rust/tantivy
1
0
Fork 0
mirror of https://github.com/quickwit-oss/tantivy.git synced 2026-01-04 08:12:54 +00:00
Code Issues Packages Projects Releases Wiki Activity

Compare commits

base: rust:columnar
Branches Tags
rust:main rust:paul.masurel/seek_into_the_danger_zone_result rust:paul.masurel/skip-first-doc-on-intersection rust:paul.masurel/search_into_danger_zone_fastfield 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: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:trinity/yoke-documents
Branches Tags
rust:paul.masurel/seek_into_the_danger_zone_result rust:paul.masurel/skip-first-doc-on-intersection rust:main rust:paul.masurel/search_into_danger_zone_fastfield 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: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

6 Commits
columnar ... trinity/yo

Author SHA1 Message Date
trinity-1686a
bcff3eb2d2 try with custom Cow<str> 2023-01-11 16:02:52 +01:00
trinity-1686a
85f2588875 implement add_borrowed_values on Document 2022-12-23 16:16:22 +01:00
trinity-1686a
db6cf65d53 make Document support Yoked inner values 2022-12-22 17:52:53 +01:00
trinity-1686a
654aa7f42c allow Value to borrow 2022-12-22 15:43:13 +01:00
François Massot
951a898633 Update bench. 2022-10-30 14:12:07 +01:00
François Massot
003722d831 Add bench to reproduce performance drop on array of texts. 2022-10-29 02:54:07 +02:00
172 changed files with 102087 additions and 8885 deletions
Expand all files Collapse all files

2
.github/workflows/test.yml vendored
View File

@@ -48,7 +48,7 @@ jobs:
strategy:
matrix:
features: [
{ label: "all", flags: "mmap,stopwords,brotli-compression,lz4-compression,snappy-compression,zstd-compression,failpoints" },
{ label: "all", flags: "mmap,brotli-compression,lz4-compression,snappy-compression,zstd-compression,failpoints" },
{ label: "quickwit", flags: "mmap,quickwit,failpoints" }
]

27
CHANGELOG.md
View File

@@ -1,37 +1,32 @@
Tantivy 0.19
================================
#### Bugfixes
- Fix missing fieldnorms for u64, i64, f64, bool, bytes and date [#1620](https://github.com/quickwit-oss/tantivy/pull/1620) (@PSeitz)
- Fix interpolation overflow in linear interpolation fastfield codec [#1480](https://github.com/quickwit-oss/tantivy/pull/1480) (@PSeitz @fulmicoton)
#### Features/Improvements
- Add support for `IN` in queryparser , e.g. `field: IN [val1 val2 val3]` [#1683](https://github.com/quickwit-oss/tantivy/pull/1683) (@trinity-1686a)
- Skip score calculation, when no scoring is required [#1646](https://github.com/quickwit-oss/tantivy/pull/1646) (@PSeitz)
- Limit fast fields to u32 (`get_val(u32)`) [#1644](https://github.com/quickwit-oss/tantivy/pull/1644) (@PSeitz)
- The `DateTime` type has been updated to hold timestamps with microseconds precision.
`DateOptions` and `DatePrecision` have been added to configure Date fields. The precision is used to hint on fast values compression. Otherwise, seconds precision is used everywhere else (i.e terms, indexing) [#1396](https://github.com/quickwit-oss/tantivy/pull/1396) (@evanxg852000)
- Major bugfix: Fix missing fieldnorms for u64, i64, f64, bool, bytes and date [#1620](https://github.com/quickwit-oss/tantivy/pull/1620) (@PSeitz)
- Updated [Date Field Type](https://github.com/quickwit-oss/tantivy/pull/1396)
The `DateTime` type has been updated to hold timestamps with microseconds precision.
`DateOptions` and `DatePrecision` have been added to configure Date fields. The precision is used to hint on fast values compression. Otherwise, seconds precision is used everywhere else (i.e terms, indexing). (@evanxg852000)
- Add IP address field type [#1553](https://github.com/quickwit-oss/tantivy/pull/1553) (@PSeitz)
- Add boolean field type [#1382](https://github.com/quickwit-oss/tantivy/pull/1382) (@boraarslan)
- Remove Searcher pool and make `Searcher` cloneable. (@PSeitz)
- Validate settings on create [#1570](https://github.com/quickwit-oss/tantivy/pull/1570) (@PSeitz)
- Validate settings on create [#1570](https://github.com/quickwit-oss/tantivy/pull/1570 (@PSeitz)
- Fix interpolation overflow in linear interpolation fastfield codec [#1480](https://github.com/quickwit-oss/tantivy/pull/1480 (@PSeitz @fulmicoton)
- Detect and apply gcd on fastfield codecs [#1418](https://github.com/quickwit-oss/tantivy/pull/1418) (@PSeitz)
- Doc store
- use separate thread to compress block store [#1389](https://github.com/quickwit-oss/tantivy/pull/1389) [#1510](https://github.com/quickwit-oss/tantivy/pull/1510) (@PSeitz @fulmicoton)
- use separate thread to compress block store [#1389](https://github.com/quickwit-oss/tantivy/pull/1389) [#1510](https://github.com/quickwit-oss/tantivy/pull/1510 (@PSeitz @fulmicoton)
- Expose doc store cache size [#1403](https://github.com/quickwit-oss/tantivy/pull/1403) (@PSeitz)
- Enable compression levels for doc store [#1378](https://github.com/quickwit-oss/tantivy/pull/1378) (@PSeitz)
- Make block size configurable [#1374](https://github.com/quickwit-oss/tantivy/pull/1374) (@kryesh)
- Make `tantivy::TantivyError` cloneable [#1402](https://github.com/quickwit-oss/tantivy/pull/1402) (@PSeitz)
- Add support for phrase slop in query language [#1393](https://github.com/quickwit-oss/tantivy/pull/1393) (@saroh)
- Aggregation
- Add aggregation support for date type [#1693](https://github.com/quickwit-oss/tantivy/pull/1693)(@PSeitz)
- Add support for keyed parameter in range and histgram aggregations [#1424](https://github.com/quickwit-oss/tantivy/pull/1424) (@k-yomo)
- Add aggregation bucket limit [#1363](https://github.com/quickwit-oss/tantivy/pull/1363) (@PSeitz)
- Faster indexing
- [#1610](https://github.com/quickwit-oss/tantivy/pull/1610) (@PSeitz)
- [#1594](https://github.com/quickwit-oss/tantivy/pull/1594) (@PSeitz)
- [#1582](https://github.com/quickwit-oss/tantivy/pull/1582) (@PSeitz)
- [#1611](https://github.com/quickwit-oss/tantivy/pull/1611) (@PSeitz)
- Added a pre-configured stop word filter for various language [#1666](https://github.com/quickwit-oss/tantivy/pull/1666) (@adamreichold)
- [#1610](https://github.com/quickwit-oss/tantivy/pull/1610 (@PSeitz)
- [#1594](https://github.com/quickwit-oss/tantivy/pull/1594 (@PSeitz)
- [#1582](https://github.com/quickwit-oss/tantivy/pull/1582 (@PSeitz)
- [#1611](https://github.com/quickwit-oss/tantivy/pull/1611 (@PSeitz)
Tantivy 0.18
================================

34
Cargo.toml
View File

@@ -1,6 +1,6 @@
[package]
name = "tantivy"
version = "0.19.0"
version = "0.19.0-dev"
authors = ["Paul Masurel <paul.masurel@gmail.com>"]
license = "MIT"
categories = ["database-implementations", "data-structures"]
@@ -15,7 +15,7 @@ rust-version = "1.62"
[dependencies]
oneshot = "0.1.5"
base64 = "0.20.0"
base64 = "0.13.0"
byteorder = "1.4.3"
crc32fast = "1.3.2"
once_cell = "1.10.0"
@@ -25,7 +25,7 @@ tantivy-fst = "0.4.0"
memmap2 = { version = "0.5.3", optional = true }
lz4_flex = { version = "0.9.2", default-features = false, features = ["checked-decode"], optional = true }
brotli = { version = "3.3.4", optional = true }
zstd = { version = "0.12", optional = true, default-features = false }
zstd = { version = "0.11", optional = true, default-features = false }
snap = { version = "1.0.5", optional = true }
tempfile = { version = "3.3.0", optional = true }
log = "0.4.16"
@@ -36,6 +36,11 @@ fs2 = { version = "0.4.3", optional = true }
levenshtein_automata = "0.2.1"
uuid = { version = "1.0.0", features = ["v4", "serde"] }
crossbeam-channel = "0.5.4"
tantivy-query-grammar = { version="0.18.0", path="./query-grammar" }
tantivy-bitpacker = { version="0.2", path="./bitpacker" }
common = { version = "0.3", path = "./common/", package = "tantivy-common" }
fastfield_codecs = { version="0.2", path="./fastfield_codecs", default-features = false }
ownedbytes = { version="0.3", path="./ownedbytes" }
stable_deref_trait = "1.2.0"
rust-stemmers = "1.2.0"
downcast-rs = "1.2.0"
@@ -53,16 +58,10 @@ lru = "0.7.5"
fastdivide = "0.4.0"
itertools = "0.10.3"
measure_time = "0.8.2"
ciborium = { version = "0.2", optional = true}
async-trait = "0.1.53"
arc-swap = "1.5.0"
sstable = { version="0.1", path="./sstable", package ="tantivy-sstable", optional = true }
stacker = { version="0.1", path="./stacker", package ="tantivy-stacker" }
tantivy-query-grammar = { version= "0.19.0", path="./query-grammar" }
tantivy-bitpacker = { version= "0.3", path="./bitpacker" }
common = { version= "0.5", path = "./common/", package = "tantivy-common" }
fastfield_codecs = { version= "0.3", path="./fastfield_codecs", default-features = false }
ownedbytes = { version= "0.5", path="./ownedbytes" }
yoke = { version = "0.6.2", features = ["derive"] }
[target.'cfg(windows)'.dependencies]
winapi = "0.3.9"
@@ -73,10 +72,10 @@ maplit = "1.0.2"
matches = "0.1.9"
pretty_assertions = "1.2.1"
proptest = "1.0.0"
criterion = "0.4"
criterion = "0.3.5"
test-log = "0.2.10"
env_logger = "0.10.0"
pprof = { version = "0.11.0", features = ["flamegraph", "criterion"] }
env_logger = "0.9.0"
pprof = { version = "0.10.0", features = ["flamegraph", "criterion"] }
futures = "0.3.21"
[dev-dependencies.fail]
@@ -93,9 +92,8 @@ debug-assertions = true
overflow-checks = true
[features]
default = ["mmap", "stopwords", "lz4-compression"]
default = ["mmap", "lz4-compression" ]
mmap = ["fs2", "tempfile", "memmap2"]
stopwords = []
brotli-compression = ["brotli"]
lz4-compression = ["lz4_flex"]
@@ -105,10 +103,10 @@ zstd-compression = ["zstd"]
failpoints = ["fail/failpoints"]
unstable = [] # useful for benches.
quickwit = ["sstable"]
quickwit = ["ciborium"]
[workspace]
members = ["query-grammar", "bitpacker", "common", "fastfield_codecs", "ownedbytes", "stacker", "sstable", "columnar"]
members = ["query-grammar", "bitpacker", "common", "fastfield_codecs", "ownedbytes"]
# Following the "fail" crate best practises, we isolate
# tests that define specific behavior in fail check points

100000
benches/hdfs_with_array.json Normal file
View File

File diff suppressed because it is too large Load Diff

221
benches/index-bench.rs
View File

@@ -1,116 +1,159 @@
use criterion::{criterion_group, criterion_main, Criterion};
use itertools::Itertools;
use pprof::criterion::{Output, PProfProfiler};
use tantivy::schema::{INDEXED, STORED, STRING, TEXT};
use tantivy::Index;
use serde_json::{self, Value as JsonValue};
use tantivy::directory::RamDirectory;
use tantivy::schema::{
FieldValue, TextFieldIndexing, TextOptions, Value, INDEXED, STORED, STRING, TEXT,
};
use tantivy::{Document, Index, IndexBuilder};
const HDFS_LOGS: &str = include_str!("hdfs.json");
const NUM_REPEATS: usize = 2;
const NUM_REPEATS: usize = 20;
pub fn hdfs_index_benchmark(c: &mut Criterion) {
let schema = {
let mut schema_builder = tantivy::schema::SchemaBuilder::new();
schema_builder.add_u64_field("timestamp", INDEXED);
schema_builder.add_text_field("body", TEXT);
schema_builder.add_text_field("severity", STRING);
schema_builder.build()
};
let schema_with_store = {
let mut schema_builder = tantivy::schema::SchemaBuilder::new();
schema_builder.add_u64_field("timestamp", INDEXED | STORED);
schema_builder.add_text_field("body", TEXT | STORED);
schema_builder.add_text_field("severity", STRING | STORED);
schema_builder.build()
};
let dynamic_schema = {
let mut schema_builder = tantivy::schema::SchemaBuilder::new();
schema_builder.add_json_field("json", TEXT);
schema_builder.build()
};
let mut schema_builder = tantivy::schema::SchemaBuilder::new();
let text_indexing_options = TextFieldIndexing::default()
.set_tokenizer("default")
.set_fieldnorms(false)
.set_index_option(tantivy::schema::IndexRecordOption::WithFreqsAndPositions);
let mut text_options = TextOptions::default().set_indexing_options(text_indexing_options);
let text_field = schema_builder.add_text_field("body", text_options);
let schema = schema_builder.build();
// prepare doc
let mut documents_no_array = Vec::new();
let mut documents_with_array = Vec::new();
for doc_json in HDFS_LOGS.trim().split("\n") {
let json_obj: serde_json::Map<String, JsonValue> = serde_json::from_str(doc_json).unwrap();
let text = json_obj.get("body").unwrap().as_str().unwrap();
let mut doc_no_array = Document::new();
doc_no_array.add_text(text_field, text);
documents_no_array.push(doc_no_array);
let mut doc_with_array = Document::new();
doc_with_array.add_borrowed_values(text.to_owned(), |text| {
text.split(' ')
.map(|text| FieldValue::new(text_field, text.into()))
.collect()
});
documents_with_array.push(doc_with_array);
}
let mut group = c.benchmark_group("index-hdfs");
group.sample_size(20);
group.bench_function("index-hdfs-no-commit", |b| {
b.iter(|| {
let index = Index::create_in_ram(schema.clone());
let index_writer = index.writer_with_num_threads(1, 100_000_000).unwrap();
let ram_directory = RamDirectory::create();
let mut index_writer = IndexBuilder::new()
.schema(schema.clone())
.single_segment_index_writer(ram_directory, 100_000_000)
.unwrap();
for _ in 0..NUM_REPEATS {
for doc_json in HDFS_LOGS.trim().split("\n") {
let doc = schema.parse_document(doc_json).unwrap();
let documents_cloned = documents_no_array.clone();
for doc in documents_cloned {
index_writer.add_document(doc).unwrap();
}
}
})
});
group.bench_function("index-hdfs-with-commit", |b| {
group.bench_function("index-hdfs-with-array-no-commit", |b| {
b.iter(|| {
let index = Index::create_in_ram(schema.clone());
let mut index_writer = index.writer_with_num_threads(1, 100_000_000).unwrap();
let ram_directory = RamDirectory::create();
let mut index_writer = IndexBuilder::new()
.schema(schema.clone())
.single_segment_index_writer(ram_directory, 100_000_000)
.unwrap();
for _ in 0..NUM_REPEATS {
for doc_json in HDFS_LOGS.trim().split("\n") {
let doc = schema.parse_document(doc_json).unwrap();
index_writer.add_document(doc).unwrap();
}
}
index_writer.commit().unwrap();
})
});
group.bench_function("index-hdfs-no-commit-with-docstore", |b| {
b.iter(|| {
let index = Index::create_in_ram(schema_with_store.clone());
let index_writer = index.writer_with_num_threads(1, 100_000_000).unwrap();
for _ in 0..NUM_REPEATS {
for doc_json in HDFS_LOGS.trim().split("\n") {
let doc = schema.parse_document(doc_json).unwrap();
let documents_with_array_cloned = documents_with_array.clone();
for doc in documents_with_array_cloned {
index_writer.add_document(doc).unwrap();
}
}
})
});
group.bench_function("index-hdfs-with-commit-with-docstore", |b| {
b.iter(|| {
let index = Index::create_in_ram(schema_with_store.clone());
let mut index_writer = index.writer_with_num_threads(1, 100_000_000).unwrap();
for _ in 0..NUM_REPEATS {
for doc_json in HDFS_LOGS.trim().split("\n") {
let doc = schema.parse_document(doc_json).unwrap();
index_writer.add_document(doc).unwrap();
}
}
index_writer.commit().unwrap();
})
});
group.bench_function("index-hdfs-no-commit-json-without-docstore", |b| {
b.iter(|| {
let index = Index::create_in_ram(dynamic_schema.clone());
let json_field = dynamic_schema.get_field("json").unwrap();
let mut index_writer = index.writer_with_num_threads(1, 100_000_000).unwrap();
for _ in 0..NUM_REPEATS {
for doc_json in HDFS_LOGS.trim().split("\n") {
let json_val: serde_json::Map<String, serde_json::Value> =
serde_json::from_str(doc_json).unwrap();
let doc = tantivy::doc!(json_field=>json_val);
index_writer.add_document(doc).unwrap();
}
}
index_writer.commit().unwrap();
})
});
group.bench_function("index-hdfs-with-commit-json-without-docstore", |b| {
b.iter(|| {
let index = Index::create_in_ram(dynamic_schema.clone());
let json_field = dynamic_schema.get_field("json").unwrap();
let mut index_writer = index.writer_with_num_threads(1, 100_000_000).unwrap();
for _ in 0..NUM_REPEATS {
for doc_json in HDFS_LOGS.trim().split("\n") {
let json_val: serde_json::Map<String, serde_json::Value> =
serde_json::from_str(doc_json).unwrap();
let doc = tantivy::doc!(json_field=>json_val);
index_writer.add_document(doc).unwrap();
}
}
index_writer.commit().unwrap();
})
});
// group.bench_function("index-hdfs-with-commit", |b| {
// b.iter(|| {
// let ram_directory = RamDirectory::create();
// let mut index_writer = IndexBuilder::new()
// .schema(schema.clone())
// .single_segment_index_writer(ram_directory, 100_000_000)
// .unwrap();
// for _ in 0..NUM_REPEATS {
// for doc_json in HDFS_LOGS.trim().split("\n") {
// let doc = schema.parse_document(doc_json).unwrap();
// index_writer.add_document(doc).unwrap();
// }
// }
// index_writer.commit().unwrap();
// })
// });
// group.bench_function("index-hdfs-no-commit-with-docstore", |b| {
// b.iter(|| {
// let ram_directory = RamDirectory::create();
// let mut index_writer = IndexBuilder::new()
// .schema(schema.clone())
// .single_segment_index_writer(ram_directory, 100_000_000)
// .unwrap();
// for _ in 0..NUM_REPEATS {
// for doc_json in HDFS_LOGS.trim().split("\n") {
// let doc = schema.parse_document(doc_json).unwrap();
// index_writer.add_document(doc).unwrap();
// }
// }
// })
// });
// group.bench_function("index-hdfs-with-commit-with-docstore", |b| {
// b.iter(|| {
// let ram_directory = RamDirectory::create();
// let mut index_writer = IndexBuilder::new()
// .schema(schema.clone())
// .single_segment_index_writer(ram_directory, 100_000_000)
// .unwrap();
// for _ in 0..NUM_REPEATS {
// for doc_json in HDFS_LOGS.trim().split("\n") {
// let doc = schema.parse_document(doc_json).unwrap();
// index_writer.add_document(doc).unwrap();
// }
// }
// index_writer.commit().unwrap();
// })
// });
// group.bench_function("index-hdfs-no-commit-json-without-docstore", |b| {
// b.iter(|| {
// let ram_directory = RamDirectory::create();
// let mut index_writer = IndexBuilder::new()
// .schema(schema.clone())
// .single_segment_index_writer(ram_directory, 100_000_000)
// .unwrap();
// for _ in 0..NUM_REPEATS {
// for doc_json in HDFS_LOGS.trim().split("\n") {
// let json_val: serde_json::Map<String, serde_json::Value> =
// serde_json::from_str(doc_json).unwrap();
// let doc = tantivy::doc!(json_field=>json_val);
// index_writer.add_document(doc).unwrap();
// }
// }
// index_writer.commit().unwrap();
// })
// });
// group.bench_function("index-hdfs-with-commit-json-without-docstore", |b| {
// b.iter(|| {
// let ram_directory = RamDirectory::create();
// let mut index_writer = IndexBuilder::new()
// .schema(schema.clone())
// .single_segment_index_writer(ram_directory, 100_000_000)
// .unwrap();
// for _ in 0..NUM_REPEATS {
// for doc_json in HDFS_LOGS.trim().split("\n") {
// let json_val: serde_json::Map<String, serde_json::Value> =
// serde_json::from_str(doc_json).unwrap();
// let doc = tantivy::doc!(json_field=>json_val);
// index_writer.add_document(doc).unwrap();
// }
// }
// index_writer.commit().unwrap();
// })
//});
}
criterion_group! {

4
bitpacker/Cargo.toml
View File

@@ -1,6 +1,6 @@
[package]
name = "tantivy-bitpacker"
version = "0.3.0"
version = "0.2.0"
edition = "2021"
authors = ["Paul Masurel <paul.masurel@gmail.com>"]
license = "MIT"
@@ -8,8 +8,6 @@ categories = []
description = """Tantivy-sub crate: bitpacking"""
repository = "https://github.com/quickwit-oss/tantivy"
keywords = []
documentation = "https://docs.rs/tantivy-bitpacker/latest/tantivy_bitpacker"
homepage = "https://github.com/quickwit-oss/tantivy"
# See more keys and their definitions at https://doc.rust-lang.org/cargo/reference/manifest.html

9
bitpacker/src/bitpacker.rs
View File

@@ -25,14 +25,15 @@ impl BitPacker {
num_bits: u8,
output: &mut TWrite,
) -> io::Result<()> {
let val_u64 = val as u64;
let num_bits = num_bits as usize;
if self.mini_buffer_written + num_bits > 64 {
self.mini_buffer |= val.wrapping_shl(self.mini_buffer_written as u32);
self.mini_buffer |= val_u64.wrapping_shl(self.mini_buffer_written as u32);
output.write_all(self.mini_buffer.to_le_bytes().as_ref())?;
self.mini_buffer = val.wrapping_shr((64 - self.mini_buffer_written) as u32);
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 << self.mini_buffer_written;
self.mini_buffer |= val_u64 << self.mini_buffer_written;
self.mini_buffer_written += num_bits;
if self.mini_buffer_written == 64 {
output.write_all(self.mini_buffer.to_le_bytes().as_ref())?;
@@ -101,7 +102,7 @@ impl BitUnpacker {
.try_into()
.unwrap();
let val_unshifted_unmasked: u64 = u64::from_le_bytes(bytes);
let val_shifted: u64 = val_unshifted_unmasked >> bit_shift;
let val_shifted = (val_unshifted_unmasked >> bit_shift) as u64;
val_shifted & self.mask
}
}

6
bitpacker/src/blocked_bitpacker.rs
View File

@@ -84,7 +84,7 @@ impl BlockedBitpacker {
#[inline]
pub fn add(&mut self, val: u64) {
self.buffer.push(val);
if self.buffer.len() == BLOCK_SIZE {
if self.buffer.len() == BLOCK_SIZE as usize {
self.flush();
}
}
@@ -126,8 +126,8 @@ impl BlockedBitpacker {
}
#[inline]
pub fn get(&self, idx: usize) -> u64 {
let metadata_pos = idx / BLOCK_SIZE;
let pos_in_block = idx % BLOCK_SIZE;
let metadata_pos = idx / BLOCK_SIZE as usize;
let pos_in_block = idx % BLOCK_SIZE as usize;
if let Some(metadata) = self.offset_and_bits.get(metadata_pos) {
let unpacked = BitUnpacker::new(metadata.num_bits()).get(
pos_in_block as u32,

26
columnar/Cargo.toml
View File

@@ -1,26 +0,0 @@
[package]
name = "tantivy-columnar"
version = "0.1.0"
edition = "2021"
[dependencies]
stacker = { path = "../stacker", package="tantivy-stacker"}
serde_json = "1"
thiserror = "1"
fnv = "1"
tantivy-fst = "0.4.0"
sstable = { path = "../sstable", package = "tantivy-sstable" }
common = { path = "../common", package = "tantivy-common" }
fastfield_codecs = { path = "../fastfield_codecs"}
ordered-float = "3.4"
itertools = "0.10"
[features]
# default = ["quickwit"]
# quickwit = ["common/quickwit"]
[dev-dependencies]
proptest = "1"

33
columnar/README.md
View File

@@ -1,33 +0,0 @@
# Columnar format
This crate describes columnar format used in tantivy.
## Goals
This format is special in the following way.
- it needs to be compact
- it does not required to be loaded in memory.
- it is designed to fit well with quickwit's strange constraint:
we need to be able to load columns rapidly.
- columns of several types can be associated with the same column name.
- it needs to support columns with different types `(str, u64, i64, f64)`
and different cardinality `(required, optional, multivalued)`.
- columns, once loaded, offer cheap random access.
# Format
A quickwit/tantivy style sstable associated
`(column names, column_cardinality, column_type) to range of bytes.
The format of the key is:
`[column_name][ZERO_BYTE][column_type_header: u8]`
Column name may not contain the zero byte.
Listing all columns associated to `column_name` can therefore
be done by listing all keys prefixed by
`[column_name][ZERO_BYTE]`
The associated range of bytes refer to a range of bytes

154
columnar/src/column_type_header.rs
View File

@@ -1,154 +0,0 @@
use crate::value::NumericalType;
#[derive(Clone, Copy, Hash, Default, Debug, PartialEq, Eq, PartialOrd, Ord)]
#[repr(u8)]
pub enum Cardinality {
#[default]
Required = 0,
Optional = 1,
Multivalued = 2,
}
impl Cardinality {
pub fn to_code(self) -> u8 {
self as u8
}
pub fn try_from_code(code: u8) -> Option<Cardinality> {
match code {
0 => Some(Cardinality::Required),
1 => Some(Cardinality::Optional),
2 => Some(Cardinality::Multivalued),
_ => None,
}
}
}
#[derive(Hash, Eq, PartialEq, Debug, Clone, Copy)]
pub enum ColumnType {
Bytes,
Numerical(NumericalType),
}
impl ColumnType {
pub fn to_code(self) -> u8 {
match self {
ColumnType::Bytes => 0u8,
ColumnType::Numerical(numerical_type) => 1u8 | (numerical_type.to_code() << 1),
}
}
pub fn try_from_code(code: u8) -> Option<ColumnType> {
if code == 0u8 {
return Some(ColumnType::Bytes);
}
if code & 1u8 == 0u8 {
return None;
}
let numerical_type = NumericalType::try_from_code(code >> 1)?;
Some(ColumnType::Numerical(numerical_type))
}
}
/// Represents the type and cardinality of a column.
/// This is encoded over one-byte and added to a column key in the
/// columnar sstable.
///
/// Cardinality is encoded as the first two highest two bits.
/// The low 6 bits encode the column type.
#[derive(Eq, Hash, PartialEq, Debug, Copy, Clone)]
pub struct ColumnTypeAndCardinality {
pub cardinality: Cardinality,
pub typ: ColumnType,
}
#[inline]
const fn compute_mask(num_bits: u8) -> u8 {
if num_bits == 8 {
u8::MAX
} else {
(1u8 << num_bits) - 1
}
}
#[inline]
fn select_bits<const START: u8, const END: u8>(code: u8) -> u8 {
assert!(START <= END);
assert!(END <= 8);
let num_bits: u8 = END - START;
let mask: u8 = compute_mask(num_bits);
(code >> START) & mask
}
#[inline]
fn place_bits<const START: u8, const END: u8>(code: u8) -> u8 {
assert!(START <= END);
assert!(END <= 8);
let num_bits: u8 = END - START;
let mask: u8 = compute_mask(num_bits);
assert!(code <= mask);
code << START
}
impl ColumnTypeAndCardinality {
pub fn to_code(self) -> u8 {
place_bits::<6, 8>(self.cardinality.to_code()) | place_bits::<0, 6>(self.typ.to_code())
}
pub fn try_from_code(code: u8) -> Option<ColumnTypeAndCardinality> {
let typ_code = select_bits::<0, 6>(code);
let cardinality_code = select_bits::<6, 8>(code);
let cardinality = Cardinality::try_from_code(cardinality_code)?;
let typ = ColumnType::try_from_code(typ_code)?;
assert_eq!(typ.to_code(), typ_code);
Some(ColumnTypeAndCardinality { cardinality, typ })
}
}
#[cfg(test)]
mod tests {
use std::collections::HashSet;
use super::ColumnTypeAndCardinality;
use crate::column_type_header::{Cardinality, ColumnType};
#[test]
fn test_column_type_header_to_code() {
let mut column_type_header_set: HashSet<ColumnTypeAndCardinality> = HashSet::new();
for code in u8::MIN..=u8::MAX {
if let Some(column_type_header) = ColumnTypeAndCardinality::try_from_code(code) {
assert_eq!(column_type_header.to_code(), code);
assert!(column_type_header_set.insert(column_type_header));
}
}
assert_eq!(
column_type_header_set.len(),
3 /* cardinality */ * (1 + 3) // column_types
);
}
#[test]
fn test_column_type_to_code() {
let mut column_type_set: HashSet<ColumnType> = HashSet::new();
for code in u8::MIN..=u8::MAX {
if let Some(column_type) = ColumnType::try_from_code(code) {
assert_eq!(column_type.to_code(), code);
assert!(column_type_set.insert(column_type));
}
}
assert_eq!(column_type_set.len(), 1 + 3);
}
#[test]
fn test_cardinality_to_code() {
let mut num_cardinality = 0;
for code in u8::MIN..=u8::MAX {
let cardinality_opt = Cardinality::try_from_code(code);
if let Some(cardinality) = cardinality_opt {
assert_eq!(cardinality.to_code(), code);
num_cardinality += 1;
}
}
assert_eq!(num_cardinality, 3);
}
}

78
columnar/src/dictionary.rs
View File

@@ -1,78 +0,0 @@
use std::io;
use fnv::FnvHashMap;
fn fst_err_into_io_err(fst_err: tantivy_fst::Error) -> io::Error {
match fst_err {
tantivy_fst::Error::Fst(fst_err) => {
io::Error::new(io::ErrorKind::Other, format!("FST Error: {:?}", fst_err))
}
tantivy_fst::Error::Io(io_err) => io_err,
}
}
/// `DictionaryBuilder` for dictionary encoding.
///
/// It stores the different terms encounterred and assigns them a temporary value
/// we call unordered id.
///
/// Upon serialization, we will sort the ids and hence build a `UnorderedId -> Term ordinal`
/// mapping.
#[derive(Default)]
pub struct DictionaryBuilder {
dict: FnvHashMap<Vec<u8>, UnorderedId>,
}
pub struct IdMapping {
unordered_to_ord: Vec<OrderedId>,
}
impl IdMapping {
pub fn to_ord(&self, unordered: UnorderedId) -> OrderedId {
self.unordered_to_ord[unordered.0 as usize]
}
}
impl DictionaryBuilder {
/// Get or allocate an unordered id.
/// (This ID is simply an auto-incremented id.)
pub fn get_or_allocate_id(&mut self, term: &[u8]) -> UnorderedId {
if let Some(term_id) = self.dict.get(term) {
return *term_id;
}
let new_id = UnorderedId(self.dict.len() as u32);
self.dict.insert(term.to_vec(), new_id);
new_id
}
/// Serialize the dictionary into an fst, and returns the
/// `UnorderedId -> TermOrdinal` map.
pub fn serialize<'a, W: io::Write + 'a>(&self, wrt: &mut W) -> io::Result<IdMapping> {
serialize_inner(&self.dict, wrt).map_err(fst_err_into_io_err)
}
}
/// Helper function just there for error conversion.
fn serialize_inner<'a, W: io::Write + 'a>(
dict: &FnvHashMap<Vec<u8>, UnorderedId>,
wrt: &mut W,
) -> tantivy_fst::Result<IdMapping> {
let mut terms: Vec<(&[u8], UnorderedId)> =
dict.iter().map(|(k, v)| (k.as_slice(), *v)).collect();
terms.sort_unstable_by_key(|(key, _)| *key);
let mut unordered_to_ord: Vec<OrderedId> = vec![OrderedId(0u32); terms.len()];
let mut fst_builder = tantivy_fst::MapBuilder::new(wrt)?;
for (ord, (key, unordered_id)) in terms.into_iter().enumerate() {
let ordered_id = OrderedId(ord as u32);
fst_builder.insert(key, ord as u64)?;
unordered_to_ord[unordered_id.0 as usize] = ordered_id;
}
fst_builder.finish()?;
Ok(IdMapping { unordered_to_ord })
}
#[derive(Clone, Copy, Debug)]
pub struct UnorderedId(pub u32);
#[derive(Clone, Copy)]
pub struct OrderedId(pub u32);

69
columnar/src/lib.rs
View File

@@ -1,69 +0,0 @@
// Copyright (C) 2022 Quickwit, Inc.
//
// Quickwit is offered under the AGPL v3.0 and as commercial software.
// For commercial licensing, contact us at hello@quickwit.io.
//
// AGPL:
// This program is free software: you can redistribute it and/or modify
// it under the terms of the GNU Affero General Public License as
// published by the Free Software Foundation, either version 3 of the
// License, or (at your option) any later version.
//
// This program is distributed in the hope that it will be useful,
// but WITHOUT ANY WARRANTY; without even the implied warranty of
// MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
// GNU Affero General Public License for more details.
//
// You should have received a copy of the GNU Affero General Public License
// along with this program. If not, see <http://www.gnu.org/licenses/>.
mod column_type_header;
mod dictionary;
mod reader;
mod serializer;
mod value;
mod writer;
pub use column_type_header::Cardinality;
pub use reader::ColumnarReader;
pub use serializer::ColumnarSerializer;
pub use writer::ColumnarWriter;
pub type DocId = u32;
#[cfg(test)]
mod tests {
use std::ops::Range;
use common::file_slice::FileSlice;
use crate::column_type_header::ColumnTypeAndCardinality;
use crate::reader::ColumnarReader;
use crate::serializer::ColumnarSerializer;
use crate::value::NumericalValue;
use crate::ColumnarWriter;
#[test]
fn test_dataframe_writer() {
let mut dataframe_writer = ColumnarWriter::default();
dataframe_writer.record_numerical(1u32, b"srical.value", NumericalValue::U64(1u64));
dataframe_writer.record_numerical(2u32, b"srical.value", NumericalValue::U64(2u64));
dataframe_writer.record_numerical(4u32, b"srical.value", NumericalValue::I64(2i64));
let mut buffer: Vec<u8> = Vec::new();
let serializer = ColumnarSerializer::new(&mut buffer);
dataframe_writer.serialize(5, serializer).unwrap();
let columnar_fileslice = FileSlice::from(buffer);
let columnar = ColumnarReader::open(columnar_fileslice).unwrap();
assert_eq!(columnar.num_columns(), 1);
let cols: Vec<(ColumnTypeAndCardinality, Range<u64>)> =
columnar.read_columns("srical.value").unwrap();
assert_eq!(cols.len(), 1);
// Right now this 31 bytes are spent as follows
//
// - header 14 bytes
// - vals 8 //< due to padding? could have been 1byte?.
// - null footer 6 bytes
// - version footer 3 bytes // Should be file-wide
assert_eq!(cols[0].1, 0..31);
}
}

66
columnar/src/reader/mod.rs
View File

@@ -1,66 +0,0 @@
use std::ops::Range;
use std::{io, mem};
use common::file_slice::FileSlice;
use common::BinarySerializable;
use sstable::{Dictionary, SSTableRange};
use crate::column_type_header::ColumnTypeAndCardinality;
fn io_invalid_data(msg: String) -> io::Error {
io::Error::new(io::ErrorKind::InvalidData, msg) // format!("Invalid key found.
// {key_bytes:?}")));
}
pub struct ColumnarReader {
column_dictionary: Dictionary<SSTableRange>,
column_data: FileSlice,
}
impl ColumnarReader {
pub fn num_columns(&self) -> usize {
self.column_dictionary.num_terms()
}
pub fn open(file_slice: FileSlice) -> io::Result<ColumnarReader> {
let (file_slice_without_sstable_len, sstable_len_bytes) =
file_slice.split_from_end(mem::size_of::<u64>());
let mut sstable_len_bytes = sstable_len_bytes.read_bytes()?;
let sstable_len = u64::deserialize(&mut sstable_len_bytes)?;
let (column_data, sstable) =
file_slice_without_sstable_len.split_from_end(sstable_len as usize);
let column_dictionary = Dictionary::open(sstable)?;
Ok(ColumnarReader {
column_dictionary,
column_data,
})
}
pub fn read_columns(
&self,
field_name: &str,
) -> io::Result<Vec<(ColumnTypeAndCardinality, Range<u64>)>> {
let mut start_key = field_name.to_string();
start_key.push('\0');
let mut end_key = field_name.to_string();
end_key.push(1u8 as char);
let mut stream = self
.column_dictionary
.range()
.ge(start_key.as_bytes())
.lt(end_key.as_bytes())
.into_stream()?;
let mut results = Vec::new();
while stream.advance() {
let key_bytes: &[u8] = stream.key();
if !key_bytes.starts_with(start_key.as_bytes()) {
return Err(io_invalid_data(format!("Invalid key found. {key_bytes:?}")));
}
let column_code: u8 = key_bytes.last().cloned().unwrap();
let column_type_and_cardinality = ColumnTypeAndCardinality::try_from_code(column_code)
.ok_or_else(|| io_invalid_data(format!("Unknown column code `{column_code}`")))?;
let range = stream.value().clone();
results.push((column_type_and_cardinality, range));
}
Ok(results)
}
}

39
columnar/src/serializer.rs
View File

@@ -1,39 +0,0 @@
use std::io;
use std::io::Write;
use std::ops::Range;
use common::CountingWriter;
use sstable::value::RangeWriter;
use sstable::SSTableRange;
pub struct ColumnarSerializer<W: io::Write> {
wrt: CountingWriter<W>,
sstable_range: sstable::Writer<Vec<u8>, RangeWriter>,
}
impl<W: io::Write> ColumnarSerializer<W> {
pub fn new(wrt: W) -> ColumnarSerializer<W> {
let sstable_range: sstable::Writer<Vec<u8>, RangeWriter> =
sstable::Dictionary::<SSTableRange>::builder(Vec::with_capacity(100_000)).unwrap();
ColumnarSerializer {
wrt: CountingWriter::wrap(wrt),
sstable_range,
}
}
pub fn record_column_offsets(&mut self, key: &[u8], byte_range: Range<u64>) -> io::Result<()> {
self.sstable_range.insert(key, &byte_range)
}
pub fn wrt(&mut self) -> &mut CountingWriter<W> {
&mut self.wrt
}
pub fn finalize(mut self) -> io::Result<()> {
let sstable_bytes: Vec<u8> = self.sstable_range.finish()?;
let sstable_num_bytes: u64 = sstable_bytes.len() as u64;
self.wrt.write_all(&sstable_bytes)?;
self.wrt.write_all(&sstable_num_bytes.to_le_bytes()[..])?;
Ok(())
}
}

123
columnar/src/value.rs
View File

@@ -1,123 +0,0 @@
use ordered_float::NotNan;
#[derive(Copy, Clone, Debug, PartialEq)]
pub enum NumericalValue {
I64(i64),
U64(u64),
F64(NotNan<f64>),
}
impl From<u64> for NumericalValue {
fn from(val: u64) -> NumericalValue {
NumericalValue::U64(val)
}
}
impl From<i64> for NumericalValue {
fn from(val: i64) -> Self {
NumericalValue::I64(val)
}
}
impl From<NotNan<f64>> for NumericalValue {
fn from(val: NotNan<f64>) -> Self {
NumericalValue::F64(val)
}
}
impl NumericalValue {
pub fn numerical_type(&self) -> NumericalType {
match self {
NumericalValue::F64(_) => NumericalType::F64,
NumericalValue::I64(_) => NumericalType::I64,
NumericalValue::U64(_) => NumericalType::U64,
}
}
}
impl Eq for NumericalValue {}
#[derive(Clone, Copy, Debug, Default, Hash, Eq, PartialEq)]
#[repr(u8)]
pub enum NumericalType {
#[default]
I64 = 0,
U64 = 1,
F64 = 2,
}
impl NumericalType {
pub fn to_code(self) -> u8 {
self as u8
}
pub fn try_from_code(code: u8) -> Option<NumericalType> {
match code {
0 => Some(NumericalType::I64),
1 => Some(NumericalType::U64),
2 => Some(NumericalType::F64),
_ => None,
}
}
}
/// We voluntarily avoid using `Into` here to keep this
/// implementation quirk as private as possible.
///
/// This coercion trait actually panics if it is used
/// to convert a loose types to a stricter type.
///
/// The level is strictness is somewhat arbitrary.
/// - i64
/// - u64
/// - f64.
pub(crate) trait Coerce {
fn coerce(numerical_value: NumericalValue) -> Self;
}
impl Coerce for i64 {
fn coerce(value: NumericalValue) -> Self {
match value {
NumericalValue::I64(val) => val,
NumericalValue::U64(val) => val as i64,
NumericalValue::F64(_) => unreachable!(),
}
}
}
impl Coerce for u64 {
fn coerce(value: NumericalValue) -> Self {
match value {
NumericalValue::I64(val) => val as u64,
NumericalValue::U64(val) => val,
NumericalValue::F64(_) => unreachable!(),
}
}
}
impl Coerce for NotNan<f64> {
fn coerce(value: NumericalValue) -> Self {
match value {
NumericalValue::I64(val) => unsafe { NotNan::new_unchecked(val as f64) },
NumericalValue::U64(val) => unsafe { NotNan::new_unchecked(val as f64) },
NumericalValue::F64(val) => val,
}
}
}
#[cfg(test)]
mod tests {
use super::NumericalType;
#[test]
fn test_numerical_type_code() {
let mut num_numerical_type = 0;
for code in u8::MIN..=u8::MAX {
if let Some(numerical_type) = NumericalType::try_from_code(code) {
assert_eq!(numerical_type.to_code(), code);
num_numerical_type += 1;
}
}
assert_eq!(num_numerical_type, 3);
}
}

321
columnar/src/writer/column_operation.rs
View File

@@ -1,321 +0,0 @@
use std::fmt;
use std::num::NonZeroU8;
use ordered_float::NotNan;
use thiserror::Error;
use crate::dictionary::UnorderedId;
use crate::value::NumericalValue;
use crate::DocId;
/// When we build a columnar dataframe, we first just group
/// all mutations per column, and append them in append-only object.
///
/// We represents all of these operations as `ColumnOperation`.
#[derive(Eq, PartialEq, Debug, Clone, Copy)]
pub(crate) enum ColumnOperation<T> {
NewDoc(DocId),
Value(T),
}
impl<T> From<T> for ColumnOperation<T> {
fn from(value: T) -> Self {
ColumnOperation::Value(value)
}
}
#[allow(clippy::from_over_into)]
pub(crate) trait SymbolValue: Into<MiniBuffer> + Clone + Copy + fmt::Debug {
fn deserialize(header: NonZeroU8, bytes: &mut &[u8]) -> Result<Self, ParseError>;
}
pub(crate) struct MiniBuffer {
pub bytes: [u8; 9],
pub len: usize,
}
impl MiniBuffer {
pub fn as_slice(&self) -> &[u8] {
&self.bytes[..self.len]
}
}
fn compute_header_byte(typ: SymbolType, len: usize) -> u8 {
assert!(len <= 9);
(len << 4) as u8 | typ as u8
}
impl SymbolValue for NumericalValue {
fn deserialize(header_byte: NonZeroU8, bytes: &mut &[u8]) -> Result<Self, ParseError> {
let (typ, len) = parse_header_byte(header_byte)?;
let value_bytes: &[u8];
(value_bytes, *bytes) = bytes.split_at(len);
let symbol: NumericalValue = match typ {
SymbolType::U64 => {
let mut octet: [u8; 8] = [0u8; 8];
octet[..value_bytes.len()].copy_from_slice(value_bytes);
let val: u64 = u64::from_le_bytes(octet);
NumericalValue::U64(val)
}
SymbolType::I64 => {
let mut octet: [u8; 8] = [0u8; 8];
octet[..value_bytes.len()].copy_from_slice(value_bytes);
let encoded: u64 = u64::from_le_bytes(octet);
let val: i64 = decode_zig_zag(encoded);
NumericalValue::I64(val)
}
SymbolType::Float => {
let octet: [u8; 8] =
value_bytes.try_into().map_err(|_| ParseError::InvalidLen {
typ: SymbolType::Float,
len,
})?;
let val_possibly_nan = f64::from_le_bytes(octet);
let val_not_nan = NotNan::new(val_possibly_nan)
.map_err(|_| ParseError::NaN)?;
NumericalValue::F64(val_not_nan)
}
};
Ok(symbol)
}
}
#[allow(clippy::from_over_into)]
impl Into<MiniBuffer> for NumericalValue {
fn into(self) -> MiniBuffer {
let mut bytes = [0u8; 9];
match self {
NumericalValue::F64(val) => {
let len = 8;
let header_byte = compute_header_byte(SymbolType::Float, len);
bytes[0] = header_byte;
bytes[1..].copy_from_slice(&val.to_le_bytes());
MiniBuffer {
bytes,
len: len + 1,
}
}
NumericalValue::U64(val) => {
let len = compute_num_bytes_for_u64(val);
let header_byte = compute_header_byte(SymbolType::U64, len);
bytes[0] = header_byte;
bytes[1..].copy_from_slice(&val.to_le_bytes());
MiniBuffer {
bytes,
len: len + 1,
}
}
NumericalValue::I64(val) => {
let encoded = encode_zig_zag(val);
let len = compute_num_bytes_for_u64(encoded);
let header_byte = compute_header_byte(SymbolType::I64, len);
bytes[0] = header_byte;
bytes[1..].copy_from_slice(&encoded.to_le_bytes());
MiniBuffer {
bytes,
len: len + 1,
}
}
}
}
}
#[allow(clippy::from_over_into)]
impl Into<MiniBuffer> for UnorderedId {
fn into(self) -> MiniBuffer {
let mut bytes = [0u8; 9];
let val = self.0 as u64;
let len = compute_num_bytes_for_u64(val) + 1;
bytes[0] = len as u8;
bytes[1..].copy_from_slice(&val.to_le_bytes());
MiniBuffer { bytes, len }
}
}
impl SymbolValue for UnorderedId {
fn deserialize(header: NonZeroU8, bytes: &mut &[u8]) -> Result<UnorderedId, ParseError> {
let len = header.get() as usize;
let symbol_bytes: &[u8];
(symbol_bytes, *bytes) = bytes.split_at(len);
let mut value_bytes = [0u8; 4];
value_bytes[..len - 1].copy_from_slice(&symbol_bytes[1..]);
let value = u32::from_le_bytes(value_bytes);
Ok(UnorderedId(value))
}
}
const HEADER_MASK: u8 = (1u8 << 4) - 1u8;
fn compute_num_bytes_for_u64(val: u64) -> usize {
let msb = (64u32 - val.leading_zeros()) as usize;
(msb + 7) / 8
}
fn parse_header_byte(byte: NonZeroU8) -> Result<(SymbolType, usize), ParseError> {
let len = (byte.get() as usize) >> 4;
let typ_code = byte.get() & HEADER_MASK;
let typ = SymbolType::try_from(typ_code)?;
Ok((typ, len))
}
#[derive(Error, Debug)]
pub enum ParseError {
#[error("Type byte unknown `{0}`")]
UnknownType(u8),
#[error("Invalid len for type `{len}` for type `{typ:?}`.")]
InvalidLen { typ: SymbolType, len: usize },
#[error("Missing bytes.")]
MissingBytes,
#[error("Not a number value.")]
NaN,
}
impl<V: SymbolValue> ColumnOperation<V> {
pub fn serialize(self) -> MiniBuffer {
match self {
ColumnOperation::NewDoc(doc) => {
let mut minibuf: [u8; 9] = [0u8; 9];
minibuf[0] = 0u8;
minibuf[1..5].copy_from_slice(&doc.to_le_bytes());
MiniBuffer {
bytes: minibuf,
len: 5,
}
}
ColumnOperation::Value(val) => val.into(),
}
}
pub fn deserialize(bytes: &mut &[u8]) -> Result<Self, ParseError> {
if bytes.is_empty() {
return Err(ParseError::MissingBytes);
}
let header_byte = bytes[0];
*bytes = &bytes[1..];
if let Some(header_byte) = NonZeroU8::new(header_byte) {
let value = V::deserialize(header_byte, bytes)?;
Ok(ColumnOperation::Value(value))
} else {
let doc_bytes: &[u8];
(doc_bytes, *bytes) = bytes.split_at(4);
let doc: u32 =
u32::from_le_bytes(doc_bytes.try_into().map_err(|_| ParseError::MissingBytes)?);
Ok(ColumnOperation::NewDoc(doc))
}
}
}
#[derive(Copy, Clone, Debug, Eq, PartialEq)]
#[repr(u8)]
pub enum SymbolType {
U64 = 1u8,
I64 = 2u8,
Float = 3u8,
}
impl TryFrom<u8> for SymbolType {
type Error = ParseError;
fn try_from(byte: u8) -> Result<Self, ParseError> {
match byte {
1u8 => Ok(SymbolType::U64),
2u8 => Ok(SymbolType::I64),
3u8 => Ok(SymbolType::Float),
_ => Err(ParseError::UnknownType(byte)),
}
}
}
fn encode_zig_zag(n: i64) -> u64 {
((n << 1) ^ (n >> 63)) as u64
}
fn decode_zig_zag(n: u64) -> i64 {
((n >> 1) as i64) ^ (-((n & 1) as i64))
}
#[cfg(test)]
mod tests {
use super::{SymbolType, *};
#[track_caller]
fn test_zig_zag_aux(val: i64) {
let encoded = super::encode_zig_zag(val);
assert_eq!(decode_zig_zag(encoded), val);
if let Some(abs_val) = val.checked_abs() {
let abs_val = abs_val as u64;
assert!(encoded <= abs_val * 2);
}
}
#[test]
fn test_zig_zag() {
assert_eq!(encode_zig_zag(0i64), 0u64);
assert_eq!(encode_zig_zag(-1i64), 1u64);
assert_eq!(encode_zig_zag(1i64), 2u64);
test_zig_zag_aux(0i64);
test_zig_zag_aux(i64::MIN);
test_zig_zag_aux(i64::MAX);
}
use proptest::prelude::any;
use proptest::proptest;
proptest! {
#[test]
fn test_proptest_zig_zag(val in any::<i64>()) {
test_zig_zag_aux(val);
}
}
#[track_caller]
fn ser_deser_header_byte_aux(symbol_type: SymbolType, len: usize) {
let header_byte = compute_header_byte(symbol_type, len);
let (serdeser_numerical_type, serdeser_len) =
parse_header_byte(NonZeroU8::new(header_byte).unwrap()).unwrap();
assert_eq!(symbol_type, serdeser_numerical_type);
assert_eq!(len, serdeser_len);
}
#[test]
fn test_header_byte_serialization() {
for len in 1..9 {
ser_deser_header_byte_aux(SymbolType::Float, len);
ser_deser_header_byte_aux(SymbolType::I64, len);
ser_deser_header_byte_aux(SymbolType::U64, len);
}
}
#[track_caller]
fn ser_deser_symbol(symbol: ColumnOperation<NumericalValue>) {
let buf = symbol.serialize();
let mut bytes = &buf.bytes[..];
let serdeser_symbol = ColumnOperation::deserialize(&mut bytes).unwrap();
assert_eq!(bytes.len() + buf.len, buf.bytes.len());
assert_eq!(symbol, serdeser_symbol);
}
#[test]
fn test_compute_num_bytes_for_u64() {
assert_eq!(compute_num_bytes_for_u64(0), 0);
assert_eq!(compute_num_bytes_for_u64(1), 1);
assert_eq!(compute_num_bytes_for_u64(255), 1);
assert_eq!(compute_num_bytes_for_u64(256), 2);
assert_eq!(compute_num_bytes_for_u64((1 << 16) - 1), 2);
assert_eq!(compute_num_bytes_for_u64(1 << 16), 3);
}
#[test]
fn test_symbol_serialization() {
ser_deser_symbol(ColumnOperation::NewDoc(0));
ser_deser_symbol(ColumnOperation::NewDoc(3));
ser_deser_symbol(ColumnOperation::Value(NumericalValue::I64(0i64)));
ser_deser_symbol(ColumnOperation::Value(NumericalValue::I64(1i64)));
ser_deser_symbol(ColumnOperation::Value(NumericalValue::U64(257u64)));
ser_deser_symbol(ColumnOperation::Value(NumericalValue::I64(-257i64)));
ser_deser_symbol(ColumnOperation::Value(NumericalValue::I64(i64::MIN)));
ser_deser_symbol(ColumnOperation::Value(NumericalValue::U64(0u64)));
ser_deser_symbol(ColumnOperation::Value(NumericalValue::U64(u64::MIN)));
ser_deser_symbol(ColumnOperation::Value(NumericalValue::U64(u64::MAX)));
}
}

675
columnar/src/writer/mod.rs
View File

@@ -1,675 +0,0 @@
mod column_operation;
mod value_index;
use std::io::{self, Write};
use column_operation::ColumnOperation;
use common::CountingWriter;
use fastfield_codecs::serialize::ValueIndexInfo;
use fastfield_codecs::{Column, MonotonicallyMappableToU64, VecColumn};
use ordered_float::NotNan;
use stacker::{Addr, ArenaHashMap, ExpUnrolledLinkedList, MemoryArena};
use crate::column_type_header::{ColumnType, ColumnTypeAndCardinality};
use crate::dictionary::{DictionaryBuilder, IdMapping, UnorderedId};
use crate::value::{Coerce, NumericalType, NumericalValue};
use crate::writer::column_operation::SymbolValue;
use crate::writer::value_index::{IndexBuilder, SpareIndexBuilders};
use crate::{Cardinality, ColumnarSerializer, DocId};
#[derive(Copy, Clone, Default)]
struct ColumnWriter {
// Detected cardinality of the column so far.
cardinality: Cardinality,
// Last document inserted.
// None if no doc has been added yet.
last_doc_opt: Option<u32>,
// Buffer containing the serialized values.
values: ExpUnrolledLinkedList,
}
#[derive(Clone, Copy, Default)]
pub struct NumericalColumnWriter {
compatible_numerical_types: CompatibleNumericalTypes,
column_writer: ColumnWriter,
}
#[derive(Clone, Copy)]
struct CompatibleNumericalTypes {
all_values_within_i64_range: bool,
all_values_within_u64_range: bool,
}
impl Default for CompatibleNumericalTypes {
fn default() -> CompatibleNumericalTypes {
CompatibleNumericalTypes {
all_values_within_i64_range: true,
all_values_within_u64_range: true,
}
}
}
impl CompatibleNumericalTypes {
pub fn accept_value(&mut self, numerical_value: NumericalValue) {
match numerical_value {
NumericalValue::I64(val_i64) => {
let value_within_u64_range = val_i64 >= 0i64;
self.all_values_within_u64_range &= value_within_u64_range;
}
NumericalValue::U64(val_u64) => {
let value_within_i64_range = val_u64 < i64::MAX as u64;
self.all_values_within_i64_range &= value_within_i64_range;
}
NumericalValue::F64(_) => {
self.all_values_within_i64_range = false;
self.all_values_within_u64_range = false;
}
}
}
pub fn to_numerical_type(self) -> NumericalType {
if self.all_values_within_i64_range {
NumericalType::I64
} else if self.all_values_within_u64_range {
NumericalType::U64
} else {
NumericalType::F64
}
}
}
impl NumericalColumnWriter {
pub fn record_numerical_value(
&mut self,
doc: DocId,
value: NumericalValue,
arena: &mut MemoryArena,
) {
self.compatible_numerical_types.accept_value(value);
self.column_writer.record(doc, value, arena);
}
}
impl ColumnWriter {
fn symbol_iterator<'a, V: SymbolValue>(
&self,
arena: &MemoryArena,
buffer: &'a mut Vec<u8>,
) -> impl Iterator<Item = ColumnOperation<V>> + 'a {
buffer.clear();
self.values.read_to_end(arena, buffer);
let mut cursor: &[u8] = &buffer[..];
std::iter::from_fn(move || {
if cursor.is_empty() {
return None;
}
let symbol = ColumnOperation::deserialize(&mut cursor)
.expect("Failed to deserialize symbol from in-memory. This should never happen.");
Some(symbol)
})
}
fn delta_with_last_doc(&self, doc: DocId) -> u32 {
self.last_doc_opt
.map(|last_doc| doc - last_doc)
.unwrap_or(doc + 1u32)
}
/// Records a change of the document being recorded.
///
/// This function will also update the cardinality of the column
/// if necessary.
fn record(&mut self, doc: DocId, value: NumericalValue, arena: &mut MemoryArena) {
// Difference between `doc` and the last doc.
match self.delta_with_last_doc(doc) {
0 => {
// This is the last encounterred document.
self.cardinality = Cardinality::Multivalued;
}
1 => {
self.last_doc_opt = Some(doc);
self.write_symbol::<NumericalValue>(ColumnOperation::NewDoc(doc), arena);
}
_ => {
self.cardinality = self.cardinality.max(Cardinality::Optional);
self.last_doc_opt = Some(doc);
self.write_symbol::<NumericalValue>(ColumnOperation::NewDoc(doc), arena);
}
}
self.write_symbol(ColumnOperation::Value(value), arena);
}
// Get the cardinality.
// The overall number of docs in the column is necessary to
// deal with the case where the all docs contain 1 value, except some documents
// at the end of the column.
fn get_cardinality(&self, num_docs: DocId) -> Cardinality {
if self.delta_with_last_doc(num_docs) > 1 {
self.cardinality.max(Cardinality::Optional)
} else {
self.cardinality
}
}
fn write_symbol<V: SymbolValue>(
&mut self,
symbol: ColumnOperation<V>,
arena: &mut MemoryArena,
) {
self.values
.writer(arena)
.extend_from_slice(symbol.serialize().as_slice());
}
}
#[derive(Copy, Clone, Default)]
pub struct BytesColumnWriter {
dictionary_id: u32,
column_writer: ColumnWriter,
}
impl BytesColumnWriter {
pub fn with_dictionary_id(dictionary_id: u32) -> BytesColumnWriter {
BytesColumnWriter {
dictionary_id,
column_writer: Default::default(),
}
}
pub fn record_bytes(
&mut self,
doc: DocId,
bytes: &[u8],
dictionaries: &mut [DictionaryBuilder],
arena: &mut MemoryArena,
) {
let unordered_id = dictionaries[self.dictionary_id as usize].get_or_allocate_id(bytes);
let numerical_value = NumericalValue::U64(unordered_id.0 as u64);
self.column_writer.record(doc, numerical_value, arena);
}
}
pub struct ColumnarWriter {
numerical_field_hash_map: ArenaHashMap,
bytes_field_hash_map: ArenaHashMap,
arena: MemoryArena,
// Dictionaries used to store dictionary-encoded values.
dictionaries: Vec<DictionaryBuilder>,
buffers: SpareBuffers,
}
#[derive(Default)]
struct SpareBuffers {
byte_buffer: Vec<u8>,
value_index_builders: SpareIndexBuilders,
i64_values: Vec<i64>,
u64_values: Vec<u64>,
f64_values: Vec<ordered_float::NotNan<f64>>,
}
impl Default for ColumnarWriter {
fn default() -> Self {
ColumnarWriter {
numerical_field_hash_map: ArenaHashMap::new(10_000),
bytes_field_hash_map: ArenaHashMap::new(10_000),
dictionaries: Vec::new(),
arena: MemoryArena::default(),
buffers: SpareBuffers::default(),
}
}
}
#[derive(Copy, Clone, Ord, PartialOrd, Eq, PartialEq, Debug)]
enum BytesOrNumerical {
Bytes,
Numerical,
}
impl ColumnarWriter {
pub fn record_numerical(&mut self, doc: DocId, key: &[u8], numerical_value: NumericalValue) {
let (hash_map, arena) = (&mut self.numerical_field_hash_map, &mut self.arena);
hash_map.mutate_or_create(key, |column_opt: Option<NumericalColumnWriter>| {
let mut column: NumericalColumnWriter = column_opt.unwrap_or_default();
column.record_numerical_value(doc, numerical_value, arena);
column
});
}
pub fn record_bytes(&mut self, doc: DocId, key: &[u8], value: &[u8]) {
let (hash_map, arena, dictionaries) = (
&mut self.bytes_field_hash_map,
&mut self.arena,
&mut self.dictionaries,
);
hash_map.mutate_or_create(key, |column_opt: Option<BytesColumnWriter>| {
let mut column: BytesColumnWriter = column_opt.unwrap_or_else(|| {
let dictionary_id = dictionaries.len() as u32;
dictionaries.push(DictionaryBuilder::default());
BytesColumnWriter::with_dictionary_id(dictionary_id)
});
column.record_bytes(doc, value, dictionaries, arena);
column
});
}
pub fn serialize<W: io::Write>(
&mut self,
num_docs: DocId,
mut serializer: ColumnarSerializer<W>,
) -> io::Result<()> {
let mut field_columns: Vec<(&[u8], BytesOrNumerical, Addr)> = self
.numerical_field_hash_map
.iter()
.map(|(term, addr, _)| (term, BytesOrNumerical::Numerical, addr))
.collect();
field_columns.extend(
self.bytes_field_hash_map
.iter()
.map(|(term, addr, _)| (term, BytesOrNumerical::Bytes, addr)),
);
let mut key_buffer = Vec::new();
field_columns.sort_unstable_by_key(|(key, col_type, _)| (*key, *col_type));
let (arena, buffers, dictionaries) = (&self.arena, &mut self.buffers, &self.dictionaries);
for (key, bytes_or_numerical, addr) in field_columns {
let wrt = serializer.wrt();
let start_offset = wrt.written_bytes();
let column_type_and_cardinality: ColumnTypeAndCardinality =
match bytes_or_numerical {
BytesOrNumerical::Bytes => {
let BytesColumnWriter { dictionary_id, column_writer } =
self.bytes_field_hash_map.read(addr);
let dictionary_builder =
&dictionaries[dictionary_id as usize];
serialize_bytes_column(
&column_writer,
num_docs,
dictionary_builder,
arena,
buffers,
wrt,
)?;
ColumnTypeAndCardinality {
cardinality: column_writer.get_cardinality(num_docs),
typ: ColumnType::Bytes,
}
}
BytesOrNumerical::Numerical => {
let NumericalColumnWriter { compatible_numerical_types, column_writer } =
self.numerical_field_hash_map.read(addr);
let cardinality = column_writer.get_cardinality(num_docs);
let numerical_type = compatible_numerical_types.to_numerical_type();
serialize_numerical_column(
cardinality,
numerical_type,
&column_writer,
num_docs,
arena,
buffers,
wrt,
)?;
ColumnTypeAndCardinality {
cardinality,
typ: ColumnType::Numerical(numerical_type),
}
}
};
let end_offset = wrt.written_bytes();
let key_with_type = prepare_key(key, column_type_and_cardinality, &mut key_buffer);
serializer.record_column_offsets(key_with_type, start_offset..end_offset)?;
}
serializer.finalize()?;
Ok(())
}
}
/// Returns a key consisting of the concatenation of the key and the column_type_and_cardinality
/// code.
fn prepare_key<'a>(
key: &[u8],
column_type_cardinality: ColumnTypeAndCardinality,
buffer: &'a mut Vec<u8>,
) -> &'a [u8] {
buffer.clear();
buffer.extend_from_slice(key);
buffer.push(0u8);
buffer.push(column_type_cardinality.to_code());
&buffer[..]
}
fn serialize_bytes_column<W: io::Write>(
column_writer: &ColumnWriter,
num_docs: DocId,
dictionary_builder: &DictionaryBuilder,
arena: &MemoryArena,
buffers: &mut SpareBuffers,
wrt: &mut CountingWriter<W>,
) -> io::Result<()> {
let start_offset = wrt.written_bytes();
let id_mapping: IdMapping = dictionary_builder.serialize(wrt)?;
let dictionary_num_bytes: u32 = (wrt.written_bytes() - start_offset) as u32;
let cardinality = column_writer.get_cardinality(num_docs);
let SpareBuffers {
byte_buffer,
value_index_builders,
u64_values,
..
} = buffers;
let symbol_iterator = column_writer
.symbol_iterator(arena, byte_buffer)
.map(|symbol: ColumnOperation<UnorderedId>| {
// We map unordered ids to ordered ids.
match symbol {
ColumnOperation::Value(unordered_id) => {
let ordered_id = id_mapping.to_ord(unordered_id);
ColumnOperation::Value(ordered_id.0 as u64)
}
ColumnOperation::NewDoc(doc) => ColumnOperation::NewDoc(doc),
}
});
serialize_column(
symbol_iterator,
cardinality,
num_docs,
value_index_builders,
u64_values,
wrt,
)?;
wrt.write_all(&dictionary_num_bytes.to_le_bytes()[..])?;
Ok(())
}
fn serialize_numerical_column<W: io::Write>(
cardinality: Cardinality,
numerical_type: NumericalType,
column_writer: &ColumnWriter,
num_docs: DocId,
arena: &MemoryArena,
buffers: &mut SpareBuffers,
wrt: &mut W,
) -> io::Result<()> {
let SpareBuffers {
byte_buffer,
value_index_builders,
u64_values,
i64_values,
f64_values,
} = buffers;
let symbol_iterator = column_writer.symbol_iterator(arena, byte_buffer);
match numerical_type {
NumericalType::I64 => {
serialize_column(
coerce_numerical_symbol::<i64>(symbol_iterator),
cardinality,
num_docs,
value_index_builders,
i64_values,
wrt,
)?;
}
NumericalType::U64 => {
serialize_column(
coerce_numerical_symbol::<u64>(symbol_iterator),
cardinality,
num_docs,
value_index_builders,
u64_values,
wrt,
)?;
}
NumericalType::F64 => {
serialize_column(
coerce_numerical_symbol::<NotNan<f64>>(symbol_iterator),
cardinality,
num_docs,
value_index_builders,
f64_values,
wrt,
)?;
}
};
Ok(())
}
fn serialize_column<
T: Copy + Ord + Default + Send + Sync + MonotonicallyMappableToU64,
W: io::Write,
>(
symbol_iterator: impl Iterator<Item = ColumnOperation<T>>,
cardinality: Cardinality,
num_docs: DocId,
value_index_builders: &mut SpareIndexBuilders,
values: &mut Vec<T>,
wrt: &mut W,
) -> io::Result<()>
where
for<'a> VecColumn<'a, T>: Column<T>,
{
match cardinality {
Cardinality::Required => {
consume_symbol_iterator(
symbol_iterator,
value_index_builders.borrow_required_index_builder(),
values,
);
fastfield_codecs::serialize(
VecColumn::from(&values[..]),
wrt,
&fastfield_codecs::ALL_CODEC_TYPES[..],
)?;
}
Cardinality::Optional => {
let optional_index_builder = value_index_builders.borrow_optional_index_builder();
consume_symbol_iterator(symbol_iterator, optional_index_builder, values);
let optional_index = optional_index_builder.finish(num_docs);
fastfield_codecs::serialize::serialize_new(
ValueIndexInfo::SingleValue(Box::new(optional_index)),
VecColumn::from(&values[..]),
wrt,
&fastfield_codecs::ALL_CODEC_TYPES[..],
)?;
}
Cardinality::Multivalued => {
let multivalued_index_builder = value_index_builders.borrow_multivalued_index_builder();
consume_symbol_iterator(symbol_iterator, multivalued_index_builder, values);
let multivalued_index = multivalued_index_builder.finish(num_docs);
fastfield_codecs::serialize::serialize_new(
ValueIndexInfo::MultiValue(Box::new(multivalued_index)),
VecColumn::from(&values[..]),
wrt,
&fastfield_codecs::ALL_CODEC_TYPES[..],
)?;
}
}
Ok(())
}
fn coerce_numerical_symbol<T>(
symbol_iterator: impl Iterator<Item = ColumnOperation<NumericalValue>>,
) -> impl Iterator<Item = ColumnOperation<T>>
where T: Coerce {
symbol_iterator.map(|symbol| match symbol {
ColumnOperation::NewDoc(doc) => ColumnOperation::NewDoc(doc),
ColumnOperation::Value(numerical_value) => {
ColumnOperation::Value(Coerce::coerce(numerical_value))
}
})
}
fn consume_symbol_iterator<T, TIndexBuilder: IndexBuilder>(
symbol_iterator: impl Iterator<Item = ColumnOperation<T>>,
index_builder: &mut TIndexBuilder,
values: &mut Vec<T>,
) {
for symbol in symbol_iterator {
match symbol {
ColumnOperation::NewDoc(doc) => {
index_builder.record_doc(doc);
}
ColumnOperation::Value(value) => {
index_builder.record_value();
values.push(value);
}
}
}
}
#[cfg(test)]
mod tests {
use ordered_float::NotNan;
use stacker::MemoryArena;
use super::prepare_key;
use crate::column_type_header::{ColumnType, ColumnTypeAndCardinality};
use crate::value::{NumericalType, NumericalValue};
use crate::writer::column_operation::ColumnOperation;
use crate::writer::CompatibleNumericalTypes;
use crate::Cardinality;
#[test]
fn test_prepare_key_bytes() {
let mut buffer: Vec<u8> = b"somegarbage".to_vec();
let column_type_and_cardinality = ColumnTypeAndCardinality {
typ: ColumnType::Bytes,
cardinality: Cardinality::Optional,
};
let prepared_key = prepare_key(b"root\0child", column_type_and_cardinality, &mut buffer);
assert_eq!(prepared_key.len(), 12);
assert_eq!(&prepared_key[..10], b"root\0child");
assert_eq!(prepared_key[10], 0u8);
assert_eq!(prepared_key[11], column_type_and_cardinality.to_code());
}
#[test]
fn test_column_writer_required_simple() {
let mut arena = MemoryArena::default();
let mut column_writer = super::ColumnWriter::default();
column_writer.record(0u32, 14i64.into(), &mut arena);
column_writer.record(1u32, 15i64.into(), &mut arena);
column_writer.record(2u32, (-16i64).into(), &mut arena);
assert_eq!(column_writer.get_cardinality(3), Cardinality::Required);
let mut buffer = Vec::new();
let symbols: Vec<ColumnOperation<NumericalValue>> = column_writer
.symbol_iterator(&mut arena, &mut buffer)
.collect();
assert_eq!(symbols.len(), 6);
assert!(matches!(symbols[0], ColumnOperation::NewDoc(0u32)));
assert!(matches!(
symbols[1],
ColumnOperation::Value(NumericalValue::I64(14i64))
));
assert!(matches!(symbols[2], ColumnOperation::NewDoc(1u32)));
assert!(matches!(
symbols[3],
ColumnOperation::Value(NumericalValue::I64(15i64))
));
assert!(matches!(symbols[4], ColumnOperation::NewDoc(2u32)));
assert!(matches!(
symbols[5],
ColumnOperation::Value(NumericalValue::I64(-16i64))
));
}
#[test]
fn test_column_writer_optional_cardinality_missing_first() {
let mut arena = MemoryArena::default();
let mut column_writer = super::ColumnWriter::default();
column_writer.record(1u32, 15i64.into(), &mut arena);
column_writer.record(2u32, (-16i64).into(), &mut arena);
assert_eq!(column_writer.get_cardinality(3), Cardinality::Optional);
let mut buffer = Vec::new();
let symbols: Vec<ColumnOperation<NumericalValue>> = column_writer
.symbol_iterator(&mut arena, &mut buffer)
.collect();
assert_eq!(symbols.len(), 4);
assert!(matches!(symbols[0], ColumnOperation::NewDoc(1u32)));
assert!(matches!(
symbols[1],
ColumnOperation::Value(NumericalValue::I64(15i64))
));
assert!(matches!(symbols[2], ColumnOperation::NewDoc(2u32)));
assert!(matches!(
symbols[3],
ColumnOperation::Value(NumericalValue::I64(-16i64))
));
}
#[test]
fn test_column_writer_optional_cardinality_missing_last() {
let mut arena = MemoryArena::default();
let mut column_writer = super::ColumnWriter::default();
column_writer.record(0u32, 15i64.into(), &mut arena);
assert_eq!(column_writer.get_cardinality(2), Cardinality::Optional);
let mut buffer = Vec::new();
let symbols: Vec<ColumnOperation<NumericalValue>> = column_writer
.symbol_iterator(&mut arena, &mut buffer)
.collect();
assert_eq!(symbols.len(), 2);
assert!(matches!(symbols[0], ColumnOperation::NewDoc(0u32)));
assert!(matches!(
symbols[1],
ColumnOperation::Value(NumericalValue::I64(15i64))
));
}
#[test]
fn test_column_writer_multivalued() {
let mut arena = MemoryArena::default();
let mut column_writer = super::ColumnWriter::default();
column_writer.record(0u32, 16i64.into(), &mut arena);
column_writer.record(0u32, 17i64.into(), &mut arena);
assert_eq!(column_writer.get_cardinality(1), Cardinality::Multivalued);
let mut buffer = Vec::new();
let symbols: Vec<ColumnOperation<NumericalValue>> = column_writer
.symbol_iterator(&mut arena, &mut buffer)
.collect();
assert_eq!(symbols.len(), 3);
assert!(matches!(symbols[0], ColumnOperation::NewDoc(0u32)));
assert!(matches!(
symbols[1],
ColumnOperation::Value(NumericalValue::I64(16i64))
));
assert!(matches!(
symbols[2],
ColumnOperation::Value(NumericalValue::I64(17i64))
));
}
#[track_caller]
fn test_column_writer_coercion_iter_aux(
values: impl Iterator<Item = NumericalValue>,
expected_numerical_type: NumericalType,
) {
let mut compatible_numerical_types = CompatibleNumericalTypes::default();
for value in values {
compatible_numerical_types.accept_value(value);
}
assert_eq!(
compatible_numerical_types.to_numerical_type(),
expected_numerical_type
);
}
#[track_caller]
fn test_column_writer_coercion_aux(
values: &[NumericalValue],
expected_numerical_type: NumericalType,
) {
test_column_writer_coercion_iter_aux(values.iter().copied(), expected_numerical_type);
test_column_writer_coercion_iter_aux(values.iter().rev().copied(), expected_numerical_type);
}
#[test]
fn test_column_writer_coercion() {
test_column_writer_coercion_aux(&[], NumericalType::I64);
test_column_writer_coercion_aux(&[1i64.into()], NumericalType::I64);
test_column_writer_coercion_aux(&[1u64.into()], NumericalType::I64);
// We don't detect exact integer at the moment. We could!
test_column_writer_coercion_aux(&[NotNan::new(1f64).unwrap().into()], NumericalType::F64);
test_column_writer_coercion_aux(&[u64::MAX.into()], NumericalType::U64);
test_column_writer_coercion_aux(&[(i64::MAX as u64).into()], NumericalType::U64);
test_column_writer_coercion_aux(&[(1u64 << 63).into()], NumericalType::U64);
test_column_writer_coercion_aux(&[1i64.into(), 1u64.into()], NumericalType::I64);
test_column_writer_coercion_aux(&[u64::MAX.into(), (-1i64).into()], NumericalType::F64);
}
}

218
columnar/src/writer/value_index.rs
View File

@@ -1,218 +0,0 @@
use fastfield_codecs::serialize::{MultiValueIndexInfo, SingleValueIndexInfo};
use crate::DocId;
/// The `IndexBuilder` interprets a sequence of
/// calls of the form:
/// (record_doc,record_value+)*
/// and can then serialize the results into an index.
///
/// It has different implementation depending on whether the
/// cardinality is required, optional, or multivalued.
pub(crate) trait IndexBuilder {
fn record_doc(&mut self, doc: DocId);
#[inline]
fn record_value(&mut self) {}
}
/// The RequiredIndexBuilder does nothing.
#[derive(Default)]
pub struct RequiredIndexBuilder;
impl IndexBuilder for RequiredIndexBuilder {
#[inline(always)]
fn record_doc(&mut self, _doc: DocId) {}
}
#[derive(Default)]
pub struct OptionalIndexBuilder {
docs: Vec<DocId>,
}
struct SingleValueArrayIndex<'a> {
docs: &'a [DocId],
num_docs: DocId,
}
impl<'a> SingleValueIndexInfo for SingleValueArrayIndex<'a> {
fn num_vals(&self) -> u32 {
self.num_docs as u32
}
fn num_non_nulls(&self) -> u32 {
self.docs.len() as u32
}
fn iter(&self) -> Box<dyn Iterator<Item = u32> + '_> {
Box::new(self.docs.iter().copied())
}
}
impl OptionalIndexBuilder {
pub fn finish(&mut self, num_docs: DocId) -> impl SingleValueIndexInfo + '_ {
debug_assert!(self
.docs
.last()
.copied()
.map(|last_doc| last_doc < num_docs)
.unwrap_or(true));
SingleValueArrayIndex {
docs: &self.docs[..],
num_docs,
}
}
fn reset(&mut self) {
self.docs.clear();
}
}
impl IndexBuilder for OptionalIndexBuilder {
#[inline(always)]
fn record_doc(&mut self, doc: DocId) {
debug_assert!(self
.docs
.last()
.copied()
.map(|prev_doc| doc > prev_doc)
.unwrap_or(true));
self.docs.push(doc);
}
}
#[derive(Default)]
pub struct MultivaluedIndexBuilder {
// TODO should we switch to `start_offset`?
end_values: Vec<DocId>,
total_num_vals_seen: u32,
}
pub struct MultivaluedValueArrayIndex<'a> {
end_offsets: &'a [DocId],
}
impl<'a> MultiValueIndexInfo for MultivaluedValueArrayIndex<'a> {
fn num_docs(&self) -> u32 {
self.end_offsets.len() as u32
}
fn num_vals(&self) -> u32 {
self.end_offsets.last().copied().unwrap_or(0u32)
}
fn iter(&self) -> Box<dyn Iterator<Item = u32> + '_> {
if self.end_offsets.is_empty() {
return Box::new(std::iter::empty());
}
let n = self.end_offsets.len();
Box::new(std::iter::once(0u32).chain(self.end_offsets[..n - 1].iter().copied()))
}
}
impl MultivaluedIndexBuilder {
pub fn finish(&mut self, num_docs: DocId) -> impl MultiValueIndexInfo + '_ {
self.end_values
.resize(num_docs as usize, self.total_num_vals_seen);
MultivaluedValueArrayIndex {
end_offsets: &self.end_values[..],
}
}
fn reset(&mut self) {
self.end_values.clear();
self.total_num_vals_seen = 0;
}
}
impl IndexBuilder for MultivaluedIndexBuilder {
fn record_doc(&mut self, doc: DocId) {
self.end_values
.resize(doc as usize, self.total_num_vals_seen);
}
fn record_value(&mut self) {
self.total_num_vals_seen += 1;
}
}
/// The `SpareIndexBuilders` is there to avoid allocating a
/// new index builder for every single column.
#[derive(Default)]
pub struct SpareIndexBuilders {
required_index_builder: RequiredIndexBuilder,
optional_index_builder: OptionalIndexBuilder,
multivalued_index_builder: MultivaluedIndexBuilder,
}
impl SpareIndexBuilders {
pub fn borrow_required_index_builder(&mut self) -> &mut RequiredIndexBuilder {
&mut self.required_index_builder
}
pub fn borrow_optional_index_builder(&mut self) -> &mut OptionalIndexBuilder {
self.optional_index_builder.reset();
&mut self.optional_index_builder
}
pub fn borrow_multivalued_index_builder(&mut self) -> &mut MultivaluedIndexBuilder {
self.multivalued_index_builder.reset();
&mut self.multivalued_index_builder
}
}
#[cfg(test)]
mod tests {
use super::*;
#[test]
fn test_optional_value_index_builder() {
let mut opt_value_index_builder = OptionalIndexBuilder::default();
opt_value_index_builder.record_doc(0u32);
opt_value_index_builder.record_value();
assert_eq!(
&opt_value_index_builder
.finish(1u32)
.iter()
.collect::<Vec<u32>>(),
&[0]
);
opt_value_index_builder.reset();
opt_value_index_builder.record_doc(1u32);
opt_value_index_builder.record_value();
assert_eq!(
&opt_value_index_builder
.finish(2u32)
.iter()
.collect::<Vec<u32>>(),
&[1]
);
}
#[test]
fn test_multivalued_value_index_builder() {
let mut multivalued_value_index_builder = MultivaluedIndexBuilder::default();
multivalued_value_index_builder.record_doc(1u32);
multivalued_value_index_builder.record_value();
multivalued_value_index_builder.record_value();
multivalued_value_index_builder.record_doc(2u32);
multivalued_value_index_builder.record_value();
assert_eq!(
multivalued_value_index_builder
.finish(4u32)
.iter()
.collect::<Vec<u32>>(),
vec![0, 0, 2, 3]
);
multivalued_value_index_builder.reset();
multivalued_value_index_builder.record_doc(2u32);
multivalued_value_index_builder.record_value();
multivalued_value_index_builder.record_value();
assert_eq!(
multivalued_value_index_builder
.finish(4u32)
.iter()
.collect::<Vec<u32>>(),
vec![0, 0, 0, 2]
);
}
}

9
common/Cargo.toml
View File

@@ -1,21 +1,16 @@
[package]
name = "tantivy-common"
version = "0.5.0"
version = "0.3.0"
authors = ["Paul Masurel <paul@quickwit.io>", "Pascal Seitz <pascal@quickwit.io>"]
license = "MIT"
edition = "2021"
description = "common traits and utility functions used by multiple tantivy subcrates"
documentation = "https://docs.rs/tantivy_common/"
homepage = "https://github.com/quickwit-oss/tantivy"
repository = "https://github.com/quickwit-oss/tantivy"
# See more keys and their definitions at https://doc.rust-lang.org/cargo/reference/manifest.html
[dependencies]
byteorder = "1.4.3"
ownedbytes = { version= "0.5", path="../ownedbytes" }
async-trait = "0.1"
ownedbytes = { version="0.3", path="../ownedbytes" }
[dev-dependencies]
proptest = "1.0.0"

6
common/src/bitset.rs
View File

@@ -151,7 +151,7 @@ impl TinySet {
if self.is_empty() {
None
} else {
let lowest = self.0.trailing_zeros();
let lowest = self.0.trailing_zeros() as u32;
self.0 ^= TinySet::singleton(lowest).0;
Some(lowest)
}
@@ -421,7 +421,7 @@ mod tests {
bitset.serialize(&mut out).unwrap();
let bitset = ReadOnlyBitSet::open(OwnedBytes::new(out));
assert_eq!(bitset.len(), i as usize);
assert_eq!(bitset.len() as usize, i as usize);
}
}
@@ -432,7 +432,7 @@ mod tests {
bitset.serialize(&mut out).unwrap();
let bitset = ReadOnlyBitSet::open(OwnedBytes::new(out));
assert_eq!(bitset.len(), 64);
assert_eq!(bitset.len() as usize, 64);
}
#[test]

3
common/src/lib.rs
View File

@@ -5,12 +5,11 @@ use std::ops::Deref;
pub use byteorder::LittleEndian as Endianness;
mod bitset;
pub mod file_slice;
mod serialize;
mod vint;
mod writer;
pub use bitset::*;
pub use ownedbytes::OwnedBytes;
pub use serialize::{BinarySerializable, DeserializeFrom, FixedSize};
pub use vint::{
deserialize_vint_u128, read_u32_vint, read_u32_vint_no_advance, serialize_vint_u128,

32
common/src/serialize.rs
View File

@@ -1,3 +1,4 @@
use std::borrow::Cow;
use std::io::{Read, Write};
use std::{fmt, io};
@@ -94,20 +95,6 @@ impl FixedSize for u32 {
const SIZE_IN_BYTES: usize = 4;
}
impl BinarySerializable for u16 {
fn serialize<W: Write>(&self, writer: &mut W) -> io::Result<()> {
writer.write_u16::<Endianness>(*self)
}
fn deserialize<R: Read>(reader: &mut R) -> io::Result<u16> {
reader.read_u16::<Endianness>()
}
}
impl FixedSize for u16 {
const SIZE_IN_BYTES: usize = 2;
}
impl BinarySerializable for u64 {
fn serialize<W: Write>(&self, writer: &mut W) -> io::Result<()> {
writer.write_u64::<Endianness>(*self)
@@ -224,6 +211,23 @@ impl BinarySerializable for String {
}
}
impl<'a> BinarySerializable for Cow<'a, str> {
fn serialize<W: Write>(&self, writer: &mut W) -> io::Result<()> {
let data: &[u8] = self.as_bytes();
VInt(data.len() as u64).serialize(writer)?;
writer.write_all(data)
}
fn deserialize<R: Read>(reader: &mut R) -> io::Result<Self> {
let string_length = VInt::deserialize(reader)?.val() as usize;
let mut result = String::with_capacity(string_length);
reader
.take(string_length as u64)
.read_to_string(&mut result)?;
Ok(Cow::Owned(result))
}
}
#[cfg(test)]
pub mod test {

2
common/src/vint.rs
View File

@@ -157,7 +157,7 @@ fn vint_len(data: &[u8]) -> usize {
/// If the buffer does not start by a valid
/// vint payload
pub fn read_u32_vint(data: &mut &[u8]) -> u32 {
let (result, vlen) = read_u32_vint_no_advance(data);
let (result, vlen) = read_u32_vint_no_advance(*data);
*data = &data[vlen..];
result
}

2
examples/aggregation.rs
View File

@@ -118,7 +118,7 @@ fn main() -> tantivy::Result<()> {
.into_iter()
.collect();
let collector = AggregationCollector::from_aggs(agg_req_1, None, index.schema());
let collector = AggregationCollector::from_aggs(agg_req_1, None);
let searcher = reader.searcher();
let agg_res: AggregationResults = searcher.search(&term_query, &collector).unwrap();

12
fastfield_codecs/Cargo.toml
View File

@@ -1,27 +1,23 @@
[package]
name = "fastfield_codecs"
version = "0.3.0"
version = "0.2.0"
authors = ["Pascal Seitz <pascal@quickwit.io>"]
license = "MIT"
edition = "2021"
description = "Fast field codecs used by tantivy"
documentation = "https://docs.rs/fastfield_codecs/"
homepage = "https://github.com/quickwit-oss/tantivy"
repository = "https://github.com/quickwit-oss/tantivy"
# See more keys and their definitions at https://doc.rust-lang.org/cargo/reference/manifest.html
[dependencies]
common = { version = "0.5", path = "../common/", package = "tantivy-common" }
tantivy-bitpacker = { version= "0.3", path = "../bitpacker/" }
ownedbytes = { version = "0.5", path = "../ownedbytes" }
common = { version = "0.3", path = "../common/", package = "tantivy-common" }
tantivy-bitpacker = { version="0.2", path = "../bitpacker/" }
ownedbytes = { version = "0.3.0", path = "../ownedbytes" }
prettytable-rs = {version="0.9.0", optional= true}
rand = {version="0.8.3", optional= true}
fastdivide = "0.4"
log = "0.4"
itertools = { version = "0.10.3" }
measure_time = { version="0.8.2", optional=true}
ordered-float = "3.4"
[dev-dependencies]
more-asserts = "0.3.0"

10
fastfield_codecs/benches/bench.rs
View File

@@ -113,7 +113,7 @@ mod tests {
b.iter(|| {
let mut positions = Vec::new();
column.get_docids_for_value_range(
column.get_positions_for_value_range(
major_item..=major_item,
0..data.len() as u32,
&mut positions,
@@ -129,7 +129,7 @@ mod tests {
b.iter(|| {
let mut positions = Vec::new();
column.get_docids_for_value_range(
column.get_positions_for_value_range(
minor_item..=minor_item,
0..data.len() as u32,
&mut positions,
@@ -145,7 +145,11 @@ mod tests {
b.iter(|| {
let mut positions = Vec::new();
column.get_docids_for_value_range(0..=u128::MAX, 0..data.len() as u32, &mut positions);
column.get_positions_for_value_range(
0..=u128::MAX,
0..data.len() as u32,
&mut positions,
);
positions
});
}

2
fastfield_codecs/src/blockwise_linear.rs
View File

@@ -75,7 +75,7 @@ impl FastFieldCodec for BlockwiseLinearCodec {
if column.num_vals() < 10 * CHUNK_SIZE as u32 {
return None;
}
let mut first_chunk: Vec<u64> = column.iter().take(CHUNK_SIZE).collect();
let mut first_chunk: Vec<u64> = column.iter().take(CHUNK_SIZE as usize).collect();
let line = Line::train(&VecColumn::from(&first_chunk));
for (i, buffer_val) in first_chunk.iter_mut().enumerate() {
let interpolated_val = line.eval(i as u32);

7
fastfield_codecs/src/column.rs
View File

@@ -35,7 +35,7 @@ pub trait Column<T: PartialOrd = u64>: Send + Sync {
///
/// Note that position == docid for single value fast fields
#[inline]
fn get_docids_for_value_range(
fn get_positions_for_value_range(
&self,
value_range: RangeInclusive<T>,
doc_id_range: Range<u32>,
@@ -222,13 +222,13 @@ where
)
}
fn get_docids_for_value_range(
fn get_positions_for_value_range(
&self,
range: RangeInclusive<Output>,
doc_id_range: Range<u32>,
positions: &mut Vec<u32>,
) {
self.from_column.get_docids_for_value_range(
self.from_column.get_positions_for_value_range(
self.monotonic_mapping.inverse(range.start().clone())
..=self.monotonic_mapping.inverse(range.end().clone()),
doc_id_range,
@@ -240,7 +240,6 @@ where
// and we do not have any specialized implementation anyway.
}
/// Wraps an iterator into a `Column`.
pub struct IterColumn<T>(T);
impl<T> From<T> for IterColumn<T>

2
fastfield_codecs/src/compact_space/build_compact_space.rs
View File

@@ -208,7 +208,7 @@ impl CompactSpaceBuilder {
};
let covered_range_len = range_mapping.range_length();
ranges_mapping.push(range_mapping);
compact_start += covered_range_len;
compact_start += covered_range_len as u64;
}
// println!("num ranges {}", ranges_mapping.len());
CompactSpace { ranges_mapping }

63
fastfield_codecs/src/compact_space/mod.rs
View File

@@ -97,7 +97,7 @@ impl BinarySerializable for CompactSpace {
};
let range_length = range_mapping.range_length();
ranges_mapping.push(range_mapping);
compact_start += range_length;
compact_start += range_length as u64;
}
Ok(Self { ranges_mapping })
@@ -306,13 +306,13 @@ impl Column<u128> for CompactSpaceDecompressor {
}
#[inline]
fn get_docids_for_value_range(
fn get_positions_for_value_range(
&self,
value_range: RangeInclusive<u128>,
positions_range: Range<u32>,
doc_id_range: Range<u32>,
positions: &mut Vec<u32>,
) {
self.get_positions_for_value_range(value_range, positions_range, positions)
self.get_positions_for_value_range(value_range, doc_id_range, positions)
}
}
@@ -351,13 +351,13 @@ impl CompactSpaceDecompressor {
pub fn get_positions_for_value_range(
&self,
value_range: RangeInclusive<u128>,
position_range: Range<u32>,
doc_id_range: Range<u32>,
positions: &mut Vec<u32>,
) {
if value_range.start() > value_range.end() {
return;
}
let position_range = position_range.start..position_range.end.min(self.num_vals());
let doc_id_range = doc_id_range.start..doc_id_range.end.min(self.num_vals());
let from_value = *value_range.start();
let to_value = *value_range.end();
assert!(to_value >= from_value);
@@ -390,10 +390,10 @@ impl CompactSpaceDecompressor {
let range = compact_from..=compact_to;
let scan_num_docs = position_range.end - position_range.start;
let scan_num_docs = doc_id_range.end - doc_id_range.start;
let step_size = 4;
let cutoff = position_range.start + scan_num_docs - scan_num_docs % step_size;
let cutoff = doc_id_range.start + scan_num_docs - scan_num_docs % step_size;
let mut push_if_in_range = |idx, val| {
if range.contains(&val) {
@@ -402,15 +402,15 @@ impl CompactSpaceDecompressor {
};
let get_val = |idx| self.params.bit_unpacker.get(idx, &self.data);
// unrolled loop
for idx in (position_range.start..cutoff).step_by(step_size as usize) {
for idx in (doc_id_range.start..cutoff).step_by(step_size as usize) {
let idx1 = idx;
let idx2 = idx + 1;
let idx3 = idx + 2;
let idx4 = idx + 3;
let val1 = get_val(idx1);
let val2 = get_val(idx2);
let val3 = get_val(idx3);
let val4 = get_val(idx4);
let val1 = get_val(idx1 as u32);
let val2 = get_val(idx2 as u32);
let val3 = get_val(idx3 as u32);
let val4 = get_val(idx4 as u32);
push_if_in_range(idx1, val1);
push_if_in_range(idx2, val2);
push_if_in_range(idx3, val3);
@@ -418,14 +418,15 @@ impl CompactSpaceDecompressor {
}
// handle rest
for idx in cutoff..position_range.end {
push_if_in_range(idx, get_val(idx));
for idx in cutoff..doc_id_range.end {
push_if_in_range(idx, get_val(idx as u32));
}
}
#[inline]
fn iter_compact(&self) -> impl Iterator<Item = u64> + '_ {
(0..self.params.num_vals).map(move |idx| self.params.bit_unpacker.get(idx, &self.data))
(0..self.params.num_vals)
.map(move |idx| self.params.bit_unpacker.get(idx, &self.data) as u64)
}
#[inline]
@@ -455,9 +456,6 @@ impl CompactSpaceDecompressor {
mod tests {
use super::*;
use crate::format_version::read_format_version;
use crate::null_index_footer::read_null_index_footer;
use crate::serialize::U128Header;
use crate::{open_u128, serialize_u128};
#[test]
@@ -503,8 +501,7 @@ mod tests {
assert_eq!(amplitude, 2);
}
fn test_all(mut data: OwnedBytes, expected: &[u128]) {
let _header = U128Header::deserialize(&mut data);
fn test_all(data: OwnedBytes, expected: &[u128]) {
let decompressor = CompactSpaceDecompressor::open(data).unwrap();
for (idx, expected_val) in expected.iter().cloned().enumerate() {
let val = decompressor.get(idx as u32);
@@ -542,10 +539,7 @@ mod tests {
.unwrap();
let data = OwnedBytes::new(out);
let (data, _format_version) = read_format_version(data).unwrap();
let (data, _null_index_footer) = read_null_index_footer(data).unwrap();
test_all(data.clone(), u128_vals);
data
}
@@ -562,13 +556,11 @@ mod tests {
4_000_211_222u128,
333u128,
];
let mut data = test_aux_vals(vals);
let _header = U128Header::deserialize(&mut data);
let data = test_aux_vals(vals);
let decomp = CompactSpaceDecompressor::open(data).unwrap();
let complete_range = 0..vals.len() as u32;
for (pos, val) in vals.iter().enumerate() {
let val = *val;
let val = *val as u128;
let pos = pos as u32;
let mut positions = Vec::new();
decomp.get_positions_for_value_range(val..=val, pos..pos + 1, &mut positions);
@@ -665,7 +657,7 @@ mod tests {
get_positions_for_value_range_helper(
&decomp,
4_000_211_221u128..=5_000_000_000u128,
complete_range
complete_range.clone()
),
vec![6, 7]
);
@@ -689,8 +681,7 @@ mod tests {
4_000_211_222u128,
333u128,
];
let mut data = test_aux_vals(vals);
let _header = U128Header::deserialize(&mut data);
let data = test_aux_vals(vals);
let decomp = CompactSpaceDecompressor::open(data).unwrap();
let complete_range = 0..vals.len() as u32;
assert_eq!(
@@ -702,7 +693,7 @@ mod tests {
vec![0]
);
assert_eq!(
get_positions_for_value_range_helper(&decomp, 0..=105, complete_range),
get_positions_for_value_range_helper(&decomp, 0..=105, complete_range.clone()),
vec![0]
);
}
@@ -713,7 +704,7 @@ mod tests {
doc_id_range: Range<u32>,
) -> Vec<u32> {
let mut positions = Vec::new();
column.get_docids_for_value_range(value_range, doc_id_range, &mut positions);
column.get_positions_for_value_range(value_range, doc_id_range, &mut positions);
positions
}
@@ -755,7 +746,11 @@ mod tests {
);
assert_eq!(
get_positions_for_value_range_helper(&*decomp, 1_000_000..=1_000_000, complete_range),
get_positions_for_value_range_helper(
&*decomp,
1_000_000..=1_000_000,
complete_range.clone()
),
vec![11]
);
}

39
fastfield_codecs/src/format_version.rs
View File

@@ -1,39 +0,0 @@
use std::io;
use common::BinarySerializable;
use ownedbytes::OwnedBytes;
const MAGIC_NUMBER: u16 = 4335u16;
const FASTFIELD_FORMAT_VERSION: u8 = 1;
pub(crate) fn append_format_version(output: &mut impl io::Write) -> io::Result<()> {
FASTFIELD_FORMAT_VERSION.serialize(output)?;
MAGIC_NUMBER.serialize(output)?;
Ok(())
}
pub(crate) fn read_format_version(data: OwnedBytes) -> io::Result<(OwnedBytes, u8)> {
let (data, magic_number_bytes) = data.rsplit(2);
let magic_number = u16::deserialize(&mut magic_number_bytes.as_slice())?;
if magic_number != MAGIC_NUMBER {
return Err(io::Error::new(
io::ErrorKind::InvalidData,
format!("magic number mismatch {} != {}", magic_number, MAGIC_NUMBER),
));
}
let (data, format_version_bytes) = data.rsplit(1);
let format_version = u8::deserialize(&mut format_version_bytes.as_slice())?;
if format_version > FASTFIELD_FORMAT_VERSION {
return Err(io::Error::new(
io::ErrorKind::InvalidData,
format!(
"Unsupported fastfield format version: {}. Max supported version: {}",
format_version, FASTFIELD_FORMAT_VERSION
),
));
}
Ok((data, format_version))
}

61
fastfield_codecs/src/lib.rs
View File

@@ -20,36 +20,28 @@ use std::sync::Arc;
use common::BinarySerializable;
use compact_space::CompactSpaceDecompressor;
use format_version::read_format_version;
use monotonic_mapping::{
StrictlyMonotonicMappingInverter, StrictlyMonotonicMappingToInternal,
StrictlyMonotonicMappingToInternalBaseval, StrictlyMonotonicMappingToInternalGCDBaseval,
};
use null_index_footer::read_null_index_footer;
use ownedbytes::OwnedBytes;
use serialize::{Header, U128Header};
use serialize::Header;
mod bitpacked;
mod blockwise_linear;
mod compact_space;
mod format_version;
mod line;
mod linear;
mod monotonic_mapping;
mod monotonic_mapping_u128;
#[allow(dead_code)]
mod null_index;
mod null_index_footer;
mod column;
mod gcd;
pub mod serialize;
pub use ordered_float;
mod serialize;
use self::bitpacked::BitpackedCodec;
use self::blockwise_linear::BlockwiseLinearCodec;
pub use self::column::{monotonic_map_column, Column, IterColumn, VecColumn};
pub use self::column::{monotonic_map_column, Column, VecColumn};
use self::linear::LinearCodec;
pub use self::monotonic_mapping::{MonotonicallyMappableToU64, StrictlyMonotonicFn};
pub use self::monotonic_mapping_u128::MonotonicallyMappableToU128;
@@ -100,49 +92,10 @@ impl FastFieldCodecType {
}
}
#[derive(PartialEq, Eq, PartialOrd, Ord, Debug, Clone, Copy)]
#[repr(u8)]
/// Available codecs to use to encode the u128 (via [`MonotonicallyMappableToU128`]) converted data.
pub enum U128FastFieldCodecType {
/// This codec takes a large number space (u128) and reduces it to a compact number space, by
/// removing the holes.
CompactSpace = 1,
}
impl BinarySerializable for U128FastFieldCodecType {
fn serialize<W: Write>(&self, wrt: &mut W) -> io::Result<()> {
self.to_code().serialize(wrt)
}
fn deserialize<R: io::Read>(reader: &mut R) -> io::Result<Self> {
let code = u8::deserialize(reader)?;
let codec_type: Self = Self::from_code(code)
.ok_or_else(|| io::Error::new(io::ErrorKind::InvalidData, "Unknown code `{code}.`"))?;
Ok(codec_type)
}
}
impl U128FastFieldCodecType {
pub(crate) fn to_code(self) -> u8 {
self as u8
}
pub(crate) fn from_code(code: u8) -> Option<Self> {
match code {
1 => Some(Self::CompactSpace),
_ => None,
}
}
}
/// Returns the correct codec reader wrapped in the `Arc` for the data.
pub fn open_u128<Item: MonotonicallyMappableToU128>(
bytes: OwnedBytes,
) -> io::Result<Arc<dyn Column<Item>>> {
let (bytes, _format_version) = read_format_version(bytes)?;
let (mut bytes, _null_index_footer) = read_null_index_footer(bytes)?;
let header = U128Header::deserialize(&mut bytes)?;
assert_eq!(header.codec_type, U128FastFieldCodecType::CompactSpace);
let reader = CompactSpaceDecompressor::open(bytes)?;
let inverted: StrictlyMonotonicMappingInverter<StrictlyMonotonicMappingToInternal<Item>> =
StrictlyMonotonicMappingToInternal::<Item>::new().into();
@@ -150,9 +103,9 @@ pub fn open_u128<Item: MonotonicallyMappableToU128>(
}
/// Returns the correct codec reader wrapped in the `Arc` for the data.
pub fn open<T: MonotonicallyMappableToU64>(bytes: OwnedBytes) -> io::Result<Arc<dyn Column<T>>> {
let (bytes, _format_version) = read_format_version(bytes)?;
let (mut bytes, _null_index_footer) = read_null_index_footer(bytes)?;
pub fn open<T: MonotonicallyMappableToU64>(
mut bytes: OwnedBytes,
) -> io::Result<Arc<dyn Column<T>>> {
let header = Header::deserialize(&mut bytes)?;
match header.codec_type {
FastFieldCodecType::Bitpacked => open_specific_codec::<BitpackedCodec, _>(bytes, &header),
@@ -265,7 +218,7 @@ mod tests {
.map(|(pos, _)| pos as u32)
.collect();
let mut positions = Vec::new();
reader.get_docids_for_value_range(
reader.get_positions_for_value_range(
data[test_rand_idx]..=data[test_rand_idx],
0..data.len() as u32,
&mut positions,

2
fastfield_codecs/src/main.rs
View File

@@ -119,7 +119,7 @@ fn bench_ip() {
for value in dataset.iter().take(1110).skip(1100).cloned() {
doc_values.clear();
print_time!("get range");
decompressor.get_docids_for_value_range(
decompressor.get_positions_for_value_range(
value..=value,
0..decompressor.num_vals(),
&mut doc_values,

34
fastfield_codecs/src/monotonic_mapping.rs
View File

@@ -1,7 +1,6 @@
use std::marker::PhantomData;
use fastdivide::DividerU64;
use ordered_float::NotNan;
use crate::MonotonicallyMappableToU128;
@@ -193,8 +192,6 @@ impl MonotonicallyMappableToU64 for bool {
}
}
// TODO remove me.
// Tantivy should refuse NaN values and work with NotNaN internally.
impl MonotonicallyMappableToU64 for f64 {
fn to_u64(self) -> u64 {
common::f64_to_u64(self)
@@ -205,42 +202,11 @@ impl MonotonicallyMappableToU64 for f64 {
}
}
impl MonotonicallyMappableToU64 for ordered_float::NotNan<f64> {
fn to_u64(self) -> u64 {
common::f64_to_u64(self.into_inner())
}
fn from_u64(val: u64) -> Self {
NotNan::new(common::u64_to_f64(val)).expect("Invalid NotNaN f64 value.")
}
}
#[cfg(test)]
mod tests {
use super::*;
#[test]
fn test_from_u64_pos_inf() {
let inf_as_u64 = common::f64_to_u64(f64::INFINITY);
let inf_back_to_f64 = NotNan::from_u64(inf_as_u64);
assert_eq!(inf_back_to_f64, NotNan::new(f64::INFINITY).unwrap());
}
#[test]
fn test_from_u64_neg_inf() {
let inf_as_u64 = common::f64_to_u64(-f64::INFINITY);
let inf_back_to_f64 = NotNan::from_u64(inf_as_u64);
assert_eq!(inf_back_to_f64, NotNan::new(-f64::INFINITY).unwrap());
}
#[test]
#[should_panic(expected = "Invalid NotNaN")]
fn test_from_u64_nan_panics() {
let nan_as_u64 = common::f64_to_u64(f64::NAN);
NotNan::from_u64(nan_as_u64);
}
#[test]
fn strictly_monotonic_test() {
// identity mapping

454
fastfield_codecs/src/null_index/dense.rs
View File

@@ -1,454 +0,0 @@
use std::convert::TryInto;
use std::io::{self, Write};
use common::BinarySerializable;
use itertools::Itertools;
use ownedbytes::OwnedBytes;
use super::{get_bit_at, set_bit_at};
/// For the `DenseCodec`, `data` which contains the encoded blocks.
/// Each block consists of [u8; 12]. The first 8 bytes is a bitvec for 64 elements.
/// The last 4 bytes are the offset, the number of set bits so far.
///
/// When translating the original index to a dense index, the correct block can be computed
/// directly `orig_idx/64`. Inside the block the position is `orig_idx%64`.
///
/// When translating a dense index to the original index, we can use the offset to find the correct
/// block. Direct computation is not possible, but we can employ a linear or binary search.
#[derive(Clone)]
pub struct DenseCodec {
// data consists of blocks of 64 bits.
//
// The format is &[(u64, u32)]
// u64 is the bitvec
// u32 is the offset of the block, the number of set bits so far.
//
// At the end one block is appended, to store the number of values in the index in offset.
data: OwnedBytes,
}
const ELEMENTS_PER_BLOCK: u32 = 64;
const BLOCK_BITVEC_SIZE: usize = 8;
const BLOCK_OFFSET_SIZE: usize = 4;
const SERIALIZED_BLOCK_SIZE: usize = BLOCK_BITVEC_SIZE + BLOCK_OFFSET_SIZE;
#[inline]
fn count_ones(bitvec: u64, pos_in_bitvec: u32) -> u32 {
if pos_in_bitvec == 63 {
bitvec.count_ones()
} else {
let mask = (1u64 << (pos_in_bitvec + 1)) - 1;
let masked_bitvec = bitvec & mask;
masked_bitvec.count_ones()
}
}
#[derive(Clone, Copy)]
struct DenseIndexBlock {
bitvec: u64,
offset: u32,
}
impl From<[u8; SERIALIZED_BLOCK_SIZE]> for DenseIndexBlock {
fn from(data: [u8; SERIALIZED_BLOCK_SIZE]) -> Self {
let bitvec = u64::from_le_bytes(data[..BLOCK_BITVEC_SIZE].try_into().unwrap());
let offset = u32::from_le_bytes(data[BLOCK_BITVEC_SIZE..].try_into().unwrap());
Self { bitvec, offset }
}
}
impl DenseCodec {
/// Open the DenseCodec from OwnedBytes
pub fn open(data: OwnedBytes) -> Self {
Self { data }
}
#[inline]
/// Check if value at position is not null.
pub fn exists(&self, idx: u32) -> bool {
let block_pos = idx / ELEMENTS_PER_BLOCK;
let bitvec = self.dense_index_block(block_pos).bitvec;
let pos_in_bitvec = idx % ELEMENTS_PER_BLOCK;
get_bit_at(bitvec, pos_in_bitvec)
}
#[inline]
fn dense_index_block(&self, block_pos: u32) -> DenseIndexBlock {
dense_index_block(&self.data, block_pos)
}
/// Return the number of non-null values in an index
pub fn num_non_nulls(&self) -> u32 {
let last_block = (self.data.len() / SERIALIZED_BLOCK_SIZE) - 1;
self.dense_index_block(last_block as u32).offset
}
#[inline]
/// Translate from the original index to the codec index.
pub fn translate_to_codec_idx(&self, idx: u32) -> Option<u32> {
let block_pos = idx / ELEMENTS_PER_BLOCK;
let index_block = self.dense_index_block(block_pos);
let pos_in_block_bit_vec = idx % ELEMENTS_PER_BLOCK;
let ones_in_block = count_ones(index_block.bitvec, pos_in_block_bit_vec);
if get_bit_at(index_block.bitvec, pos_in_block_bit_vec) {
// -1 is ok, since idx does exist, so there's at least one
Some(index_block.offset + ones_in_block - 1)
} else {
None
}
}
/// Translate positions from the codec index to the original index.
///
/// # Panics
///
/// May panic if any `idx` is greater than the max codec index.
pub fn translate_codec_idx_to_original_idx<'a>(
&'a self,
iter: impl Iterator<Item = u32> + 'a,
) -> impl Iterator<Item = u32> + 'a {
let mut block_pos = 0u32;
iter.map(move |dense_idx| {
// update block_pos to limit search scope
block_pos = find_block(dense_idx, block_pos, &self.data);
let index_block = self.dense_index_block(block_pos);
// The next offset is higher than dense_idx and therefore:
// dense_idx <= offset + num_set_bits in block
let mut num_set_bits = 0;
for idx_in_bitvec in 0..ELEMENTS_PER_BLOCK {
if get_bit_at(index_block.bitvec, idx_in_bitvec) {
num_set_bits += 1;
}
if num_set_bits == (dense_idx - index_block.offset + 1) {
let orig_idx = block_pos * ELEMENTS_PER_BLOCK + idx_in_bitvec;
return orig_idx;
}
}
panic!("Internal Error: Offset calculation in dense idx seems to be wrong.");
})
}
}
#[inline]
fn dense_index_block(data: &[u8], block_pos: u32) -> DenseIndexBlock {
let data_start_pos = block_pos as usize * SERIALIZED_BLOCK_SIZE;
let block_data: [u8; SERIALIZED_BLOCK_SIZE] = data[data_start_pos..][..SERIALIZED_BLOCK_SIZE]
.try_into()
.unwrap();
block_data.into()
}
#[inline]
/// Finds the block position containing the dense_idx.
///
/// # Correctness
/// dense_idx needs to be smaller than the number of values in the index
///
/// The last offset number is equal to the number of values in the index.
fn find_block(dense_idx: u32, mut block_pos: u32, data: &[u8]) -> u32 {
loop {
let offset = dense_index_block(data, block_pos).offset;
if offset > dense_idx {
return block_pos - 1;
}
block_pos += 1;
}
}
/// Iterator over all values, true if set, otherwise false
pub fn serialize_dense_codec(
iter: impl Iterator<Item = bool>,
mut out: impl Write,
) -> io::Result<()> {
let mut offset: u32 = 0;
for chunk in &iter.chunks(ELEMENTS_PER_BLOCK as usize) {
let mut block: u64 = 0;
for (pos, is_bit_set) in chunk.enumerate() {
if is_bit_set {
set_bit_at(&mut block, pos as u64);
}
}
block.serialize(&mut out)?;
offset.serialize(&mut out)?;
offset += block.count_ones();
}
// Add sentinal block for the offset
let block: u64 = 0;
block.serialize(&mut out)?;
offset.serialize(&mut out)?;
Ok(())
}
#[cfg(test)]
mod tests {
use proptest::prelude::{any, prop, *};
use proptest::strategy::Strategy;
use proptest::{prop_oneof, proptest};
use super::*;
fn random_bitvec() -> BoxedStrategy<Vec<bool>> {
prop_oneof![
1 => prop::collection::vec(proptest::bool::weighted(1.0), 0..100),
1 => prop::collection::vec(proptest::bool::weighted(1.0), 0..64),
1 => prop::collection::vec(proptest::bool::weighted(0.0), 0..100),
1 => prop::collection::vec(proptest::bool::weighted(0.0), 0..64),
8 => vec![any::<bool>()],
2 => prop::collection::vec(any::<bool>(), 0..50),
]
.boxed()
}
proptest! {
#![proptest_config(ProptestConfig::with_cases(500))]
#[test]
fn test_with_random_bitvecs(bitvec1 in random_bitvec(), bitvec2 in random_bitvec(), bitvec3 in random_bitvec()) {
let mut bitvec = Vec::new();
bitvec.extend_from_slice(&bitvec1);
bitvec.extend_from_slice(&bitvec2);
bitvec.extend_from_slice(&bitvec3);
test_null_index(bitvec);
}
}
#[test]
fn dense_codec_test_one_block_false() {
let mut iter = vec![false; 64];
iter.push(true);
test_null_index(iter);
}
fn test_null_index(data: Vec<bool>) {
let mut out = vec![];
serialize_dense_codec(data.iter().cloned(), &mut out).unwrap();
let null_index = DenseCodec::open(OwnedBytes::new(out));
let orig_idx_with_value: Vec<u32> = data
.iter()
.enumerate()
.filter(|(_pos, val)| **val)
.map(|(pos, _val)| pos as u32)
.collect();
assert_eq!(
null_index
.translate_codec_idx_to_original_idx(0..orig_idx_with_value.len() as u32)
.collect_vec(),
orig_idx_with_value
);
for (dense_idx, orig_idx) in orig_idx_with_value.iter().enumerate() {
assert_eq!(
null_index.translate_to_codec_idx(*orig_idx),
Some(dense_idx as u32)
);
}
for (pos, value) in data.iter().enumerate() {
assert_eq!(null_index.exists(pos as u32), *value);
}
}
#[test]
fn dense_codec_test_translation() {
let mut out = vec![];
let iter = ([true, false, true, false]).iter().cloned();
serialize_dense_codec(iter, &mut out).unwrap();
let null_index = DenseCodec::open(OwnedBytes::new(out));
assert_eq!(
null_index
.translate_codec_idx_to_original_idx(0..2)
.collect_vec(),
vec![0, 2]
);
}
#[test]
fn dense_codec_translate() {
let mut out = vec![];
let iter = ([true, false, true, false]).iter().cloned();
serialize_dense_codec(iter, &mut out).unwrap();
let null_index = DenseCodec::open(OwnedBytes::new(out));
assert_eq!(null_index.translate_to_codec_idx(0), Some(0));
assert_eq!(null_index.translate_to_codec_idx(2), Some(1));
}
#[test]
fn dense_codec_test_small() {
let mut out = vec![];
let iter = ([true, false, true, false]).iter().cloned();
serialize_dense_codec(iter, &mut out).unwrap();
let null_index = DenseCodec::open(OwnedBytes::new(out));
assert!(null_index.exists(0));
assert!(!null_index.exists(1));
assert!(null_index.exists(2));
assert!(!null_index.exists(3));
}
#[test]
fn dense_codec_test_large() {
let mut docs = vec![];
docs.extend((0..1000).map(|_idx| false));
docs.extend((0..=1000).map(|_idx| true));
let iter = docs.iter().cloned();
let mut out = vec![];
serialize_dense_codec(iter, &mut out).unwrap();
let null_index = DenseCodec::open(OwnedBytes::new(out));
assert!(!null_index.exists(0));
assert!(!null_index.exists(100));
assert!(!null_index.exists(999));
assert!(null_index.exists(1000));
assert!(null_index.exists(1999));
assert!(null_index.exists(2000));
assert!(!null_index.exists(2001));
}
#[test]
fn test_count_ones() {
let mut block = 0;
set_bit_at(&mut block, 0);
set_bit_at(&mut block, 2);
assert_eq!(count_ones(block, 0), 1);
assert_eq!(count_ones(block, 1), 1);
assert_eq!(count_ones(block, 2), 2);
}
}
#[cfg(all(test, feature = "unstable"))]
mod bench {
use rand::rngs::StdRng;
use rand::{Rng, SeedableRng};
use test::Bencher;
use super::*;
const TOTAL_NUM_VALUES: u32 = 1_000_000;
fn gen_bools(fill_ratio: f64) -> DenseCodec {
let mut out = Vec::new();
let mut rng: StdRng = StdRng::from_seed([1u8; 32]);
let bools: Vec<_> = (0..TOTAL_NUM_VALUES)
.map(|_| rng.gen_bool(fill_ratio))
.collect();
serialize_dense_codec(bools.into_iter(), &mut out).unwrap();
let codec = DenseCodec::open(OwnedBytes::new(out));
codec
}
fn random_range_iterator(start: u32, end: u32, step_size: u32) -> impl Iterator<Item = u32> {
let mut rng: StdRng = StdRng::from_seed([1u8; 32]);
let mut current = start;
std::iter::from_fn(move || {
current += rng.gen_range(1..step_size + 1);
if current >= end {
None
} else {
Some(current)
}
})
}
fn walk_over_data(codec: &DenseCodec, max_step_size: u32) -> Option<u32> {
walk_over_data_from_positions(
codec,
random_range_iterator(0, TOTAL_NUM_VALUES, max_step_size),
)
}
fn walk_over_data_from_positions(
codec: &DenseCodec,
positions: impl Iterator<Item = u32>,
) -> Option<u32> {
let mut dense_idx: Option<u32> = None;
for idx in positions {
dense_idx = dense_idx.or(codec.translate_to_codec_idx(idx));
}
dense_idx
}
#[bench]
fn bench_dense_codec_translate_orig_to_codec_90percent_filled_random_stride(
bench: &mut Bencher,
) {
let codec = gen_bools(0.9f64);
bench.iter(|| walk_over_data(&codec, 100));
}
#[bench]
fn bench_dense_codec_translate_orig_to_codec_50percent_filled_random_stride(
bench: &mut Bencher,
) {
let codec = gen_bools(0.5f64);
bench.iter(|| walk_over_data(&codec, 100));
}
#[bench]
fn bench_dense_codec_translate_orig_to_codec_full_scan_10percent(bench: &mut Bencher) {
let codec = gen_bools(0.1f64);
bench.iter(|| walk_over_data_from_positions(&codec, 0..TOTAL_NUM_VALUES));
}
#[bench]
fn bench_dense_codec_translate_orig_to_codec_full_scan_90percent(bench: &mut Bencher) {
let codec = gen_bools(0.9f64);
bench.iter(|| walk_over_data_from_positions(&codec, 0..TOTAL_NUM_VALUES));
}
#[bench]
fn bench_dense_codec_translate_orig_to_codec_10percent_filled_random_stride(
bench: &mut Bencher,
) {
let codec = gen_bools(0.1f64);
bench.iter(|| walk_over_data(&codec, 100));
}
#[bench]
fn bench_dense_codec_translate_codec_to_orig_90percent_filled_random_stride_big_step(
bench: &mut Bencher,
) {
let codec = gen_bools(0.9f64);
let num_vals = codec.num_non_nulls();
bench.iter(|| {
codec
.translate_codec_idx_to_original_idx(random_range_iterator(0, num_vals, 50_000))
.last()
});
}
#[bench]
fn bench_dense_codec_translate_codec_to_orig_90percent_filled_random_stride(
bench: &mut Bencher,
) {
let codec = gen_bools(0.9f64);
let num_vals = codec.num_non_nulls();
bench.iter(|| {
codec
.translate_codec_idx_to_original_idx(random_range_iterator(0, num_vals, 100))
.last()
});
}
#[bench]
fn bench_dense_codec_translate_codec_to_orig_90percent_filled_full_scan(bench: &mut Bencher) {
let codec = gen_bools(0.9f64);
let num_vals = codec.num_non_nulls();
bench.iter(|| {
codec
.translate_codec_idx_to_original_idx(0..num_vals)
.last()
});
}
}

14
fastfield_codecs/src/null_index/mod.rs
View File

@@ -1,14 +0,0 @@
pub use dense::{serialize_dense_codec, DenseCodec};
mod dense;
mod sparse;
#[inline]
fn get_bit_at(input: u64, n: u32) -> bool {
input & (1 << n) != 0
}
#[inline]
fn set_bit_at(input: &mut u64, n: u64) {
*input |= 1 << n;
}

752
fastfield_codecs/src/null_index/sparse.rs
View File

@@ -1,752 +0,0 @@
use std::io::{self, Write};
use common::BitSet;
use ownedbytes::OwnedBytes;
use super::{serialize_dense_codec, DenseCodec};
/// `SparseCodec` is the codec for data, when only few documents have values.
/// In contrast to `DenseCodec` opening a `SparseCodec` causes runtime data to be produced, for
/// faster access.
///
/// The lower 16 bits of doc ids are stored as u16 while the upper 16 bits are given by the block
/// id. Each block contains 1<<16 docids.
///
/// # Serialized Data Layout
/// The data starts with the block data. Each block is either dense or sparse encoded, depending on
/// the number of values in the block. A block is sparse when it contains less than
/// DENSE_BLOCK_THRESHOLD (6144) values.
/// [Sparse data block | dense data block, .. #repeat*; Desc: Either a sparse or dense encoded
/// block]
/// ### Sparse block data
/// [u16 LE, .. #repeat*; Desc: Positions with values in a block]
/// ### Dense block data
/// [Dense codec for the whole block; Desc: Similar to a bitvec(0..ELEMENTS_PER_BLOCK) + Metadata
/// for faster lookups. See dense.rs]
///
/// The data is followed by block metadata, to know which area of the raw block data belongs to
/// which block. Only metadata for blocks with elements is recorded to
/// keep the overhead low for scenarios with many very sparse columns. The block metadata consists
/// of the block index and the number of values in the block. Since we don't store empty blocks
/// num_vals is incremented by 1, e.g. 0 means 1 value.
///
/// The last u16 is storing the number of metadata blocks.
/// [u16 LE, .. #repeat*; Desc: Positions with values in a block][(u16 LE, u16 LE), .. #repeat*;
/// Desc: (Block Id u16, Num Elements u16)][u16 LE; Desc: num blocks with values u16]
///
/// # Opening
/// When opening the data layout, the data is expanded to `Vec<SparseCodecBlockVariant>`, where the
/// index is the block index. For each block `byte_start` and `offset` is computed.
pub struct SparseCodec {
data: OwnedBytes,
blocks: Vec<SparseCodecBlockVariant>,
}
/// The threshold for for number of elements after which we switch to dense block encoding
const DENSE_BLOCK_THRESHOLD: u32 = 6144;
const ELEMENTS_PER_BLOCK: u32 = u16::MAX as u32 + 1;
/// 1.5 bit per Element + 12 bytes for the sentinal block
const NUM_BYTES_DENSE_BLOCK: u32 = (ELEMENTS_PER_BLOCK + ELEMENTS_PER_BLOCK / 2 + 64 + 32) / 8;
#[derive(Clone)]
enum SparseCodecBlockVariant {
Empty { offset: u32 },
Dense(DenseBlock),
Sparse(SparseBlock),
}
impl SparseCodecBlockVariant {
/// The number of non-null values that preceeded that block.
fn offset(&self) -> u32 {
match self {
SparseCodecBlockVariant::Empty { offset } => *offset,
SparseCodecBlockVariant::Dense(dense) => dense.offset,
SparseCodecBlockVariant::Sparse(sparse) => sparse.offset,
}
}
}
/// A block consists of max u16 values
#[derive(Clone)]
struct DenseBlock {
/// The number of values set before the block
offset: u32,
/// The data for the dense encoding
codec: DenseCodec,
}
impl DenseBlock {
pub fn exists(&self, idx: u32) -> bool {
self.codec.exists(idx)
}
pub fn translate_to_codec_idx(&self, idx: u32) -> Option<u32> {
self.codec.translate_to_codec_idx(idx)
}
pub fn translate_codec_idx_to_original_idx(&self, idx: u32) -> u32 {
self.codec
.translate_codec_idx_to_original_idx(idx..=idx)
.next()
.unwrap()
}
}
/// A block consists of max u16 values
#[derive(Debug, Copy, Clone)]
struct SparseBlock {
/// The number of values in the block
num_vals: u32,
/// The number of values set before the block
offset: u32,
/// The start position of the data for the block
byte_start: u32,
}
impl SparseBlock {
fn empty_block(offset: u32) -> Self {
Self {
num_vals: 0,
byte_start: 0,
offset,
}
}
#[inline]
fn value_at_idx(&self, data: &[u8], idx: u16) -> u16 {
let start_offset: usize = self.byte_start as usize + (idx as u32 as usize * 2);
get_u16(data, start_offset)
}
#[inline]
#[allow(clippy::comparison_chain)]
// Looks for the element in the block. Returns the positions if found.
fn binary_search(&self, data: &[u8], target: u16) -> Option<u16> {
let mut size = self.num_vals as u16;
let mut left = 0;
let mut right = size;
// TODO try different implem.
// e.g. exponential search into binary search
while left < right {
let mid = left + size / 2;
// TODO do boundary check only once, and then use an
// unsafe `value_at_idx`
let mid_val = self.value_at_idx(data, mid);
if target > mid_val {
left = mid + 1;
} else if target < mid_val {
right = mid;
} else {
return Some(mid);
}
size = right - left;
}
None
}
}
#[inline]
fn get_u16(data: &[u8], byte_position: usize) -> u16 {
let bytes: [u8; 2] = data[byte_position..byte_position + 2].try_into().unwrap();
u16::from_le_bytes(bytes)
}
const SERIALIZED_BLOCK_METADATA_SIZE: usize = 4;
fn deserialize_sparse_codec_block(data: &OwnedBytes) -> Vec<SparseCodecBlockVariant> {
// The number of vals so far
let mut offset = 0;
let mut sparse_codec_blocks = Vec::new();
let num_blocks = get_u16(data, data.len() - 2);
let block_data_index_start =
data.len() - 2 - num_blocks as usize * SERIALIZED_BLOCK_METADATA_SIZE;
let mut byte_start = 0;
for block_num in 0..num_blocks as usize {
let block_data_index = block_data_index_start + SERIALIZED_BLOCK_METADATA_SIZE * block_num;
let block_idx = get_u16(data, block_data_index);
let num_vals = get_u16(data, block_data_index + 2) as u32 + 1;
sparse_codec_blocks.resize(
block_idx as usize,
SparseCodecBlockVariant::Empty { offset },
);
if is_sparse(num_vals) {
let block = SparseBlock {
num_vals,
offset,
byte_start,
};
sparse_codec_blocks.push(SparseCodecBlockVariant::Sparse(block));
byte_start += 2 * num_vals;
} else {
let block = DenseBlock {
offset,
codec: DenseCodec::open(data.slice(byte_start as usize..data.len()).clone()),
};
sparse_codec_blocks.push(SparseCodecBlockVariant::Dense(block));
// Dense blocks have a fixed size spanning ELEMENTS_PER_BLOCK.
byte_start += NUM_BYTES_DENSE_BLOCK;
}
offset += num_vals;
}
sparse_codec_blocks.push(SparseCodecBlockVariant::Empty { offset });
sparse_codec_blocks
}
/// Splits a value address into lower and upper 16bits.
/// The lower 16 bits are the value in the block
/// The upper 16 bits are the block index
#[derive(Debug, Clone, Copy)]
struct ValueAddr {
block_idx: u16,
value_in_block: u16,
}
/// Splits a idx into block index and value in the block
fn value_addr(idx: u32) -> ValueAddr {
/// Static assert number elements per block this method expects
#[allow(clippy::assertions_on_constants)]
const _: () = assert!(ELEMENTS_PER_BLOCK == (1 << 16));
let value_in_block = idx as u16;
let block_idx = (idx >> 16) as u16;
ValueAddr {
block_idx,
value_in_block,
}
}
impl SparseCodec {
/// Open the SparseCodec from OwnedBytes
pub fn open(data: OwnedBytes) -> Self {
let blocks = deserialize_sparse_codec_block(&data);
Self { data, blocks }
}
#[inline]
/// Check if value at position is not null.
pub fn exists(&self, idx: u32) -> bool {
let value_addr = value_addr(idx);
// There may be trailing nulls without data, those are not stored as blocks. It would be
// possible to create empty blocks, but for that we would need to serialize the number of
// values or pass them when opening
if let Some(block) = self.blocks.get(value_addr.block_idx as usize) {
match block {
SparseCodecBlockVariant::Empty { offset: _ } => false,
SparseCodecBlockVariant::Dense(block) => {
block.exists(value_addr.value_in_block as u32)
}
SparseCodecBlockVariant::Sparse(block) => block
.binary_search(&self.data, value_addr.value_in_block)
.is_some(),
}
} else {
false
}
}
/// Return the number of non-null values in an index
pub fn num_non_nulls(&self) -> u32 {
self.blocks.last().map(|block| block.offset()).unwrap_or(0)
}
#[inline]
/// Translate from the original index to the codec index.
pub fn translate_to_codec_idx(&self, idx: u32) -> Option<u32> {
let value_addr = value_addr(idx);
let block = self.blocks.get(value_addr.block_idx as usize)?;
match block {
SparseCodecBlockVariant::Empty { offset: _ } => None,
SparseCodecBlockVariant::Dense(block) => block
.translate_to_codec_idx(value_addr.value_in_block as u32)
.map(|pos_in_block| pos_in_block + block.offset),
SparseCodecBlockVariant::Sparse(block) => {
let pos_in_block = block.binary_search(&self.data, value_addr.value_in_block);
pos_in_block.map(|pos_in_block: u16| block.offset + pos_in_block as u32)
}
}
}
fn find_block(&self, dense_idx: u32, mut block_pos: u32) -> u32 {
loop {
let offset = self.blocks[block_pos as usize].offset();
if offset > dense_idx {
return block_pos - 1;
}
block_pos += 1;
}
}
/// Translate positions from the codec index to the original index.
///
/// # Panics
///
/// May panic if any `idx` is greater than the max codec index.
pub fn translate_codec_idx_to_original_idx<'a>(
&'a self,
iter: impl Iterator<Item = u32> + 'a,
) -> impl Iterator<Item = u32> + 'a {
// TODO: There's a big potential performance gain, by using iterators per block instead of
// random access for each element in a block
// group_by itertools won't help though, since it requires a temporary local variable
let mut block_pos = 0u32;
iter.map(move |codec_idx| {
// update block_pos to limit search scope
block_pos = self.find_block(codec_idx, block_pos);
let block_doc_idx_start = block_pos * ELEMENTS_PER_BLOCK;
let block = &self.blocks[block_pos as usize];
let idx_in_block = codec_idx - block.offset();
match block {
SparseCodecBlockVariant::Empty { offset: _ } => {
panic!(
"invalid input, cannot translate to original index. associated empty \
block with dense idx. block_pos {}, idx_in_block {}",
block_pos, idx_in_block
)
}
SparseCodecBlockVariant::Dense(dense) => {
dense.translate_codec_idx_to_original_idx(idx_in_block) + block_doc_idx_start
}
SparseCodecBlockVariant::Sparse(block) => {
block.value_at_idx(&self.data, idx_in_block as u16) as u32 + block_doc_idx_start
}
}
})
}
}
fn is_sparse(num_elem_in_block: u32) -> bool {
num_elem_in_block < DENSE_BLOCK_THRESHOLD
}
#[derive(Default)]
struct BlockDataSerialized {
block_idx: u16,
num_vals: u32,
}
/// Iterator over positions of set values.
pub fn serialize_sparse_codec<W: Write>(
mut iter: impl Iterator<Item = u32>,
mut out: W,
) -> io::Result<()> {
let mut block_metadata: Vec<BlockDataSerialized> = Vec::new();
let mut current_block = Vec::new();
// This if-statement for the first element ensures that
// `block_metadata` is not empty in the loop below.
if let Some(idx) = iter.next() {
let value_addr = value_addr(idx);
block_metadata.push(BlockDataSerialized {
block_idx: value_addr.block_idx,
num_vals: 1,
});
current_block.push(value_addr.value_in_block);
}
let flush_block = |current_block: &mut Vec<u16>, out: &mut W| -> io::Result<()> {
let is_sparse = is_sparse(current_block.len() as u32);
if is_sparse {
for val_in_block in current_block.iter() {
out.write_all(val_in_block.to_le_bytes().as_ref())?;
}
} else {
let mut bitset = BitSet::with_max_value(ELEMENTS_PER_BLOCK + 1);
for val_in_block in current_block.iter() {
bitset.insert(*val_in_block as u32);
}
let iter = (0..ELEMENTS_PER_BLOCK).map(|idx| bitset.contains(idx));
serialize_dense_codec(iter, out)?;
}
current_block.clear();
Ok(())
};
for idx in iter {
let value_addr = value_addr(idx);
if block_metadata[block_metadata.len() - 1].block_idx == value_addr.block_idx {
let last_idx_metadata = block_metadata.len() - 1;
block_metadata[last_idx_metadata].num_vals += 1;
} else {
// flush prev block
flush_block(&mut current_block, &mut out)?;
block_metadata.push(BlockDataSerialized {
block_idx: value_addr.block_idx,
num_vals: 1,
});
}
current_block.push(value_addr.value_in_block);
}
// handle last block
flush_block(&mut current_block, &mut out)?;
for block in &block_metadata {
out.write_all(block.block_idx.to_le_bytes().as_ref())?;
// We don't store empty blocks, therefore we can subtract 1.
// This way we will be able to use u16 when the number of elements is 1 << 16 or u16::MAX+1
out.write_all(((block.num_vals - 1) as u16).to_le_bytes().as_ref())?;
}
out.write_all((block_metadata.len() as u16).to_le_bytes().as_ref())?;
Ok(())
}
#[cfg(test)]
mod tests {
use itertools::Itertools;
use proptest::prelude::{any, prop, *};
use proptest::strategy::Strategy;
use proptest::{prop_oneof, proptest};
use super::*;
fn random_bitvec() -> BoxedStrategy<Vec<bool>> {
prop_oneof![
1 => prop::collection::vec(proptest::bool::weighted(1.0), 0..100),
1 => prop::collection::vec(proptest::bool::weighted(0.00), 0..(ELEMENTS_PER_BLOCK as usize * 3)), // empty blocks
1 => prop::collection::vec(proptest::bool::weighted(1.00), 0..(ELEMENTS_PER_BLOCK as usize + 10)), // full block
1 => prop::collection::vec(proptest::bool::weighted(0.01), 0..100),
1 => prop::collection::vec(proptest::bool::weighted(0.01), 0..u16::MAX as usize),
8 => vec![any::<bool>()],
]
.boxed()
}
proptest! {
#![proptest_config(ProptestConfig::with_cases(50))]
#[test]
fn test_with_random_bitvecs(bitvec1 in random_bitvec(), bitvec2 in random_bitvec(), bitvec3 in random_bitvec()) {
let mut bitvec = Vec::new();
bitvec.extend_from_slice(&bitvec1);
bitvec.extend_from_slice(&bitvec2);
bitvec.extend_from_slice(&bitvec3);
test_null_index(bitvec);
}
}
#[test]
fn sparse_codec_test_one_block_false() {
let mut iter = vec![false; ELEMENTS_PER_BLOCK as usize];
iter.push(true);
test_null_index(iter);
}
#[test]
fn sparse_codec_test_one_block_true() {
let mut iter = vec![true; ELEMENTS_PER_BLOCK as usize];
iter.push(true);
test_null_index(iter);
}
fn test_null_index(data: Vec<bool>) {
let mut out = vec![];
serialize_sparse_codec(
data.iter()
.cloned()
.enumerate()
.filter(|(_pos, val)| *val)
.map(|(pos, _val)| pos as u32),
&mut out,
)
.unwrap();
let null_index = SparseCodec::open(OwnedBytes::new(out));
let orig_idx_with_value: Vec<u32> = data
.iter()
.enumerate()
.filter(|(_pos, val)| **val)
.map(|(pos, _val)| pos as u32)
.collect();
assert_eq!(
null_index
.translate_codec_idx_to_original_idx(0..orig_idx_with_value.len() as u32)
.collect_vec(),
orig_idx_with_value
);
let step_size = (orig_idx_with_value.len() / 100).max(1);
for (dense_idx, orig_idx) in orig_idx_with_value.iter().enumerate().step_by(step_size) {
assert_eq!(
null_index.translate_to_codec_idx(*orig_idx),
Some(dense_idx as u32)
);
}
// 100 samples
let step_size = (data.len() / 100).max(1);
for (pos, value) in data.iter().enumerate().step_by(step_size) {
assert_eq!(null_index.exists(pos as u32), *value);
}
}
#[test]
fn sparse_codec_test_translation() {
let mut out = vec![];
let iter = ([true, false, true, false]).iter().cloned();
serialize_sparse_codec(
iter.enumerate()
.filter(|(_pos, val)| *val)
.map(|(pos, _val)| pos as u32),
&mut out,
)
.unwrap();
let null_index = SparseCodec::open(OwnedBytes::new(out));
assert_eq!(
null_index
.translate_codec_idx_to_original_idx(0..2)
.collect_vec(),
vec![0, 2]
);
}
#[test]
fn sparse_codec_translate() {
let mut out = vec![];
let iter = ([true, false, true, false]).iter().cloned();
serialize_sparse_codec(
iter.enumerate()
.filter(|(_pos, val)| *val)
.map(|(pos, _val)| pos as u32),
&mut out,
)
.unwrap();
let null_index = SparseCodec::open(OwnedBytes::new(out));
assert_eq!(null_index.translate_to_codec_idx(0), Some(0));
assert_eq!(null_index.translate_to_codec_idx(2), Some(1));
}
#[test]
fn sparse_codec_test_small() {
let mut out = vec![];
let iter = ([true, false, true, false]).iter().cloned();
serialize_sparse_codec(
iter.enumerate()
.filter(|(_pos, val)| *val)
.map(|(pos, _val)| pos as u32),
&mut out,
)
.unwrap();
let null_index = SparseCodec::open(OwnedBytes::new(out));
assert!(null_index.exists(0));
assert!(!null_index.exists(1));
assert!(null_index.exists(2));
assert!(!null_index.exists(3));
}
#[test]
fn sparse_codec_test_large() {
let mut docs = vec![];
docs.extend((0..ELEMENTS_PER_BLOCK).map(|_idx| false));
docs.extend((0..=1).map(|_idx| true));
let iter = docs.iter().cloned();
let mut out = vec![];
serialize_sparse_codec(
iter.enumerate()
.filter(|(_pos, val)| *val)
.map(|(pos, _val)| pos as u32),
&mut out,
)
.unwrap();
let null_index = SparseCodec::open(OwnedBytes::new(out));
assert!(!null_index.exists(0));
assert!(!null_index.exists(100));
assert!(!null_index.exists(ELEMENTS_PER_BLOCK - 1));
assert!(null_index.exists(ELEMENTS_PER_BLOCK));
assert!(null_index.exists(ELEMENTS_PER_BLOCK + 1));
}
}
#[cfg(all(test, feature = "unstable"))]
mod bench {
use rand::rngs::StdRng;
use rand::{Rng, SeedableRng};
use test::Bencher;
use super::*;
const TOTAL_NUM_VALUES: u32 = 1_000_000;
fn gen_bools(fill_ratio: f64) -> SparseCodec {
let mut out = Vec::new();
let mut rng: StdRng = StdRng::from_seed([1u8; 32]);
serialize_sparse_codec(
(0..TOTAL_NUM_VALUES)
.map(|_| rng.gen_bool(fill_ratio))
.enumerate()
.filter(|(_pos, val)| *val)
.map(|(pos, _val)| pos as u32),
&mut out,
)
.unwrap();
let codec = SparseCodec::open(OwnedBytes::new(out));
codec
}
fn random_range_iterator(start: u32, end: u32, step_size: u32) -> impl Iterator<Item = u32> {
let mut rng: StdRng = StdRng::from_seed([1u8; 32]);
let mut current = start;
std::iter::from_fn(move || {
current += rng.gen_range(1..step_size + 1);
if current >= end {
None
} else {
Some(current)
}
})
}
fn walk_over_data(codec: &SparseCodec, max_step_size: u32) -> Option<u32> {
walk_over_data_from_positions(
codec,
random_range_iterator(0, TOTAL_NUM_VALUES, max_step_size),
)
}
fn walk_over_data_from_positions(
codec: &SparseCodec,
positions: impl Iterator<Item = u32>,
) -> Option<u32> {
let mut dense_idx: Option<u32> = None;
for idx in positions {
dense_idx = dense_idx.or(codec.translate_to_codec_idx(idx));
}
dense_idx
}
#[bench]
fn bench_sparse_codec_translate_orig_to_codec_1percent_filled_random_stride(
bench: &mut Bencher,
) {
let codec = gen_bools(0.01f64);
bench.iter(|| walk_over_data(&codec, 100));
}
#[bench]
fn bench_sparse_codec_translate_orig_to_codec_5percent_filled_random_stride(
bench: &mut Bencher,
) {
let codec = gen_bools(0.05f64);
bench.iter(|| walk_over_data(&codec, 100));
}
#[bench]
fn bench_sparse_codec_translate_orig_to_codec_full_scan_10percent(bench: &mut Bencher) {
let codec = gen_bools(0.1f64);
bench.iter(|| walk_over_data_from_positions(&codec, 0..TOTAL_NUM_VALUES));
}
#[bench]
fn bench_sparse_codec_translate_orig_to_codec_full_scan_90percent(bench: &mut Bencher) {
let codec = gen_bools(0.9f64);
bench.iter(|| walk_over_data_from_positions(&codec, 0..TOTAL_NUM_VALUES));
}
#[bench]
fn bench_sparse_codec_translate_orig_to_codec_full_scan_1percent(bench: &mut Bencher) {
let codec = gen_bools(0.01f64);
bench.iter(|| walk_over_data_from_positions(&codec, 0..TOTAL_NUM_VALUES));
}
#[bench]
fn bench_sparse_codec_translate_orig_to_codec_10percent_filled_random_stride(
bench: &mut Bencher,
) {
let codec = gen_bools(0.1f64);
bench.iter(|| walk_over_data(&codec, 100));
}
#[bench]
fn bench_sparse_codec_translate_orig_to_codec_90percent_filled_random_stride(
bench: &mut Bencher,
) {
let codec = gen_bools(0.9f64);
bench.iter(|| walk_over_data(&codec, 100));
}
#[bench]
fn bench_sparse_codec_translate_codec_to_orig_1percent_filled_random_stride_big_step(
bench: &mut Bencher,
) {
let codec = gen_bools(0.01f64);
let num_vals = codec.num_non_nulls();
bench.iter(|| {
codec
.translate_codec_idx_to_original_idx(random_range_iterator(0, num_vals, 50_000))
.last()
});
}
#[bench]
fn bench_sparse_codec_translate_codec_to_orig_1percent_filled_random_stride(
bench: &mut Bencher,
) {
let codec = gen_bools(0.01f64);
let num_vals = codec.num_non_nulls();
bench.iter(|| {
codec
.translate_codec_idx_to_original_idx(random_range_iterator(0, num_vals, 100))
.last()
});
}
#[bench]
fn bench_sparse_codec_translate_codec_to_orig_1percent_filled_full_scan(bench: &mut Bencher) {
let codec = gen_bools(0.01f64);
let num_vals = codec.num_non_nulls();
bench.iter(|| {
codec
.translate_codec_idx_to_original_idx(0..num_vals)
.last()
});
}
#[bench]
fn bench_sparse_codec_translate_codec_to_orig_90percent_filled_random_stride_big_step(
bench: &mut Bencher,
) {
let codec = gen_bools(0.90f64);
let num_vals = codec.num_non_nulls();
bench.iter(|| {
codec
.translate_codec_idx_to_original_idx(random_range_iterator(0, num_vals, 50_000))
.last()
});
}
#[bench]
fn bench_sparse_codec_translate_codec_to_orig_90percent_filled_random_stride(
bench: &mut Bencher,
) {
let codec = gen_bools(0.9f64);
let num_vals = codec.num_non_nulls();
bench.iter(|| {
codec
.translate_codec_idx_to_original_idx(random_range_iterator(0, num_vals, 100))
.last()
});
}
#[bench]
fn bench_sparse_codec_translate_codec_to_orig_90percent_filled_full_scan(bench: &mut Bencher) {
let codec = gen_bools(0.9f64);
let num_vals = codec.num_non_nulls();
bench.iter(|| {
codec
.translate_codec_idx_to_original_idx(0..num_vals)
.last()
});
}
}

146
fastfield_codecs/src/null_index_footer.rs
View File

@@ -1,146 +0,0 @@
use std::io::{self, Write};
use std::ops::Range;
use common::{BinarySerializable, CountingWriter, VInt};
use ownedbytes::OwnedBytes;
#[derive(Debug, Clone, Copy, Eq, PartialEq)]
pub(crate) enum FastFieldCardinality {
Single = 1,
Multi = 2,
}
impl BinarySerializable for FastFieldCardinality {
fn serialize<W: Write>(&self, wrt: &mut W) -> io::Result<()> {
self.to_code().serialize(wrt)
}
fn deserialize<R: io::Read>(reader: &mut R) -> io::Result<Self> {
let code = u8::deserialize(reader)?;
let codec_type: Self = Self::from_code(code)
.ok_or_else(|| io::Error::new(io::ErrorKind::InvalidData, "Unknown code `{code}.`"))?;
Ok(codec_type)
}
}
impl FastFieldCardinality {
pub(crate) fn to_code(self) -> u8 {
self as u8
}
pub(crate) fn from_code(code: u8) -> Option<Self> {
match code {
1 => Some(Self::Single),
2 => Some(Self::Multi),
_ => None,
}
}
}
#[derive(Debug, Clone, Copy, PartialEq, Eq)]
pub(crate) enum NullIndexCodec {
Full = 1,
}
impl BinarySerializable for NullIndexCodec {
fn serialize<W: Write>(&self, wrt: &mut W) -> io::Result<()> {
self.to_code().serialize(wrt)
}
fn deserialize<R: io::Read>(reader: &mut R) -> io::Result<Self> {
let code = u8::deserialize(reader)?;
let codec_type: Self = Self::from_code(code)
.ok_or_else(|| io::Error::new(io::ErrorKind::InvalidData, "Unknown code `{code}.`"))?;
Ok(codec_type)
}
}
impl NullIndexCodec {
pub(crate) fn to_code(self) -> u8 {
self as u8
}
pub(crate) fn from_code(code: u8) -> Option<Self> {
match code {
1 => Some(Self::Full),
_ => None,
}
}
}
#[derive(Debug, Clone, Eq, PartialEq)]
pub(crate) struct NullIndexFooter {
pub(crate) cardinality: FastFieldCardinality,
pub(crate) null_index_codec: NullIndexCodec,
// Unused for NullIndexCodec::Full
pub(crate) null_index_byte_range: Range<u64>,
}
impl BinarySerializable for NullIndexFooter {
fn serialize<W: Write>(&self, writer: &mut W) -> io::Result<()> {
self.cardinality.serialize(writer)?;
self.null_index_codec.serialize(writer)?;
VInt(self.null_index_byte_range.start).serialize(writer)?;
VInt(self.null_index_byte_range.end - self.null_index_byte_range.start)
.serialize(writer)?;
Ok(())
}
fn deserialize<R: io::Read>(reader: &mut R) -> io::Result<Self> {
let cardinality = FastFieldCardinality::deserialize(reader)?;
let null_index_codec = NullIndexCodec::deserialize(reader)?;
let null_index_byte_range_start = VInt::deserialize(reader)?.0;
let null_index_byte_range_end = VInt::deserialize(reader)?.0 + null_index_byte_range_start;
Ok(Self {
cardinality,
null_index_codec,
null_index_byte_range: null_index_byte_range_start..null_index_byte_range_end,
})
}
}
pub(crate) fn append_null_index_footer(
output: &mut impl io::Write,
null_index_footer: NullIndexFooter,
) -> io::Result<()> {
let mut counting_write = CountingWriter::wrap(output);
null_index_footer.serialize(&mut counting_write)?;
let footer_payload_len = counting_write.written_bytes();
BinarySerializable::serialize(&(footer_payload_len as u16), &mut counting_write)?;
Ok(())
}
pub(crate) fn read_null_index_footer(
data: OwnedBytes,
) -> io::Result<(OwnedBytes, NullIndexFooter)> {
let (data, null_footer_length_bytes) = data.rsplit(2);
let footer_length = u16::deserialize(&mut null_footer_length_bytes.as_slice())?;
let (data, null_index_footer_bytes) = data.rsplit(footer_length as usize);
let null_index_footer = NullIndexFooter::deserialize(&mut null_index_footer_bytes.as_ref())?;
Ok((data, null_index_footer))
}
#[cfg(test)]
mod tests {
use super::*;
#[test]
fn null_index_footer_deser_test() {
let null_index_footer = NullIndexFooter {
cardinality: FastFieldCardinality::Single,
null_index_codec: NullIndexCodec::Full,
null_index_byte_range: 100..120,
};
let mut out = vec![];
null_index_footer.serialize(&mut out).unwrap();
assert_eq!(
null_index_footer,
NullIndexFooter::deserialize(&mut &out[..]).unwrap()
);
}
}

161
fastfield_codecs/src/serialize.rs
View File

@@ -1,3 +1,22 @@
// Copyright (C) 2022 Quickwit, Inc.
//
// Quickwit is offered under the AGPL v3.0 and as commercial software.
// For commercial licensing, contact us at hello@quickwit.io.
//
// AGPL:
// This program is free software: you can redistribute it and/or modify
// it under the terms of the GNU Affero General Public License as
// published by the Free Software Foundation, either version 3 of the
// License, or (at your option) any later version.
//
// This program is distributed in the hope that it will be useful,
// but WITHOUT ANY WARRANTY; without even the implied warranty of
// MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
// GNU Affero General Public License for more details.
//
// You should have received a copy of the GNU Affero General Public License
// along with this program. If not, see <http://www.gnu.org/licenses/>.
use std::io;
use std::num::NonZeroU64;
use std::sync::Arc;
@@ -9,18 +28,14 @@ use ownedbytes::OwnedBytes;
use crate::bitpacked::BitpackedCodec;
use crate::blockwise_linear::BlockwiseLinearCodec;
use crate::compact_space::CompactSpaceCompressor;
use crate::format_version::append_format_version;
use crate::linear::LinearCodec;
use crate::monotonic_mapping::{
StrictlyMonotonicFn, StrictlyMonotonicMappingToInternal,
StrictlyMonotonicMappingToInternalGCDBaseval,
};
use crate::null_index_footer::{
append_null_index_footer, FastFieldCardinality, NullIndexCodec, NullIndexFooter,
};
use crate::{
monotonic_map_column, Column, FastFieldCodec, FastFieldCodecType, MonotonicallyMappableToU64,
U128FastFieldCodecType, VecColumn, ALL_CODEC_TYPES,
VecColumn, ALL_CODEC_TYPES,
};
/// The normalized header gives some parameters after applying the following
@@ -83,29 +98,6 @@ impl Header {
}
}
#[derive(Debug, Copy, Clone, PartialEq, Eq)]
pub(crate) struct U128Header {
pub num_vals: u32,
pub codec_type: U128FastFieldCodecType,
}
impl BinarySerializable for U128Header {
fn serialize<W: io::Write>(&self, writer: &mut W) -> io::Result<()> {
VInt(self.num_vals as u64).serialize(writer)?;
self.codec_type.serialize(writer)?;
Ok(())
}
fn deserialize<R: io::Read>(reader: &mut R) -> io::Result<Self> {
let num_vals = VInt::deserialize(reader)?.0 as u32;
let codec_type = U128FastFieldCodecType::deserialize(reader)?;
Ok(U128Header {
num_vals,
codec_type,
})
}
}
pub fn normalize_column<C: Column>(
from_column: C,
min_value: u64,
@@ -175,84 +167,10 @@ pub fn serialize_u128<F: Fn() -> I, I: Iterator<Item = u128>>(
num_vals: u32,
output: &mut impl io::Write,
) -> io::Result<()> {
serialize_u128_new(ValueIndexInfo::default(), iter_gen, num_vals, output)
}
#[allow(dead_code)]
pub enum ValueIndexInfo<'a> {
MultiValue(Box<dyn MultiValueIndexInfo + 'a>),
SingleValue(Box<dyn SingleValueIndexInfo + 'a>),
}
impl Default for ValueIndexInfo<'static> {
fn default() -> Self {
struct Dummy {}
impl SingleValueIndexInfo for Dummy {
fn num_vals(&self) -> u32 {
todo!()
}
fn num_non_nulls(&self) -> u32 {
todo!()
}
fn iter(&self) -> Box<dyn Iterator<Item = u32>> {
todo!()
}
}
Self::SingleValue(Box::new(Dummy {}))
}
}
impl<'a> ValueIndexInfo<'a> {
fn get_cardinality(&self) -> FastFieldCardinality {
match self {
ValueIndexInfo::MultiValue(_) => FastFieldCardinality::Multi,
ValueIndexInfo::SingleValue(_) => FastFieldCardinality::Single,
}
}
}
pub trait MultiValueIndexInfo {
/// The number of docs in the column.
fn num_docs(&self) -> u32;
/// The number of values in the column.
fn num_vals(&self) -> u32;
/// Return the start index of the values for each doc
fn iter(&self) -> Box<dyn Iterator<Item = u32> + '_>;
}
pub trait SingleValueIndexInfo {
/// The number of values including nulls in the column.
fn num_vals(&self) -> u32;
/// The number of non-null values in the column.
fn num_non_nulls(&self) -> u32;
/// Return a iterator of the positions of docs with a value
fn iter(&self) -> Box<dyn Iterator<Item = u32> + '_>;
}
/// Serializes u128 values with the compact space codec.
pub fn serialize_u128_new<F: Fn() -> I, I: Iterator<Item = u128>>(
value_index: ValueIndexInfo,
iter_gen: F,
num_vals: u32,
output: &mut impl io::Write,
) -> io::Result<()> {
let header = U128Header {
num_vals,
codec_type: U128FastFieldCodecType::CompactSpace,
};
header.serialize(output)?;
// TODO write header, to later support more codecs
let compressor = CompactSpaceCompressor::train_from(iter_gen(), num_vals);
compressor.compress_into(iter_gen(), output).unwrap();
let null_index_footer = NullIndexFooter {
cardinality: value_index.get_cardinality(),
null_index_codec: NullIndexCodec::Full,
null_index_byte_range: 0..0,
};
append_null_index_footer(output, null_index_footer)?;
append_format_version(output)?;
Ok(())
}
@@ -261,16 +179,6 @@ pub fn serialize<T: MonotonicallyMappableToU64>(
typed_column: impl Column<T>,
output: &mut impl io::Write,
codecs: &[FastFieldCodecType],
) -> io::Result<()> {
serialize_new(ValueIndexInfo::default(), typed_column, output, codecs)
}
/// Serializes the column with the codec with the best estimate on the data.
pub fn serialize_new<T: MonotonicallyMappableToU64>(
value_index: ValueIndexInfo,
typed_column: impl Column<T>,
output: &mut impl io::Write,
codecs: &[FastFieldCodecType],
) -> io::Result<()> {
let column = monotonic_map_column(typed_column, StrictlyMonotonicMappingToInternal::<T>::new());
let header = Header::compute_header(&column, codecs).ok_or_else(|| {
@@ -286,15 +194,6 @@ pub fn serialize_new<T: MonotonicallyMappableToU64>(
let normalized_column = header.normalize_column(column);
assert_eq!(normalized_column.min_value(), 0u64);
serialize_given_codec(normalized_column, header.codec_type, output)?;
let null_index_footer = NullIndexFooter {
cardinality: value_index.get_cardinality(),
null_index_codec: NullIndexCodec::Full,
null_index_byte_range: 0..0,
};
append_null_index_footer(output, null_index_footer)?;
append_format_version(output)?;
Ok(())
}
@@ -359,18 +258,6 @@ pub fn serialize_and_load<T: MonotonicallyMappableToU64 + Ord + Default>(
mod tests {
use super::*;
#[test]
fn test_serialize_deserialize_u128_header() {
let original = U128Header {
num_vals: 11,
codec_type: U128FastFieldCodecType::CompactSpace,
};
let mut out = Vec::new();
original.serialize(&mut out).unwrap();
let restored = U128Header::deserialize(&mut &out[..]).unwrap();
assert_eq!(restored, original);
}
#[test]
fn test_serialize_deserialize() {
let original = [1u64, 5u64, 10u64];
@@ -384,7 +271,7 @@ mod tests {
let col = VecColumn::from(&[false, true][..]);
serialize(col, &mut buffer, &ALL_CODEC_TYPES).unwrap();
// 5 bytes of header, 1 byte of value, 7 bytes of padding.
assert_eq!(buffer.len(), 3 + 5 + 8 + 4 + 2);
assert_eq!(buffer.len(), 5 + 8);
}
#[test]
@@ -393,7 +280,7 @@ mod tests {
let col = VecColumn::from(&[true][..]);
serialize(col, &mut buffer, &ALL_CODEC_TYPES).unwrap();
// 5 bytes of header, 0 bytes of value, 7 bytes of padding.
assert_eq!(buffer.len(), 3 + 5 + 7 + 4 + 2);
assert_eq!(buffer.len(), 5 + 7);
}
#[test]
@@ -403,6 +290,6 @@ mod tests {
let col = VecColumn::from(&vals[..]);
serialize(col, &mut buffer, &[FastFieldCodecType::Bitpacked]).unwrap();
// Values are stored over 3 bits.
assert_eq!(buffer.len(), 3 + 7 + (3 * 80 / 8) + 7 + 4 + 2);
assert_eq!(buffer.len(), 7 + (3 * 80 / 8) + 7);
}
}

6
ownedbytes/Cargo.toml
View File

@@ -1,14 +1,10 @@
[package]
authors = ["Paul Masurel <paul@quickwit.io>", "Pascal Seitz <pascal@quickwit.io>"]
name = "ownedbytes"
version = "0.5.0"
version = "0.3.0"
edition = "2021"
description = "Expose data as static slice"
license = "MIT"
documentation = "https://docs.rs/ownedbytes/"
homepage = "https://github.com/quickwit-oss/tantivy"
repository = "https://github.com/quickwit-oss/tantivy"
# See more keys and their definitions at https://doc.rust-lang.org/cargo/reference/manifest.html
[dependencies]

17
ownedbytes/src/lib.rs
View File

@@ -3,7 +3,7 @@ use std::ops::{Deref, Range};
use std::sync::Arc;
use std::{fmt, io, mem};
pub use stable_deref_trait::StableDeref;
use stable_deref_trait::StableDeref;
/// An OwnedBytes simply wraps an object that owns a slice of data and exposes
/// this data as a slice.
@@ -80,21 +80,6 @@ impl OwnedBytes {
(left, right)
}
/// Splits the OwnedBytes into two OwnedBytes `(left, right)`.
///
/// Right will hold `split_len` bytes.
///
/// This operation is cheap and does not require to copy any memory.
/// On the other hand, both `left` and `right` retain a handle over
/// the entire slice of memory. In other words, the memory will only
/// be released when both left and right are dropped.
#[inline]
#[must_use]
pub fn rsplit(self, split_len: usize) -> (OwnedBytes, OwnedBytes) {
let data_len = self.data.len();
self.split(data_len - split_len)
}
/// Splits the right part of the `OwnedBytes` at the given offset.
///
/// `self` is truncated to `split_len`, left with the remaining bytes.

2
query-grammar/Cargo.toml
View File

@@ -1,6 +1,6 @@
[package]
name = "tantivy-query-grammar"
version = "0.19.0"
version = "0.18.0"
authors = ["Paul Masurel <paul.masurel@gmail.com>"]
license = "MIT"
categories = ["database-implementations", "data-structures"]

23
query-grammar/src/query_grammar.rs
View File

@@ -5,8 +5,7 @@ use combine::parser::range::{take_while, take_while1};
use combine::parser::repeat::escaped;
use combine::parser::Parser;
use combine::{
attempt, between, choice, eof, many, many1, one_of, optional, parser, satisfy, sep_by,
skip_many1, value,
attempt, choice, eof, many, many1, one_of, optional, parser, satisfy, skip_many1, value,
};
use once_cell::sync::Lazy;
use regex::Regex;
@@ -265,17 +264,6 @@ fn range<'a>() -> impl Parser<&'a str, Output = UserInputLeaf> {
})
}
/// Function that parses a set out of a Stream
/// Supports ranges like: `IN [val1 val2 val3]`
fn set<'a>() -> impl Parser<&'a str, Output = UserInputLeaf> {
let term_list = between(char('['), char(']'), sep_by(term_val(), spaces()));
let set_content = ((string("IN"), spaces()), term_list).map(|(_, elements)| elements);
(optional(attempt(field_name().skip(spaces()))), set_content)
.map(|(field, elements)| UserInputLeaf::Set { field, elements })
}
fn negate(expr: UserInputAst) -> UserInputAst {
expr.unary(Occur::MustNot)
}
@@ -290,7 +278,6 @@ fn leaf<'a>() -> impl Parser<&'a str, Output = UserInputAst> {
string("NOT").skip(spaces1()).with(leaf()).map(negate),
))
.or(attempt(range().map(UserInputAst::from)))
.or(attempt(set().map(UserInputAst::from)))
.or(literal().map(UserInputAst::from))
.parse_stream(input)
.into_result()
@@ -760,14 +747,6 @@ mod test {
test_parse_query_to_ast_helper("+(a b) +d", "(+(*\"a\" *\"b\") +\"d\")");
}
#[test]
fn test_parse_test_query_set() {
test_parse_query_to_ast_helper("abc: IN [a b c]", r#""abc": IN ["a" "b" "c"]"#);
test_parse_query_to_ast_helper("abc: IN [1]", r#""abc": IN ["1"]"#);
test_parse_query_to_ast_helper("abc: IN []", r#""abc": IN []"#);
test_parse_query_to_ast_helper("IN [1 2]", r#"IN ["1" "2"]"#);
}
#[test]
fn test_parse_test_query_other() {
test_parse_query_to_ast_helper("(+a +b) d", "(*(+\"a\" +\"b\") *\"d\")");

17
query-grammar/src/user_input_ast.rs
View File

@@ -12,10 +12,6 @@ pub enum UserInputLeaf {
lower: UserInputBound,
upper: UserInputBound,
},
Set {
field: Option<String>,
elements: Vec<String>,
},
}
impl Debug for UserInputLeaf {
@@ -35,19 +31,6 @@ impl Debug for UserInputLeaf {
upper.display_upper(formatter)?;
Ok(())
}
UserInputLeaf::Set { field, elements } => {
if let Some(ref field) = field {
write!(formatter, "\"{}\": ", field)?;
}
write!(formatter, "IN [")?;
for (i, element) in elements.iter().enumerate() {
if i != 0 {
write!(formatter, " ")?;
}
write!(formatter, "\"{}\"", element)?;
}
write!(formatter, "]")
}
UserInputLeaf::All => write!(formatter, "*"),
}
}

10
src/aggregation/agg_req_with_accessor.rs
View File

@@ -11,7 +11,7 @@ use super::bucket::{HistogramAggregation, RangeAggregation, TermsAggregation};
use super::metric::{AverageAggregation, StatsAggregation};
use super::segment_agg_result::BucketCount;
use super::VecWithNames;
use crate::fastfield::{type_and_cardinality, MultiValuedFastFieldReader};
use crate::fastfield::{type_and_cardinality, FastType, MultiValuedFastFieldReader};
use crate::schema::{Cardinality, Type};
use crate::{InvertedIndexReader, SegmentReader, TantivyError};
@@ -194,7 +194,13 @@ fn get_ff_reader_and_validate(
.ok_or_else(|| TantivyError::FieldNotFound(field_name.to_string()))?;
let field_type = reader.schema().get_field_entry(field).field_type();
if let Some((_ff_type, field_cardinality)) = type_and_cardinality(field_type) {
if let Some((ff_type, field_cardinality)) = type_and_cardinality(field_type) {
if ff_type == FastType::Date {
return Err(TantivyError::InvalidArgument(
"Unsupported field type date in aggregation".to_string(),
));
}
if cardinality != field_cardinality {
return Err(TantivyError::InvalidArgument(format!(
"Invalid field cardinality on field {} expected {:?}, but got {:?}",

21
src/aggregation/agg_result.rs
View File

@@ -4,6 +4,8 @@
//! intermediate average results, which is the sum and the number of values. The actual average is
//! calculated on the step from intermediate to final aggregation result tree.
use std::collections::HashMap;
use rustc_hash::FxHashMap;
use serde::{Deserialize, Serialize};
@@ -12,12 +14,11 @@ use super::bucket::GetDocCount;
use super::intermediate_agg_result::{IntermediateBucketResult, IntermediateMetricResult};
use super::metric::{SingleMetricResult, Stats};
use super::Key;
use crate::schema::Schema;
use crate::TantivyError;
#[derive(Clone, Default, Debug, PartialEq, Serialize, Deserialize)]
/// The final aggegation result.
pub struct AggregationResults(pub FxHashMap<String, AggregationResult>);
pub struct AggregationResults(pub HashMap<String, AggregationResult>);
impl AggregationResults {
pub(crate) fn get_value_from_aggregation(
@@ -130,12 +131,9 @@ pub enum BucketResult {
}
impl BucketResult {
pub(crate) fn empty_from_req(
req: &BucketAggregationInternal,
schema: &Schema,
) -> crate::Result<Self> {
pub(crate) fn empty_from_req(req: &BucketAggregationInternal) -> crate::Result<Self> {
let empty_bucket = IntermediateBucketResult::empty_from_req(&req.bucket_agg);
empty_bucket.into_final_bucket_result(req, schema)
empty_bucket.into_final_bucket_result(req)
}
}
@@ -178,9 +176,6 @@ pub enum BucketEntries<T> {
/// ```
#[derive(Clone, Debug, PartialEq, Serialize, Deserialize)]
pub struct BucketEntry {
#[serde(skip_serializing_if = "Option::is_none")]
/// The string representation of the bucket.
pub key_as_string: Option<String>,
/// The identifier of the bucket.
pub key: Key,
/// Number of documents in the bucket.
@@ -245,10 +240,4 @@ pub struct RangeBucketEntry {
/// The to range of the bucket. Equals `f64::MAX` when `None`.
#[serde(skip_serializing_if = "Option::is_none")]
pub to: Option<f64>,
/// The optional string representation for the `from` range.
#[serde(skip_serializing_if = "Option::is_none")]
pub from_as_string: Option<String>,
/// The optional string representation for the `to` range.
#[serde(skip_serializing_if = "Option::is_none")]
pub to_as_string: Option<String>,
}

164
src/aggregation/bucket/histogram/histogram.rs
View File

@@ -10,12 +10,12 @@ use crate::aggregation::agg_req_with_accessor::{
AggregationsWithAccessor, BucketAggregationWithAccessor,
};
use crate::aggregation::agg_result::BucketEntry;
use crate::aggregation::f64_from_fastfield_u64;
use crate::aggregation::intermediate_agg_result::{
IntermediateAggregationResults, IntermediateBucketResult, IntermediateHistogramBucketEntry,
};
use crate::aggregation::segment_agg_result::SegmentAggregationResultsCollector;
use crate::aggregation::{f64_from_fastfield_u64, format_date};
use crate::schema::{Schema, Type};
use crate::schema::Type;
use crate::{DocId, TantivyError};
/// Histogram is a bucket aggregation, where buckets are created dynamically for given `interval`.
@@ -206,7 +206,6 @@ pub struct SegmentHistogramCollector {
field_type: Type,
interval: f64,
offset: f64,
min_doc_count: u64,
first_bucket_num: i64,
bounds: HistogramBounds,
}
@@ -216,30 +215,6 @@ impl SegmentHistogramCollector {
self,
agg_with_accessor: &BucketAggregationWithAccessor,
) -> crate::Result<IntermediateBucketResult> {
// Compute the number of buckets to validate against max num buckets
// Note: We use min_doc_count here, but it's only an lowerbound here, since were are on the
// intermediate level and after merging the number of documents of a bucket could exceed
// `min_doc_count`.
{
let cut_off_buckets_front = self
.buckets
.iter()
.take_while(|bucket| bucket.doc_count <= self.min_doc_count)
.count();
let cut_off_buckets_back = self.buckets[cut_off_buckets_front..]
.iter()
.rev()
.take_while(|bucket| bucket.doc_count <= self.min_doc_count)
.count();
let estimate_num_buckets =
self.buckets.len() - cut_off_buckets_front - cut_off_buckets_back;
agg_with_accessor
.bucket_count
.add_count(estimate_num_buckets as u32);
agg_with_accessor.bucket_count.validate_bucket_count()?;
}
let mut buckets = Vec::with_capacity(
self.buckets
.iter()
@@ -276,6 +251,11 @@ impl SegmentHistogramCollector {
);
};
agg_with_accessor
.bucket_count
.add_count(buckets.len() as u32);
agg_with_accessor.bucket_count.validate_bucket_count()?;
Ok(IntermediateBucketResult::Histogram { buckets })
}
@@ -328,7 +308,6 @@ impl SegmentHistogramCollector {
first_bucket_num,
bounds,
sub_aggregations,
min_doc_count: req.min_doc_count(),
})
}
@@ -401,7 +380,7 @@ impl SegmentHistogramCollector {
debug_assert_eq!(
self.buckets[bucket_pos].key,
get_bucket_val(val, self.interval, self.offset)
get_bucket_val(val, self.interval, self.offset) as f64
);
self.increment_bucket(bucket_pos, doc, &bucket_with_accessor.sub_aggregation)?;
}
@@ -428,7 +407,7 @@ impl SegmentHistogramCollector {
if bounds.contains(val) {
debug_assert_eq!(
self.buckets[bucket_pos].key,
get_bucket_val(val, self.interval, self.offset)
get_bucket_val(val, self.interval, self.offset) as f64
);
self.increment_bucket(bucket_pos, doc, bucket_with_accessor)?;
@@ -472,7 +451,6 @@ fn intermediate_buckets_to_final_buckets_fill_gaps(
buckets: Vec<IntermediateHistogramBucketEntry>,
histogram_req: &HistogramAggregation,
sub_aggregation: &AggregationsInternal,
schema: &Schema,
) -> crate::Result<Vec<BucketEntry>> {
// Generate the full list of buckets without gaps.
//
@@ -513,9 +491,7 @@ fn intermediate_buckets_to_final_buckets_fill_gaps(
sub_aggregation: empty_sub_aggregation.clone(),
},
})
.map(|intermediate_bucket| {
intermediate_bucket.into_final_bucket_entry(sub_aggregation, schema)
})
.map(|intermediate_bucket| intermediate_bucket.into_final_bucket_entry(sub_aggregation))
.collect::<crate::Result<Vec<_>>>()
}
@@ -524,43 +500,20 @@ pub(crate) fn intermediate_histogram_buckets_to_final_buckets(
buckets: Vec<IntermediateHistogramBucketEntry>,
histogram_req: &HistogramAggregation,
sub_aggregation: &AggregationsInternal,
schema: &Schema,
) -> crate::Result<Vec<BucketEntry>> {
let mut buckets = if histogram_req.min_doc_count() == 0 {
if histogram_req.min_doc_count() == 0 {
// With min_doc_count != 0, we may need to add buckets, so that there are no
// gaps, since intermediate result does not contain empty buckets (filtered to
// reduce serialization size).
intermediate_buckets_to_final_buckets_fill_gaps(
buckets,
histogram_req,
sub_aggregation,
schema,
)?
intermediate_buckets_to_final_buckets_fill_gaps(buckets, histogram_req, sub_aggregation)
} else {
buckets
.into_iter()
.filter(|histogram_bucket| histogram_bucket.doc_count >= histogram_req.min_doc_count())
.map(|histogram_bucket| {
histogram_bucket.into_final_bucket_entry(sub_aggregation, schema)
})
.collect::<crate::Result<Vec<_>>>()?
};
// If we have a date type on the histogram buckets, we add the `key_as_string` field as rfc339
let field = schema
.get_field(&histogram_req.field)
.ok_or_else(|| TantivyError::FieldNotFound(histogram_req.field.to_string()))?;
if schema.get_field_entry(field).field_type().is_date() {
for bucket in buckets.iter_mut() {
if let crate::aggregation::Key::F64(val) = bucket.key {
let key_as_string = format_date(val as i64)?;
bucket.key_as_string = Some(key_as_string);
}
}
.map(|histogram_bucket| histogram_bucket.into_final_bucket_entry(sub_aggregation))
.collect::<crate::Result<Vec<_>>>()
}
Ok(buckets)
}
/// Applies req extended_bounds/hard_bounds on the min_max value
@@ -1419,63 +1372,6 @@ mod tests {
Ok(())
}
#[test]
fn histogram_date_test_single_segment() -> crate::Result<()> {
histogram_date_test_with_opt(true)
}
#[test]
fn histogram_date_test_multi_segment() -> crate::Result<()> {
histogram_date_test_with_opt(false)
}
fn histogram_date_test_with_opt(merge_segments: bool) -> crate::Result<()> {
let index = get_test_index_2_segments(merge_segments)?;
let agg_req: Aggregations = vec![(
"histogram".to_string(),
Aggregation::Bucket(BucketAggregation {
bucket_agg: BucketAggregationType::Histogram(HistogramAggregation {
field: "date".to_string(),
interval: 86400000000.0, // one day in microseconds
..Default::default()
}),
sub_aggregation: Default::default(),
}),
)]
.into_iter()
.collect();
let agg_res = exec_request(agg_req, &index)?;
let res: Value = serde_json::from_str(&serde_json::to_string(&agg_res)?)?;
assert_eq!(res["histogram"]["buckets"][0]["key"], 1546300800000000.0);
assert_eq!(
res["histogram"]["buckets"][0]["key_as_string"],
"2019-01-01T00:00:00Z"
);
assert_eq!(res["histogram"]["buckets"][0]["doc_count"], 1);
assert_eq!(res["histogram"]["buckets"][1]["key"], 1546387200000000.0);
assert_eq!(
res["histogram"]["buckets"][1]["key_as_string"],
"2019-01-02T00:00:00Z"
);
assert_eq!(res["histogram"]["buckets"][1]["doc_count"], 5);
assert_eq!(res["histogram"]["buckets"][2]["key"], 1546473600000000.0);
assert_eq!(
res["histogram"]["buckets"][2]["key_as_string"],
"2019-01-03T00:00:00Z"
);
assert_eq!(res["histogram"]["buckets"][3], Value::Null);
Ok(())
}
#[test]
fn histogram_invalid_request() -> crate::Result<()> {
let index = get_test_index_2_segments(true)?;
@@ -1542,36 +1438,4 @@ mod tests {
Ok(())
}
#[test]
fn histogram_test_max_buckets_segments() -> crate::Result<()> {
let values = vec![0.0, 70000.0];
let index = get_test_index_from_values(true, &values)?;
let agg_req: Aggregations = vec![(
"my_interval".to_string(),
Aggregation::Bucket(BucketAggregation {
bucket_agg: BucketAggregationType::Histogram(HistogramAggregation {
field: "score_f64".to_string(),
interval: 1.0,
..Default::default()
}),
sub_aggregation: Default::default(),
}),
)]
.into_iter()
.collect();
let res = exec_request(agg_req, &index);
assert_eq!(
res.unwrap_err().to_string(),
"An invalid argument was passed: 'Aborting aggregation because too many buckets were \
created'"
.to_string()
);
Ok(())
}
}

108
src/aggregation/bucket/range.rs
View File

@@ -1,7 +1,6 @@
use std::fmt::Debug;
use std::ops::Range;
use fastfield_codecs::MonotonicallyMappableToU64;
use rustc_hash::FxHashMap;
use serde::{Deserialize, Serialize};
@@ -12,9 +11,7 @@ use crate::aggregation::intermediate_agg_result::{
IntermediateBucketResult, IntermediateRangeBucketEntry, IntermediateRangeBucketResult,
};
use crate::aggregation::segment_agg_result::{BucketCount, SegmentAggregationResultsCollector};
use crate::aggregation::{
f64_from_fastfield_u64, f64_to_fastfield_u64, format_date, Key, SerializedKey,
};
use crate::aggregation::{f64_from_fastfield_u64, f64_to_fastfield_u64, Key, SerializedKey};
use crate::schema::Type;
use crate::{DocId, TantivyError};
@@ -184,7 +181,7 @@ impl SegmentRangeCollector {
.into_iter()
.map(move |range_bucket| {
Ok((
range_to_string(&range_bucket.range, &field_type)?,
range_to_string(&range_bucket.range, &field_type),
range_bucket
.bucket
.into_intermediate_bucket_entry(&agg_with_accessor.sub_aggregation)?,
@@ -212,8 +209,8 @@ impl SegmentRangeCollector {
let key = range
.key
.clone()
.map(|key| Ok(Key::Str(key)))
.unwrap_or_else(|| range_to_key(&range.range, &field_type))?;
.map(Key::Str)
.unwrap_or_else(|| range_to_key(&range.range, &field_type));
let to = if range.range.end == u64::MAX {
None
} else {
@@ -231,7 +228,6 @@ impl SegmentRangeCollector {
sub_aggregation,
)?)
};
Ok(SegmentRangeAndBucketEntry {
range: range.range.clone(),
bucket: SegmentRangeBucketEntry {
@@ -406,45 +402,34 @@ fn extend_validate_ranges(
Ok(converted_buckets)
}
pub(crate) fn range_to_string(range: &Range<u64>, field_type: &Type) -> crate::Result<String> {
pub(crate) fn range_to_string(range: &Range<u64>, field_type: &Type) -> String {
// is_start is there for malformed requests, e.g. ig the user passes the range u64::MIN..0.0,
// it should be rendered as "*-0" and not "*-*"
let to_str = |val: u64, is_start: bool| {
if (is_start && val == u64::MIN) || (!is_start && val == u64::MAX) {
Ok("*".to_string())
} else if *field_type == Type::Date {
let val = i64::from_u64(val);
format_date(val)
"*".to_string()
} else {
Ok(f64_from_fastfield_u64(val, field_type).to_string())
f64_from_fastfield_u64(val, field_type).to_string()
}
};
Ok(format!(
"{}-{}",
to_str(range.start, true)?,
to_str(range.end, false)?
))
format!("{}-{}", to_str(range.start, true), to_str(range.end, false))
}
pub(crate) fn range_to_key(range: &Range<u64>, field_type: &Type) -> crate::Result<Key> {
Ok(Key::Str(range_to_string(range, field_type)?))
pub(crate) fn range_to_key(range: &Range<u64>, field_type: &Type) -> Key {
Key::Str(range_to_string(range, field_type))
}
#[cfg(test)]
mod tests {
use fastfield_codecs::MonotonicallyMappableToU64;
use serde_json::Value;
use super::*;
use crate::aggregation::agg_req::{
Aggregation, Aggregations, BucketAggregation, BucketAggregationType,
};
use crate::aggregation::tests::{
exec_request, exec_request_with_query, get_test_index_2_segments,
get_test_index_with_num_docs,
};
use crate::aggregation::tests::{exec_request_with_query, get_test_index_with_num_docs};
pub fn get_collector_from_ranges(
ranges: Vec<RangeAggregationRange>,
@@ -582,77 +567,6 @@ mod tests {
Ok(())
}
#[test]
fn range_date_test_single_segment() -> crate::Result<()> {
range_date_test_with_opt(true)
}
#[test]
fn range_date_test_multi_segment() -> crate::Result<()> {
range_date_test_with_opt(false)
}
fn range_date_test_with_opt(merge_segments: bool) -> crate::Result<()> {
let index = get_test_index_2_segments(merge_segments)?;
let agg_req: Aggregations = vec![(
"date_ranges".to_string(),
Aggregation::Bucket(BucketAggregation {
bucket_agg: BucketAggregationType::Range(RangeAggregation {
field: "date".to_string(),
ranges: vec![
RangeAggregationRange {
key: None,
from: None,
to: Some(1546300800000000.0f64),
},
RangeAggregationRange {
key: None,
from: Some(1546300800000000.0f64),
to: Some(1546387200000000.0f64),
},
],
keyed: false,
}),
sub_aggregation: Default::default(),
}),
)]
.into_iter()
.collect();
let agg_res = exec_request(agg_req, &index)?;
let res: Value = serde_json::from_str(&serde_json::to_string(&agg_res)?)?;
assert_eq!(
res["date_ranges"]["buckets"][0]["from_as_string"],
Value::Null
);
assert_eq!(
res["date_ranges"]["buckets"][0]["key"],
"*-2019-01-01T00:00:00Z"
);
assert_eq!(
res["date_ranges"]["buckets"][1]["from_as_string"],
"2019-01-01T00:00:00Z"
);
assert_eq!(
res["date_ranges"]["buckets"][1]["to_as_string"],
"2019-01-02T00:00:00Z"
);
assert_eq!(
res["date_ranges"]["buckets"][2]["from_as_string"],
"2019-01-02T00:00:00Z"
);
assert_eq!(
res["date_ranges"]["buckets"][2]["to_as_string"],
Value::Null
);
Ok(())
}
#[test]
fn range_custom_key_keyed_buckets_test() -> crate::Result<()> {
let index = get_test_index_with_num_docs(false, 100)?;

7
src/aggregation/collector.rs
View File

@@ -7,7 +7,6 @@ use super::intermediate_agg_result::IntermediateAggregationResults;
use super::segment_agg_result::SegmentAggregationResultsCollector;
use crate::aggregation::agg_req_with_accessor::get_aggs_with_accessor_and_validate;
use crate::collector::{Collector, SegmentCollector};
use crate::schema::Schema;
use crate::{SegmentReader, TantivyError};
/// The default max bucket count, before the aggregation fails.
@@ -17,7 +16,6 @@ pub const MAX_BUCKET_COUNT: u32 = 65000;
///
/// The collector collects all aggregations by the underlying aggregation request.
pub struct AggregationCollector {
schema: Schema,
agg: Aggregations,
max_bucket_count: u32,
}
@@ -27,9 +25,8 @@ impl AggregationCollector {
///
/// Aggregation fails when the total bucket count is higher than max_bucket_count.
/// max_bucket_count will default to `MAX_BUCKET_COUNT` (65000) when unset
pub fn from_aggs(agg: Aggregations, max_bucket_count: Option<u32>, schema: Schema) -> Self {
pub fn from_aggs(agg: Aggregations, max_bucket_count: Option<u32>) -> Self {
Self {
schema,
agg,
max_bucket_count: max_bucket_count.unwrap_or(MAX_BUCKET_COUNT),
}
@@ -116,7 +113,7 @@ impl Collector for AggregationCollector {
segment_fruits: Vec<<Self::Child as SegmentCollector>::Fruit>,
) -> crate::Result<Self::Fruit> {
let res = merge_fruits(segment_fruits)?;
res.into_final_bucket_result(self.agg.clone(), &self.schema)
res.into_final_bucket_result(self.agg.clone())
}
}

18
src/aggregation/date.rs
View File

@@ -1,18 +0,0 @@
use time::format_description::well_known::Rfc3339;
use time::OffsetDateTime;
use crate::TantivyError;
pub(crate) fn format_date(val: i64) -> crate::Result<String> {
let datetime =
OffsetDateTime::from_unix_timestamp_nanos(1_000 * (val as i128)).map_err(|err| {
TantivyError::InvalidArgument(format!(
"Could not convert {:?} to OffsetDateTime, err {:?}",
val, err
))
})?;
let key_as_string = datetime
.format(&Rfc3339)
.map_err(|_err| TantivyError::InvalidArgument("Could not serialize date".to_string()))?;
Ok(key_as_string)
}

87
src/aggregation/intermediate_agg_result.rs
View File

@@ -3,6 +3,7 @@
//! indices.
use std::cmp::Ordering;
use std::collections::HashMap;
use itertools::Itertools;
use rustc_hash::FxHashMap;
@@ -10,7 +11,7 @@ use serde::{Deserialize, Serialize};
use super::agg_req::{
Aggregations, AggregationsInternal, BucketAggregationInternal, BucketAggregationType,
MetricAggregation, RangeAggregation,
MetricAggregation,
};
use super::agg_result::{AggregationResult, BucketResult, RangeBucketEntry};
use super::bucket::{
@@ -19,11 +20,9 @@ use super::bucket::{
};
use super::metric::{IntermediateAverage, IntermediateStats};
use super::segment_agg_result::SegmentMetricResultCollector;
use super::{format_date, Key, SerializedKey, VecWithNames};
use super::{Key, SerializedKey, VecWithNames};
use crate::aggregation::agg_result::{AggregationResults, BucketEntries, BucketEntry};
use crate::aggregation::bucket::TermsAggregationInternal;
use crate::schema::Schema;
use crate::TantivyError;
/// Contains the intermediate aggregation result, which is optimized to be merged with other
/// intermediate results.
@@ -37,12 +36,8 @@ pub struct IntermediateAggregationResults {
impl IntermediateAggregationResults {
/// Convert intermediate result and its aggregation request to the final result.
pub fn into_final_bucket_result(
self,
req: Aggregations,
schema: &Schema,
) -> crate::Result<AggregationResults> {
self.into_final_bucket_result_internal(&(req.into()), schema)
pub fn into_final_bucket_result(self, req: Aggregations) -> crate::Result<AggregationResults> {
self.into_final_bucket_result_internal(&(req.into()))
}
/// Convert intermediate result and its aggregation request to the final result.
@@ -52,19 +47,18 @@ impl IntermediateAggregationResults {
pub(crate) fn into_final_bucket_result_internal(
self,
req: &AggregationsInternal,
schema: &Schema,
) -> crate::Result<AggregationResults> {
// Important assumption:
// When the tree contains buckets/metric, we expect it to have all buckets/metrics from the
// request
let mut results: FxHashMap<String, AggregationResult> = FxHashMap::default();
let mut results: HashMap<String, AggregationResult> = HashMap::new();
if let Some(buckets) = self.buckets {
convert_and_add_final_buckets_to_result(&mut results, buckets, &req.buckets, schema)?
convert_and_add_final_buckets_to_result(&mut results, buckets, &req.buckets)?
} else {
// When there are no buckets, we create empty buckets, so that the serialized json
// format is constant
add_empty_final_buckets_to_result(&mut results, &req.buckets, schema)?
add_empty_final_buckets_to_result(&mut results, &req.buckets)?
};
if let Some(metrics) = self.metrics {
@@ -138,7 +132,7 @@ impl IntermediateAggregationResults {
}
fn convert_and_add_final_metrics_to_result(
results: &mut FxHashMap<String, AggregationResult>,
results: &mut HashMap<String, AggregationResult>,
metrics: VecWithNames<IntermediateMetricResult>,
) {
results.extend(
@@ -149,7 +143,7 @@ fn convert_and_add_final_metrics_to_result(
}
fn add_empty_final_metrics_to_result(
results: &mut FxHashMap<String, AggregationResult>,
results: &mut HashMap<String, AggregationResult>,
req_metrics: &VecWithNames<MetricAggregation>,
) -> crate::Result<()> {
results.extend(req_metrics.iter().map(|(key, req)| {
@@ -163,30 +157,27 @@ fn add_empty_final_metrics_to_result(
}
fn add_empty_final_buckets_to_result(
results: &mut FxHashMap<String, AggregationResult>,
results: &mut HashMap<String, AggregationResult>,
req_buckets: &VecWithNames<BucketAggregationInternal>,
schema: &Schema,
) -> crate::Result<()> {
let requested_buckets = req_buckets.iter();
for (key, req) in requested_buckets {
let empty_bucket =
AggregationResult::BucketResult(BucketResult::empty_from_req(req, schema)?);
let empty_bucket = AggregationResult::BucketResult(BucketResult::empty_from_req(req)?);
results.insert(key.to_string(), empty_bucket);
}
Ok(())
}
fn convert_and_add_final_buckets_to_result(
results: &mut FxHashMap<String, AggregationResult>,
results: &mut HashMap<String, AggregationResult>,
buckets: VecWithNames<IntermediateBucketResult>,
req_buckets: &VecWithNames<BucketAggregationInternal>,
schema: &Schema,
) -> crate::Result<()> {
assert_eq!(buckets.len(), req_buckets.len());
let buckets_with_request = buckets.into_iter().zip(req_buckets.values());
for ((key, bucket), req) in buckets_with_request {
let result = AggregationResult::BucketResult(bucket.into_final_bucket_result(req, schema)?);
let result = AggregationResult::BucketResult(bucket.into_final_bucket_result(req)?);
results.insert(key, result);
}
Ok(())
@@ -276,21 +267,13 @@ impl IntermediateBucketResult {
pub(crate) fn into_final_bucket_result(
self,
req: &BucketAggregationInternal,
schema: &Schema,
) -> crate::Result<BucketResult> {
match self {
IntermediateBucketResult::Range(range_res) => {
let mut buckets: Vec<RangeBucketEntry> = range_res
.buckets
.into_values()
.map(|bucket| {
bucket.into_final_bucket_entry(
&req.sub_aggregation,
schema,
req.as_range()
.expect("unexpected aggregation, expected histogram aggregation"),
)
})
.into_iter()
.map(|(_, bucket)| bucket.into_final_bucket_entry(&req.sub_aggregation))
.collect::<crate::Result<Vec<_>>>()?;
buckets.sort_by(|left, right| {
@@ -321,7 +304,6 @@ impl IntermediateBucketResult {
req.as_histogram()
.expect("unexpected aggregation, expected histogram aggregation"),
&req.sub_aggregation,
schema,
)?;
let buckets = if req.as_histogram().unwrap().keyed {
@@ -340,7 +322,6 @@ impl IntermediateBucketResult {
req.as_term()
.expect("unexpected aggregation, expected term aggregation"),
&req.sub_aggregation,
schema,
),
}
}
@@ -431,7 +412,6 @@ impl IntermediateTermBucketResult {
self,
req: &TermsAggregation,
sub_aggregation_req: &AggregationsInternal,
schema: &Schema,
) -> crate::Result<BucketResult> {
let req = TermsAggregationInternal::from_req(req);
let mut buckets: Vec<BucketEntry> = self
@@ -440,12 +420,11 @@ impl IntermediateTermBucketResult {
.filter(|bucket| bucket.1.doc_count >= req.min_doc_count)
.map(|(key, entry)| {
Ok(BucketEntry {
key_as_string: None,
key: Key::Str(key),
doc_count: entry.doc_count,
sub_aggregation: entry
.sub_aggregation
.into_final_bucket_result_internal(sub_aggregation_req, schema)?,
.into_final_bucket_result_internal(sub_aggregation_req)?,
})
})
.collect::<crate::Result<_>>()?;
@@ -550,15 +529,13 @@ impl IntermediateHistogramBucketEntry {
pub(crate) fn into_final_bucket_entry(
self,
req: &AggregationsInternal,
schema: &Schema,
) -> crate::Result<BucketEntry> {
Ok(BucketEntry {
key_as_string: None,
key: Key::F64(self.key),
doc_count: self.doc_count,
sub_aggregation: self
.sub_aggregation
.into_final_bucket_result_internal(req, schema)?,
.into_final_bucket_result_internal(req)?,
})
}
}
@@ -595,38 +572,16 @@ impl IntermediateRangeBucketEntry {
pub(crate) fn into_final_bucket_entry(
self,
req: &AggregationsInternal,
schema: &Schema,
range_req: &RangeAggregation,
) -> crate::Result<RangeBucketEntry> {
let mut range_bucket_entry = RangeBucketEntry {
Ok(RangeBucketEntry {
key: self.key,
doc_count: self.doc_count,
sub_aggregation: self
.sub_aggregation
.into_final_bucket_result_internal(req, schema)?,
.into_final_bucket_result_internal(req)?,
to: self.to,
from: self.from,
to_as_string: None,
from_as_string: None,
};
// If we have a date type on the histogram buckets, we add the `key_as_string` field as
// rfc339
let field = schema
.get_field(&range_req.field)
.ok_or_else(|| TantivyError::FieldNotFound(range_req.field.to_string()))?;
if schema.get_field_entry(field).field_type().is_date() {
if let Some(val) = range_bucket_entry.to {
let key_as_string = format_date(val as i64)?;
range_bucket_entry.to_as_string = Some(key_as_string);
}
if let Some(val) = range_bucket_entry.from {
let key_as_string = format_date(val as i64)?;
range_bucket_entry.from_as_string = Some(key_as_string);
}
}
Ok(range_bucket_entry)
})
}
}

4
src/aggregation/metric/stats.rs
View File

@@ -222,7 +222,7 @@ mod tests {
.into_iter()
.collect();
let collector = AggregationCollector::from_aggs(agg_req_1, None, index.schema());
let collector = AggregationCollector::from_aggs(agg_req_1, None);
let reader = index.reader()?;
let searcher = reader.searcher();
@@ -300,7 +300,7 @@ mod tests {
.into_iter()
.collect();
let collector = AggregationCollector::from_aggs(agg_req_1, None, index.schema());
let collector = AggregationCollector::from_aggs(agg_req_1, None);
let searcher = reader.searcher();
let agg_res: AggregationResults = searcher.search(&term_query, &collector).unwrap();

76
src/aggregation/mod.rs
View File

@@ -12,7 +12,7 @@
//!
//! ## Prerequisite
//! Currently aggregations work only on [fast fields](`crate::fastfield`). Single value fast fields
//! of type `u64`, `f64`, `i64`, `date` and fast fields on text fields.
//! of type `u64`, `f64`, `i64` and fast fields on text fields.
//!
//! ## Usage
//! To use aggregations, build an aggregation request by constructing
@@ -53,10 +53,9 @@
//! use tantivy::query::AllQuery;
//! use tantivy::aggregation::agg_result::AggregationResults;
//! use tantivy::IndexReader;
//! use tantivy::schema::Schema;
//!
//! # #[allow(dead_code)]
//! fn aggregate_on_index(reader: &IndexReader, schema: Schema) {
//! fn aggregate_on_index(reader: &IndexReader) {
//! let agg_req: Aggregations = vec![
//! (
//! "average".to_string(),
@@ -68,7 +67,7 @@
//! .into_iter()
//! .collect();
//!
//! let collector = AggregationCollector::from_aggs(agg_req, None, schema);
//! let collector = AggregationCollector::from_aggs(agg_req, None);
//!
//! let searcher = reader.searcher();
//! let agg_res: AggregationResults = searcher.search(&AllQuery, &collector).unwrap();
@@ -158,7 +157,6 @@ mod agg_req_with_accessor;
pub mod agg_result;
pub mod bucket;
mod collector;
mod date;
pub mod intermediate_agg_result;
pub mod metric;
mod segment_agg_result;
@@ -169,7 +167,6 @@ pub use collector::{
AggregationCollector, AggregationSegmentCollector, DistributedAggregationCollector,
MAX_BUCKET_COUNT,
};
pub(crate) use date::format_date;
use fastfield_codecs::MonotonicallyMappableToU64;
use itertools::Itertools;
use serde::{Deserialize, Serialize};
@@ -286,11 +283,11 @@ impl Display for Key {
/// Inverse of `to_fastfield_u64`. Used to convert to `f64` for metrics.
///
/// # Panics
/// Only `u64`, `f64`, `date`, and `i64` are supported.
/// Only `u64`, `f64`, and `i64` are supported.
pub(crate) fn f64_from_fastfield_u64(val: u64, field_type: &Type) -> f64 {
match field_type {
Type::U64 => val as f64,
Type::I64 | Type::Date => i64::from_u64(val) as f64,
Type::I64 => i64::from_u64(val) as f64,
Type::F64 => f64::from_u64(val),
_ => {
panic!("unexpected type {:?}. This should not happen", field_type)
@@ -298,9 +295,10 @@ pub(crate) fn f64_from_fastfield_u64(val: u64, field_type: &Type) -> f64 {
}
}
/// Converts the `f64` value to fast field value space, which is always u64.
/// Converts the `f64` value to fast field value space.
///
/// If the fast field has `u64`, values are stored unchanged as `u64` in the fast field.
/// If the fast field has `u64`, values are stored as `u64` in the fast field.
/// A `f64` value of e.g. `2.0` therefore needs to be converted to `1u64`.
///
/// If the fast field has `f64` values are converted and stored to `u64` using a
/// monotonic mapping.
@@ -310,7 +308,7 @@ pub(crate) fn f64_from_fastfield_u64(val: u64, field_type: &Type) -> f64 {
pub(crate) fn f64_to_fastfield_u64(val: f64, field_type: &Type) -> Option<u64> {
match field_type {
Type::U64 => Some(val as u64),
Type::I64 | Type::Date => Some((val as i64).to_u64()),
Type::I64 => Some((val as i64).to_u64()),
Type::F64 => Some(val.to_u64()),
_ => None,
}
@@ -319,7 +317,6 @@ pub(crate) fn f64_to_fastfield_u64(val: f64, field_type: &Type) -> Option<u64> {
#[cfg(test)]
mod tests {
use serde_json::Value;
use time::OffsetDateTime;
use super::agg_req::{Aggregation, Aggregations, BucketAggregation};
use super::bucket::RangeAggregation;
@@ -335,7 +332,7 @@ mod tests {
use crate::aggregation::DistributedAggregationCollector;
use crate::query::{AllQuery, TermQuery};
use crate::schema::{Cardinality, IndexRecordOption, Schema, TextFieldIndexing, FAST, STRING};
use crate::{DateTime, Index, Term};
use crate::{Index, Term};
fn get_avg_req(field_name: &str) -> Aggregation {
Aggregation::Metric(MetricAggregation::Average(
@@ -361,7 +358,7 @@ mod tests {
index: &Index,
query: Option<(&str, &str)>,
) -> crate::Result<Value> {
let collector = AggregationCollector::from_aggs(agg_req, None, index.schema());
let collector = AggregationCollector::from_aggs(agg_req, None);
let reader = index.reader()?;
let searcher = reader.searcher();
@@ -451,9 +448,9 @@ mod tests {
text_field_id => term.to_string(),
string_field_id => term.to_string(),
score_field => i as u64,
score_field_f64 => i,
score_field_f64 => i as f64,
score_field_i64 => i as i64,
fraction_field => i/100.0,
fraction_field => i as f64/100.0,
))?;
}
index_writer.commit()?;
@@ -555,10 +552,10 @@ mod tests {
let searcher = reader.searcher();
let intermediate_agg_result = searcher.search(&AllQuery, &collector).unwrap();
intermediate_agg_result
.into_final_bucket_result(agg_req, &index.schema())
.into_final_bucket_result(agg_req)
.unwrap()
} else {
let collector = AggregationCollector::from_aggs(agg_req, None, index.schema());
let collector = AggregationCollector::from_aggs(agg_req, None);
let searcher = reader.searcher();
searcher.search(&AllQuery, &collector).unwrap()
@@ -651,7 +648,6 @@ mod tests {
.set_fast()
.set_stored();
let text_field = schema_builder.add_text_field("text", text_fieldtype);
let date_field = schema_builder.add_date_field("date", FAST);
schema_builder.add_text_field("dummy_text", STRING);
let score_fieldtype =
crate::schema::NumericOptions::default().set_fast(Cardinality::SingleValue);
@@ -669,7 +665,6 @@ mod tests {
// writing the segment
index_writer.add_document(doc!(
text_field => "cool",
date_field => DateTime::from_utc(OffsetDateTime::from_unix_timestamp(1_546_300_800).unwrap()),
score_field => 1u64,
score_field_f64 => 1f64,
score_field_i64 => 1i64,
@@ -678,7 +673,6 @@ mod tests {
))?;
index_writer.add_document(doc!(
text_field => "cool",
date_field => DateTime::from_utc(OffsetDateTime::from_unix_timestamp(1_546_300_800 + 86400).unwrap()),
score_field => 3u64,
score_field_f64 => 3f64,
score_field_i64 => 3i64,
@@ -687,21 +681,18 @@ mod tests {
))?;
index_writer.add_document(doc!(
text_field => "cool",
date_field => DateTime::from_utc(OffsetDateTime::from_unix_timestamp(1_546_300_800 + 86400).unwrap()),
score_field => 5u64,
score_field_f64 => 5f64,
score_field_i64 => 5i64,
))?;
index_writer.add_document(doc!(
text_field => "nohit",
date_field => DateTime::from_utc(OffsetDateTime::from_unix_timestamp(1_546_300_800 + 86400).unwrap()),
score_field => 6u64,
score_field_f64 => 6f64,
score_field_i64 => 6i64,
))?;
index_writer.add_document(doc!(
text_field => "cool",
date_field => DateTime::from_utc(OffsetDateTime::from_unix_timestamp(1_546_300_800 + 86400).unwrap()),
score_field => 7u64,
score_field_f64 => 7f64,
score_field_i64 => 7i64,
@@ -709,14 +700,12 @@ mod tests {
index_writer.commit()?;
index_writer.add_document(doc!(
text_field => "cool",
date_field => DateTime::from_utc(OffsetDateTime::from_unix_timestamp(1_546_300_800 + 86400).unwrap()),
score_field => 11u64,
score_field_f64 => 11f64,
score_field_i64 => 11i64,
))?;
index_writer.add_document(doc!(
text_field => "cool",
date_field => DateTime::from_utc(OffsetDateTime::from_unix_timestamp(1_546_300_800 + 86400 + 86400).unwrap()),
score_field => 14u64,
score_field_f64 => 14f64,
score_field_i64 => 14i64,
@@ -724,7 +713,6 @@ mod tests {
index_writer.add_document(doc!(
text_field => "cool",
date_field => DateTime::from_utc(OffsetDateTime::from_unix_timestamp(1_546_300_800 + 86400 + 86400).unwrap()),
score_field => 44u64,
score_field_f64 => 44.5f64,
score_field_i64 => 44i64,
@@ -735,7 +723,6 @@ mod tests {
// no hits segment
index_writer.add_document(doc!(
text_field => "nohit",
date_field => DateTime::from_utc(OffsetDateTime::from_unix_timestamp(1_546_300_800 + 86400 + 86400).unwrap()),
score_field => 44u64,
score_field_f64 => 44.5f64,
score_field_i64 => 44i64,
@@ -808,7 +795,7 @@ mod tests {
.into_iter()
.collect();
let collector = AggregationCollector::from_aggs(agg_req_1, None, index.schema());
let collector = AggregationCollector::from_aggs(agg_req_1, None);
let searcher = reader.searcher();
let agg_res: AggregationResults = searcher.search(&term_query, &collector).unwrap();
@@ -1008,10 +995,9 @@ mod tests {
// Test de/serialization roundtrip on intermediate_agg_result
let res: IntermediateAggregationResults =
serde_json::from_str(&serde_json::to_string(&res).unwrap()).unwrap();
res.into_final_bucket_result(agg_req.clone(), &index.schema())
.unwrap()
res.into_final_bucket_result(agg_req.clone()).unwrap()
} else {
let collector = AggregationCollector::from_aggs(agg_req.clone(), None, index.schema());
let collector = AggregationCollector::from_aggs(agg_req.clone(), None);
let searcher = reader.searcher();
searcher.search(&term_query, &collector).unwrap()
@@ -1069,7 +1055,7 @@ mod tests {
);
// Test empty result set
let collector = AggregationCollector::from_aggs(agg_req, None, index.schema());
let collector = AggregationCollector::from_aggs(agg_req, None);
let searcher = reader.searcher();
searcher.search(&query_with_no_hits, &collector).unwrap();
@@ -1134,7 +1120,7 @@ mod tests {
.into_iter()
.collect();
let collector = AggregationCollector::from_aggs(agg_req_1, None, index.schema());
let collector = AggregationCollector::from_aggs(agg_req_1, None);
let searcher = reader.searcher();
@@ -1247,7 +1233,7 @@ mod tests {
.into_iter()
.collect();
let collector = AggregationCollector::from_aggs(agg_req_1, None, index.schema());
let collector = AggregationCollector::from_aggs(agg_req_1, None);
let searcher = reader.searcher();
let agg_res: AggregationResults =
@@ -1278,7 +1264,7 @@ mod tests {
.into_iter()
.collect();
let collector = AggregationCollector::from_aggs(agg_req_1, None, index.schema());
let collector = AggregationCollector::from_aggs(agg_req_1, None);
let searcher = reader.searcher();
let agg_res: AggregationResults =
@@ -1309,7 +1295,7 @@ mod tests {
.into_iter()
.collect();
let collector = AggregationCollector::from_aggs(agg_req_1, None, index.schema());
let collector = AggregationCollector::from_aggs(agg_req_1, None);
let searcher = reader.searcher();
let agg_res: AggregationResults =
@@ -1348,7 +1334,7 @@ mod tests {
.into_iter()
.collect();
let collector = AggregationCollector::from_aggs(agg_req_1, None, index.schema());
let collector = AggregationCollector::from_aggs(agg_req_1, None);
let searcher = reader.searcher();
let agg_res: AggregationResults =
@@ -1377,7 +1363,7 @@ mod tests {
.into_iter()
.collect();
let collector = AggregationCollector::from_aggs(agg_req, None, index.schema());
let collector = AggregationCollector::from_aggs(agg_req, None);
let searcher = reader.searcher();
let agg_res: AggregationResults =
@@ -1406,7 +1392,7 @@ mod tests {
.into_iter()
.collect();
let collector = AggregationCollector::from_aggs(agg_req, None, index.schema());
let collector = AggregationCollector::from_aggs(agg_req, None);
let searcher = reader.searcher();
let agg_res: AggregationResults =
@@ -1443,7 +1429,7 @@ mod tests {
.into_iter()
.collect();
let collector = AggregationCollector::from_aggs(agg_req_1, None, index.schema());
let collector = AggregationCollector::from_aggs(agg_req_1, None);
let searcher = reader.searcher();
let agg_res: AggregationResults =
@@ -1478,7 +1464,7 @@ mod tests {
.into_iter()
.collect();
let collector = AggregationCollector::from_aggs(agg_req_1, None, index.schema());
let collector = AggregationCollector::from_aggs(agg_req_1, None);
let searcher = reader.searcher();
let agg_res: AggregationResults =
@@ -1517,7 +1503,7 @@ mod tests {
.into_iter()
.collect();
let collector = AggregationCollector::from_aggs(agg_req_1, None, index.schema());
let collector = AggregationCollector::from_aggs(agg_req_1, None);
let searcher = reader.searcher();
let agg_res: AggregationResults =
@@ -1547,7 +1533,7 @@ mod tests {
.into_iter()
.collect();
let collector = AggregationCollector::from_aggs(agg_req_1, None, index.schema());
let collector = AggregationCollector::from_aggs(agg_req_1, None);
let searcher = reader.searcher();
let agg_res: AggregationResults =
@@ -1604,7 +1590,7 @@ mod tests {
.into_iter()
.collect();
let collector = AggregationCollector::from_aggs(agg_req_1, None, index.schema());
let collector = AggregationCollector::from_aggs(agg_req_1, None);
let searcher = reader.searcher();
let agg_res: AggregationResults =

2
src/aggregation/segment_agg_result.rs
View File

@@ -305,7 +305,7 @@ impl BucketCount {
}
pub(crate) fn add_count(&self, count: u32) {
self.bucket_count
.fetch_add(count, std::sync::atomic::Ordering::Relaxed);
.fetch_add(count as u32, std::sync::atomic::Ordering::Relaxed);
}
pub(crate) fn get_count(&self) -> u32 {
self.bucket_count.load(std::sync::atomic::Ordering::Relaxed)

4
src/collector/facet_collector.rs
View File

@@ -357,7 +357,7 @@ impl SegmentCollector for FacetSegmentCollector {
let mut facet = vec![];
let facet_ord = self.collapse_facet_ords[collapsed_facet_ord];
// TODO handle errors.
if facet_dict.ord_to_term(facet_ord, &mut facet).is_ok() {
if facet_dict.ord_to_term(facet_ord as u64, &mut facet).is_ok() {
if let Ok(facet) = Facet::from_encoded(facet) {
facet_counts.insert(facet, count);
}
@@ -616,7 +616,7 @@ mod tests {
.map(|mut doc| {
doc.add_facet(
facet_field,
&format!("/facet/{}", thread_rng().sample(uniform)),
&format!("/facet/{}", thread_rng().sample(&uniform)),
);
doc
})

36
src/collector/mod.rs
View File

@@ -170,35 +170,19 @@ pub trait Collector: Sync + Send {
segment_ord: u32,
reader: &SegmentReader,
) -> crate::Result<<Self::Child as SegmentCollector>::Fruit> {
let mut segment_collector = self.for_segment(segment_ord, reader)?;
let mut segment_collector = self.for_segment(segment_ord as u32, reader)?;
match (reader.alive_bitset(), self.requires_scoring()) {
(Some(alive_bitset), true) => {
weight.for_each(reader, &mut |doc, score| {
if alive_bitset.is_alive(doc) {
segment_collector.collect(doc, score);
}
})?;
}
(Some(alive_bitset), false) => {
weight.for_each_no_score(reader, &mut |doc| {
if alive_bitset.is_alive(doc) {
segment_collector.collect(doc, 0.0);
}
})?;
}
(None, true) => {
weight.for_each(reader, &mut |doc, score| {
if let Some(alive_bitset) = reader.alive_bitset() {
weight.for_each(reader, &mut |doc, score| {
if alive_bitset.is_alive(doc) {
segment_collector.collect(doc, score);
})?;
}
(None, false) => {
weight.for_each_no_score(reader, &mut |doc| {
segment_collector.collect(doc, 0.0);
})?;
}
}
})?;
} else {
weight.for_each(reader, &mut |doc, score| {
segment_collector.collect(doc, score);
})?;
}
Ok(segment_collector.harvest())
}
}

5
src/core/index.rs
View File

@@ -149,8 +149,7 @@ impl IndexBuilder {
/// Creates a new index using the [`RamDirectory`].
///
/// The index will be allocated in anonymous memory.
/// This is useful for indexing small set of documents
/// for instances like unit test or temporary in memory index.
/// This should only be used for unit tests.
pub fn create_in_ram(self) -> Result<Index, TantivyError> {
let ram_directory = RamDirectory::create();
self.create(ram_directory)
@@ -813,7 +812,7 @@ mod tests {
let field = schema.get_field("num_likes").unwrap();
let tempdir = TempDir::new().unwrap();
let tempdir_path = PathBuf::from(tempdir.path());
let index = Index::create_in_dir(tempdir_path, schema).unwrap();
let index = Index::create_in_dir(&tempdir_path, schema).unwrap();
let reader = index
.reader_builder()
.reload_policy(ReloadPolicy::OnCommit)

2
src/core/index_meta.rs
View File

@@ -133,7 +133,7 @@ impl SegmentMeta {
/// associated with a segment component.
pub fn relative_path(&self, component: SegmentComponent) -> PathBuf {
let mut path = self.id().uuid_string();
path.push_str(&match component {
path.push_str(&*match component {
SegmentComponent::Postings => ".idx".to_string(),
SegmentComponent::Positions => ".pos".to_string(),
SegmentComponent::Terms => ".term".to_string(),

27
src/core/inverted_index_reader.rs
View File

@@ -200,7 +200,10 @@ impl InvertedIndexReader {
#[cfg(feature = "quickwit")]
impl InvertedIndexReader {
pub(crate) async fn get_term_info_async(&self, term: &Term) -> io::Result<Option<TermInfo>> {
pub(crate) async fn get_term_info_async(
&self,
term: &Term,
) -> crate::AsyncIoResult<Option<TermInfo>> {
self.termdict.get_async(term.value_bytes()).await
}
@@ -208,8 +211,12 @@ impl InvertedIndexReader {
/// This method is for an advanced usage only.
///
/// Most users should prefer using [`Self::read_postings()`] instead.
pub async fn warm_postings(&self, term: &Term, with_positions: bool) -> io::Result<()> {
let term_info_opt: Option<TermInfo> = self.get_term_info_async(term).await?;
pub async fn warm_postings(
&self,
term: &Term,
with_positions: bool,
) -> crate::AsyncIoResult<()> {
let term_info_opt = self.get_term_info_async(term).await?;
if let Some(term_info) = term_info_opt {
self.postings_file_slice
.read_bytes_slice_async(term_info.postings_range.clone())
@@ -223,20 +230,8 @@ impl InvertedIndexReader {
Ok(())
}
/// Read the block postings for all terms.
/// This method is for an advanced usage only.
///
/// If you know which terms to pre-load, prefer using [`Self::warm_postings`] instead.
pub async fn warm_postings_full(&self, with_positions: bool) -> io::Result<()> {
self.postings_file_slice.read_bytes_async().await?;
if with_positions {
self.positions_file_slice.read_bytes_async().await?;
}
Ok(())
}
/// Returns the number of documents containing the term asynchronously.
pub async fn doc_freq_async(&self, term: &Term) -> io::Result<u32> {
pub async fn doc_freq_async(&self, term: &Term) -> crate::AsyncIoResult<u32> {
Ok(self
.get_term_info_async(term)
.await?

9
src/core/searcher.rs
View File

@@ -4,7 +4,7 @@ use std::{fmt, io};
use crate::collector::Collector;
use crate::core::{Executor, SegmentReader};
use crate::query::{EnableScoring, Query};
use crate::query::Query;
use crate::schema::{Document, Schema, Term};
use crate::space_usage::SearcherSpaceUsage;
use crate::store::{CacheStats, StoreReader};
@@ -199,12 +199,7 @@ impl Searcher {
executor: &Executor,
) -> crate::Result<C::Fruit> {
let scoring_enabled = collector.requires_scoring();
let enabled_scoring = if scoring_enabled {
EnableScoring::Enabled(self)
} else {
EnableScoring::Disabled(self.schema())
};
let weight = query.weight(enabled_scoring)?;
let weight = query.weight(self, scoring_enabled)?;
let segment_readers = self.segment_readers();
let fruits = executor.map(
|(segment_ord, segment_reader)| {

2
src/directory/composite_file.rs
View File

@@ -75,7 +75,7 @@ impl<W: TerminatingWrite + Write> CompositeWrite<W> {
let mut prev_offset = 0;
for (file_addr, offset) in self.offsets {
VInt(offset - prev_offset).serialize(&mut self.write)?;
VInt((offset - prev_offset) as u64).serialize(&mut self.write)?;
file_addr.serialize(&mut self.write)?;
prev_offset = offset;
}

2
src/directory/directory.rs
View File

@@ -55,7 +55,7 @@ impl<T: Send + Sync + 'static> From<Box<T>> for DirectoryLock {
impl Drop for DirectoryLockGuard {
fn drop(&mut self) {
if let Err(e) = self.directory.delete(&self.path) {
if let Err(e) = self.directory.delete(&*self.path) {
error!("Failed to remove the lock file. {:?}", e);
}
}

75
common/src/file_slice.rs → src/directory/file_slice.rs
View File

@@ -1,18 +1,19 @@
use std::ops::{Deref, Range, RangeBounds};
use std::ops::{Deref, Range};
use std::sync::Arc;
use std::{fmt, io};
use async_trait::async_trait;
use ownedbytes::{OwnedBytes, StableDeref};
use common::HasLen;
use stable_deref_trait::StableDeref;
use crate::HasLen;
use crate::directory::OwnedBytes;
/// Objects that represents files sections in tantivy.
///
/// By contract, whatever happens to the directory file, as long as a FileHandle
/// is alive, the data associated with it cannot be altered or destroyed.
///
/// The underlying behavior is therefore specific to the `Directory` that
/// The underlying behavior is therefore specific to the [`Directory`](crate::Directory) that
/// created it. Despite its name, a [`FileSlice`] may or may not directly map to an actual file
/// on the filesystem.
@@ -23,9 +24,13 @@ pub trait FileHandle: 'static + Send + Sync + HasLen + fmt::Debug {
/// This method may panic if the range requested is invalid.
fn read_bytes(&self, range: Range<usize>) -> io::Result<OwnedBytes>;
#[cfg(feature = "quickwit")]
#[doc(hidden)]
async fn read_bytes_async(&self, byte_range: Range<usize>) -> io::Result<OwnedBytes> {
self.read_bytes(byte_range)
async fn read_bytes_async(
&self,
_byte_range: Range<usize>,
) -> crate::AsyncIoResult<OwnedBytes> {
Err(crate::error::AsyncIoError::AsyncUnsupported)
}
}
@@ -37,7 +42,7 @@ impl FileHandle for &'static [u8] {
}
#[cfg(feature = "quickwit")]
async fn read_bytes_async(&self, byte_range: Range<usize>) -> io::Result<OwnedBytes> {
async fn read_bytes_async(&self, byte_range: Range<usize>) -> crate::AsyncIoResult<OwnedBytes> {
Ok(self.read_bytes(byte_range)?)
}
}
@@ -65,25 +70,6 @@ impl fmt::Debug for FileSlice {
}
}
#[inline]
fn combine_ranges<R: RangeBounds<usize>>(orig_range: Range<usize>, rel_range: R) -> Range<usize> {
let start: usize = orig_range.start
+ match rel_range.start_bound().cloned() {
std::ops::Bound::Included(rel_start) => rel_start,
std::ops::Bound::Excluded(rel_start) => rel_start + 1,
std::ops::Bound::Unbounded => 0,
};
assert!(start <= orig_range.end);
let end: usize = match rel_range.end_bound().cloned() {
std::ops::Bound::Included(rel_end) => orig_range.start + rel_end + 1,
std::ops::Bound::Excluded(rel_end) => orig_range.start + rel_end,
std::ops::Bound::Unbounded => orig_range.end,
};
assert!(end >= start);
assert!(end <= orig_range.end);
start..end
}
impl FileSlice {
/// Wraps a FileHandle.
pub fn new(file_handle: Arc<dyn FileHandle>) -> Self {
@@ -107,11 +93,11 @@ impl FileSlice {
///
/// Panics if `byte_range.end` exceeds the filesize.
#[must_use]
#[inline]
pub fn slice<R: RangeBounds<usize>>(&self, byte_range: R) -> FileSlice {
pub fn slice(&self, byte_range: Range<usize>) -> FileSlice {
assert!(byte_range.end <= self.len());
FileSlice {
data: self.data.clone(),
range: combine_ranges(self.range.clone(), byte_range),
range: self.range.start + byte_range.start..self.range.start + byte_range.end,
}
}
@@ -131,8 +117,9 @@ impl FileSlice {
self.data.read_bytes(self.range.clone())
}
#[cfg(feature = "quickwit")]
#[doc(hidden)]
pub async fn read_bytes_async(&self) -> io::Result<OwnedBytes> {
pub async fn read_bytes_async(&self) -> crate::AsyncIoResult<OwnedBytes> {
self.data.read_bytes_async(self.range.clone()).await
}
@@ -150,8 +137,12 @@ impl FileSlice {
.read_bytes(self.range.start + range.start..self.range.start + range.end)
}
#[cfg(feature = "quickwit")]
#[doc(hidden)]
pub async fn read_bytes_slice_async(&self, byte_range: Range<usize>) -> io::Result<OwnedBytes> {
pub async fn read_bytes_slice_async(
&self,
byte_range: Range<usize>,
) -> crate::AsyncIoResult<OwnedBytes> {
assert!(
self.range.start + byte_range.end <= self.range.end,
"`to` exceeds the fileslice length"
@@ -214,7 +205,7 @@ impl FileHandle for FileSlice {
}
#[cfg(feature = "quickwit")]
async fn read_bytes_async(&self, byte_range: Range<usize>) -> io::Result<OwnedBytes> {
async fn read_bytes_async(&self, byte_range: Range<usize>) -> crate::AsyncIoResult<OwnedBytes> {
self.read_bytes_slice_async(byte_range).await
}
}
@@ -232,7 +223,7 @@ impl FileHandle for OwnedBytes {
}
#[cfg(feature = "quickwit")]
async fn read_bytes_async(&self, range: Range<usize>) -> io::Result<OwnedBytes> {
async fn read_bytes_async(&self, range: Range<usize>) -> crate::AsyncIoResult<OwnedBytes> {
let bytes = self.read_bytes(range)?;
Ok(bytes)
}
@@ -243,9 +234,9 @@ mod tests {
use std::io;
use std::sync::Arc;
use common::HasLen;
use super::{FileHandle, FileSlice};
use crate::file_slice::combine_ranges;
use crate::HasLen;
#[test]
fn test_file_slice() -> io::Result<()> {
@@ -316,18 +307,4 @@ mod tests {
b"bcd"
);
}
#[test]
fn test_combine_range() {
assert_eq!(combine_ranges(1..3, 0..1), 1..2);
assert_eq!(combine_ranges(1..3, 1..), 2..3);
assert_eq!(combine_ranges(1..4, ..2), 1..3);
assert_eq!(combine_ranges(3..10, 2..5), 5..8);
}
#[test]
#[should_panic]
fn test_combine_range_panics() {
let _ = combine_ranges(3..5, 1..4);
}
}

9
src/directory/footer.rs
View File

@@ -38,7 +38,7 @@ impl Footer {
counting_write.write_all(serde_json::to_string(&self)?.as_ref())?;
let footer_payload_len = counting_write.written_bytes();
BinarySerializable::serialize(&(footer_payload_len as u32), write)?;
BinarySerializable::serialize(&FOOTER_MAGIC_NUMBER, write)?;
BinarySerializable::serialize(&(FOOTER_MAGIC_NUMBER as u32), write)?;
Ok(())
}
@@ -90,10 +90,9 @@ impl Footer {
));
}
let footer: Footer =
serde_json::from_slice(&file.read_bytes_slice(
file.len() - total_footer_size..file.len() - footer_metadata_len,
)?)?;
let footer: Footer = serde_json::from_slice(&file.read_bytes_slice(
file.len() - total_footer_size..file.len() - footer_metadata_len as usize,
)?)?;
let body = file.slice_to(file.len() - total_footer_size);
Ok((footer, body))

2
src/directory/managed_directory.rs
View File

@@ -388,7 +388,7 @@ mod tests_mmap_specific {
let tempdir_path = PathBuf::from(tempdir.path());
let living_files = HashSet::new();
let mmap_directory = MmapDirectory::open(tempdir_path).unwrap();
let mmap_directory = MmapDirectory::open(&tempdir_path).unwrap();
let mut managed_directory = ManagedDirectory::wrap(Box::new(mmap_directory)).unwrap();
let mut write = managed_directory.open_write(test_path1).unwrap();
write.write_all(&[0u8, 1u8]).unwrap();

6
src/directory/mmap_directory.rs
View File

@@ -341,7 +341,7 @@ impl Directory for MmapDirectory {
/// removed before the file is deleted.
fn delete(&self, path: &Path) -> result::Result<(), DeleteError> {
let full_path = self.resolve_path(path);
fs::remove_file(full_path).map_err(|e| {
fs::remove_file(&full_path).map_err(|e| {
if e.kind() == io::ErrorKind::NotFound {
DeleteError::FileDoesNotExist(path.to_owned())
} else {
@@ -395,7 +395,7 @@ impl Directory for MmapDirectory {
fn atomic_read(&self, path: &Path) -> Result<Vec<u8>, OpenReadError> {
let full_path = self.resolve_path(path);
let mut buffer = Vec::new();
match File::open(full_path) {
match File::open(&full_path) {
Ok(mut file) => {
file.read_to_end(&mut buffer).map_err(|io_error| {
OpenReadError::wrap_io_error(io_error, path.to_path_buf())
@@ -425,7 +425,7 @@ impl Directory for MmapDirectory {
let file: File = OpenOptions::new()
.write(true)
.create(true) //< if the file does not exist yet, create it.
.open(full_path)
.open(&full_path)
.map_err(LockError::wrap_io_error)?;
if lock.is_blocking {
file.lock_exclusive().map_err(LockError::wrap_io_error)?;

3
src/directory/mod.rs
View File

@@ -5,6 +5,7 @@ mod mmap_directory;
mod directory;
mod directory_lock;
mod file_slice;
mod file_watcher;
mod footer;
mod managed_directory;
@@ -19,13 +20,13 @@ mod composite_file;
use std::io::BufWriter;
use std::path::PathBuf;
pub use common::file_slice::{FileHandle, FileSlice};
pub use common::{AntiCallToken, TerminatingWrite};
pub use ownedbytes::OwnedBytes;
pub(crate) use self::composite_file::{CompositeFile, CompositeWrite};
pub use self::directory::{Directory, DirectoryClone, DirectoryLock};
pub use self::directory_lock::{Lock, INDEX_WRITER_LOCK, META_LOCK};
pub use self::file_slice::{FileHandle, FileSlice};
pub use self::ram_directory::RamDirectory;
pub use self::watch_event_router::{WatchCallback, WatchCallbackList, WatchHandle};

22
src/error.rs
View File

@@ -104,6 +104,28 @@ pub enum TantivyError {
InternalError(String),
}
#[cfg(feature = "quickwit")]
#[derive(Error, Debug)]
#[doc(hidden)]
pub enum AsyncIoError {
#[error("io::Error `{0}`")]
Io(#[from] io::Error),
#[error("Asynchronous API is unsupported by this directory")]
AsyncUnsupported,
}
#[cfg(feature = "quickwit")]
impl From<AsyncIoError> for TantivyError {
fn from(async_io_err: AsyncIoError) -> Self {
match async_io_err {
AsyncIoError::Io(io_err) => TantivyError::from(io_err),
AsyncIoError::AsyncUnsupported => {
TantivyError::SystemError(format!("{:?}", async_io_err))
}
}
}
}
impl From<io::Error> for TantivyError {
fn from(io_err: io::Error) -> TantivyError {
TantivyError::IoError(Arc::new(io_err))

7
src/fastfield/bytes/mod.rs
View File

@@ -6,7 +6,7 @@ pub use self::writer::BytesFastFieldWriter;
#[cfg(test)]
mod tests {
use crate::query::{EnableScoring, TermQuery};
use crate::query::TermQuery;
use crate::schema::{BytesOptions, IndexRecordOption, Schema, Value, FAST, INDEXED, STORED};
use crate::{DocAddress, DocSet, Index, Searcher, Term};
@@ -82,7 +82,7 @@ mod tests {
let field = searcher.schema().get_field("string_bytes").unwrap();
let term = Term::from_field_bytes(field, b"lucene".as_ref());
let term_query = TermQuery::new(term, IndexRecordOption::Basic);
let term_weight = term_query.specialized_weight(EnableScoring::Enabled(&searcher))?;
let term_weight = term_query.specialized_weight(&searcher, true)?;
let term_scorer = term_weight.specialized_scorer(searcher.segment_reader(0), 1.0)?;
assert_eq!(term_scorer.doc(), 0u32);
Ok(())
@@ -95,8 +95,7 @@ mod tests {
let field = searcher.schema().get_field("string_bytes").unwrap();
let term = Term::from_field_bytes(field, b"lucene".as_ref());
let term_query = TermQuery::new(term, IndexRecordOption::Basic);
let term_weight_err =
term_query.specialized_weight(EnableScoring::Disabled(searcher.schema()));
let term_weight_err = term_query.specialized_weight(&searcher, false);
assert!(matches!(
term_weight_err,
Err(crate::TantivyError::SchemaError(_))

37
src/fastfield/bytes/reader.rs
View File

@@ -1,9 +1,10 @@
use std::ops::Range;
use std::sync::Arc;
use fastfield_codecs::Column;
use crate::directory::{FileSlice, OwnedBytes};
use crate::fastfield::MultiValueIndex;
use crate::fastfield::MultiValueLength;
use crate::DocId;
/// Reader for byte array fast fields
@@ -18,7 +19,7 @@ use crate::DocId;
/// and the start index for the next document, and keeping the bytes in between.
#[derive(Clone)]
pub struct BytesFastFieldReader {
idx_reader: MultiValueIndex,
idx_reader: Arc<dyn Column<u64>>,
values: OwnedBytes,
}
@@ -28,31 +29,41 @@ impl BytesFastFieldReader {
values_file: FileSlice,
) -> crate::Result<BytesFastFieldReader> {
let values = values_file.read_bytes()?;
Ok(BytesFastFieldReader {
idx_reader: MultiValueIndex::new(idx_reader),
values,
})
Ok(BytesFastFieldReader { idx_reader, values })
}
/// returns the multivalue index
pub fn get_index_reader(&self) -> &MultiValueIndex {
&self.idx_reader
fn range(&self, doc: DocId) -> Range<u32> {
let start = self.idx_reader.get_val(doc) as u32;
let end = self.idx_reader.get_val(doc + 1) as u32;
start..end
}
/// Returns the bytes associated with the given `doc`
pub fn get_bytes(&self, doc: DocId) -> &[u8] {
let range = self.idx_reader.range(doc);
let range = self.range(doc);
&self.values.as_slice()[range.start as usize..range.end as usize]
}
/// Returns the length of the bytes associated with the given `doc`
pub fn num_bytes(&self, doc: DocId) -> u64 {
let range = self.idx_reader.range(doc);
let range = self.range(doc);
(range.end - range.start) as u64
}
/// Returns the overall number of bytes in this bytes fast field.
pub fn total_num_bytes(&self) -> u32 {
self.values.len() as u32
pub fn total_num_bytes(&self) -> u64 {
self.values.len() as u64
}
}
impl MultiValueLength for BytesFastFieldReader {
fn get_range(&self, doc_id: DocId) -> std::ops::Range<u32> {
self.range(doc_id)
}
fn get_len(&self, doc_id: DocId) -> u64 {
self.num_bytes(doc_id)
}
fn get_total_len(&self) -> u64 {
self.total_num_bytes()
}
}

4
src/fastfield/facet_reader.rs
View File

@@ -64,7 +64,9 @@ impl FacetReader {
facet_ord: TermOrdinal,
output: &mut Facet,
) -> crate::Result<()> {
let found_term = self.term_dict.ord_to_term(facet_ord, &mut self.buffer)?;
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[..])
.map_err(|utf8_err| DataCorruption::comment_only(utf8_err.to_string()))?;

41
src/fastfield/mod.rs
View File

@@ -27,16 +27,16 @@ pub use self::error::{FastFieldNotAvailableError, Result};
pub use self::facet_reader::FacetReader;
pub(crate) use self::multivalued::{get_fastfield_codecs_for_multivalue, MultivalueStartIndex};
pub use self::multivalued::{
MultiValueIndex, MultiValueU128FastFieldWriter, MultiValuedFastFieldReader,
MultiValuedFastFieldWriter, MultiValuedU128FastFieldReader,
MultiValueU128FastFieldWriter, MultiValuedFastFieldReader, MultiValuedFastFieldWriter,
MultiValuedU128FastFieldReader,
};
pub(crate) use self::readers::type_and_cardinality;
pub use self::readers::FastFieldReaders;
pub(crate) use self::readers::{type_and_cardinality, FastType};
pub use self::serializer::{Column, CompositeFastFieldSerializer};
use self::writer::unexpected_value;
pub use self::writer::{FastFieldsWriter, IntFastFieldWriter};
use crate::schema::{Type, Value};
use crate::DateTime;
use crate::{DateTime, DocId};
mod alive_bitset;
mod bytes;
@@ -47,6 +47,17 @@ mod readers;
mod serializer;
mod writer;
/// Trait for `BytesFastFieldReader` and `MultiValuedFastFieldReader` to return the length of data
/// for a doc_id
pub trait MultiValueLength {
/// returns the positions for a docid
fn get_range(&self, doc_id: DocId) -> std::ops::Range<u32>;
/// returns the num of values associated with a doc_id
fn get_len(&self, doc_id: DocId) -> u64;
/// returns the sum of num values for all doc_ids
fn get_total_len(&self) -> u64;
}
/// Trait for types that are allowed for fast fields:
/// (u64, i64 and f64, bool, DateTime).
pub trait FastValue:
@@ -207,7 +218,7 @@ mod tests {
serializer.close().unwrap();
}
let file = directory.open_read(path).unwrap();
assert_eq!(file.len(), 34);
assert_eq!(file.len(), 25);
let composite_file = CompositeFile::open(&file)?;
let fast_field_bytes = composite_file.open_read(*FIELD).unwrap().read_bytes()?;
let fast_field_reader = open::<u64>(fast_field_bytes)?;
@@ -256,7 +267,7 @@ mod tests {
serializer.close()?;
}
let file = directory.open_read(path)?;
assert_eq!(file.len(), 62);
assert_eq!(file.len(), 53);
{
let fast_fields_composite = CompositeFile::open(&file)?;
let data = fast_fields_composite
@@ -297,7 +308,7 @@ mod tests {
serializer.close().unwrap();
}
let file = directory.open_read(path).unwrap();
assert_eq!(file.len(), 35);
assert_eq!(file.len(), 26);
{
let fast_fields_composite = CompositeFile::open(&file).unwrap();
let data = fast_fields_composite
@@ -336,7 +347,7 @@ mod tests {
serializer.close().unwrap();
}
let file = directory.open_read(path).unwrap();
assert_eq!(file.len(), 80049);
assert_eq!(file.len(), 80040);
{
let fast_fields_composite = CompositeFile::open(&file)?;
let data = fast_fields_composite
@@ -378,7 +389,7 @@ mod tests {
serializer.close().unwrap();
}
let file = directory.open_read(path).unwrap();
assert_eq!(file.len(), 49_usize);
assert_eq!(file.len(), 40_usize);
{
let fast_fields_composite = CompositeFile::open(&file)?;
@@ -473,7 +484,7 @@ mod tests {
let fast_field_reader = open::<u64>(data)?;
for a in 0..n {
assert_eq!(fast_field_reader.get_val(a as u32), permutation[a]);
assert_eq!(fast_field_reader.get_val(a as u32), permutation[a as usize]);
}
}
Ok(())
@@ -822,7 +833,7 @@ mod tests {
serializer.close().unwrap();
}
let file = directory.open_read(path).unwrap();
assert_eq!(file.len(), 33);
assert_eq!(file.len(), 24);
let composite_file = CompositeFile::open(&file)?;
let data = composite_file.open_read(field).unwrap().read_bytes()?;
let fast_field_reader = open::<bool>(data)?;
@@ -860,7 +871,7 @@ mod tests {
serializer.close().unwrap();
}
let file = directory.open_read(path).unwrap();
assert_eq!(file.len(), 45);
assert_eq!(file.len(), 36);
let composite_file = CompositeFile::open(&file)?;
let data = composite_file.open_read(field).unwrap().read_bytes()?;
let fast_field_reader = open::<bool>(data)?;
@@ -892,7 +903,7 @@ mod tests {
}
let file = directory.open_read(path).unwrap();
let composite_file = CompositeFile::open(&file)?;
assert_eq!(file.len(), 32);
assert_eq!(file.len(), 23);
let data = composite_file.open_read(field).unwrap().read_bytes()?;
let fast_field_reader = open::<bool>(data)?;
assert_eq!(fast_field_reader.get_val(0), false);
@@ -926,10 +937,10 @@ mod tests {
pub fn test_gcd_date() -> crate::Result<()> {
let size_prec_sec =
test_gcd_date_with_codec(FastFieldCodecType::Bitpacked, DatePrecision::Seconds)?;
assert_eq!(size_prec_sec, 5 + 4 + 28 + (1_000 * 13) / 8); // 13 bits per val = ceil(log_2(number of seconds in 2hours);
assert_eq!(size_prec_sec, 28 + (1_000 * 13) / 8); // 13 bits per val = ceil(log_2(number of seconds in 2hours);
let size_prec_micro =
test_gcd_date_with_codec(FastFieldCodecType::Bitpacked, DatePrecision::Microseconds)?;
assert_eq!(size_prec_micro, 5 + 4 + 26 + (1_000 * 33) / 8); // 33 bits per val = ceil(log_2(number of microsecsseconds in 2hours);
assert_eq!(size_prec_micro, 26 + (1_000 * 33) / 8); // 33 bits per val = ceil(log_2(number of microsecsseconds in 2hours);
Ok(())
}

148
src/fastfield/multivalued/index.rs
View File

@@ -1,148 +0,0 @@
use std::ops::Range;
use std::sync::Arc;
use fastfield_codecs::Column;
use crate::DocId;
#[derive(Clone)]
/// Index to resolve value range for given doc_id.
/// Starts at 0.
pub struct MultiValueIndex {
idx: Arc<dyn Column<u64>>,
}
impl MultiValueIndex {
pub(crate) fn new(idx: Arc<dyn Column<u64>>) -> Self {
Self { idx }
}
/// Returns `[start, end)`, such that the values associated with
/// the given document are `start..end`.
#[inline]
pub(crate) fn range(&self, doc: DocId) -> Range<u32> {
let start = self.idx.get_val(doc) as u32;
let end = self.idx.get_val(doc + 1) as u32;
start..end
}
/// Given a range of documents, returns the Range of value offsets fo
/// these documents.
///
/// For instance, `given start_doc..end_doc`,
/// if we assume Document #start_doc end #end_doc both
/// have values, this function returns `start..end`
/// such that `value_column.get(start_doc)` is the first value of
/// `start_doc` (well, if there is one), and `value_column.get(end_doc - 1)`
/// is the last value of `end_doc`.
///
/// The passed end range is allowed to be out of bounds, in which case
/// it will be clipped to make it valid.
#[inline]
pub(crate) fn docid_range_to_position_range(&self, range: Range<DocId>) -> Range<u32> {
let end_docid = range.end.min(self.num_docs() - 1) + 1;
let start_docid = range.start.min(end_docid);
let start = self.idx.get_val(start_docid) as u32;
let end = self.idx.get_val(end_docid) as u32;
assert!(start <= end);
start..end
}
/// returns the num of values associated with a doc_id
pub(crate) fn num_vals_for_doc(&self, doc: DocId) -> u32 {
let range = self.range(doc);
range.end - range.start
}
/// Returns the overall number of values in this field.
#[inline]
pub fn total_num_vals(&self) -> u32 {
self.idx.max_value() as u32
}
/// Returns the number of documents in the index.
#[inline]
pub fn num_docs(&self) -> u32 {
self.idx.num_vals() - 1
}
/// Converts a list of positions of values in a 1:n index to the corresponding list of DocIds.
/// Positions are converted inplace to docids.
///
/// Since there is no index for value pos -> docid, but docid -> value pos range, we scan the
/// index.
///
/// Correctness: positions needs to be sorted. idx_reader needs to contain monotonically
/// increasing positions.
///
///
/// TODO: Instead of a linear scan we can employ a exponential search into binary search to
/// match a docid to its value position.
pub(crate) fn positions_to_docids(&self, doc_id_range: Range<u32>, positions: &mut Vec<u32>) {
if positions.is_empty() {
return;
}
let mut cur_doc = doc_id_range.start;
let mut last_doc = None;
assert!(self.idx.get_val(doc_id_range.start) as u32 <= positions[0]);
let mut write_doc_pos = 0;
for i in 0..positions.len() {
let pos = positions[i];
loop {
let end = self.idx.get_val(cur_doc + 1) as u32;
if end > pos {
positions[write_doc_pos] = cur_doc;
write_doc_pos += if last_doc == Some(cur_doc) { 0 } else { 1 };
last_doc = Some(cur_doc);
break;
}
cur_doc += 1;
}
}
positions.truncate(write_doc_pos);
}
}
#[cfg(test)]
mod tests {
use std::ops::Range;
use std::sync::Arc;
use fastfield_codecs::IterColumn;
use crate::fastfield::MultiValueIndex;
fn index_to_pos_helper(
index: &MultiValueIndex,
doc_id_range: Range<u32>,
positions: &[u32],
) -> Vec<u32> {
let mut positions = positions.to_vec();
index.positions_to_docids(doc_id_range, &mut positions);
positions
}
#[test]
fn test_positions_to_docid() {
let offsets = vec![0, 10, 12, 15, 22, 23]; // docid values are [0..10, 10..12, 12..15, etc.]
let column = IterColumn::from(offsets.into_iter());
let index = MultiValueIndex::new(Arc::new(column));
assert_eq!(index.num_docs(), 5);
{
let positions = vec![10u32, 11, 15, 20, 21, 22];
assert_eq!(index_to_pos_helper(&index, 0..5, &positions), vec![1, 3, 4]);
assert_eq!(index_to_pos_helper(&index, 1..5, &positions), vec![1, 3, 4]);
assert_eq!(index_to_pos_helper(&index, 0..5, &[9]), vec![0]);
assert_eq!(index_to_pos_helper(&index, 1..5, &[10]), vec![1]);
assert_eq!(index_to_pos_helper(&index, 1..5, &[11]), vec![1]);
assert_eq!(index_to_pos_helper(&index, 2..5, &[12]), vec![2]);
assert_eq!(index_to_pos_helper(&index, 2..5, &[12, 14]), vec![2]);
assert_eq!(index_to_pos_helper(&index, 2..5, &[12, 14, 15]), vec![2, 3]);
}
}
}

2
src/fastfield/multivalued/mod.rs
View File

@@ -1,9 +1,7 @@
mod index;
mod reader;
mod writer;
use fastfield_codecs::FastFieldCodecType;
pub use index::MultiValueIndex;
pub use self::reader::{MultiValuedFastFieldReader, MultiValuedU128FastFieldReader};
pub(crate) use self::writer::MultivalueStartIndex;

174
src/fastfield/multivalued/reader.rs
View File

@@ -3,8 +3,7 @@ use std::sync::Arc;
use fastfield_codecs::{Column, MonotonicallyMappableToU128};
use super::MultiValueIndex;
use crate::fastfield::FastValue;
use crate::fastfield::{FastValue, MultiValueLength};
use crate::DocId;
/// Reader for a multivalued `u64` fast field.
@@ -14,10 +13,9 @@ use crate::DocId;
/// The `vals_reader` will access the concatenated list of all
/// values for all reader.
/// The `idx_reader` associated, for each document, the index of its first value.
/// Stores the start position for each document.
#[derive(Clone)]
pub struct MultiValuedFastFieldReader<Item: FastValue> {
idx_reader: MultiValueIndex,
idx_reader: Arc<dyn Column<u64>>,
vals_reader: Arc<dyn Column<Item>>,
}
@@ -27,11 +25,20 @@ impl<Item: FastValue> MultiValuedFastFieldReader<Item> {
vals_reader: Arc<dyn Column<Item>>,
) -> MultiValuedFastFieldReader<Item> {
MultiValuedFastFieldReader {
idx_reader: MultiValueIndex::new(idx_reader),
idx_reader,
vals_reader,
}
}
/// Returns `[start, end)`, such that the values associated with
/// the given document are `start..end`.
#[inline]
fn range(&self, doc: DocId) -> Range<u32> {
let start = self.idx_reader.get_val(doc) as u32;
let end = self.idx_reader.get_val(doc + 1) as u32;
start..end
}
/// Returns the array of values associated with the given `doc`.
#[inline]
fn get_vals_for_range(&self, range: Range<u32>, vals: &mut Vec<Item>) {
@@ -44,15 +51,10 @@ impl<Item: FastValue> MultiValuedFastFieldReader<Item> {
/// Returns the array of values associated with the given `doc`.
#[inline]
pub fn get_vals(&self, doc: DocId, vals: &mut Vec<Item>) {
let range = self.idx_reader.range(doc);
let range = self.range(doc);
self.get_vals_for_range(range, vals);
}
/// returns the multivalue index
pub fn get_index_reader(&self) -> &MultiValueIndex {
&self.idx_reader
}
/// Returns the minimum value for this fast field.
///
/// The min value does not take in account of possible
@@ -73,14 +75,28 @@ impl<Item: FastValue> MultiValuedFastFieldReader<Item> {
/// Returns the number of values associated with the document `DocId`.
#[inline]
pub fn num_vals(&self, doc: DocId) -> u32 {
self.idx_reader.num_vals_for_doc(doc)
pub fn num_vals(&self, doc: DocId) -> usize {
let range = self.range(doc);
(range.end - range.start) as usize
}
/// Returns the overall number of values in this field.
/// Returns the overall number of values in this field .
#[inline]
pub fn total_num_vals(&self) -> u32 {
self.idx_reader.total_num_vals()
pub fn total_num_vals(&self) -> u64 {
self.idx_reader.max_value()
}
}
impl<Item: FastValue> MultiValueLength for MultiValuedFastFieldReader<Item> {
fn get_range(&self, doc_id: DocId) -> Range<u32> {
self.range(doc_id)
}
fn get_len(&self, doc_id: DocId) -> u64 {
self.num_vals(doc_id) as u64
}
fn get_total_len(&self) -> u64 {
self.total_num_vals() as u64
}
}
@@ -93,7 +109,7 @@ impl<Item: FastValue> MultiValuedFastFieldReader<Item> {
/// The `idx_reader` associated, for each document, the index of its first value.
#[derive(Clone)]
pub struct MultiValuedU128FastFieldReader<T: MonotonicallyMappableToU128> {
idx_reader: MultiValueIndex,
idx_reader: Arc<dyn Column<u64>>,
vals_reader: Arc<dyn Column<T>>,
}
@@ -103,15 +119,24 @@ impl<T: MonotonicallyMappableToU128> MultiValuedU128FastFieldReader<T> {
vals_reader: Arc<dyn Column<T>>,
) -> MultiValuedU128FastFieldReader<T> {
Self {
idx_reader: MultiValueIndex::new(idx_reader),
idx_reader,
vals_reader,
}
}
/// Returns `[start, end)`, such that the values associated
/// to the given document are `start..end`.
#[inline]
fn range(&self, doc: DocId) -> Range<u32> {
let start = self.idx_reader.get_val(doc) as u32;
let end = self.idx_reader.get_val(doc + 1) as u32;
start..end
}
/// Returns the array of values associated to the given `doc`.
#[inline]
pub fn get_first_val(&self, doc: DocId) -> Option<T> {
let range = self.idx_reader.range(doc);
let range = self.range(doc);
if range.is_empty() {
return None;
}
@@ -127,18 +152,26 @@ impl<T: MonotonicallyMappableToU128> MultiValuedU128FastFieldReader<T> {
.get_range(range.start as u64, &mut vals[..]);
}
/// Returns the index reader
pub fn get_index_reader(&self) -> &MultiValueIndex {
&self.idx_reader
}
/// Returns the array of values associated to the given `doc`.
#[inline]
pub fn get_vals(&self, doc: DocId, vals: &mut Vec<T>) {
let range = self.idx_reader.range(doc);
let range = self.range(doc);
self.get_vals_for_range(range, vals);
}
/// Returns all docids which are in the provided value range
pub fn get_positions_for_value_range(
&self,
value_range: RangeInclusive<T>,
doc_id_range: Range<u32>,
) -> Vec<DocId> {
let mut positions = Vec::new(); // TODO replace
self.vals_reader
.get_positions_for_value_range(value_range, doc_id_range, &mut positions);
positions_to_docids(&positions, self.idx_reader.as_ref())
}
/// Iterates over all elements in the fast field
pub fn iter(&self) -> impl Iterator<Item = T> + '_ {
self.vals_reader.iter()
@@ -164,44 +197,85 @@ impl<T: MonotonicallyMappableToU128> MultiValuedU128FastFieldReader<T> {
/// Returns the number of values associated with the document `DocId`.
#[inline]
pub fn num_vals(&self, doc: DocId) -> u32 {
self.idx_reader.num_vals_for_doc(doc)
pub fn num_vals(&self, doc: DocId) -> usize {
let range = self.range(doc);
(range.end - range.start) as usize
}
/// Returns the overall number of values in this field. It does not include deletes.
/// Returns the overall number of values in this field.
#[inline]
pub fn total_num_vals(&self) -> u32 {
assert_eq!(
self.vals_reader.num_vals(),
self.get_index_reader().total_num_vals()
);
self.idx_reader.total_num_vals()
pub fn total_num_vals(&self) -> u64 {
self.idx_reader.max_value()
}
}
impl<T: MonotonicallyMappableToU128> MultiValueLength for MultiValuedU128FastFieldReader<T> {
fn get_range(&self, doc_id: DocId) -> std::ops::Range<u32> {
self.range(doc_id)
}
fn get_len(&self, doc_id: DocId) -> u64 {
self.num_vals(doc_id) as u64
}
fn get_total_len(&self) -> u64 {
self.total_num_vals() as u64
}
}
/// Converts a list of positions of values in a 1:n index to the corresponding list of DocIds.
///
/// Since there is no index for value pos -> docid, but docid -> value pos range, we scan the index.
///
/// Correctness: positions needs to be sorted. idx_reader needs to contain monotonically increasing
/// positions.
///
/// TODO: Instead of a linear scan we can employ a expotential search into binary search to match a
/// docid to its value position.
fn positions_to_docids<C: Column + ?Sized>(positions: &[u32], idx_reader: &C) -> Vec<DocId> {
let mut docs = vec![];
let mut cur_doc = 0u32;
let mut last_doc = None;
for pos in positions {
loop {
let end = idx_reader.get_val(cur_doc + 1) as u32;
if end > *pos {
// avoid duplicates
if Some(cur_doc) == last_doc {
break;
}
docs.push(cur_doc);
last_doc = Some(cur_doc);
break;
}
cur_doc += 1;
}
}
/// Returns the docids matching given doc_id_range and value_range.
#[inline]
pub fn get_docids_for_value_range(
&self,
value_range: RangeInclusive<T>,
doc_id_range: Range<u32>,
positions: &mut Vec<u32>,
) {
let position_range = self
.get_index_reader()
.docid_range_to_position_range(doc_id_range.clone());
self.vals_reader
.get_docids_for_value_range(value_range, position_range, positions);
self.idx_reader.positions_to_docids(doc_id_range, positions);
}
docs
}
#[cfg(test)]
mod tests {
use fastfield_codecs::VecColumn;
use crate::core::Index;
use crate::fastfield::multivalued::reader::positions_to_docids;
use crate::schema::{Cardinality, Facet, FacetOptions, NumericOptions, Schema};
#[test]
fn test_positions_to_docid() {
let positions = vec![10u32, 11, 15, 20, 21, 22];
let offsets = vec![0, 10, 12, 15, 22, 23];
{
let column = VecColumn::from(&offsets);
let docids = positions_to_docids(&positions, &column);
assert_eq!(docids, vec![1, 3, 4]);
}
}
#[test]
fn test_multifastfield_reader() -> crate::Result<()> {
let mut schema_builder = Schema::builder();

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

@@ -264,7 +264,7 @@ fn iter_remapped_multivalue_index<'a, C: Column>(
std::iter::once(0).chain(doc_id_map.iter_old_doc_ids().map(move |old_doc| {
let num_vals_for_doc = column.get_val(old_doc + 1) - column.get_val(old_doc);
offset += num_vals_for_doc;
offset
offset as u64
}))
}

14
src/fastfield/writer.rs
View File

@@ -360,10 +360,20 @@ impl U128FastFieldWriter {
.map(|idx| self.vals[idx as usize])
};
serializer.create_u128_fast_field_with_idx(self.field, iter_gen, self.val_count, 0)?;
serializer.create_u128_fast_field_with_idx(
self.field,
iter_gen,
self.val_count as u32,
0,
)?;
} else {
let iter_gen = || self.vals.iter().cloned();
serializer.create_u128_fast_field_with_idx(self.field, iter_gen, self.val_count, 0)?;
serializer.create_u128_fast_field_with_idx(
self.field,
iter_gen,
self.val_count as u32,
0,
)?;
}
Ok(())

6
src/fieldnorm/mod.rs
View File

@@ -34,7 +34,7 @@ mod tests {
use crate::directory::{CompositeFile, Directory, RamDirectory, WritePtr};
use crate::fieldnorm::{FieldNormReader, FieldNormsSerializer, FieldNormsWriter};
use crate::query::{EnableScoring, Query, TermQuery};
use crate::query::{Query, TermQuery};
use crate::schema::{
Field, IndexRecordOption, Schema, TextFieldIndexing, TextOptions, STORED, TEXT,
};
@@ -112,7 +112,7 @@ mod tests {
Term::from_field_text(text, "hello"),
IndexRecordOption::WithFreqs,
);
let weight = query.weight(EnableScoring::Enabled(&searcher))?;
let weight = query.weight(&searcher, true)?;
let mut scorer = weight.scorer(searcher.segment_reader(0), 1.0f32)?;
assert_eq!(scorer.doc(), 0);
assert!((scorer.score() - 0.22920431).abs() < 0.001f32);
@@ -141,7 +141,7 @@ mod tests {
Term::from_field_text(text, "hello"),
IndexRecordOption::WithFreqs,
);
let weight = query.weight(EnableScoring::Enabled(&searcher))?;
let weight = query.weight(&searcher, true)?;
let mut scorer = weight.scorer(searcher.segment_reader(0), 1.0f32)?;
assert_eq!(scorer.doc(), 0);
assert!((scorer.score() - 0.22920431).abs() < 0.001f32);

2
src/fieldnorm/writer.rs
View File

@@ -9,7 +9,7 @@ use crate::DocId;
/// The `FieldNormsWriter` is in charge of tracking the fieldnorm byte
/// of each document for each field with field norms.
///
/// `FieldNormsWriter` stores a `Vec<u8>` for each tracked field, using a
/// `FieldNormsWriter` stores a Vec<u8> for each tracked field, using a
/// byte per document per field.
pub struct FieldNormsWriter {
fieldnorms_buffers: Vec<Option<Vec<u8>>>,

4
src/indexer/demuxer.rs
View File

@@ -252,8 +252,8 @@ mod tests {
&demux_mapping,
target_settings,
vec![
Box::<RamDirectory>::default(),
Box::<RamDirectory>::default(),
Box::new(RamDirectory::default()),
Box::new(RamDirectory::default()),
],
)?;

179
src/indexer/index_writer.rs
View File

@@ -19,9 +19,9 @@ use crate::indexer::index_writer_status::IndexWriterStatus;
use crate::indexer::operation::DeleteOperation;
use crate::indexer::stamper::Stamper;
use crate::indexer::{MergePolicy, SegmentEntry, SegmentWriter};
use crate::query::{EnableScoring, Query, TermQuery};
use crate::query::{Query, TermQuery};
use crate::schema::{Document, IndexRecordOption, Term};
use crate::{FutureResult, Opstamp};
use crate::{FutureResult, IndexReader, Opstamp};
// Size of the margin for the `memory_arena`. A segment is closed when the remaining memory
// in the `memory_arena` goes below MARGIN_IN_BYTES.
@@ -57,6 +57,7 @@ pub struct IndexWriter {
_directory_lock: Option<DirectoryLock>,
index: Index,
index_reader: IndexReader,
memory_arena_in_bytes_per_thread: usize,
@@ -94,7 +95,7 @@ fn compute_deleted_bitset(
// document that were inserted before it.
delete_op
.target
.for_each_no_score(segment_reader, &mut |doc_matching_delete_query| {
.for_each(segment_reader, &mut |doc_matching_delete_query, _| {
if doc_opstamps.is_deleted(doc_matching_delete_query, delete_op.opstamp) {
alive_bitset.remove(doc_matching_delete_query);
might_have_changed = true;
@@ -152,7 +153,7 @@ pub(crate) fn advance_deletes(
let num_deleted_docs = max_doc - num_alive_docs;
if num_deleted_docs > num_deleted_docs_before {
// There are new deletes. We need to write a new delete file.
segment = segment.with_delete_meta(num_deleted_docs, target_opstamp);
segment = segment.with_delete_meta(num_deleted_docs as u32, target_opstamp);
let mut alive_doc_file = segment.open_write(SegmentComponent::Delete)?;
write_alive_bitset(&alive_bitset, &mut alive_doc_file)?;
alive_doc_file.terminate()?;
@@ -297,6 +298,8 @@ impl IndexWriter {
memory_arena_in_bytes_per_thread,
index: index.clone(),
index_reader: index.reader()?,
index_writer_status: IndexWriterStatus::from(document_receiver),
operation_sender: document_sender,
@@ -678,7 +681,8 @@ impl IndexWriter {
/// only after calling `commit()`.
#[doc(hidden)]
pub fn delete_query(&self, query: Box<dyn Query>) -> crate::Result<Opstamp> {
let weight = query.weight(EnableScoring::Disabled(&self.index.schema()))?;
let weight = query.weight(&self.index_reader.searcher(), false)?;
let opstamp = self.stamper.stamp();
let delete_operation = DeleteOperation {
opstamp,
@@ -759,7 +763,8 @@ impl IndexWriter {
match user_op {
UserOperation::Delete(term) => {
let query = TermQuery::new(term, IndexRecordOption::Basic);
let weight = query.weight(EnableScoring::Disabled(&self.index.schema()))?;
let weight = query.weight(&self.index_reader.searcher(), false)?;
let delete_operation = DeleteOperation {
opstamp,
target: weight,
@@ -984,7 +989,7 @@ mod tests {
"LogMergePolicy { min_num_segments: 8, max_docs_before_merge: 10000000, \
min_layer_size: 10000, level_log_size: 0.75, del_docs_ratio_before_merge: 1.0 }"
);
let merge_policy = Box::<NoMergePolicy>::default();
let merge_policy = Box::new(NoMergePolicy::default());
index_writer.set_merge_policy(merge_policy);
assert_eq!(
format!("{:?}", index_writer.get_merge_policy()),
@@ -1586,25 +1591,6 @@ mod tests {
(existing_ids, deleted_ids)
}
fn get_id_list(ops: &[IndexingOp]) -> Vec<u64> {
let mut id_list = Vec::new();
for &op in ops {
match op {
IndexingOp::AddDoc { id } => {
id_list.push(id);
}
IndexingOp::DeleteDoc { id } => {
id_list.retain(|el| *el != id);
}
IndexingOp::DeleteDocQuery { id } => {
id_list.retain(|el| *el != id);
}
_ => {}
}
}
id_list
}
fn test_operation_strategy(
ops: &[IndexingOp],
sort_index: bool,
@@ -1614,9 +1600,7 @@ mod tests {
let ip_field = schema_builder.add_ip_addr_field("ip", FAST | INDEXED | STORED);
let ips_field = schema_builder.add_ip_addr_field(
"ips",
IpAddrOptions::default()
.set_fast(Cardinality::MultiValues)
.set_indexed(),
IpAddrOptions::default().set_fast(Cardinality::MultiValues),
);
let id_field = schema_builder.add_u64_field("id", FAST | INDEXED | STORED);
let i64_field = schema_builder.add_i64_field("i64", INDEXED);
@@ -1681,13 +1665,11 @@ mod tests {
// rotate right
let multi_text_field_text3 = "test3 test1 test2 test3 test1 test2";
let ip_from_id = |id| Ipv6Addr::from_u128(id as u128);
for &op in ops {
match op {
IndexingOp::AddDoc { id } => {
let facet = Facet::from(&("/cola/".to_string() + &id.to_string()));
let ip = ip_from_id(id);
let ip_from_id = Ipv6Addr::from_u128(id as u128);
if !ip_exists(id) {
// every 3rd doc has no ip field
@@ -1711,9 +1693,9 @@ mod tests {
} else {
index_writer.add_document(doc!(id_field=>id,
bytes_field => id.to_le_bytes().as_slice(),
ip_field => ip,
ips_field => ip,
ips_field => ip,
ip_field => ip_from_id,
ips_field => ip_from_id,
ips_field => ip_from_id,
multi_numbers=> id,
multi_numbers => id,
bool_field => (id % 2u64) != 0,
@@ -1756,7 +1738,6 @@ mod tests {
index_writer.commit()?;
let searcher = index.reader()?.searcher();
let num_segments_before_merge = searcher.segment_readers().len();
if force_end_merge {
index_writer.wait_merging_threads()?;
let mut index_writer = index.writer_for_tests()?;
@@ -1768,7 +1749,6 @@ mod tests {
assert!(index_writer.wait_merging_threads().is_ok());
}
}
let num_segments_after_merge = searcher.segment_readers().len();
old_reader.reload()?;
let old_searcher = old_reader.searcher();
@@ -1796,25 +1776,9 @@ mod tests {
.collect();
let (expected_ids_and_num_occurrences, deleted_ids) = expected_ids(ops);
let id_list = get_id_list(ops);
// multivalue fast field content
let mut all_ips = Vec::new();
let mut num_ips = 0;
for segment_reader in searcher.segment_readers().iter() {
let ip_reader = segment_reader.fast_fields().ip_addrs(ips_field).unwrap();
for doc in segment_reader.doc_ids_alive() {
let mut vals = vec![];
ip_reader.get_vals(doc, &mut vals);
all_ips.extend_from_slice(&vals);
}
num_ips += ip_reader.total_num_vals();
}
let num_docs_expected = expected_ids_and_num_occurrences
.values()
.map(|id_occurrences| *id_occurrences as usize)
.iter()
.map(|(_, id_occurrences)| *id_occurrences as usize)
.sum::<usize>();
assert_eq!(searcher.num_docs() as usize, num_docs_expected);
assert_eq!(old_searcher.num_docs() as usize, num_docs_expected);
@@ -1833,30 +1797,6 @@ mod tests {
.collect::<HashSet<_>>()
);
if force_end_merge && num_segments_before_merge > 1 && num_segments_after_merge == 1 {
let mut expected_multi_ips: Vec<_> = id_list
.iter()
.filter(|id| ip_exists(**id))
.flat_map(|id| vec![ip_from_id(*id), ip_from_id(*id)])
.collect();
assert_eq!(num_ips, expected_multi_ips.len() as u32);
expected_multi_ips.sort();
all_ips.sort();
assert_eq!(expected_multi_ips, all_ips);
// Test fastfield num_docs
let num_docs: usize = searcher
.segment_readers()
.iter()
.map(|segment_reader| {
let ff_reader = segment_reader.fast_fields().ip_addrs(ips_field).unwrap();
ff_reader.get_index_reader().num_docs() as usize
})
.sum();
assert_eq!(num_docs, num_docs_expected);
}
// Load all ips addr
let ips: HashSet<Ipv6Addr> = searcher
.segment_readers()
@@ -2060,51 +2000,6 @@ mod tests {
assert_eq!(do_search_ip_field(&format!("\"{}\"", ip_addr)), count);
}
}
// assert data is like expected
//
for (existing_id, count) in expected_ids_and_num_occurrences.iter().take(10) {
let (existing_id, count) = (*existing_id, *count);
if !ip_exists(existing_id) {
continue;
}
let gen_query_inclusive = |field: &str, from: Ipv6Addr, to: Ipv6Addr| {
format!("{}:[{} TO {}]", field, &from.to_string(), &to.to_string())
};
let ip = ip_from_id(existing_id);
let do_search_ip_field = |term: &str| do_search(term, ip_field).len() as u64;
// Range query on single value field
// let query = gen_query_inclusive("ip", ip, ip);
// assert_eq!(do_search_ip_field(&query), count);
// Range query on multi value field
let query = gen_query_inclusive("ips", ip, ip);
assert_eq!(do_search_ip_field(&query), count);
}
// ip range query on fast field
//
for (existing_id, count) in expected_ids_and_num_occurrences.iter().take(10) {
let (existing_id, count) = (*existing_id, *count);
if !ip_exists(existing_id) {
continue;
}
let gen_query_inclusive = |field: &str, from: Ipv6Addr, to: Ipv6Addr| {
format!("{}:[{} TO {}]", field, &from.to_string(), &to.to_string())
};
let ip = ip_from_id(existing_id);
let do_search_ip_field = |term: &str| do_search(term, ip_field).len() as u64;
// Range query on single value field
// let query = gen_query_inclusive("ip", ip, ip);
// assert_eq!(do_search_ip_field(&query), count);
// Range query on multi value field
let query = gen_query_inclusive("ips", ip, ip);
assert_eq!(do_search_ip_field(&query), count);
}
// test facets
for segment_reader in searcher.segment_readers().iter() {
let mut facet_reader = segment_reader.facet_reader(facet_field).unwrap();
@@ -2126,40 +2021,6 @@ mod tests {
Ok(())
}
#[test]
fn test_ip_range_query_multivalue_bug() {
assert!(test_operation_strategy(
&[
IndexingOp::AddDoc { id: 2 },
IndexingOp::Commit,
IndexingOp::AddDoc { id: 1 },
IndexingOp::AddDoc { id: 1 },
IndexingOp::Commit,
IndexingOp::Merge
],
true,
false
)
.is_ok());
}
#[test]
fn test_ff_num_ips_regression() {
assert!(test_operation_strategy(
&[
IndexingOp::AddDoc { id: 13 },
IndexingOp::AddDoc { id: 1 },
IndexingOp::Commit,
IndexingOp::DeleteDocQuery { id: 13 },
IndexingOp::AddDoc { id: 1 },
IndexingOp::Commit,
],
false,
true
)
.is_ok());
}
#[test]
fn test_minimal() {
assert!(test_operation_strategy(
@@ -2169,7 +2030,7 @@ mod tests {
IndexingOp::DeleteDoc { id: 13 }
],
true,
true
false
)
.is_ok());

159
src/indexer/json_term_writer.rs
View File

@@ -67,12 +67,11 @@ pub(crate) fn index_json_values<'a>(
doc: DocId,
json_values: impl Iterator<Item = crate::Result<&'a serde_json::Map<String, serde_json::Value>>>,
text_analyzer: &TextAnalyzer,
expand_dots_enabled: bool,
term_buffer: &mut Term,
postings_writer: &mut dyn PostingsWriter,
ctx: &mut IndexingContext,
) -> crate::Result<()> {
let mut json_term_writer = JsonTermWriter::wrap(term_buffer, expand_dots_enabled);
let mut json_term_writer = JsonTermWriter::wrap(term_buffer);
let mut positions_per_path: IndexingPositionsPerPath = Default::default();
for json_value_res in json_values {
let json_value = json_value_res?;
@@ -260,65 +259,29 @@ pub(crate) fn set_string_and_get_terms(
pub struct JsonTermWriter<'a> {
term_buffer: &'a mut Term,
path_stack: Vec<usize>,
expand_dots_enabled: bool,
}
/// Splits a json path supplied to the query parser in such a way that
/// `.` can be escaped.
///
/// In other words,
/// - `k8s.node` ends up as `["k8s", "node"]`.
/// - `k8s\.node` ends up as `["k8s.node"]`.
fn split_json_path(json_path: &str) -> Vec<String> {
let mut escaped_state: bool = false;
let mut json_path_segments = Vec::new();
let mut buffer = String::new();
for ch in json_path.chars() {
if escaped_state {
buffer.push(ch);
escaped_state = false;
continue;
}
match ch {
'\\' => {
escaped_state = true;
}
'.' => {
let new_segment = std::mem::take(&mut buffer);
json_path_segments.push(new_segment);
}
_ => {
buffer.push(ch);
}
}
}
json_path_segments.push(buffer);
json_path_segments
}
impl<'a> JsonTermWriter<'a> {
pub fn from_field_and_json_path(
field: Field,
json_path: &str,
expand_dots_enabled: bool,
term_buffer: &'a mut Term,
) -> Self {
term_buffer.set_field_and_type(field, Type::Json);
let mut json_term_writer = Self::wrap(term_buffer, expand_dots_enabled);
for segment in split_json_path(json_path) {
json_term_writer.push_path_segment(&segment);
let mut json_term_writer = Self::wrap(term_buffer);
for segment in json_path.split('.') {
json_term_writer.push_path_segment(segment);
}
json_term_writer
}
pub fn wrap(term_buffer: &'a mut Term, expand_dots_enabled: bool) -> Self {
pub fn wrap(term_buffer: &'a mut Term) -> Self {
term_buffer.clear_with_type(Type::Json);
let mut path_stack = Vec::with_capacity(10);
path_stack.push(0);
Self {
term_buffer,
path_stack,
expand_dots_enabled,
}
}
@@ -340,24 +303,11 @@ impl<'a> JsonTermWriter<'a> {
self.trim_to_end_of_path();
let buffer = self.term_buffer.value_bytes_mut();
let buffer_len = buffer.len();
if self.path_stack.len() > 1 {
buffer[buffer_len - 1] = JSON_PATH_SEGMENT_SEP;
}
if self.expand_dots_enabled && segment.as_bytes().contains(&b'.') {
// We need to replace `.` by JSON_PATH_SEGMENT_SEP.
self.term_buffer
.append_bytes(segment.as_bytes())
.iter_mut()
.for_each(|byte| {
if *byte == b'.' {
*byte = JSON_PATH_SEGMENT_SEP;
}
});
} else {
self.term_buffer.append_bytes(segment.as_bytes());
}
self.term_buffer.push_byte(JSON_PATH_SEGMENT_SEP);
self.term_buffer.append_bytes(segment.as_bytes());
self.term_buffer.append_bytes(&[JSON_PATH_SEGMENT_SEP]);
self.path_stack.push(self.term_buffer.len_bytes());
}
@@ -400,7 +350,7 @@ impl<'a> JsonTermWriter<'a> {
#[cfg(test)]
mod tests {
use super::{split_json_path, JsonTermWriter};
use super::JsonTermWriter;
use crate::schema::{Field, Type};
use crate::Term;
@@ -408,7 +358,7 @@ mod tests {
fn test_json_writer() {
let field = Field::from_field_id(1);
let mut term = Term::with_type_and_field(Type::Json, field);
let mut json_writer = JsonTermWriter::wrap(&mut term, false);
let mut json_writer = JsonTermWriter::wrap(&mut term);
json_writer.push_path_segment("attributes");
json_writer.push_path_segment("color");
json_writer.set_str("red");
@@ -442,7 +392,7 @@ mod tests {
fn test_string_term() {
let field = Field::from_field_id(1);
let mut term = Term::with_type_and_field(Type::Json, field);
let mut json_writer = JsonTermWriter::wrap(&mut term, false);
let mut json_writer = JsonTermWriter::wrap(&mut term);
json_writer.push_path_segment("color");
json_writer.set_str("red");
assert_eq!(
@@ -455,7 +405,7 @@ mod tests {
fn test_i64_term() {
let field = Field::from_field_id(1);
let mut term = Term::with_type_and_field(Type::Json, field);
let mut json_writer = JsonTermWriter::wrap(&mut term, false);
let mut json_writer = JsonTermWriter::wrap(&mut term);
json_writer.push_path_segment("color");
json_writer.set_fast_value(-4i64);
assert_eq!(
@@ -468,7 +418,7 @@ mod tests {
fn test_u64_term() {
let field = Field::from_field_id(1);
let mut term = Term::with_type_and_field(Type::Json, field);
let mut json_writer = JsonTermWriter::wrap(&mut term, false);
let mut json_writer = JsonTermWriter::wrap(&mut term);
json_writer.push_path_segment("color");
json_writer.set_fast_value(4u64);
assert_eq!(
@@ -481,7 +431,7 @@ mod tests {
fn test_f64_term() {
let field = Field::from_field_id(1);
let mut term = Term::with_type_and_field(Type::Json, field);
let mut json_writer = JsonTermWriter::wrap(&mut term, false);
let mut json_writer = JsonTermWriter::wrap(&mut term);
json_writer.push_path_segment("color");
json_writer.set_fast_value(4.0f64);
assert_eq!(
@@ -494,7 +444,7 @@ mod tests {
fn test_bool_term() {
let field = Field::from_field_id(1);
let mut term = Term::with_type_and_field(Type::Json, field);
let mut json_writer = JsonTermWriter::wrap(&mut term, false);
let mut json_writer = JsonTermWriter::wrap(&mut term);
json_writer.push_path_segment("color");
json_writer.set_fast_value(true);
assert_eq!(
@@ -507,7 +457,7 @@ mod tests {
fn test_push_after_set_path_segment() {
let field = Field::from_field_id(1);
let mut term = Term::with_type_and_field(Type::Json, field);
let mut json_writer = JsonTermWriter::wrap(&mut term, false);
let mut json_writer = JsonTermWriter::wrap(&mut term);
json_writer.push_path_segment("attribute");
json_writer.set_str("something");
json_writer.push_path_segment("color");
@@ -522,7 +472,7 @@ mod tests {
fn test_pop_segment() {
let field = Field::from_field_id(1);
let mut term = Term::with_type_and_field(Type::Json, field);
let mut json_writer = JsonTermWriter::wrap(&mut term, false);
let mut json_writer = JsonTermWriter::wrap(&mut term);
json_writer.push_path_segment("color");
json_writer.push_path_segment("hue");
json_writer.pop_path_segment();
@@ -537,7 +487,7 @@ mod tests {
fn test_json_writer_path() {
let field = Field::from_field_id(1);
let mut term = Term::with_type_and_field(Type::Json, field);
let mut json_writer = JsonTermWriter::wrap(&mut term, false);
let mut json_writer = JsonTermWriter::wrap(&mut term);
json_writer.push_path_segment("color");
assert_eq!(json_writer.path(), b"color");
json_writer.push_path_segment("hue");
@@ -545,79 +495,4 @@ mod tests {
json_writer.set_str("pink");
assert_eq!(json_writer.path(), b"color\x01hue");
}
#[test]
fn test_json_path_expand_dots_disabled() {
let field = Field::from_field_id(1);
let mut term = Term::with_type_and_field(Type::Json, field);
let mut json_writer = JsonTermWriter::wrap(&mut term, false);
json_writer.push_path_segment("color.hue");
assert_eq!(json_writer.path(), b"color.hue");
}
#[test]
fn test_json_path_expand_dots_enabled() {
let field = Field::from_field_id(1);
let mut term = Term::with_type_and_field(Type::Json, field);
let mut json_writer = JsonTermWriter::wrap(&mut term, true);
json_writer.push_path_segment("color.hue");
assert_eq!(json_writer.path(), b"color\x01hue");
}
#[test]
fn test_json_path_expand_dots_enabled_pop_segment() {
let field = Field::from_field_id(1);
let mut term = Term::with_type_and_field(Type::Json, field);
let mut json_writer = JsonTermWriter::wrap(&mut term, true);
json_writer.push_path_segment("hello");
assert_eq!(json_writer.path(), b"hello");
json_writer.push_path_segment("color.hue");
assert_eq!(json_writer.path(), b"hello\x01color\x01hue");
json_writer.pop_path_segment();
assert_eq!(json_writer.path(), b"hello");
}
#[test]
fn test_split_json_path_simple() {
let json_path = split_json_path("titi.toto");
assert_eq!(&json_path, &["titi", "toto"]);
}
#[test]
fn test_split_json_path_single_segment() {
let json_path = split_json_path("toto");
assert_eq!(&json_path, &["toto"]);
}
#[test]
fn test_split_json_path_trailing_dot() {
let json_path = split_json_path("toto.");
assert_eq!(&json_path, &["toto", ""]);
}
#[test]
fn test_split_json_path_heading_dot() {
let json_path = split_json_path(".toto");
assert_eq!(&json_path, &["", "toto"]);
}
#[test]
fn test_split_json_path_escaped_dot() {
let json_path = split_json_path(r#"toto\.titi"#);
assert_eq!(&json_path, &["toto.titi"]);
let json_path_2 = split_json_path(r#"k8s\.container\.name"#);
assert_eq!(&json_path_2, &["k8s.container.name"]);
}
#[test]
fn test_split_json_path_escaped_backslash() {
let json_path = split_json_path(r#"toto\\titi"#);
assert_eq!(&json_path, &[r#"toto\titi"#]);
}
#[test]
fn test_split_json_path_escaped_normal_letter() {
let json_path = split_json_path(r#"toto\titi"#);
assert_eq!(&json_path, &[r#"tototiti"#]);
}
}

61
src/indexer/merger.rs
View File

@@ -13,7 +13,7 @@ use crate::docset::{DocSet, TERMINATED};
use crate::error::DataCorruption;
use crate::fastfield::{
get_fastfield_codecs_for_multivalue, AliveBitSet, Column, CompositeFastFieldSerializer,
MultiValueIndex, MultiValuedFastFieldReader, MultiValuedU128FastFieldReader,
MultiValueLength, MultiValuedFastFieldReader, MultiValuedU128FastFieldReader,
};
use crate::fieldnorm::{FieldNormReader, FieldNormReaders, FieldNormsSerializer, FieldNormsWriter};
use crate::indexer::doc_id_mapping::{expect_field_id_for_sort_field, SegmentDocIdMapping};
@@ -348,29 +348,9 @@ impl IndexMerger {
field,
fast_field_serializer,
doc_id_mapping,
&segment_and_ff_readers
.iter()
.map(|(segment_reader, u64s_reader)| {
(*segment_reader, u64s_reader.get_index_reader())
})
.collect::<Vec<_>>(),
&segment_and_ff_readers,
)?;
let num_vals = segment_and_ff_readers
.iter()
.map(|(segment_reader, reader)| {
// TODO implement generic version, implement reverse scan, all - deletes
if let Some(alive_bitset) = segment_reader.alive_bitset() {
alive_bitset
.iter_alive()
.map(|doc| reader.num_vals(doc))
.sum()
} else {
reader.total_num_vals()
}
})
.sum();
let fast_field_readers = segment_and_ff_readers
.into_iter()
.map(|(_, ff_reader)| ff_reader)
@@ -385,7 +365,12 @@ impl IndexMerger {
})
};
fast_field_serializer.create_u128_fast_field_with_idx(field, iter_gen, num_vals, 1)?;
fast_field_serializer.create_u128_fast_field_with_idx(
field,
iter_gen,
doc_id_mapping.len() as u32,
1,
)?;
Ok(())
}
@@ -544,11 +529,11 @@ impl IndexMerger {
// Creating the index file to point into the data, generic over `BytesFastFieldReader` and
// `MultiValuedFastFieldReader`
//
fn write_1_n_fast_field_idx_generic(
fn write_1_n_fast_field_idx_generic<T: MultiValueLength + Send + Sync>(
field: Field,
fast_field_serializer: &mut CompositeFastFieldSerializer,
doc_id_mapping: &SegmentDocIdMapping,
segment_and_ff_readers: &[(&SegmentReader, &MultiValueIndex)],
segment_and_ff_readers: &[(&SegmentReader, T)],
) -> crate::Result<()> {
let column =
RemappedDocIdMultiValueIndexColumn::new(segment_and_ff_readers, doc_id_mapping);
@@ -582,12 +567,7 @@ impl IndexMerger {
field,
fast_field_serializer,
doc_id_mapping,
&segment_and_ff_readers
.iter()
.map(|(segment_reader, u64s_reader)| {
(*segment_reader, u64s_reader.get_index_reader())
})
.collect::<Vec<_>>(),
&segment_and_ff_readers,
)
}
@@ -717,12 +697,7 @@ impl IndexMerger {
field,
fast_field_serializer,
doc_id_mapping,
&segment_and_ff_readers
.iter()
.map(|(segment_reader, u64s_reader)| {
(*segment_reader, u64s_reader.get_index_reader())
})
.collect::<Vec<_>>(),
&segment_and_ff_readers,
)?;
let mut serialize_vals = fast_field_serializer.new_bytes_fast_field(field);
@@ -829,7 +804,7 @@ impl IndexMerger {
// Let's compute the list of non-empty posting lists
for (segment_ord, term_info) in merged_terms.current_segment_ords_and_term_infos() {
let segment_reader = &self.readers[segment_ord];
let inverted_index: &InvertedIndexReader = &field_readers[segment_ord];
let inverted_index: &InvertedIndexReader = &*field_readers[segment_ord];
let segment_postings = inverted_index
.read_postings_from_terminfo(&term_info, segment_postings_option)?;
let alive_bitset_opt = segment_reader.alive_bitset();
@@ -968,7 +943,7 @@ impl IndexMerger {
let doc_bytes = doc_bytes_res?;
store_writer.store_bytes(&doc_bytes)?;
} else {
return Err(DataCorruption::comment_only(format!(
return Err(DataCorruption::comment_only(&format!(
"unexpected missing document in docstore on merge, doc address \
{old_doc_addr:?}",
))
@@ -1064,7 +1039,7 @@ mod tests {
};
use crate::collector::{Count, FacetCollector};
use crate::core::Index;
use crate::query::{AllQuery, BooleanQuery, EnableScoring, Scorer, TermQuery};
use crate::query::{AllQuery, BooleanQuery, Scorer, TermQuery};
use crate::schema::{
Cardinality, Document, Facet, FacetOptions, IndexRecordOption, NumericOptions, Term,
TextFieldIndexing, INDEXED, TEXT,
@@ -1977,7 +1952,7 @@ mod tests {
let reader = index.reader()?;
let searcher = reader.searcher();
let mut term_scorer = term_query
.specialized_weight(EnableScoring::Enabled(&searcher))?
.specialized_weight(&searcher, true)?
.specialized_scorer(searcher.segment_reader(0u32), 1.0)?;
assert_eq!(term_scorer.doc(), 0);
assert_nearly_equals!(term_scorer.block_max_score(), 0.0079681855);
@@ -1992,7 +1967,7 @@ mod tests {
assert_eq!(searcher.segment_readers().len(), 2);
for segment_reader in searcher.segment_readers() {
let mut term_scorer = term_query
.specialized_weight(EnableScoring::Enabled(&searcher))?
.specialized_weight(&searcher, true)?
.specialized_scorer(segment_reader, 1.0)?;
// the difference compared to before is intrinsic to the bm25 formula. no worries
// there.
@@ -2017,7 +1992,7 @@ mod tests {
let segment_reader = searcher.segment_reader(0u32);
let mut term_scorer = term_query
.specialized_weight(EnableScoring::Enabled(&searcher))?
.specialized_weight(&searcher, true)?
.specialized_scorer(segment_reader, 1.0)?;
// the difference compared to before is intrinsic to the bm25 formula. no worries there.
for doc in segment_reader.doc_ids_alive() {

49
src/indexer/mod.rs
View File

@@ -58,15 +58,13 @@ type AddBatchReceiver = channel::Receiver<AddBatch>;
#[cfg(feature = "mmap")]
#[cfg(test)]
mod tests_mmap {
use crate::collector::Count;
use crate::query::QueryParser;
use crate::schema::{JsonObjectOptions, Schema, TEXT};
use crate::schema::{self, Schema};
use crate::{Index, Term};
#[test]
fn test_advance_delete_bug() -> crate::Result<()> {
let mut schema_builder = Schema::builder();
let text_field = schema_builder.add_text_field("text", TEXT);
let text_field = schema_builder.add_text_field("text", schema::TEXT);
let index = Index::create_from_tempdir(schema_builder.build())?;
let mut index_writer = index.writer_for_tests()?;
// there must be one deleted document in the segment
@@ -77,48 +75,7 @@ mod tests_mmap {
index_writer.add_document(doc!(text_field=>"c"))?;
}
index_writer.commit()?;
index_writer.commit()?;
Ok(())
}
#[test]
fn test_json_field_expand_dots_disabled_dot_escaped_required() {
let mut schema_builder = Schema::builder();
let json_field = schema_builder.add_json_field("json", TEXT);
let index = Index::create_in_ram(schema_builder.build());
let mut index_writer = index.writer_for_tests().unwrap();
let json = serde_json::json!({"k8s.container.name": "prometheus", "val": "hello"});
index_writer.add_document(doc!(json_field=>json)).unwrap();
index_writer.commit().unwrap();
let reader = index.reader().unwrap();
let searcher = reader.searcher();
assert_eq!(searcher.num_docs(), 1);
let parse_query = QueryParser::for_index(&index, Vec::new());
let query = parse_query
.parse_query(r#"json.k8s\.container\.name:prometheus"#)
.unwrap();
let num_docs = searcher.search(&query, &Count).unwrap();
assert_eq!(num_docs, 1);
}
#[test]
fn test_json_field_expand_dots_enabled_dot_escape_not_required() {
let mut schema_builder = Schema::builder();
let json_options: JsonObjectOptions =
JsonObjectOptions::from(TEXT).set_expand_dots_enabled();
let json_field = schema_builder.add_json_field("json", json_options);
let index = Index::create_in_ram(schema_builder.build());
let mut index_writer = index.writer_for_tests().unwrap();
let json = serde_json::json!({"k8s.container.name": "prometheus", "val": "hello"});
index_writer.add_document(doc!(json_field=>json)).unwrap();
index_writer.commit().unwrap();
let reader = index.reader().unwrap();
let searcher = reader.searcher();
assert_eq!(searcher.num_docs(), 1);
let parse_query = QueryParser::for_index(&index, Vec::new());
let query = parse_query
.parse_query(r#"json.k8s.container.name:prometheus"#)
.unwrap();
let num_docs = searcher.search(&query, &Count).unwrap();
assert_eq!(num_docs, 1);
}
}

2
src/indexer/segment_updater.rs
View File

@@ -866,7 +866,7 @@ mod tests {
}
assert_eq!(indices.len(), 3);
let output_directory: Box<dyn Directory> = Box::<RamDirectory>::default();
let output_directory: Box<dyn Directory> = Box::new(RamDirectory::default());
let index = merge_indices(&indices, output_directory)?;
assert_eq!(index.schema(), schema);

59
src/indexer/segment_writer.rs
View File

@@ -158,7 +158,6 @@ impl SegmentWriter {
let doc_id = self.max_doc;
let vals_grouped_by_field = doc
.field_values()
.iter()
.sorted_by_key(|el| el.field())
.group_by(|el| el.field());
for (field, field_values) in &vals_grouped_by_field {
@@ -180,7 +179,7 @@ impl SegmentWriter {
self.per_field_postings_writers.get_for_field_mut(field);
term_buffer.clear_with_field_and_type(field_entry.field_type().value_type(), field);
match field_entry.field_type() {
match *field_entry.field_type() {
FieldType::Facet(_) => {
for value in values {
let facet = value.as_facet().ok_or_else(make_schema_error)?;
@@ -307,7 +306,7 @@ impl SegmentWriter {
self.fieldnorms_writer.record(doc_id, field, num_vals);
}
}
FieldType::JsonObject(json_options) => {
FieldType::JsonObject(_) => {
let text_analyzer = &self.per_field_text_analyzers[field.field_id() as usize];
let json_values_it =
values.map(|value| value.as_json().ok_or_else(make_schema_error));
@@ -315,7 +314,6 @@ impl SegmentWriter {
doc_id,
json_values_it,
text_analyzer,
json_options.is_expand_dots_enabled(),
term_buffer,
postings_writer,
ctx,
@@ -503,9 +501,17 @@ mod tests {
let reader = StoreReader::open(directory.open_read(path).unwrap(), 0).unwrap();
let doc = reader.get(0).unwrap();
assert_eq!(doc.field_values().len(), 2);
assert_eq!(doc.field_values()[0].value().as_text(), Some("A"));
assert_eq!(doc.field_values()[1].value().as_text(), Some("title"));
assert_eq!(doc.value_count(), 2);
let mut field_value_iter = doc.field_values();
assert_eq!(
field_value_iter.next().unwrap().value().as_text(),
Some("A")
);
assert_eq!(
field_value_iter.next().unwrap().value().as_text(),
Some("title")
);
assert!(field_value_iter.next().is_none());
}
#[test]
@@ -558,7 +564,7 @@ mod tests {
let mut term = Term::with_type_and_field(Type::Json, json_field);
let mut term_stream = term_dict.stream().unwrap();
let mut json_term_writer = JsonTermWriter::wrap(&mut term, false);
let mut json_term_writer = JsonTermWriter::wrap(&mut term);
json_term_writer.push_path_segment("bool");
json_term_writer.set_fast_value(true);
@@ -649,7 +655,7 @@ mod tests {
let segment_reader = searcher.segment_reader(0u32);
let inv_index = segment_reader.inverted_index(json_field).unwrap();
let mut term = Term::with_type_and_field(Type::Json, json_field);
let mut json_term_writer = JsonTermWriter::wrap(&mut term, false);
let mut json_term_writer = JsonTermWriter::wrap(&mut term);
json_term_writer.push_path_segment("mykey");
json_term_writer.set_str("token");
let term_info = inv_index
@@ -693,7 +699,7 @@ mod tests {
let segment_reader = searcher.segment_reader(0u32);
let inv_index = segment_reader.inverted_index(json_field).unwrap();
let mut term = Term::with_type_and_field(Type::Json, json_field);
let mut json_term_writer = JsonTermWriter::wrap(&mut term, false);
let mut json_term_writer = JsonTermWriter::wrap(&mut term);
json_term_writer.push_path_segment("mykey");
json_term_writer.set_str("two tokens");
let term_info = inv_index
@@ -738,7 +744,7 @@ mod tests {
let reader = index.reader().unwrap();
let searcher = reader.searcher();
let mut term = Term::with_type_and_field(Type::Json, json_field);
let mut json_term_writer = JsonTermWriter::wrap(&mut term, false);
let mut json_term_writer = JsonTermWriter::wrap(&mut term);
json_term_writer.push_path_segment("mykey");
json_term_writer.push_path_segment("field");
json_term_writer.set_str("hello");
@@ -834,20 +840,23 @@ mod tests {
// This is a bit of a contrived example.
let tokens = PreTokenizedString {
text: "contrived-example".to_string(), //< I can't think of a use case where this corner case happens in real life.
tokens: vec![Token { // Not the last token, yet ends after the last token.
offset_from: 0,
offset_to: 14,
position: 0,
text: "long_token".to_string(),
position_length: 3,
},
Token {
offset_from: 0,
offset_to: 14,
position: 1,
text: "short".to_string(),
position_length: 1,
}],
tokens: vec![
Token {
// Not the last token, yet ends after the last token.
offset_from: 0,
offset_to: 14,
position: 0,
text: "long_token".to_string(),
position_length: 3,
},
Token {
offset_from: 0,
offset_to: 14,
position: 1,
text: "short".to_string(),
position_length: 1,
},
],
};
doc.add_pre_tokenized_text(text, tokens);
doc.add_text(text, "hello");

24
src/indexer/sorted_doc_id_multivalue_column.rs
View File

@@ -3,7 +3,7 @@ use std::cmp;
use fastfield_codecs::Column;
use super::flat_map_with_buffer::FlatMapWithBufferIter;
use crate::fastfield::{MultiValueIndex, MultiValuedFastFieldReader};
use crate::fastfield::{MultiValueLength, MultiValuedFastFieldReader};
use crate::indexer::doc_id_mapping::SegmentDocIdMapping;
use crate::schema::Field;
use crate::{DocAddress, SegmentReader};
@@ -94,17 +94,17 @@ impl<'a> Column for RemappedDocIdMultiValueColumn<'a> {
}
}
pub(crate) struct RemappedDocIdMultiValueIndexColumn<'a> {
pub(crate) struct RemappedDocIdMultiValueIndexColumn<'a, T: MultiValueLength> {
doc_id_mapping: &'a SegmentDocIdMapping,
multi_value_length_readers: Vec<&'a MultiValueIndex>,
multi_value_length_readers: Vec<&'a T>,
min_value: u64,
max_value: u64,
num_vals: u32,
}
impl<'a> RemappedDocIdMultiValueIndexColumn<'a> {
impl<'a, T: MultiValueLength> RemappedDocIdMultiValueIndexColumn<'a, T> {
pub(crate) fn new(
segment_and_ff_readers: &'a [(&'a SegmentReader, &'a MultiValueIndex)],
segment_and_ff_readers: &'a [(&'a SegmentReader, T)],
doc_id_mapping: &'a SegmentDocIdMapping,
) -> Self {
// We go through a complete first pass to compute the minimum and the
@@ -115,19 +115,17 @@ impl<'a> RemappedDocIdMultiValueIndexColumn<'a> {
let mut multi_value_length_readers = Vec::with_capacity(segment_and_ff_readers.len());
for segment_and_ff_reader in segment_and_ff_readers {
let segment_reader = segment_and_ff_reader.0;
let multi_value_length_reader = segment_and_ff_reader.1;
let multi_value_length_reader = &segment_and_ff_reader.1;
if !segment_reader.has_deletes() {
max_value += multi_value_length_reader.total_num_vals() as u64;
max_value += multi_value_length_reader.get_total_len();
} else {
for doc in segment_reader.doc_ids_alive() {
max_value += multi_value_length_reader.num_vals_for_doc(doc) as u64;
max_value += multi_value_length_reader.get_len(doc);
}
}
num_vals += segment_reader.num_docs();
multi_value_length_readers.push(multi_value_length_reader);
}
// The value range is always get_val(doc)..get_val(doc + 1)
num_vals += 1;
Self {
doc_id_mapping,
multi_value_length_readers,
@@ -138,7 +136,7 @@ impl<'a> RemappedDocIdMultiValueIndexColumn<'a> {
}
}
impl<'a> Column for RemappedDocIdMultiValueIndexColumn<'a> {
impl<'a, T: MultiValueLength + Send + Sync> Column for RemappedDocIdMultiValueIndexColumn<'a, T> {
fn get_val(&self, _pos: u32) -> u64 {
unimplemented!()
}
@@ -150,8 +148,8 @@ impl<'a> Column for RemappedDocIdMultiValueIndexColumn<'a> {
move |old_doc_addr| {
let ff_reader =
&self.multi_value_length_readers[old_doc_addr.segment_ord as usize];
offset += ff_reader.num_vals_for_doc(old_doc_addr.doc_id);
offset as u64
offset += ff_reader.get_len(old_doc_addr.doc_id);
offset
},
)),
)

6
src/indexer/stamper.rs
View File

@@ -16,11 +16,11 @@ mod atomic_impl {
impl AtomicU64Wrapper {
pub fn new(first_opstamp: Opstamp) -> AtomicU64Wrapper {
AtomicU64Wrapper(AtomicU64::new(first_opstamp))
AtomicU64Wrapper(AtomicU64::new(first_opstamp as u64))
}
pub fn fetch_add(&self, val: u64, order: Ordering) -> u64 {
self.0.fetch_add(val, order) as u64
self.0.fetch_add(val as u64, order) as u64
}
pub fn revert(&self, val: u64, order: Ordering) -> u64 {
@@ -77,7 +77,7 @@ impl Stamper {
}
pub fn stamp(&self) -> Opstamp {
self.0.fetch_add(1u64, Ordering::SeqCst)
self.0.fetch_add(1u64, Ordering::SeqCst) as u64
}
/// Given a desired count `n`, `stamps` returns an iterator that

8
src/lib.rs
View File

@@ -177,7 +177,7 @@ impl DateTime {
/// The given date/time is converted to UTC and the actual
/// time zone is discarded.
pub const fn from_utc(dt: OffsetDateTime) -> Self {
let timestamp_micros = dt.unix_timestamp() * 1_000_000 + dt.microsecond() as i64;
let timestamp_micros = dt.unix_timestamp() as i64 * 1_000_000 + dt.microsecond() as i64;
Self { timestamp_micros }
}
@@ -259,6 +259,10 @@ pub use crate::future_result::FutureResult;
/// and instead, refer to this as `crate::Result<T>`.
pub type Result<T> = std::result::Result<T, TantivyError>;
/// Result for an Async io operation.
#[cfg(feature = "quickwit")]
pub type AsyncIoResult<T> = std::result::Result<T, crate::error::AsyncIoError>;
mod core;
mod indexer;
@@ -273,8 +277,6 @@ pub mod fastfield;
pub mod fieldnorm;
pub mod positions;
pub mod postings;
/// Module containing the different query implementations.
pub mod query;
pub mod schema;
pub mod space_usage;

2
src/positions/reader.rs
View File

@@ -71,7 +71,7 @@ impl PositionReader {
.map(|num_bits| num_bits as usize)
.sum();
let num_bytes_to_skip = num_bits * COMPRESSION_BLOCK_SIZE / 8;
self.bit_widths.advance(num_blocks);
self.bit_widths.advance(num_blocks as usize);
self.positions.advance(num_bytes_to_skip);
self.anchor_offset += (num_blocks * COMPRESSION_BLOCK_SIZE) as u64;
}

8
src/postings/indexing_context.rs
View File

@@ -1,11 +1,11 @@
use stacker::{ArenaHashMap, MemoryArena};
use crate::postings::stacker::{MemoryArena, TermHashMap};
/// IndexingContext contains all of the transient memory arenas
/// required for building the inverted index.
pub(crate) struct IndexingContext {
/// The term index is an adhoc hashmap,
/// itself backed by a dedicated memory arena.
pub term_index: ArenaHashMap,
pub term_index: TermHashMap,
/// Arena is a memory arena that stores posting lists / term frequencies / positions.
pub arena: MemoryArena,
}
@@ -13,9 +13,9 @@ pub(crate) struct IndexingContext {
impl IndexingContext {
/// Create a new IndexingContext given the size of the term hash map.
pub(crate) fn new(table_size: usize) -> IndexingContext {
let term_index = ArenaHashMap::new(table_size);
let term_index = TermHashMap::new(table_size);
IndexingContext {
arena: MemoryArena::default(),
arena: MemoryArena::new(),
term_index,
}
}

3
src/postings/json_postings_writer.rs
View File

@@ -1,11 +1,10 @@
use std::io;
use stacker::Addr;
use crate::fastfield::MultiValuedFastFieldWriter;
use crate::indexer::doc_id_mapping::DocIdMapping;
use crate::postings::postings_writer::SpecializedPostingsWriter;
use crate::postings::recorder::{BufferLender, DocIdRecorder, Recorder};
use crate::postings::stacker::Addr;
use crate::postings::{
FieldSerializer, IndexingContext, IndexingPosition, PostingsWriter, UnorderedTermId,
};

6
src/postings/mod.rs
View File

@@ -15,10 +15,9 @@ mod recorder;
mod segment_postings;
mod serializer;
mod skip;
mod stacker;
mod term_info;
pub(crate) use stacker::compute_table_size;
pub use self::block_segment_postings::BlockSegmentPostings;
pub(crate) use self::indexing_context::IndexingContext;
pub(crate) use self::per_field_postings_writer::PerFieldPostingsWriter;
@@ -27,9 +26,10 @@ pub(crate) use self::postings_writer::{serialize_postings, IndexingPosition, Pos
pub use self::segment_postings::SegmentPostings;
pub use self::serializer::{FieldSerializer, InvertedIndexSerializer};
pub(crate) use self::skip::{BlockInfo, SkipReader};
pub(crate) use self::stacker::compute_table_size;
pub use self::term_info::TermInfo;
pub(crate) type UnorderedTermId = stacker::UnorderedId;
pub(crate) type UnorderedTermId = u64;
#[allow(clippy::enum_variant_names)]
#[derive(Debug, PartialEq, Clone, Copy, Eq)]

2
src/postings/per_field_postings_writer.rs
View File

@@ -51,7 +51,7 @@ fn posting_writer_from_field_entry(field_entry: &FieldEntry) -> Box<dyn Postings
| FieldType::Date(_)
| FieldType::Bytes(_)
| FieldType::IpAddr(_)
| FieldType::Facet(_) => Box::<SpecializedPostingsWriter<DocIdRecorder>>::default(),
| FieldType::Facet(_) => Box::new(SpecializedPostingsWriter::<DocIdRecorder>::default()),
FieldType::JsonObject(ref json_object_options) => {
if let Some(text_indexing_option) = json_object_options.get_text_indexing_options() {
match text_indexing_option.index_option() {

8
src/postings/postings_writer.rs
View File

@@ -4,8 +4,8 @@ use std::marker::PhantomData;
use std::ops::Range;
use rustc_hash::FxHashMap;
use stacker::Addr;
use super::stacker::Addr;
use crate::fastfield::MultiValuedFastFieldWriter;
use crate::fieldnorm::FieldNormReaders;
use crate::indexer::doc_id_mapping::DocIdMapping;
@@ -59,11 +59,7 @@ pub(crate) fn serialize_postings(
) -> crate::Result<HashMap<Field, FxHashMap<UnorderedTermId, TermOrdinal>>> {
let mut term_offsets: Vec<(Term<&[u8]>, Addr, UnorderedTermId)> =
Vec::with_capacity(ctx.term_index.len());
term_offsets.extend(
ctx.term_index
.iter()
.map(|(bytes, addr, unordered_id)| (Term::wrap(bytes), addr, unordered_id)),
);
term_offsets.extend(ctx.term_index.iter());
term_offsets.sort_unstable_by_key(|(k, _, _)| k.clone());
let mut unordered_term_mappings: HashMap<Field, FxHashMap<UnorderedTermId, TermOrdinal>> =
HashMap::new();

19
src/postings/recorder.rs
View File

@@ -1,6 +1,6 @@
use common::read_u32_vint;
use stacker::{ExpUnrolledLinkedList, MemoryArena};
use super::stacker::{ExpUnrolledLinkedList, MemoryArena};
use crate::indexer::doc_id_mapping::DocIdMapping;
use crate::postings::FieldSerializer;
use crate::DocId;
@@ -91,7 +91,7 @@ pub struct DocIdRecorder {
impl Default for DocIdRecorder {
fn default() -> Self {
DocIdRecorder {
stack: ExpUnrolledLinkedList::default(),
stack: ExpUnrolledLinkedList::new(),
current_doc: u32::MAX,
}
}
@@ -144,7 +144,7 @@ impl Recorder for DocIdRecorder {
}
/// Recorder encoding document ids, and term frequencies
#[derive(Clone, Copy, Default)]
#[derive(Clone, Copy)]
pub struct TermFrequencyRecorder {
stack: ExpUnrolledLinkedList,
current_doc: DocId,
@@ -152,6 +152,17 @@ pub struct TermFrequencyRecorder {
term_doc_freq: u32,
}
impl Default for TermFrequencyRecorder {
fn default() -> Self {
TermFrequencyRecorder {
stack: ExpUnrolledLinkedList::new(),
current_doc: 0,
current_tf: 0u32,
term_doc_freq: 0u32,
}
}
}
impl Recorder for TermFrequencyRecorder {
fn current_doc(&self) -> DocId {
self.current_doc
@@ -218,7 +229,7 @@ pub struct TfAndPositionRecorder {
impl Default for TfAndPositionRecorder {
fn default() -> Self {
TfAndPositionRecorder {
stack: ExpUnrolledLinkedList::default(),
stack: ExpUnrolledLinkedList::new(),
current_doc: u32::MAX,
term_doc_freq: 0u32,
}

30
src/postings/serializer.rs
View File

@@ -213,21 +213,21 @@ impl<'a> FieldSerializer<'a> {
fail_point!("FieldSerializer::close_term", |msg: Option<String>| {
Err(io::Error::new(io::ErrorKind::Other, format!("{:?}", msg)))
});
if !self.term_open {
return Ok(());
if self.term_open {
self.postings_serializer
.close_term(self.current_term_info.doc_freq)?;
self.current_term_info.postings_range.end =
self.postings_serializer.written_bytes() as usize;
if let Some(positions_serializer) = self.positions_serializer_opt.as_mut() {
positions_serializer.close_term()?;
self.current_term_info.positions_range.end =
positions_serializer.written_bytes() as usize;
}
self.term_dictionary_builder
.insert_value(&self.current_term_info)?;
self.term_open = false;
}
self.postings_serializer
.close_term(self.current_term_info.doc_freq)?;
self.current_term_info.postings_range.end =
self.postings_serializer.written_bytes() as usize;
if let Some(positions_serializer) = self.positions_serializer_opt.as_mut() {
positions_serializer.close_term()?;
self.current_term_info.positions_range.end =
positions_serializer.written_bytes() as usize;
}
self.term_dictionary_builder
.insert_value(&self.current_term_info)?;
self.term_open = false;
Ok(())
}
@@ -465,7 +465,7 @@ impl<W: Write> PostingsSerializer<W> {
/// When called after writing the postings of a term, this value is used as a
/// end offset.
fn written_bytes(&self) -> u64 {
self.output_write.written_bytes()
self.output_write.written_bytes() as u64
}
fn clear(&mut self) {

32
stacker/src/expull.rs → src/postings/stacker/expull.rs
View File

@@ -2,8 +2,8 @@ use std::mem;
use common::serialize_vint_u32;
use crate::memory_arena::{load, store};
use crate::{Addr, MemoryArena};
use super::{Addr, MemoryArena};
use crate::postings::stacker::memory_arena::{load, store};
const MAX_BLOCK_LEN: u32 = 1u32 << 15;
const FIRST_BLOCK: usize = 16;
@@ -62,7 +62,7 @@ fn len_to_capacity(len: u32) -> CapacityResult {
pub struct ExpUnrolledLinkedList {
len: u32,
tail: Addr,
inlined_data: [u8; INLINED_BLOCK_LEN],
inlined_data: [u8; INLINED_BLOCK_LEN as usize],
}
pub struct ExpUnrolledLinkedListWriter<'a> {
@@ -120,17 +120,15 @@ impl<'a> ExpUnrolledLinkedListWriter<'a> {
}
}
impl Default for ExpUnrolledLinkedList {
fn default() -> ExpUnrolledLinkedList {
impl ExpUnrolledLinkedList {
pub fn new() -> ExpUnrolledLinkedList {
ExpUnrolledLinkedList {
len: 0u32,
tail: Addr::null_pointer(),
inlined_data: [0u8; INLINED_BLOCK_LEN],
inlined_data: [0u8; INLINED_BLOCK_LEN as usize],
}
}
}
impl ExpUnrolledLinkedList {
#[inline]
pub fn writer<'a>(&'a mut self, arena: &'a mut MemoryArena) -> ExpUnrolledLinkedListWriter<'a> {
ExpUnrolledLinkedListWriter { eull: self, arena }
@@ -171,8 +169,8 @@ mod tests {
#[test]
fn test_eull() {
let mut arena = MemoryArena::default();
let mut stack = ExpUnrolledLinkedList::default();
let mut arena = MemoryArena::new();
let mut stack = ExpUnrolledLinkedList::new();
stack.writer(&mut arena).extend_from_slice(&[1u8]);
stack.writer(&mut arena).extend_from_slice(&[2u8]);
stack.writer(&mut arena).extend_from_slice(&[3u8, 4u8]);
@@ -186,8 +184,8 @@ mod tests {
#[test]
fn test_eull_long() {
let mut arena = MemoryArena::default();
let mut eull = ExpUnrolledLinkedList::default();
let mut arena = MemoryArena::new();
let mut eull = ExpUnrolledLinkedList::new();
let data: Vec<u32> = (0..100).collect();
for &el in &data {
eull.writer(&mut arena).write_u32_vint(el);
@@ -204,9 +202,9 @@ mod tests {
#[test]
fn test_eull_interlaced() {
let mut eull = MemoryArena::default();
let mut stack = ExpUnrolledLinkedList::default();
let mut stack2 = ExpUnrolledLinkedList::default();
let mut eull = MemoryArena::new();
let mut stack = ExpUnrolledLinkedList::new();
let mut stack2 = ExpUnrolledLinkedList::new();
let mut vec1: Vec<u8> = vec![];
let mut vec2: Vec<u8> = vec![];
@@ -308,9 +306,9 @@ mod bench {
#[bench]
fn bench_push_stack(bench: &mut Bencher) {
bench.iter(|| {
let mut arena = MemoryArena::default();
let mut arena = MemoryArena::new();
let mut stacks: Vec<ExpUnrolledLinkedList> =
iter::repeat_with(ExpUnrolledLinkedList::default)
iter::repeat_with(ExpUnrolledLinkedList::new)
.take(NUM_STACK)
.collect();
for s in 0..NUM_STACK {

11
stacker/src/memory_arena.rs → src/postings/stacker/memory_arena.rs
View File

@@ -86,16 +86,15 @@ pub struct MemoryArena {
pages: Vec<Page>,
}
impl Default for MemoryArena {
fn default() -> MemoryArena {
impl MemoryArena {
/// Creates a new memory arena.
pub fn new() -> MemoryArena {
let first_page = Page::new(0);
MemoryArena {
pages: vec![first_page],
}
}
}
impl MemoryArena {
fn add_page(&mut self) -> &mut Page {
let new_page_id = self.pages.len();
self.pages.push(Page::new(new_page_id));
@@ -198,7 +197,7 @@ mod tests {
#[test]
fn test_arena_allocate_slice() {
let mut arena = MemoryArena::default();
let mut arena = MemoryArena::new();
let a = b"hello";
let b = b"happy tax payer";
@@ -221,7 +220,7 @@ mod tests {
#[test]
fn test_store_object() {
let mut arena = MemoryArena::default();
let mut arena = MemoryArena::new();
let a = MyTest {
a: 143,
b: 21,

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