
.reload_policy(ReloadPolicy::Manual)
.try_into()
.unwrap();
// writing the segment
let mut index_writer = index.writer_with_num_threads(2, 40_000_000).unwrap();
let mut index_writer = index.writer_with_num_threads(2, 6_000_000).unwrap();
let add_document = |index_writer: &mut IndexWriter, val: &'static str| {
let doc = doc!(text_field=>val);
@@ -737,20 +721,20 @@ mod tests {
remove_document(&mut index_writer, "38");
remove_document(&mut index_writer, "34");
index_writer.commit().unwrap();
index.load_searchers().unwrap();
let searcher = index.searcher();
reader.reload().unwrap();
let searcher = reader.searcher();
assert_eq!(searcher.num_docs(), 6);
}
#[test]
fn test_termfreq() {
let mut schema_builder = SchemaBuilder::default();
let mut schema_builder = Schema::builder();
let text_field = schema_builder.add_text_field("text", TEXT);
let schema = schema_builder.build();
let index = Index::create_in_ram(schema);
{
// writing the segment
let mut index_writer = index.writer_with_num_threads(1, 40_000_000).unwrap();
let mut index_writer = index.writer_with_num_threads(1, 3_000_000).unwrap();
{
let doc = doc!(text_field=>"af af af bc bc");
index_writer.add_document(doc);
@@ -758,16 +742,14 @@ mod tests {
index_writer.commit().unwrap();
}
{
index.load_searchers().unwrap();
let searcher = index.searcher();
let index_reader = index.reader().unwrap();
let searcher = index_reader.searcher();
let reader = searcher.segment_reader(0);
let inverted_index = reader.inverted_index(text_field);
let term_abcd = Term::from_field_text(text_field, "abcd");
assert!(
inverted_index
.read_postings(&term_abcd, IndexRecordOption::WithFreqsAndPositions)
.is_none()
);
assert!(inverted_index
.read_postings(&term_abcd, IndexRecordOption::WithFreqsAndPositions)
.is_none());
let term_af = Term::from_field_text(text_field, "af");
let mut postings = inverted_index
.read_postings(&term_af, IndexRecordOption::WithFreqsAndPositions)
@@ -781,109 +763,84 @@ mod tests {
#[test]
fn test_searcher_1() {
let mut schema_builder = SchemaBuilder::default();
let mut schema_builder = Schema::builder();
let text_field = schema_builder.add_text_field("text", TEXT);
let schema = schema_builder.build();
let index = Index::create_in_ram(schema);
let reader = index.reader().unwrap();
{
// writing the segment
let mut index_writer = index.writer_with_num_threads(1, 40_000_000).unwrap();
{
let doc = doc!(text_field=>"af af af b");
index_writer.add_document(doc);
}
{
let doc = doc!(text_field=>"a b c");
index_writer.add_document(doc);
}
{
let doc = doc!(text_field=>"a b c d");
index_writer.add_document(doc);
}
let mut index_writer = index.writer_with_num_threads(1, 3_000_000).unwrap();
index_writer.add_document(doc!(text_field=>"af af af b"));
index_writer.add_document(doc!(text_field=>"a b c"));
index_writer.add_document(doc!(text_field=>"a b c d"));
index_writer.commit().unwrap();
}
{
index.load_searchers().unwrap();
let searcher = index.searcher();
reader.reload().unwrap();
let searcher = reader.searcher();
let get_doc_ids = |terms: Vec<Term>| {
let query = BooleanQuery::new_multiterms_query(terms);
let mut collector = TestCollector::default();
assert!(searcher.search(&query, &mut collector).is_ok());
collector.docs()
let topdocs = searcher.search(&query, &TestCollector).unwrap();
topdocs.docs().to_vec()
};
{
assert_eq!(
get_doc_ids(vec![Term::from_field_text(text_field, "a")]),
vec![1, 2]
);
}
{
assert_eq!(
get_doc_ids(vec![Term::from_field_text(text_field, "af")]),
vec![0]
);
}
{
assert_eq!(
get_doc_ids(vec![Term::from_field_text(text_field, "b")]),
vec![0, 1, 2]
);
}
{
assert_eq!(
get_doc_ids(vec![Term::from_field_text(text_field, "c")]),
vec![1, 2]
);
}
{
assert_eq!(
get_doc_ids(vec![Term::from_field_text(text_field, "d")]),
vec![2]
);
}
{
assert_eq!(
get_doc_ids(vec![
Term::from_field_text(text_field, "b"),
Term::from_field_text(text_field, "a"),
]),
vec![0, 1, 2]
);
}
assert_eq!(
get_doc_ids(vec![Term::from_field_text(text_field, "a")]),
vec![DocAddress(0, 1), DocAddress(0, 2)]
);
assert_eq!(
get_doc_ids(vec![Term::from_field_text(text_field, "af")]),
vec![DocAddress(0, 0)]
);
assert_eq!(
get_doc_ids(vec![Term::from_field_text(text_field, "b")]),
vec![DocAddress(0, 0), DocAddress(0, 1), DocAddress(0, 2)]
);
assert_eq!(
get_doc_ids(vec![Term::from_field_text(text_field, "c")]),
vec![DocAddress(0, 1), DocAddress(0, 2)]
);
assert_eq!(
get_doc_ids(vec![Term::from_field_text(text_field, "d")]),
vec![DocAddress(0, 2)]
);
assert_eq!(
get_doc_ids(vec![
Term::from_field_text(text_field, "b"),
Term::from_field_text(text_field, "a"),
]),
vec![DocAddress(0, 0), DocAddress(0, 1), DocAddress(0, 2)]
);
}
}
#[test]
fn test_searcher_2() {
let mut schema_builder = SchemaBuilder::default();
let mut schema_builder = Schema::builder();
let text_field = schema_builder.add_text_field("text", TEXT);
let schema = schema_builder.build();
let index = Index::create_in_ram(schema);
let reader = index
.reader_builder()
.reload_policy(ReloadPolicy::Manual)
.try_into()
.unwrap();
assert_eq!(reader.searcher().num_docs(), 0u64);
{
// writing the segment
let mut index_writer = index.writer_with_num_threads(1, 40_000_000).unwrap();
{
let doc = doc!(text_field=>"af b");
index_writer.add_document(doc);
}
{
let doc = doc!(text_field=>"a b c");
index_writer.add_document(doc);
}
{
let doc = doc!(text_field=>"a b c d");
index_writer.add_document(doc);
}
let mut index_writer = index.writer_with_num_threads(1, 3_000_000).unwrap();
index_writer.add_document(doc!(text_field=>"af b"));
index_writer.add_document(doc!(text_field=>"a b c"));
index_writer.add_document(doc!(text_field=>"a b c d"));
index_writer.commit().unwrap();
}
index.searcher();
reader.reload().unwrap();
assert_eq!(reader.searcher().num_docs(), 3u64);
}
#[test]
fn test_doc_macro() {
let mut schema_builder = SchemaBuilder::default();
let mut schema_builder = Schema::builder();
let text_field = schema_builder.add_text_field("text", TEXT);
let other_text_field = schema_builder.add_text_field("text2", TEXT);
let document = doc!(text_field => "tantivy",
@@ -892,20 +849,20 @@ mod tests {
assert_eq!(document.len(), 3);
let values = document.get_all(text_field);
assert_eq!(values.len(), 2);
assert_eq!(values[0].text(), "tantivy");
assert_eq!(values[1].text(), "some other value");
assert_eq!(values[0].text(), Some("tantivy"));
assert_eq!(values[1].text(), Some("some other value"));
let values = document.get_all(other_text_field);
assert_eq!(values.len(), 1);
assert_eq!(values[0].text(), "short");
assert_eq!(values[0].text(), Some("short"));
}
#[test]
fn test_wrong_fast_field_type() {
let mut schema_builder = SchemaBuilder::default();
let mut schema_builder = Schema::builder();
let fast_field_unsigned = schema_builder.add_u64_field("unsigned", FAST);
let fast_field_signed = schema_builder.add_i64_field("signed", FAST);
let text_field = schema_builder.add_text_field("text", TEXT);
let stored_int_field = schema_builder.add_u64_field("text", INT_STORED);
let stored_int_field = schema_builder.add_u64_field("text", STORED);
let schema = schema_builder.build();
let index = Index::create_in_ram(schema);
@@ -915,33 +872,32 @@ mod tests {
index_writer.add_document(document);
index_writer.commit().unwrap();
}
index.load_searchers().unwrap();
let searcher = index.searcher();
let reader = index.reader().unwrap();
let searcher = reader.searcher();
let segment_reader: &SegmentReader = searcher.segment_reader(0);
{
let fast_field_reader_res = segment_reader.fast_field_reader::<u64>(text_field);
assert!(fast_field_reader_res.is_err());
let fast_field_reader_opt = segment_reader.fast_fields().u64(text_field);
assert!(fast_field_reader_opt.is_none());
}
{
let fast_field_reader_res = segment_reader.fast_field_reader::<u64>(stored_int_field);
assert!(fast_field_reader_res.is_err());
let fast_field_reader_opt = segment_reader.fast_fields().u64(stored_int_field);
assert!(fast_field_reader_opt.is_none());
}
{
let fast_field_reader_res = segment_reader.fast_field_reader::<u64>(fast_field_signed);
assert!(fast_field_reader_res.is_err());
let fast_field_reader_opt = segment_reader.fast_fields().u64(fast_field_signed);
assert!(fast_field_reader_opt.is_none());
}
{
let fast_field_reader_res = segment_reader.fast_field_reader::<i64>(fast_field_signed);
assert!(fast_field_reader_res.is_ok());
let fast_field_reader = fast_field_reader_res.unwrap();
let fast_field_reader_opt = segment_reader.fast_fields().i64(fast_field_signed);
assert!(fast_field_reader_opt.is_some());
let fast_field_reader = fast_field_reader_opt.unwrap();
assert_eq!(fast_field_reader.get(0), 4i64)
}
{
let fast_field_reader_res = segment_reader.fast_field_reader::<i64>(fast_field_signed);
assert!(fast_field_reader_res.is_ok());
let fast_field_reader = fast_field_reader_res.unwrap();
let fast_field_reader_opt = segment_reader.fast_fields().i64(fast_field_signed);
assert!(fast_field_reader_opt.is_some());
let fast_field_reader = fast_field_reader_opt.unwrap();
assert_eq!(fast_field_reader.get(0), 4i64)
}
}

28
src/macros.rs
View File

@@ -26,12 +26,12 @@
/// #[macro_use]
/// extern crate tantivy;
///
/// use tantivy::schema::{SchemaBuilder, TEXT, FAST};
/// use tantivy::schema::{Schema, TEXT, FAST};
///
/// //...
///
/// # fn main() {
/// let mut schema_builder = SchemaBuilder::new();
/// let mut schema_builder = Schema::builder();
/// let title = schema_builder.add_text_field("title", TEXT);
/// let author = schema_builder.add_text_field("text", TEXT);
/// let likes = schema_builder.add_u64_field("num_u64", FAST);
@@ -61,39 +61,39 @@ macro_rules! doc(
};
// if there is a trailing comma retry with the trailing comma stripped.
($($field:expr => $value:expr),+ ,) => {
doc!( $( $field => $value ), *);
doc!( $( $field => $value ), *)
};
);
#[cfg(test)]
mod test {
use schema::{SchemaBuilder, FAST, TEXT};
use schema::{Schema, FAST, TEXT};
#[test]
fn test_doc_basic() {
let mut schema_builder = SchemaBuilder::new();
let mut schema_builder = Schema::builder();
let title = schema_builder.add_text_field("title", TEXT);
let author = schema_builder.add_text_field("text", TEXT);
let likes = schema_builder.add_u64_field("num_u64", FAST);
let _schema = schema_builder.build();
let _doc = doc!(
title => "Life Aquatic",
author => "Wes Anderson",
likes => 4u64
);
title => "Life Aquatic",
author => "Wes Anderson",
likes => 4u64
);
}
#[test]
fn test_doc_trailing_comma() {
let mut schema_builder = SchemaBuilder::new();
let mut schema_builder = Schema::builder();
let title = schema_builder.add_text_field("title", TEXT);
let author = schema_builder.add_text_field("text", TEXT);
let likes = schema_builder.add_u64_field("num_u64", FAST);
let _schema = schema_builder.build();
let _doc = doc!(
title => "Life Aquatic",
author => "Wes Anderson",
likes => 4u64,
);
title => "Life Aquatic",
author => "Wes Anderson",
likes => 4u64,
);
}
}

32
src/positions/mod.rs
View File

@@ -1,4 +1,3 @@
/// Positions are stored in three parts and over two files.
//
/// The `SegmentComponent::POSITIONS` file contains all of the bitpacked positions delta,
@@ -24,29 +23,24 @@
/// The long skip structure makes it possible to skip rapidly to the a checkpoint close to this
/// value, and then skip normally.
///
mod reader;
mod serializer;
pub use self::reader::PositionReader;
pub use self::serializer::PositionSerializer;
use bitpacking::{BitPacker4x, BitPacker};
use bitpacking::{BitPacker, BitPacker4x};
const COMPRESSION_BLOCK_SIZE: usize = BitPacker4x::BLOCK_LEN;
const LONG_SKIP_IN_BLOCKS: usize = 1_024;
const LONG_SKIP_INTERVAL: u64 = (LONG_SKIP_IN_BLOCKS * COMPRESSION_BLOCK_SIZE) as u64;
lazy_static! {
static ref BIT_PACKER: BitPacker4x = BitPacker4x::new();
}
#[cfg(test)]
pub mod tests {
use std::iter;
use super::{PositionSerializer, PositionReader};
use super::{PositionReader, PositionSerializer};
use directory::ReadOnlySource;
use positions::COMPRESSION_BLOCK_SIZE;
use std::iter;
fn create_stream_buffer(vals: &[u32]) -> (ReadOnlySource, ReadOnlySource) {
let mut skip_buffer = vec![];
@@ -59,7 +53,10 @@ pub mod tests {
}
serializer.close().unwrap();
}
(ReadOnlySource::from(stream_buffer), ReadOnlySource::from(skip_buffer))
(
ReadOnlySource::from(stream_buffer),
ReadOnlySource::from(skip_buffer),
)
}
#[test]
@@ -103,7 +100,7 @@ pub mod tests {
assert_eq!(skip.len(), 12);
assert_eq!(stream.len(), 1168);
let mut position_reader = PositionReader::new(stream,skip, 0u64);
let mut position_reader = PositionReader::new(stream, skip, 0u64);
let mut buf = [0u32; 7];
let mut c = 0;
for _ in 0..100 {
@@ -125,7 +122,7 @@ pub mod tests {
let (stream, skip) = create_stream_buffer(&v[..]);
assert_eq!(skip.len(), 15_749);
assert_eq!(stream.len(), 1_000_000);
let mut position_reader = PositionReader::new(stream,skip, 128 * 1024);
let mut position_reader = PositionReader::new(stream, skip, 128 * 1024);
let mut buf = [0u32; 1];
position_reader.read(&mut buf);
assert_eq!(buf[0], CONST_VAL);
@@ -137,12 +134,17 @@ pub mod tests {
let (stream, skip) = create_stream_buffer(&v[..]);
assert_eq!(skip.len(), 15_749);
assert_eq!(stream.len(), 4_987_872);
for &offset in &[10, 128 * 1024, 128 * 1024 - 1, 128 * 1024 + 7, 128 * 10 * 1024 + 10] {
let mut position_reader = PositionReader::new(stream.clone(),skip.clone(), offset);
for &offset in &[
10,
128 * 1024,
128 * 1024 - 1,
128 * 1024 + 7,
128 * 10 * 1024 + 10,
] {
let mut position_reader = PositionReader::new(stream.clone(), skip.clone(), offset);
let mut buf = [0u32; 1];
position_reader.read(&mut buf);
assert_eq!(buf[0], offset as u32);
}
}
}

190
src/positions/reader.rs
View File

@@ -1,16 +1,91 @@
use bitpacking::{BitPacker4x, BitPacker};
use owned_read::OwnedRead;
/// Positions works as a long sequence of compressed block.
/// All terms are chained one after the other.
///
/// When accessing the position of a term, we get a positions_idx from the `Terminfo`.
/// This means we need to skip to the `nth` positions efficiently.
///
/// This is done thanks to two levels of skiping that we refer to in the code
/// as `long_skip` and `short_skip`.
///
/// The `long_skip` makes it possible to skip every 1_024 compression blocks (= 131_072 positions).
/// Skipping offset are simply stored one after as an offset stored over 8 bytes.
///
/// We find the number of long skips, as `n / long_skip`.
///
/// Blocks are compressed using bitpacking, so `skip_read` contains the number of bytes
/// (values can go from 0bit to 32 bits) required to decompressed every block.
///
/// A given block obviously takes `(128 x num_bit_for_the_block / num_bits_in_a_byte)`,
/// so skipping a block without decompressing it is just a matter of advancing that many
/// bytes.
use bitpacking::{BitPacker, BitPacker4x};
use common::{BinarySerializable, FixedSize};
use postings::compression::compressed_block_size;
use directory::ReadOnlySource;
use owned_read::OwnedRead;
use positions::COMPRESSION_BLOCK_SIZE;
use positions::LONG_SKIP_IN_BLOCKS;
use positions::LONG_SKIP_INTERVAL;
use super::BIT_PACKER;
use positions::LONG_SKIP_IN_BLOCKS;
use postings::compression::compressed_block_size;
struct Positions {
bit_packer: BitPacker4x,
skip_source: ReadOnlySource,
position_source: ReadOnlySource,
long_skip_source: ReadOnlySource,
}
impl Positions {
pub fn new(position_source: ReadOnlySource, skip_source: ReadOnlySource) -> Positions {
let skip_len = skip_source.len();
let (body, footer) = skip_source.split(skip_len - u32::SIZE_IN_BYTES);
let num_long_skips = u32::deserialize(&mut footer.as_slice()).expect("Index corrupted");
let body_split = body.len() - u64::SIZE_IN_BYTES * (num_long_skips as usize);
let (skip_source, long_skip_source) = body.split(body_split);
Positions {
bit_packer: BitPacker4x::new(),
skip_source,
long_skip_source,
position_source,
}
}
/// Returns the offset of the block associated to the given `long_skip_id`.
///
/// One `long_skip_id` means `LONG_SKIP_IN_BLOCKS` blocks.
fn long_skip(&self, long_skip_id: usize) -> u64 {
if long_skip_id == 0 {
return 0;
}
let long_skip_slice = self.long_skip_source.as_slice();
let mut long_skip_blocks: &[u8] = &long_skip_slice[(long_skip_id - 1) * 8..][..8];
u64::deserialize(&mut long_skip_blocks).expect("Index corrupted")
}
fn reader(&self, offset: u64) -> PositionReader {
let long_skip_id = (offset / LONG_SKIP_INTERVAL) as usize;
let small_skip = (offset % LONG_SKIP_INTERVAL) as usize;
let offset_num_bytes: u64 = self.long_skip(long_skip_id);
let mut position_read = OwnedRead::new(self.position_source.clone());
position_read.advance(offset_num_bytes as usize);
let mut skip_read = OwnedRead::new(self.skip_source.clone());
skip_read.advance(long_skip_id * LONG_SKIP_IN_BLOCKS);
let mut position_reader = PositionReader {
bit_packer: self.bit_packer,
skip_read,
position_read,
inner_offset: 0,
buffer: Box::new([0u32; 128]),
ahead: None,
};
position_reader.skip(small_skip);
position_reader
}
}
pub struct PositionReader {
skip_read: OwnedRead,
position_read: OwnedRead,
bit_packer: BitPacker4x,
inner_offset: usize,
buffer: Box<[u32; 128]>,
ahead: Option<usize>, // if None, no block is loaded.
@@ -18,7 +93,6 @@ pub struct PositionReader {
// of the block of the next int to read.
}
// `ahead` represents the offset of the block currently loaded
// compared to the cursor of the actual stream.
//
@@ -28,67 +102,44 @@ pub struct PositionReader {
// If the requested number of els ends exactly at a given block, the next
// block is not decompressed.
fn read_impl(
bit_packer: BitPacker4x,
mut position: &[u8],
buffer: &mut [u32; 128],
mut inner_offset: usize,
num_bits: &[u8],
output: &mut [u32]) -> usize {
output: &mut [u32],
) -> usize {
let mut output_start = 0;
let mut output_len = output.len();
let mut ahead = 0;
loop {
let available_len = 128 - inner_offset;
let available_len = COMPRESSION_BLOCK_SIZE - inner_offset;
// We have enough elements in the current block.
// Let's copy the requested elements in the output buffer,
// and return.
if output_len <= available_len {
output[output_start..].copy_from_slice(&buffer[inner_offset..][..output_len]);
return ahead;
} else {
output[output_start..][..available_len].copy_from_slice(&buffer[inner_offset..]);
output_len -= available_len;
output_start += available_len;
inner_offset = 0;
let num_bits = num_bits[ahead];
BitPacker4x::new()
.decompress(position, &mut buffer[..], num_bits);
let block_len = compressed_block_size(num_bits);
position = &position[block_len..];
ahead += 1;
}
output[output_start..][..available_len].copy_from_slice(&buffer[inner_offset..]);
output_len -= available_len;
output_start += available_len;
inner_offset = 0;
let num_bits = num_bits[ahead];
bit_packer.decompress(position, &mut buffer[..], num_bits);
let block_len = compressed_block_size(num_bits);
position = &position[block_len..];
ahead += 1;
}
}
impl PositionReader {
pub fn new(position_source: ReadOnlySource,
skip_source: ReadOnlySource,
offset: u64) -> PositionReader {
let skip_len = skip_source.len();
let (body, footer) = skip_source.split(skip_len - u32::SIZE_IN_BYTES);
let num_long_skips = u32::deserialize(&mut footer.as_slice()).expect("Index corrupted");
let body_split = body.len() - u64::SIZE_IN_BYTES * (num_long_skips as usize);
let (skip_body, long_skips) = body.split(body_split);
let long_skip_id = (offset / LONG_SKIP_INTERVAL) as usize;
let small_skip = (offset - (long_skip_id as u64) * (LONG_SKIP_INTERVAL as u64)) as usize;
let offset_num_bytes: u64 = {
if long_skip_id > 0 {
let mut long_skip_blocks: &[u8] = &long_skips.as_slice()[(long_skip_id - 1) * 8..][..8];
u64::deserialize(&mut long_skip_blocks).expect("Index corrupted") * 16
} else {
0
}
};
let mut position_read = OwnedRead::new(position_source);
position_read.advance(offset_num_bytes as usize);
let mut skip_read = OwnedRead::new(skip_body);
skip_read.advance(long_skip_id * LONG_SKIP_IN_BLOCKS);
let mut position_reader = PositionReader {
skip_read,
position_read,
inner_offset: 0,
buffer: Box::new([0u32; 128]),
ahead: None
};
position_reader.skip(small_skip);
position_reader
pub fn new(
position_source: ReadOnlySource,
skip_source: ReadOnlySource,
offset: u64,
) -> PositionReader {
Positions::new(position_source, skip_source).reader(offset)
}
/// Fills a buffer with the next `output.len()` integers.
@@ -100,15 +151,19 @@ impl PositionReader {
if self.ahead != Some(0) {
// the block currently available is not the block
// for the current position
BIT_PACKER.decompress(position_data, self.buffer.as_mut(), num_bits);
self.bit_packer
.decompress(position_data, self.buffer.as_mut(), num_bits);
self.ahead = Some(0);
}
let block_len = compressed_block_size(num_bits);
self.ahead = Some(read_impl(
self.bit_packer,
&position_data[block_len..],
self.buffer.as_mut(),
self.inner_offset,
&skip_data[1..],
output));
output,
));
}
/// Skip the next `skip_len` integer.
@@ -118,29 +173,26 @@ impl PositionReader {
///
/// May panic if the end of the stream is reached.
pub fn skip(&mut self, skip_len: usize) {
let skip_len_plus_inner_offset = skip_len + self.inner_offset;
let num_blocks_to_advance = skip_len_plus_inner_offset / COMPRESSION_BLOCK_SIZE;
self.inner_offset = skip_len_plus_inner_offset % COMPRESSION_BLOCK_SIZE;
self.ahead = self.ahead
.and_then(|num_blocks| {
if num_blocks >= num_blocks_to_advance {
Some(num_blocks_to_advance - num_blocks_to_advance)
} else {
None
}
});
self.ahead = self.ahead.and_then(|num_blocks| {
if num_blocks >= num_blocks_to_advance {
Some(num_blocks - num_blocks_to_advance)
} else {
None
}
});
let skip_len = self.skip_read
.as_ref()[..num_blocks_to_advance]
let skip_len_in_bits = self.skip_read.as_ref()[..num_blocks_to_advance]
.iter()
.cloned()
.map(|num_bit| num_bit as usize)
.sum::<usize>() * (COMPRESSION_BLOCK_SIZE / 8);
.map(|num_bits| *num_bits as usize)
.sum::<usize>()
* COMPRESSION_BLOCK_SIZE;
let skip_len_in_bytes = skip_len_in_bits / 8;
self.skip_read.advance(num_blocks_to_advance);
self.position_read.advance(skip_len);
self.position_read.advance(skip_len_in_bytes);
}
}

27
src/positions/serializer.rs
View File

@@ -1,29 +1,30 @@
use std::io;
use bitpacking::BitPacker;
use positions::{COMPRESSION_BLOCK_SIZE, LONG_SKIP_INTERVAL};
use bitpacking::BitPacker4x;
use common::BinarySerializable;
use super::BIT_PACKER;
use common::CountingWriter;
use positions::{COMPRESSION_BLOCK_SIZE, LONG_SKIP_INTERVAL};
use std::io::{self, Write};
pub struct PositionSerializer<W: io::Write> {
write_stream: W,
bit_packer: BitPacker4x,
write_stream: CountingWriter<W>,
write_skiplist: W,
block: Vec<u32>,
buffer: Vec<u8>,
num_ints: u64,
long_skips: Vec<u64>,
cumulated_num_bits: u64,
}
impl<W: io::Write> PositionSerializer<W> {
pub fn new(write_stream: W, write_skiplist: W) -> PositionSerializer<W> {
PositionSerializer {
write_stream,
bit_packer: BitPacker4x::new(),
write_stream: CountingWriter::wrap(write_stream),
write_skiplist,
block: Vec::with_capacity(128),
buffer: vec![0u8; 128 * 4],
num_ints: 0u64,
long_skips: Vec::new(),
cumulated_num_bits: 0u64
}
}
@@ -31,7 +32,6 @@ impl<W: io::Write> PositionSerializer<W> {
self.num_ints
}
fn remaining_block_len(&self) -> usize {
COMPRESSION_BLOCK_SIZE - self.block.len()
}
@@ -51,14 +51,15 @@ impl<W: io::Write> PositionSerializer<W> {
}
fn flush_block(&mut self) -> io::Result<()> {
let num_bits = BIT_PACKER.num_bits(&self.block[..]);
self.cumulated_num_bits += num_bits as u64;
self.write_skiplist.write(&[num_bits])?;
let written_len = BIT_PACKER.compress(&self.block[..], &mut self.buffer, num_bits);
let num_bits = self.bit_packer.num_bits(&self.block[..]);
self.write_skiplist.write_all(&[num_bits])?;
let written_len = self
.bit_packer
.compress(&self.block[..], &mut self.buffer, num_bits);
self.write_stream.write_all(&self.buffer[..written_len])?;
self.block.clear();
if (self.num_ints % LONG_SKIP_INTERVAL) == 0u64 {
self.long_skips.push(self.cumulated_num_bits);
self.long_skips.push(self.write_stream.written_bytes());
}
Ok(())
}

249
src/postings/block_search.rs Normal file
View File

@@ -0,0 +1,249 @@
use postings::compression::AlignedBuffer;
/// This modules define the logic used to search for a doc in a given
/// block. (at most 128 docs)
///
/// Searching within a block is a hotspot when running intersection.
/// so it was worth defining it in its own module.
#[cfg(target_arch = "x86_64")]
mod sse2 {
use postings::compression::{AlignedBuffer, COMPRESSION_BLOCK_SIZE};
use std::arch::x86_64::__m128i as DataType;
use std::arch::x86_64::_mm_add_epi32 as op_add;
use std::arch::x86_64::_mm_cmplt_epi32 as op_lt;
use std::arch::x86_64::_mm_load_si128 as op_load; // requires 128-bits alignment
use std::arch::x86_64::_mm_set1_epi32 as set1;
use std::arch::x86_64::_mm_setzero_si128 as set0;
use std::arch::x86_64::_mm_sub_epi32 as op_sub;
use std::arch::x86_64::{_mm_cvtsi128_si32, _mm_shuffle_epi32};
const MASK1: i32 = 78;
const MASK2: i32 = 177;
/// Performs an exhaustive linear search over the
///
/// There is no early exit here. We simply count the
/// number of elements that are `< target`.
pub(crate) fn linear_search_sse2_128(arr: &AlignedBuffer, target: u32) -> usize {
unsafe {
let ptr = arr as *const AlignedBuffer as *const DataType;
let vkey = set1(target as i32);
let mut cnt = set0();
// We work over 4 `__m128i` at a time.
// A single `__m128i` actual contains 4 `u32`.
for i in 0..(COMPRESSION_BLOCK_SIZE as isize) / (4 * 4) {
let cmp1 = op_lt(op_load(ptr.offset(i * 4)), vkey);
let cmp2 = op_lt(op_load(ptr.offset(i * 4 + 1)), vkey);
let cmp3 = op_lt(op_load(ptr.offset(i * 4 + 2)), vkey);
let cmp4 = op_lt(op_load(ptr.offset(i * 4 + 3)), vkey);
let sum = op_add(op_add(cmp1, cmp2), op_add(cmp3, cmp4));
cnt = op_sub(cnt, sum);
}
cnt = op_add(cnt, _mm_shuffle_epi32(cnt, MASK1));
cnt = op_add(cnt, _mm_shuffle_epi32(cnt, MASK2));
_mm_cvtsi128_si32(cnt) as usize
}
}
#[cfg(test)]
mod test {
use super::linear_search_sse2_128;
use postings::compression::{AlignedBuffer, COMPRESSION_BLOCK_SIZE};
#[test]
fn test_linear_search_sse2_128_u32() {
let mut block = [0u32; COMPRESSION_BLOCK_SIZE];
for el in 0u32..128u32 {
block[el as usize] = el * 2 + 1 << 18;
}
let target = block[64] + 1;
assert_eq!(linear_search_sse2_128(&AlignedBuffer(block), target), 65);
}
}
}
/// This `linear search` browser exhaustively through the array.
/// but the early exit is very difficult to predict.
///
/// Coupled with `exponential search` this function is likely
/// to be called with the same `len`
fn linear_search(arr: &[u32], target: u32) -> usize {
arr.iter().map(|&el| if el < target { 1 } else { 0 }).sum()
}
fn exponential_search(arr: &[u32], target: u32) -> (usize, usize) {
let end = arr.len();
let mut begin = 0;
for &pivot in &[1, 3, 7, 15, 31, 63] {
if pivot >= end {
break;
}
if arr[pivot] > target {
return (begin, pivot);
}
begin = pivot;
}
(begin, end)
}
fn galloping(block_docs: &[u32], target: u32) -> usize {
let (start, end) = exponential_search(&block_docs, target);
start + linear_search(&block_docs[start..end], target)
}
/// Tantivy may rely on SIMD instructions to search for a specific document within
/// a given block.
#[derive(Clone, Copy, PartialEq)]
pub enum BlockSearcher {
#[cfg(target_arch = "x86_64")]
SSE2,
Scalar,
}
impl BlockSearcher {
/// Search the first index containing an element greater or equal to
/// the target.
///
/// The results should be equivalent to
/// ```ignore
/// block[..]
// .iter()
// .take_while(|&&val| val < target)
// .count()
/// ```
///
/// The `start` argument is just used to hint that the response is
/// greater than beyond `start`. The implementation may or may not use
/// it for optimization.
///
/// # Assumption
///
/// The array len is > start.
/// The block is sorted
/// The target is assumed greater or equal to the `arr[start]`.
/// The target is assumed smaller or equal to the last element of the block.
///
/// Currently the scalar implementation starts by an exponential search, and
/// then operates a linear search in the result subarray.
///
/// If SSE2 instructions are available in the `(platform, running CPU)`,
/// then we use a different implementation that does an exhaustive linear search over
/// the full block whenever the block is full (`len == 128`). It is surprisingly faster, most likely because of the lack
/// of branch.
pub(crate) fn search_in_block(
self,
block_docs: &AlignedBuffer,
len: usize,
start: usize,
target: u32,
) -> usize {
#[cfg(target_arch = "x86_64")]
{
use postings::compression::COMPRESSION_BLOCK_SIZE;
if self == BlockSearcher::SSE2 && len == COMPRESSION_BLOCK_SIZE {
return sse2::linear_search_sse2_128(block_docs, target);
}
}
start + galloping(&block_docs.0[start..len], target)
}
}
impl Default for BlockSearcher {
fn default() -> BlockSearcher {
#[cfg(target_arch = "x86_64")]
{
if is_x86_feature_detected!("sse2") {
return BlockSearcher::SSE2;
}
}
BlockSearcher::Scalar
}
}
#[cfg(test)]
mod tests {
use super::exponential_search;
use super::linear_search;
use super::BlockSearcher;
use postings::compression::{AlignedBuffer, COMPRESSION_BLOCK_SIZE};
#[test]
fn test_linear_search() {
let len: usize = 50;
let arr: Vec<u32> = (0..len).map(|el| 1u32 + (el as u32) * 2).collect();
for target in 1..*arr.last().unwrap() {
let res = linear_search(&arr[..], target);
if res > 0 {
assert!(arr[res - 1] < target);
}
if res < len {
assert!(arr[res] >= target);
}
}
}
#[test]
fn test_exponentiel_search() {
assert_eq!(exponential_search(&[1, 2], 0), (0, 1));
assert_eq!(exponential_search(&[1, 2], 1), (0, 1));
assert_eq!(
exponential_search(&[1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11], 7),
(3, 7)
);
}
fn util_test_search_in_block(block_searcher: BlockSearcher, block: &[u32], target: u32) {
let cursor = search_in_block_trivial_but_slow(block, target);
assert!(block.len() < COMPRESSION_BLOCK_SIZE);
let mut output_buffer = [u32::max_value(); COMPRESSION_BLOCK_SIZE];
output_buffer[..block.len()].copy_from_slice(block);
for i in 0..cursor {
assert_eq!(
block_searcher.search_in_block(
&AlignedBuffer(output_buffer),
block.len(),
i,
target
),
cursor
);
}
}
fn util_test_search_in_block_all(block_searcher: BlockSearcher, block: &[u32]) {
use std::collections::HashSet;
let mut targets = HashSet::new();
for (i, val) in block.iter().cloned().enumerate() {
if i > 0 {
targets.insert(val - 1);
}
targets.insert(val);
}
for target in targets {
util_test_search_in_block(block_searcher, block, target);
}
}
fn search_in_block_trivial_but_slow(block: &[u32], target: u32) -> usize {
block.iter().take_while(|&&val| val < target).count()
}
fn test_search_in_block_util(block_searcher: BlockSearcher) {
for len in 1u32..128u32 {
let v: Vec<u32> = (0..len).map(|i| i * 2).collect();
util_test_search_in_block_all(block_searcher, &v[..]);
}
}
#[test]
fn test_search_in_block_scalar() {
test_search_in_block_util(BlockSearcher::Scalar);
}
#[cfg(target_arch = "x86_64")]
#[test]
fn test_search_in_block_sse2() {
test_search_in_block_util(BlockSearcher::SSE2);
}
}

80
src/postings/compression/mod.rs
View File

@@ -28,22 +28,29 @@ impl BlockEncoder {
pub fn compress_block_sorted(&mut self, block: &[u32], offset: u32) -> (u8, &[u8]) {
let num_bits = self.bitpacker.num_bits_sorted(offset, block);
let written_size = self.bitpacker
.compress_sorted(offset, block, &mut self.output[..], num_bits);
let written_size =
self.bitpacker
.compress_sorted(offset, block, &mut self.output[..], num_bits);
(num_bits, &self.output[..written_size])
}
pub fn compress_block_unsorted(&mut self, block: &[u32]) -> (u8, &[u8]) {
let num_bits = self.bitpacker.num_bits(block);
let written_size = self.bitpacker
let written_size = self
.bitpacker
.compress(block, &mut self.output[..], num_bits);
(num_bits, &self.output[..written_size])
}
}
/// We ensure that the OutputBuffer is align on 128 bits
/// in order to run SSE2 linear search on it.
#[repr(align(128))]
pub(crate) struct AlignedBuffer(pub [u32; COMPRESSION_BLOCK_SIZE]);
pub struct BlockDecoder {
bitpacker: BitPacker4x,
pub output: [u32; COMPRESSION_BLOCK_SIZE + 1],
output: AlignedBuffer,
pub output_len: usize,
}
@@ -53,42 +60,46 @@ impl BlockDecoder {
}
pub fn with_val(val: u32) -> BlockDecoder {
let mut output = [val; COMPRESSION_BLOCK_SIZE + 1];
output[COMPRESSION_BLOCK_SIZE] = 0u32;
BlockDecoder {
bitpacker: BitPacker4x::new(),
output,
output: AlignedBuffer([val; COMPRESSION_BLOCK_SIZE]),
output_len: 0,
}
}
pub fn uncompress_block_sorted(&mut self, compressed_data: &[u8], offset: u32, num_bits: u8) -> usize {
pub fn uncompress_block_sorted(
&mut self,
compressed_data: &[u8],
offset: u32,
num_bits: u8,
) -> usize {
self.output_len = COMPRESSION_BLOCK_SIZE;
self.bitpacker.decompress_sorted(
offset,
&compressed_data,
&mut self.output,
num_bits,
)
self.bitpacker
.decompress_sorted(offset, &compressed_data, &mut self.output.0, num_bits)
}
pub fn uncompress_block_unsorted(&mut self, compressed_data: &[u8], num_bits: u8) -> usize {
self.output_len = COMPRESSION_BLOCK_SIZE;
self.bitpacker.decompress(&compressed_data, &mut self.output, num_bits)
self.bitpacker
.decompress(&compressed_data, &mut self.output.0, num_bits)
}
#[inline]
pub fn output_array(&self) -> &[u32] {
&self.output[..self.output_len]
&self.output.0[..self.output_len]
}
#[inline]
pub(crate) fn output_aligned(&self) -> (&AlignedBuffer, usize) {
(&self.output, self.output_len)
}
#[inline]
pub fn output(&self, idx: usize) -> u32 {
self.output[idx]
self.output.0[idx]
}
}
pub trait VIntEncoder {
/// Compresses an array of `u32` integers,
/// using [delta-encoding](https://en.wikipedia.org/wiki/Delta_ encoding)
@@ -156,12 +167,12 @@ impl VIntDecoder for BlockDecoder {
num_els: usize,
) -> usize {
self.output_len = num_els;
vint::uncompress_sorted(compressed_data, &mut self.output[..num_els], offset)
vint::uncompress_sorted(compressed_data, &mut self.output.0[..num_els], offset)
}
fn uncompress_vint_unsorted<'a>(&mut self, compressed_data: &'a [u8], num_els: usize) -> usize {
self.output_len = num_els;
vint::uncompress_unsorted(compressed_data, &mut self.output[..num_els])
vint::uncompress_unsorted(compressed_data, &mut self.output.0[..num_els])
}
}
@@ -263,21 +274,17 @@ pub mod tests {
mod bench {
use super::*;
use rand::Rng;
use rand::SeedableRng;
use rand::XorShiftRng;
use rand::{Rng, XorShiftRng};
use test::Bencher;
fn generate_array_with_seed(n: usize, ratio: f32, seed_val: u32) -> Vec<u32> {
let seed: &[u32; 4] = &[1, 2, 3, seed_val];
fn generate_array_with_seed(n: usize, ratio: f64, seed_val: u8) -> Vec<u32> {
let seed: &[u8; 16] = &[0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, seed_val];
let mut rng: XorShiftRng = XorShiftRng::from_seed(*seed);
(0..u32::max_value())
.filter(|_| rng.next_f32() < ratio)
.take(n)
.collect()
(0u32..).filter(|_| rng.gen_bool(ratio)).take(n).collect()
}
pub fn generate_array(n: usize, ratio: f32) -> Vec<u32> {
pub fn generate_array(n: usize, ratio: f64) -> Vec<u32> {
generate_array_with_seed(n, ratio, 4)
}
@@ -294,24 +301,23 @@ mod bench {
fn bench_uncompress(b: &mut Bencher) {
let mut encoder = BlockEncoder::new();
let data = generate_array(COMPRESSION_BLOCK_SIZE, 0.1);
let (_, compressed) = encoder.compress_block_sorted(&data, 0u32);
let (num_bits, compressed) = encoder.compress_block_sorted(&data, 0u32);
let mut decoder = BlockDecoder::new();
b.iter(|| {
decoder.uncompress_block_sorted(compressed, 0u32);
decoder.uncompress_block_sorted(compressed, 0u32, num_bits);
});
}
#[test]
fn test_all_docs_compression_numbits() {
for num_bits in 0..33 {
for expected_num_bits in 0u8.. {
let mut data = [0u32; 128];
if num_bits > 0 {
data[0] = 1 << (num_bits - 1);
if expected_num_bits > 0 {
data[0] = (1u64 << (expected_num_bits as usize) - 1) as u32;
}
let mut encoder = BlockEncoder::new();
let compressed = encoder.compress_block_unsorted(&data);
assert_eq!(compressed[0] as usize, num_bits);
assert_eq!(compressed.len(), compressed_block_size(compressed[0]));
let (num_bits, compressed) = encoder.compress_block_unsorted(&data);
assert_eq!(compressed.len(), compressed_block_size(num_bits));
}
}

28
src/postings/compression/vint.rs
View File

@@ -1,9 +1,5 @@
#[inline(always)]
pub fn compress_sorted<'a>(
input: &[u32],
output: &'a mut [u8],
mut offset: u32,
) -> &'a [u8] {
pub fn compress_sorted<'a>(input: &[u32], output: &'a mut [u8], mut offset: u32) -> &'a [u8] {
let mut byte_written = 0;
for &v in input {
let mut to_encode: u32 = v - offset;
@@ -46,47 +42,41 @@ pub(crate) fn compress_unsorted<'a>(input: &[u32], output: &'a mut [u8]) -> &'a
}
#[inline(always)]
pub fn uncompress_sorted<'a>(
compressed_data: &'a [u8],
output: &mut [u32],
offset: u32,
) -> usize {
pub fn uncompress_sorted<'a>(compressed_data: &'a [u8], output: &mut [u32], offset: u32) -> usize {
let mut read_byte = 0;
let mut result = offset;
let num_els = output.len();
for i in 0..num_els {
for output_mut in output.iter_mut() {
let mut shift = 0u32;
loop {
let cur_byte = compressed_data[read_byte];
read_byte += 1;
result += ((cur_byte % 128u8) as u32) << shift;
result += u32::from(cur_byte % 128u8) << shift;
if cur_byte & 128u8 != 0u8 {
break;
}
shift += 7;
}
output[i] = result;
*output_mut = result;
}
read_byte
}
#[inline(always)]
pub(crate) fn uncompress_unsorted<'a>(compressed_data: &'a [u8], output: &mut [u32]) -> usize {
pub(crate) fn uncompress_unsorted(compressed_data: &[u8], output_arr: &mut [u32]) -> usize {
let mut read_byte = 0;
let num_els = output.len();
for i in 0..num_els {
for output_mut in output_arr.iter_mut() {
let mut result = 0u32;
let mut shift = 0u32;
loop {
let cur_byte = compressed_data[read_byte];
read_byte += 1;
result += ((cur_byte % 128u8) as u32) << shift;
result += u32::from(cur_byte % 128u8) << shift;
if cur_byte & 128u8 != 0u8 {
break;
}
shift += 7;
}
output[i] = result;
*output_mut = result;
}
read_byte
}

Some files were not shown because too many files have changed in this diff Show More