rust/tantivy
1
0
Fork 0
mirror of https://github.com/quickwit-oss/tantivy.git synced 2026-06-11 13:00:42 +00:00
Code Issues Packages Projects Releases Wiki Activity

Compare commits

base: rust:0.26.1
Branches Tags
rust:main rust:faster_histo rust:dependabot/github_actions/github/codeql-action-4.36.2 rust:postings-writer-enum-code-june2026 rust:bench/alyze-vs-unicode-seg-tokenizer rust:larger-collect-block-buffer rust:faster_union rust:paul.masurel/upstream-datafusion rust:PSeitz-patch-1 rust:dependabot/cargo/fs4-1.1.0 rust:dependabot/cargo/lru-0.18.0 rust:paul.masurel/bench-nested rust:paul.masurel/2892-nested-term-agg-opt rust:paul.masurel/nested-aggregation-slow rust:dependabot/cargo/rand_distr-0.6 rust:storage_abstraction rust:bench_0.25_compare rust:dependabot/cargo/rand-0.10 rust:dependabot/cargo/oneshot-0.2.1 rust:composite-agg-rebase rust:composite-agg rust:postings-writer-enum-codec5 rust:congxie/supportAgg rust:postings-writer-enum-codec4 rust:segmentreader_trait rust:stuhood.intersection-termination-upstream rust:postings-writer-enum-codec3 rust:postings-writer-enum-codec2 rust:postings-writer-enum rust:paul.masurel/skip-first-doc-on-intersection rust:stuhood.lazy-scorers-blocks-upstream rust:paul.masurel/seek_into_the_danger_zone_result rust:paul.masurel/search_into_danger_zone_fastfield rust:stuhood.lazy-scorers-blocks 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:tagged-user-input-ast rust:1686a/parse-agg-res rust:issue/2411-adding-panic-handler-to-rayon-merge-thread-pool rust:dependabot/cargo/nom-8 rust:upgrade_deps rust:test_parser 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:0.26.1 rust:0.26.0 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:faster_histo
Branches Tags
rust:faster_histo rust:main rust:dependabot/github_actions/github/codeql-action-4.36.2 rust:postings-writer-enum-code-june2026 rust:bench/alyze-vs-unicode-seg-tokenizer rust:larger-collect-block-buffer rust:faster_union rust:paul.masurel/upstream-datafusion rust:PSeitz-patch-1 rust:dependabot/cargo/fs4-1.1.0 rust:dependabot/cargo/lru-0.18.0 rust:paul.masurel/bench-nested rust:paul.masurel/2892-nested-term-agg-opt rust:paul.masurel/nested-aggregation-slow rust:dependabot/cargo/rand_distr-0.6 rust:storage_abstraction rust:bench_0.25_compare rust:dependabot/cargo/rand-0.10 rust:dependabot/cargo/oneshot-0.2.1 rust:composite-agg-rebase rust:composite-agg rust:postings-writer-enum-codec5 rust:congxie/supportAgg rust:postings-writer-enum-codec4 rust:segmentreader_trait rust:stuhood.intersection-termination-upstream rust:postings-writer-enum-codec3 rust:postings-writer-enum-codec2 rust:postings-writer-enum rust:paul.masurel/skip-first-doc-on-intersection rust:stuhood.lazy-scorers-blocks-upstream rust:paul.masurel/seek_into_the_danger_zone_result rust:paul.masurel/search_into_danger_zone_fastfield rust:stuhood.lazy-scorers-blocks 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:tagged-user-input-ast rust:1686a/parse-agg-res rust:issue/2411-adding-panic-handler-to-rayon-merge-thread-pool rust:dependabot/cargo/nom-8 rust:upgrade_deps rust:test_parser 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:0.26.1 rust:0.26.0 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

74 Commits
0.26.1 ... faster_his

Author SHA1 Message Date
Pascal Seitz
4681282cac derive term counts 2026-06-11 12:18:49 +02:00
Pascal Seitz
15eb58ebf5 add specialized TermHistogram 2026-06-10 17:23:46 +02:00
Pascal Seitz
b467779028 add dense histogram, optional sub-buckets 2026-06-10 15:03:23 +02:00
Pascal Seitz
b51f104668 add no sub-gg fastpath 2026-06-10 15:03:23 +02:00
Pascal Seitz
5ce9bf7fda use get_range when possible 2026-06-10 15:03:23 +02:00
Pascal Seitz
c863264731 add status -> date histogram bench 2026-06-10 15:02:53 +02:00
Pascal Seitz
b19f0ddc77 fix clippy 2026-06-09 23:14:12 +08:00
Pascal Seitz
b4acfcf881 cleanup AggregationsSegmentCtx 
The metric/cardinality/histogram _mut getters had no callers needing
mutation; their two uses already pass the resulting reference as &T.

simplify req_data ownership: clone into collectors, Rc only for filter BitSet

Replace Vec<Option<Box<T>>> + take/put-back round-trip with Vec<T> +
direct clone into collector. Collectors now own their per-segment
request data outright, removing the borrow-checker dance that the
take/put-back pattern existed to satisfy.

The structural clones are cheap (Column<u64> is Arc-internal) except
for the filter aggregation, whose DocumentQueryEvaluator carries a
precomputed per-segment BitSet sized by max_doc. Wrap that in
Rc<DocumentQueryEvaluator> so FilterAggReqData::clone() bumps a
refcount instead of duplicating the BitSet. Move SegmentFilterCollector's
matching_docs_buffer out of FilterAggReqData so its pre-allocated
capacity is preserved per collector instead of being lost on every clone.
 2026-06-09 23:14:12 +08:00
dependabot[bot]
3a8240b123 Bump codecov/codecov-action from 6.0.0 to 7.0.0 
Bumps [codecov/codecov-action](https://github.com/codecov/codecov-action) from 6.0.0 to 7.0.0.
- [Release notes](https://github.com/codecov/codecov-action/releases)
- [Changelog](https://github.com/codecov/codecov-action/blob/main/CHANGELOG.md)
- [Commits](57e3a136b7...fb8b3582c8)

---
updated-dependencies:
- dependency-name: codecov/codecov-action
  dependency-version: 7.0.0
  dependency-type: direct:production
  update-type: version-update:semver-major
...

Signed-off-by: dependabot[bot] <support@github.com>
 2026-06-09 14:48:17 +08:00
dependabot[bot]
fd9713e1ca Bump actions/checkout from 6.0.2 to 6.0.3 (#2949) 
Bumps [actions/checkout](https://github.com/actions/checkout) from 6.0.2 to 6.0.3.
- [Release notes](https://github.com/actions/checkout/releases)
- [Changelog](https://github.com/actions/checkout/blob/main/CHANGELOG.md)
- [Commits](de0fac2e45...df4cb1c069)

---
updated-dependencies:
- dependency-name: actions/checkout
  dependency-version: 6.0.3
  dependency-type: direct:production
  update-type: version-update:semver-patch
...

Signed-off-by: dependabot[bot] <support@github.com>
Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com>
 2026-06-08 10:55:54 +02:00
dependabot[bot]
96f3784f79 Bump github/codeql-action from 4.35.2 to 4.36.1 (#2948) 
Bumps [github/codeql-action](https://github.com/github/codeql-action) from 4.35.2 to 4.36.1.
- [Release notes](https://github.com/github/codeql-action/releases)
- [Changelog](https://github.com/github/codeql-action/blob/main/CHANGELOG.md)
- [Commits](95e58e9a2c...87557b9c84)

---
updated-dependencies:
- dependency-name: github/codeql-action
  dependency-version: 4.36.1
  dependency-type: direct:production
  update-type: version-update:semver-minor
...

Signed-off-by: dependabot[bot] <support@github.com>
Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com>
 2026-06-08 10:49:04 +02:00
dependabot[bot]
87a6679a79 Bump actions/upload-artifact from 7.0.0 to 7.0.1 (#2917) 
Bumps [actions/upload-artifact](https://github.com/actions/upload-artifact) from 7.0.0 to 7.0.1.
- [Release notes](https://github.com/actions/upload-artifact/releases)
- [Commits](bbbca2ddaa...043fb46d1a)

---
updated-dependencies:
- dependency-name: actions/upload-artifact
  dependency-version: 7.0.1
  dependency-type: direct:production
  update-type: version-update:semver-patch
...

Signed-off-by: dependabot[bot] <support@github.com>
Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com>
 2026-06-08 10:48:48 +02:00
dependabot[bot]
864a6aa72c Update murmurhash32 requirement from 0.3 to 0.4 (#2894) 
Updates the requirements on [murmurhash32](https://github.com/quickwit-inc/murmurhash32) to permit the latest version.
- [Commits](https://github.com/quickwit-inc/murmurhash32/commits)

---
updated-dependencies:
- dependency-name: murmurhash32
  dependency-version: 0.4.0
  dependency-type: direct:production
...

Signed-off-by: dependabot[bot] <support@github.com>
Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com>
 2026-06-08 10:48:32 +02:00
Paul Masurel
abcf6754a2 CR comments from https://github.com/quickwit-oss/tantivy/pull/2940 (#2952) 
Co-authored-by: Paul Masurel <paul.masurel@datadoghq.com>
 2026-06-08 10:47:58 +02:00
Kanishk Sachan
70a8e56ee5 test(postings): add unit tests for TermFrequencyRecorder 
Closes #2285

The TermFrequencyRecorder was completely untested. Add five focused tests:

- term_frequency_recorder_has_term_freq: verifies the recorder
  correctly advertises term-frequency support via has_term_freq()
- term_frequency_recorder_zero_docs: term_doc_freq() returns Some(0)
  before any documents are recorded
- term_frequency_recorder_term_doc_freq_single_doc: one document with
  two occurrences yields term_doc_freq() == Some(1)
- term_frequency_recorder_term_doc_freq_multiple_docs: three documents
  with varying term frequencies yield term_doc_freq() == Some(3),
  confirming the count tracks documents, not occurrences
- term_frequency_recorder_single_occurrence_per_doc: each of three
  documents has exactly one occurrence
- term_frequency_recorder_high_frequency_doc: a single document with
  1000 occurrences still yields term_doc_freq() == Some(1)
 2026-06-06 14:44:51 +08:00
Paul Masurel
62705526e8 Add sve + neon filter vec implementation as spotted by Adam (#2940) 
* Add filter_vec benchmarks (dense, sparse, full coverage)

Uses get_ids_for_value_range to exercise both the bitpacking decode and
the filter_vec SIMD path together under realistic cache conditions.

* Add NEON and SVE implementations for filter_vec

Adds aarch64-specific SIMD paths (NEON always available on aarch64;
SVE gated on nightly + non-Apple target) with routing logic in mod.rs
that selects the best available instruction set at runtime.

* Using asm! to workaround the lack of stabilized SVE intrinsics

* showing instruction set

* improved proptesting

* removing build.rs

---------

Co-authored-by: Paul Masurel <paul.masurel@datadoghq.com>
 2026-06-04 17:51:26 +02:00
Paul Masurel
a27c64998f Cargo clippy fix (#2943) 
Co-authored-by: Paul Masurel <paul.masurel@datadoghq.com>
 2026-06-01 14:39:44 +02:00
Paul Masurel
46b3fb9ed3 Relying on upstream version of datasketch and stop using HLL 4. (#2936) 
We were relying on a fork for:

a bugfix in LIST serialization
a better API exposing a new Coupon type, required for caching coupons.
We also stop using HLL8 in hope to fix
https://datadoghq.atlassian.net/browse/CLOUDPREM-625

Co-authored-by: Paul Masurel <paul.masurel@datadoghq.com>
 2026-05-19 13:29:35 +02:00
trinity-1686a
fbe620b9b4 Merge pull request #2933 from quickwit-oss/1686a/sstable-opt 
optimise sstable index access pattern
 2026-05-19 11:43:17 +02:00
trinity-1686a
95d8a3989a cr 2026-05-19 11:38:48 +02:00
trinity-1686a
ea61a68db4 skip sstable index binary search when ordinal is in same block 2026-05-16 11:35:38 +02:00
trinity-1686a
c367df37c1 refactor sstable index 2026-05-16 11:30:02 +02:00
Mohammad Dashti
d99a5d4e91 Rename validate_aggregation_fields to validate_aggregation_fields_exist 
Applies @PSeitz's review suggestion to make the function name more
descriptive of what it checks. Also adds a doc note clarifying why
validation is opt-in rather than enforced by default.
 2026-05-16 15:45:20 +08:00
Mohammad Dashti
2de6f075ce Fixed the example 2026-05-16 15:45:20 +08:00
Mohammad Dashti
18080067c7 Applied PR comment: 
I would move it outside of the aggregation. You can fetch the fields from the aggregation request and do a validation in a helper function
 2026-05-16 15:45:20 +08:00
Mohammad Dashti
95db7d2e5c Revert "Revert all impl." 
This reverts commit d5e0991549a05bf80f19f853f7689ad69f96e7e5.
 2026-05-16 15:45:20 +08:00
Mohammad Dashti
fc017c4c74 Applied PR comments. 2026-05-16 15:45:20 +08:00
Mohammad Dashti
141c91d028 Added a flag: strict_validation 2026-05-16 15:45:20 +08:00
Mohammad Dashti
36a83e7c1a Fixed agg validation 2026-05-16 15:45:20 +08:00
jinhelin
be11f8a6a1 Fix opening positions file error 2026-05-14 15:55:59 +08:00
dependabot[bot]
4305e4029e Update binggan requirement from 0.16.1 to 0.17.0 
Updates the requirements on [binggan](https://github.com/pseitz/binggan) to permit the latest version.
- [Changelog](https://github.com/PSeitz/binggan/blob/main/CHANGELOG.md)
- [Commits](https://github.com/pseitz/binggan/commits)

---
updated-dependencies:
- dependency-name: binggan
  dependency-version: 0.17.0
  dependency-type: direct:production
...

Signed-off-by: dependabot[bot] <support@github.com>
 2026-05-12 15:10:20 +08:00
Pascal Seitz
edfb02b47e switch to enum, fix mixed types for cardinality agg 2026-05-05 16:39:51 +08:00
Pascal Seitz
d0fad88bac use bitsets for card agg 2026-05-05 16:39:51 +08:00
Pascal Seitz
351280c0b4 add card bench for high card 2026-05-05 16:39:51 +08:00
James Sewell
4480cf0a98 Enable BMW for single-scorer boolean queries by removing early return in scorer_union (#2915) 
The early return for `scorers.len() == 1` in `scorer_union` short-circuits a single TermScorer into `SpecializedScorer::Other`, bypassing the `TermUnion` path that enables block-max WAND (BMW) in `for_each_pruning`.

This was originally addressed in PR #2898 (backed out), which added a special case in `BooleanWeight::for_each_pruning`. PR #2912 (merged as d27ca164a) added a single-scorer fast path inside `block_wand` itself, but did not remove this early return — so a single SHOULD TermScorer still never reaches the BMW path.

Removing the early return lets a single TermScorer with freq reading flow through to `SpecializedScorer::TermUnion`, where `block_wand` → `block_wand_single_scorer` handles it efficiently.
 2026-04-28 14:49:53 -07:00
Pascal Seitz
d47abdf104 early cut off for order by sub agg in term agg 2026-04-28 16:59:59 +02:00
Pascal Seitz
c11952eb7c add order by agg benchmark 2026-04-28 16:59:59 +02:00
trinity-1686a
09667ee9c8 Merge pull request #2909 from osyniakov/claude/add-ossf-scorecard-1z6Vn 
Add OpenSSF Scorecard workflow
 2026-04-28 11:57:36 +02:00
trinity-1686a
333ccf5300 Merge pull request #2896 from osyniakov/claude/fix-issues-5945-5937-eQm1Q 
ci: pin GitHub Actions to full commit SHAs and restrict token permissions
 2026-04-28 11:57:18 +02:00
Oleksii Syniakov
60a39a4689 Merge branch 'main' into claude/fix-issues-5945-5937-eQm1Q 2026-04-28 10:28:23 +02:00
Oleksii Syniakov
f8f3e4277f remove not neeeded permissions for the public repo 2026-04-28 10:09:30 +02:00
Oleksii Syniakov
ff1433713a bump upload-sarif -> 4.35.2 
Co-authored-by: trinity-1686a <trinity.pointard@gmail.com>
 2026-04-28 10:07:45 +02:00
trinity-1686a
ca139d8eb1 Merge pull request #2910 from quickwit-oss/abdul.andha/composite-agg-after 
Composite aggregations: send after key on last page
 2026-04-27 23:38:52 +02:00
Abdul Andha
ac508108aa address pr comment 2026-04-27 12:39:38 -04:00
Paul Masurel
63da5a21b2 Optimizing top K using Adrien Grand's ideas (#2865) 
* Optimizing top K using Adrien Grand's ideas

https://jpountz.github.io/2025/08/28/compiled-vs-vectorized-search-engine-edition.html

* Suffix-sum pruning for multi-term intersection candidates

After scoring each secondary in Phase 2, check whether remaining
secondaries' block_max scores can still beat the threshold. Skip
to the next candidate early if impossible, avoiding expensive seeks
into later secondaries.

Improves three-term intersection by ~8% on the balanced benchmark
while keeping two-term performance neutral.

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

* Claude CR comment

* Removed 16 term scorer limit.

---------

Co-authored-by: Paul Masurel <paul.masurel@datadoghq.com>
Co-authored-by: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
 2026-04-26 12:14:40 +02:00
lif
54cd5bba98 fix: skip sentinel facet ords in harvest to prevent wrong root (#2867) 
When a document has the exact registered facet path (not a child),
compute_collapse_mapping_one maps it to a sentinel (u64::MAX, 0).
Without filtering, harvest() passes u64::MAX to ord_to_term which
resolves to the last dictionary entry, producing a spurious facet
from an unrelated branch.

Skip entries where facet_ord == u64::MAX in harvest().

Closes #2494

Signed-off-by: majiayu000 <1835304752@qq.com>
 2026-04-25 22:23:30 +02:00
Paul Masurel
d27ca164a9 block_wand: use single-scorer path when there is only one scorer 2026-04-25 16:35:00 +02:00
dependabot[bot]
2f5a48e8b1 Update criterion requirement from 0.5 to 0.8 (#2873) 
Updates the requirements on [criterion](https://github.com/criterion-rs/criterion.rs) to permit the latest version.
- [Release notes](https://github.com/criterion-rs/criterion.rs/releases)
- [Changelog](https://github.com/criterion-rs/criterion.rs/blob/master/CHANGELOG.md)
- [Commits](https://github.com/criterion-rs/criterion.rs/compare/0.5.0...criterion-v0.8.2)

---
updated-dependencies:
- dependency-name: criterion
  dependency-version: 0.8.2
  dependency-type: direct:production
...

Signed-off-by: dependabot[bot] <support@github.com>
Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com>
 2026-04-25 14:15:53 +02:00
dependabot[bot]
ae0ab907fe Bump actions/checkout from 4 to 6 (#2875) 
Bumps [actions/checkout](https://github.com/actions/checkout) from 4 to 6.
- [Release notes](https://github.com/actions/checkout/releases)
- [Changelog](https://github.com/actions/checkout/blob/main/CHANGELOG.md)
- [Commits](https://github.com/actions/checkout/compare/v4...v6)

---
updated-dependencies:
- dependency-name: actions/checkout
  dependency-version: '6'
  dependency-type: direct:production
  update-type: version-update:semver-major
...

Signed-off-by: dependabot[bot] <support@github.com>
Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com>
 2026-04-25 14:15:27 +02:00
dependabot[bot]
7d62e084e7 Bump codecov/codecov-action from 3 to 6 (#2876) 
Bumps [codecov/codecov-action](https://github.com/codecov/codecov-action) from 3 to 6.
- [Release notes](https://github.com/codecov/codecov-action/releases)
- [Changelog](https://github.com/codecov/codecov-action/blob/main/CHANGELOG.md)
- [Commits](https://github.com/codecov/codecov-action/compare/v3...v6)

---
updated-dependencies:
- dependency-name: codecov/codecov-action
  dependency-version: '6'
  dependency-type: direct:production
  update-type: version-update:semver-major
...

Signed-off-by: dependabot[bot] <support@github.com>
Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com>
 2026-04-25 14:14:54 +02:00
James Sewell
322286ee16 Tighen Block-Max in single-scorer (#2897) 
In the Block-Max WAND single-scorer, it uses block_max_score() < threshold,
whereas the multi-term one uses  block_max_score_upperbound <= threshold.

As both of these are guarded later on with if score > threshold we can
use the more efficent form in single-scorer.

Single-scorer block skip (<, should be <=): https://github.com/quickwit-oss/tantivy/blob/main/src/query/boolean_query/block_wand.rs#L231
Multi-scorer block skip (already <=): https://github.com/quickwit-oss/tantivy/blob/main/src/query/boolean_query/block_wand.rs#L179
Single-scorer per-doc guard (>): https://github.com/quickwit-oss/tantivy/blob/main/src/query/boolean_query/block_wand.rs#L246
Multi-scorer per-doc guard (>): https://github.com/quickwit-oss/tantivy/blob/main/src/query/boolean_query/block_wand.rs#L206

This will improve performance when there are many identical scores.
 2026-04-25 14:13:07 +02:00
RJ Barman
73ad18fa1e fix: Add space for missing sentinel in allowed bitset when a missing key is provided (#119) (#2907) 
## Bug Overview
Under certain conditions, a `terms` aggregation request can cause a
bounds-check panic. Those conditions are:
- The queried field must be a text field
- There must be a segment where the number of distinct terms in it's
dictionary for the queried field is divisible by 64 (i.e.e where
`count(term_dict.keys) % 64 == 0`)
- That same segment must contain at least one document that does not
contain this field.
- The request contain a `missing` key that is a string.
- The request must contain an `include` or `exclude` filter.
For example:
```json
{
    "my_bool": {
        "terms": {
            "field": "title",
            "include": "foo",
            "missing": "__NULL__",
        }
    }
}
```
Check out the added tests in `src/aggregation/bucket/term_agg.rs` to see
this in action

## How the bug happens
### Preparation
While preparing the aggregation nodes:
1) When we've provided a `missing` key, we derive a missing sentinel.
For string keys this column's max value (which for string keys is always
the number of terms in this segment) + 1.
2) for string columns only, we optionally prep an "allowed" `BitSet` for
allowed term ids. (`build_allowed_term_ids_for_str` in
`src/aggregation/agg_data.rs`)
- If no `include` or `exclude` filter is provided, this just returns
`None`, causing this check to be skipped down the line
- Otherwise the bitset is initialized to be able to hold the exact
number of terms in the segments term dictionary, and the bits are set to
signify which terms are to be included in the results.

### Collection
If we have an "allowed" `BitSet`, filter documents against that. For
each document, we check if the `BitSet` contains the documents term id.
For documents without the field, this is the missing sentinel we derived
earlier, minus 1 (to account for zero-based indexing): `(num_terms + 1)
- 1`.However, the `BitSet`s size is only `num_terms`. Normally, this
slips by without a problem, but if `num_terms % 64 == 0`, this will
cause a panic.

### Why `BitSet` panics
`BitSet` is represented under the hood by a boxed slice of `u64`s. When
you go to check a bit using `BitSet::contains`, it must determine which
of those `u64`s the bit is in, and then the position within that `u64`
of the bit.

In cases where the number of terms is not divisible by 64, the `BitSet`
must waste some bits. When we then look up the missing sentinel's bit,
it happens to be one of those wasted bits, for which `BitSet` is happy
to return the value of. For example, if the number of terms was 63:
```rust
let bitset_init_size = 63; // so BitSet's boxed slice has a length of 1, capable of holding 64 bits, term id [0, 62]
let missing_sentinel = 63; // num_terms + 1 - 1;
let byte_pos = missing_sentinel / 64; // 0 - within the valid slice
let bit_pos = missing_sentinel % 64; // 63 - hits the 1 wasted bit
```

But if the number of terms is indeed divisible by 64, then the `BitSet`
is perfectly aligned to the byte boundary:
```rust
let bitset_init_size = 64; // so BitSet's boxed slice has a length of 1, capable of holding 64 bits, term ids [0, 63]
let missing_sentinel = 64; // num_terms + 1 - 1, 
let byte_pos = missing_sentinel / 64; // 1 - idx 1 >= slice length 1
let bit_pos = missing_sentinel % 64; // 0 
```
We try to access a byte outside of the bounds of the boxed slice,
causing a panic from the bounds check to failing.

## Fixing it
The fix is simple. If we need to account for the missing sentinel,
initialize the `BitSet` with capacity for one more bit.

## Tests
- Added a bunch of unit tests that hit these conditions. I ensured they
failed without the fix, and that they now pass.
- All unit tests pass with the fix in place

## Other
- The investigation that led to finding this bug began with
https://github.com/paradedb/paradedb/issues/4746.
 2026-04-25 14:11:47 +02:00
Abdul Andha
4fbae92187 send after key on last page 2026-04-24 15:33:26 -04:00
Cameron
89f0cef807 Fix O(2^n) query parser regression for deeply-nested queries (#2905) 
* Fix O(2^n) query parser regression for deeply-nested queries

The top-level `ast()` parser used `alt((boolean_expr, single_leaf))` at
every group level. When the group contained a single leaf with no
trailing operand, `boolean_expr` would parse `occur_leaf` (recursing
into the inner group), fail at `multispace1`, backtrack, and then
`single_leaf` would re-parse `occur_leaf` from scratch. Every nesting
level doubled the work, giving O(2^n) time for queries like
`(((((title:test)))))`.

Parse `occur_leaf` once and peek ahead for a trailing operand instead
of backtracking. This keeps parsing O(n) and also avoids the duplicate
parse for simple single-leaf queries.

Fixes #2498.

Measured on the issue reproducer (release build):

    depth   before     after
       20   0.87 s   <1 us
       25  28.23 s   <1 us
       60  (years)   ~5 us

Non-pathological queries are unaffected or slightly faster:

    query                     before     after
    hello                     650 ns     308 ns
    a AND b AND c            1380 ns    1364 ns
    title:rust AND (...)     3426 ns    3460 ns

All 53 existing grammar tests and 56 query_parser tests pass. Adds a
regression test at depth 60 that would not complete under the old
parser.

* Add ignored benchmark for nested query parsing at depth 20/21

Matches the depths from issue #2498 which reported 0.87 s / 1.72 s
under the regression. With the fix these parse in single-digit
microseconds. Runs via:

  cargo test -p tantivy-query-grammar --release bench_deeply_nested \
      -- --ignored --nocapture

* Propagate Err::Failure and Err::Incomplete from operand parser

`alt((boolean_expr, single_leaf))` only retried on `Err::Error` and
propagated `Err::Failure` and `Err::Incomplete`. The replacement was
catching all three with `Err(_)`, which would silently fall back to
a single leaf if any cut point were ever added to `operand_leaf` or
its descendants. Match specifically on `Err::Error` to preserve the
original `alt` semantics.

* Replace inline bench with binggan bench in benches/

Move the nested-query benchmark out of the query-grammar test module
and into a proper binggan benchmark at benches/query_parser_nested.rs,
registered as a harnessless bench in Cargo.toml. Keeps the correctness
regression test (depth 60) in place.

Run with: cargo bench --bench query_parser_nested

* Fix rustfmt import ordering in query_parser_nested bench
 2026-04-24 03:54:00 -04:00
Claude
a5d297c75f Add OpenSSF Scorecard workflow 
Runs weekly security analysis and uploads SARIF results to GitHub code
scanning. Third-party actions are pinned by commit SHA. Adds the Scorecard
badge to the README.

Based on quickwit-oss/quickwit#5969.
 2026-04-24 06:56:58 +00:00
Pascal Seitz
2e16243f9a fix memory consumption for histogram 2026-04-21 13:58:39 +02:00
Pascal Seitz
e015abab8e docs: add 0.26.1 changelog entry for aggregation perf fix 2026-04-21 11:12:37 +02:00
Pascal Seitz
73c711ec74 perf(agg): only measure active parent bucket in composite collect 
Same change as 26a589e for SegmentCompositeCollector: get_memory_consumption
summed across all parent_buckets on every block, scaling with outer bucket
cardinality. Pass parent_bucket_id and index the single bucket.
 2026-04-21 07:26:58 +02:00
Pascal Seitz
cb037c8079 add inline 2026-04-21 07:26:58 +02:00
Pascal Seitz
ed3453606b agg fix: compute memory consumption only for current bucket 2026-04-21 07:26:58 +02:00
Pascal Seitz
e9641f99c5 add nested term benchmark 2026-04-21 07:26:58 +02:00
Paul Masurel
13d74c3c20 Update binggan requirement from 0.16.0 to 0.16.1 (#2899) 2026-04-20 11:59:47 +02:00
Claude
3a6a3de8d7 ci: update pinned Action SHAs to current latest versions 
The previous commit pinned actions to commit SHAs but used stale
version tags (v4.2.2, v2.7.5, old nextest/cargo-llvm-cov refs).
Update to the actual current HEAD of each pinned tag:

  actions/checkout        v4.2.2 → v4.3.1  (34e114876b0b...)
  Swatinem/rust-cache     v2.7.5 → v2.9.1  (c19371144df3...)
  taiki-e/install-action  nextest           (56cc9adf3a3e...)
  taiki-e/install-action  cargo-llvm-cov    (e4b3a0453201...)

actions-rs/toolchain, actions-rs/clippy-check, and
codecov/codecov-action SHAs were already correct.

https://claude.ai/code/session_01VD7Bo8upj3cQwWDf9ni2Ln
 2026-04-16 06:49:47 +00:00
Claude
af3c6c0070 ci: pin GitHub Actions to full commit SHAs and restrict token permissions 
Fixes two supply chain / token security issues:

- Pin all third-party Actions to immutable full commit SHAs instead of
  mutable version tags (addresses unpinned-dependencies risk, analogous
  to quickwit-oss/quickwit#5937):
    actions/checkout v4.2.2
    actions-rs/toolchain v1.0.7
    Swatinem/rust-cache v2.7.5
    taiki-e/install-action nextest / cargo-llvm-cov
    actions-rs/clippy-check v1.0.7
    codecov/codecov-action v3.1.6

- Add explicit least-privilege `permissions` blocks at workflow and job
  level (addresses excessive GITHUB_TOKEN permissions, analogous to
  quickwit-oss/quickwit#5945):
    default: contents: read
    check job: also grants checks: write (required by clippy-check)

https://claude.ai/code/session_01VD7Bo8upj3cQwWDf9ni2Ln
 2026-04-15 20:55:43 +00:00
dependabot[bot]
058afff8b7 Update binggan requirement from 0.15.3 to 0.16.0 
Updates the requirements on [binggan](https://github.com/pseitz/binggan) to permit the latest version.
- [Changelog](https://github.com/PSeitz/binggan/blob/main/CHANGELOG.md)
- [Commits](https://github.com/pseitz/binggan/commits)

---
updated-dependencies:
- dependency-name: binggan
  dependency-version: 0.16.0
  dependency-type: direct:production
...

Signed-off-by: dependabot[bot] <support@github.com>
 2026-04-15 08:58:03 +02:00
Paul Masurel
58aa4b7074 Fix cardinality aggregation using invalid coupons (#2893) 
Previously, coupons were computed via murmurhash32 and fed as raw u32
to the HLL sketch, bypassing the sketch's internal hashing and producing
invalid (slot, value) pairs. Switch to Coupon::from_hash from the
datasketches crate which correctly derives coupons, and drop the
now-unused murmurhash32 dependency.

Co-authored-by: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
 2026-04-13 19:14:30 +02:00
Paul Masurel
04beab3b29 Performance improvement for nested cardinality aggregation 
When a string cardinality aggregation is nested it end up being applied to different buckets.
Dictionary encoding relies on a different dictionaries for each segment.

As a result, during segment collection, we only collect term ordinals in a HashSet, and decode them in the
term dictionary at the end of collection.

Before this PR, this decoding phase was done once for each bucket, causing the same work to be done over and over. This PR introduce a coupon cache. The HLL sketch relies on a hash of the string values.

We populate the cache before bucket collection, and get our values from it.

This PR also rename "caching" "buffering" in aggregation (it was never caching), and does several cleanups.
 2026-04-10 14:51:00 +02:00
alexanderbianchi
3cd9011f87 Make BucketEntries::iter, PercentileValuesVecEntry fields, and TopNComputer::threshold public (#2890) 
These items need to be accessible from the tantivy-datafusion crate:
- BucketEntries::iter() for iterating aggregation bucket results
- PercentileValuesVecEntry.key/.value for reading percentile results
- TopNComputer.threshold for Block-WAND score pruning in the inverted index provider

Co-authored-by: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Co-authored-by: Paul Masurel <paul@quickwit.io>
 2026-04-09 13:32:31 +02:00
Paul Masurel
d2c1b8bc2c Optimized intersection count using a bitset when the first leg is dense 2026-04-06 12:01:52 -04:00
nuri
a65107135a Use BinaryHeap for score-based top-K collection (#2881) 
* Use BinaryHeap for score-based top-K collection

* Use peek_mut and add proptest for TopNHeap

---------

Co-authored-by: nryoo <nryoo@nryooui-MacBookPro.local>
 2026-04-04 19:49:05 +02:00
Pascal Seitz
5c344db1bf chore: Release 2026-03-31 17:15:34 +08:00
Pascal Seitz
dc0f31554d unbump for release and update Changelog.md 2026-03-31 17:15:34 +08:00
trinity-1686a
a28ce3ee54 Merge pull request #2869 from quickwit-oss/trinity.pointard/maint 
add dependabot cooldown
 2026-03-31 09:52:22 +02:00
trinity Pointard
cf9800f981 add dependabot cooldown 2026-03-30 11:36:04 +02:00
72 changed files with 5785 additions and 1217 deletions
Expand all files Collapse all files

4
.github/dependabot.yml vendored
View File

@@ -6,6 +6,8 @@ updates:
interval: daily
time: "20:00"
open-pull-requests-limit: 10
cooldown:
default-days: 2
- package-ecosystem: "github-actions"
directory: "/"
@@ -13,3 +15,5 @@ updates:
interval: daily
time: "20:00"
open-pull-requests-limit: 10
cooldown:
default-days: 2

15
.github/workflows/coverage.yml vendored
View File

@@ -4,6 +4,9 @@ on:
push:
branches: [main]
permissions:
contents: read
# Ensures that we cancel running jobs for the same PR / same workflow.
concurrency:
group: ${{ github.workflow }}-${{ github.event.pull_request.number || github.ref }}
@@ -12,16 +15,20 @@ concurrency:
jobs:
coverage:
runs-on: ubuntu-latest
permissions:
contents: read
steps:
- uses: actions/checkout@v4
- uses: actions/checkout@df4cb1c069e1874edd31b4311f1884172cec0e10 # v6.0.3
- name: Install Rust
run: rustup toolchain install nightly-2025-12-01 --profile minimal --component llvm-tools-preview
- uses: Swatinem/rust-cache@v2
- uses: taiki-e/install-action@cargo-llvm-cov
- uses: Swatinem/rust-cache@c19371144df3bb44fab255c43d04cbc2ab54d1c4 # v2.9.1
- uses: taiki-e/install-action@e4b3a0453201addddc06d3a72db90326aad87084 # cargo-llvm-cov
- name: Generate code coverage
run: cargo +nightly-2025-12-01 llvm-cov --all-features --workspace --doctests --lcov --output-path lcov.info
- name: Upload coverage to Codecov
uses: codecov/codecov-action@v3
uses: codecov/codecov-action@fb8b3582c8e4def4969c97caa2f19720cb33a72f # v7.0.0
continue-on-error: true
with:
token: ${{ secrets.CODECOV_TOKEN }} # not required for public repos

10
.github/workflows/long_running.yml vendored
View File

@@ -8,6 +8,9 @@ env:
CARGO_TERM_COLOR: always
NUM_FUNCTIONAL_TEST_ITERATIONS: 20000
permissions:
contents: read
# Ensures that we cancel running jobs for the same PR / same workflow.
concurrency:
group: ${{ github.workflow }}-${{ github.event.pull_request.number || github.ref }}
@@ -18,10 +21,13 @@ jobs:
runs-on: ubuntu-latest
permissions:
contents: read
steps:
- uses: actions/checkout@v4
- uses: actions/checkout@df4cb1c069e1874edd31b4311f1884172cec0e10 # v6.0.3
- name: Install stable
uses: actions-rs/toolchain@v1
uses: actions-rs/toolchain@16499b5e05bf2e26879000db0c1d13f7e13fa3af # v1.0.7
with:
toolchain: stable
profile: minimal

49
.github/workflows/scorecard.yml vendored Normal file
View File

@@ -0,0 +1,49 @@
name: OpenSSF Scorecard
on:
schedule:
- cron: '0 0 * * 0'
push:
branches:
- main
permissions:
contents: read
jobs:
analysis:
name: Scorecards analysis
runs-on: ubuntu-latest
permissions:
# Needed to upload the results to code-scanning dashboard.
security-events: write
# Needed to publish results
id-token: write
steps:
- name: 'Checkout code'
uses: actions/checkout@df4cb1c069e1874edd31b4311f1884172cec0e10 # v6.0.3
with:
persist-credentials: false
- name: 'Run analysis'
uses: ossf/scorecard-action@4eaacf0543bb3f2c246792bd56e8cdeffafb205a # v2.4.3
with:
results_file: results.sarif
results_format: sarif
repo_token: ${{ secrets.GITHUB_TOKEN }}
publish_results: true
# Upload the results as artifacts.
- name: 'Upload artifact'
uses: actions/upload-artifact@043fb46d1a93c77aae656e7c1c64a875d1fc6a0a # v7.0.1
with:
name: SARIF file
path: results.sarif
retention-days: 5
# Upload the results to GitHub's code scanning dashboard.
- name: 'Upload to code-scanning'
uses: github/codeql-action/upload-sarif@87557b9c84dde89fdd9b10e88954ac2f4248e463 # v4.36.1
with:
sarif_file: results.sarif

28
.github/workflows/test.yml vendored
View File

@@ -9,6 +9,9 @@ on:
env:
CARGO_TERM_COLOR: always
permissions:
contents: read
# Ensures that we cancel running jobs for the same PR / same workflow.
concurrency:
group: ${{ github.workflow }}-${{ github.event.pull_request.number || github.ref }}
@@ -19,23 +22,27 @@ jobs:
runs-on: ubuntu-latest
permissions:
contents: read
checks: write
steps:
- uses: actions/checkout@v4
- uses: actions/checkout@df4cb1c069e1874edd31b4311f1884172cec0e10 # v6.0.3
- name: Install nightly
uses: actions-rs/toolchain@v1
uses: actions-rs/toolchain@16499b5e05bf2e26879000db0c1d13f7e13fa3af # v1.0.7
with:
toolchain: nightly
profile: minimal
components: rustfmt
- name: Install stable
uses: actions-rs/toolchain@v1
uses: actions-rs/toolchain@16499b5e05bf2e26879000db0c1d13f7e13fa3af # v1.0.7
with:
toolchain: stable
profile: minimal
components: clippy
- uses: Swatinem/rust-cache@v2
- uses: Swatinem/rust-cache@c19371144df3bb44fab255c43d04cbc2ab54d1c4 # v2.9.1
- name: Check Formatting
run: cargo +nightly fmt --all -- --check
@@ -47,7 +54,7 @@ jobs:
- name: Check Bench Compilation
run: cargo +nightly bench --no-run --profile=dev --all-features
- uses: actions-rs/clippy-check@v1
- uses: actions-rs/clippy-check@b5b5f21f4797c02da247df37026fcd0a5024aa4d # v1.0.7
with:
toolchain: stable
token: ${{ secrets.GITHUB_TOKEN }}
@@ -57,6 +64,9 @@ jobs:
runs-on: ubuntu-latest
permissions:
contents: read
strategy:
matrix:
features:
@@ -67,17 +77,17 @@ jobs:
name: test-${{ matrix.features.label}}
steps:
- uses: actions/checkout@v4
- uses: actions/checkout@df4cb1c069e1874edd31b4311f1884172cec0e10 # v6.0.3
- name: Install stable
uses: actions-rs/toolchain@v1
uses: actions-rs/toolchain@16499b5e05bf2e26879000db0c1d13f7e13fa3af # v1.0.7
with:
toolchain: stable
profile: minimal
override: true
- uses: taiki-e/install-action@nextest
- uses: Swatinem/rust-cache@v2
- uses: taiki-e/install-action@56cc9adf3a3e2c23eafb56e8acaf9d0373cb845a # nextest
- uses: Swatinem/rust-cache@c19371144df3bb44fab255c43d04cbc2ab54d1c4 # v2.9.1
- name: Run tests
run: |

5
CHANGELOG.md
View File

@@ -1,9 +1,8 @@
Tantivy 0.26.1
================================
## Bugfixes
- Fix memory consumption accounting in nested term aggregation to only scan the active parent bucket (@PSeitz)
- Fix memory consumption accounting in composite aggregation to only scan the active parent bucket (@PSeitz)
## Performance
- Fix quadratic runtime in nested term and composite aggregations: memory accounting scanned all parent buckets on every collect instead of just the current parent (@PSeitz @fulmicoton)
Tantivy 0.26 (Unreleased)
================================

16
Cargo.toml
View File

@@ -1,6 +1,6 @@
[package]
name = "tantivy"
version = "0.26.1"
version = "0.26.0"
authors = ["Paul Masurel <paul.masurel@gmail.com>"]
license = "MIT"
categories = ["database-implementations", "data-structures"]
@@ -65,7 +65,7 @@ tantivy-bitpacker = { version = "0.10", path = "./bitpacker" }
common = { version = "0.11", path = "./common/", package = "tantivy-common" }
tokenizer-api = { version = "0.7", path = "./tokenizer-api", package = "tantivy-tokenizer-api" }
sketches-ddsketch = { version = "0.4", features = ["use_serde"] }
datasketches = "0.2.0"
datasketches = { version = "0.3.0", features = ["hll"] }
futures-util = { version = "0.3.28", optional = true }
futures-channel = { version = "0.3.28", optional = true }
fnv = "1.0.7"
@@ -75,7 +75,7 @@ typetag = "0.2.21"
winapi = "0.3.9"
[dev-dependencies]
binggan = "0.15.3"
binggan = "0.17.0"
rand = "0.9"
maplit = "1.0.2"
matches = "0.1.9"
@@ -92,7 +92,7 @@ postcard = { version = "1.0.4", features = [
], default-features = false }
[target.'cfg(not(windows))'.dev-dependencies]
criterion = { version = "0.5", default-features = false }
criterion = { version = "0.8", default-features = false }
[dev-dependencies.fail]
version = "0.5.0"
@@ -201,3 +201,11 @@ harness = false
[[bench]]
name = "regex_all_terms"
harness = false
[[bench]]
name = "query_parser_nested"
harness = false
[[bench]]
name = "intersection_bench"
harness = false

1
README.md
View File

@@ -1,6 +1,7 @@
[![Docs](https://docs.rs/tantivy/badge.svg)](https://docs.rs/crate/tantivy/)
[![Build Status](https://github.com/quickwit-oss/tantivy/actions/workflows/test.yml/badge.svg)](https://github.com/quickwit-oss/tantivy/actions/workflows/test.yml)
[![codecov](https://codecov.io/gh/quickwit-oss/tantivy/branch/main/graph/badge.svg)](https://codecov.io/gh/quickwit-oss/tantivy)
[![OpenSSF Scorecard](https://api.scorecard.dev/projects/github.com/quickwit-oss/tantivy/badge)](https://scorecard.dev/viewer/?uri=github.com/quickwit-oss/tantivy)
[![Join the chat at https://discord.gg/MT27AG5EVE](https://shields.io/discord/908281611840282624?label=chat%20on%20discord)](https://discord.gg/MT27AG5EVE)
[![License: MIT](https://img.shields.io/badge/License-MIT-yellow.svg)](https://opensource.org/licenses/MIT)
[![Crates.io](https://img.shields.io/crates/v/tantivy.svg)](https://crates.io/crates/tantivy)

170
benches/agg_bench.rs
View File

@@ -63,7 +63,11 @@ fn bench_agg(mut group: InputGroup<Index>) {
register!(group, terms_all_unique_with_avg_sub_agg);
register!(group, terms_many_with_avg_sub_agg);
register!(group, terms_status_with_avg_sub_agg);
register!(group, terms_status_with_terms_zipf_1000_sub_agg);
register!(group, terms_zipf_1000_with_terms_status_sub_agg);
register!(group, terms_status_with_histogram);
register!(group, terms_status_with_date_histogram);
register!(group, terms_status_with_date_histogram_and_sibling_terms);
register!(group, terms_zipf_1000);
register!(group, terms_zipf_1000_with_histogram);
register!(group, terms_zipf_1000_with_avg_sub_agg);
@@ -77,7 +81,12 @@ fn bench_agg(mut group: InputGroup<Index>) {
register!(group, composite_histogram_calendar);
register!(group, cardinality_agg);
register!(group, cardinality_agg_high_card);
register!(group, cardinality_agg_low_card);
register!(group, terms_status_with_cardinality_agg);
register!(group, terms_100_buckets_with_cardinality_agg);
register!(group, terms_many_with_single_term_order_by_card);
register!(group, terms_many_with_single_term_2_order_by_card);
register!(group, range_agg);
register!(group, range_agg_with_avg_sub_agg);
@@ -165,10 +174,52 @@ fn cardinality_agg(index: &Index) {
});
execute_agg(index, agg_req);
}
// Full-scan cardinality on a near-1M-cardinality string field.
// Hits the dense (PagedBitset) path: every doc has a unique term,
// so the bucket promotes from FxHashSet shortly into the scan.
fn cardinality_agg_high_card(index: &Index) {
let agg_req = json!({
"cardinality": {
"cardinality": {
"field": "text_all_unique_terms"
},
}
});
execute_agg(index, agg_req);
}
// Full-scan cardinality on a tiny-cardinality string field (7 distinct
// values). Stays on the FxHashSet path — the promotion threshold is
// never crossed. Validates no regression on the sparse path.
fn cardinality_agg_low_card(index: &Index) {
let agg_req = json!({
"cardinality": {
"cardinality": {
"field": "text_few_terms_status"
},
}
});
execute_agg(index, agg_req);
}
fn terms_status_with_cardinality_agg(index: &Index) {
let agg_req = json!({
"my_texts": {
"terms": { "field": "text_few_terms_status" },
"aggs": {
"cardinality": {
"cardinality": {
"field": "text_few_terms_status"
},
}
}
},
});
execute_agg(index, agg_req);
}
fn terms_100_buckets_with_cardinality_agg(index: &Index) {
let agg_req = json!({
"my_texts": {
"terms": { "field": "text_1000_terms_zipf", "size": 100 },
"aggs": {
"cardinality": {
"cardinality": {
@@ -181,6 +232,58 @@ fn terms_status_with_cardinality_agg(index: &Index) {
execute_agg(index, agg_req);
}
fn terms_many_with_single_term_order_by_card(index: &Index) {
let agg_req = json!({
"my_texts": {
"terms": { "field": "text_many_terms" },
"aggs": {
"nested_terms": {
"terms": {
"field": "single_term",
"order": { "cardinality": "desc" }
},
"aggs": {
"cardinality": {
"cardinality": { "field": "text_few_terms" }
}
}
}
}
},
});
execute_agg(index, agg_req);
}
// Two-level terms ordered by cardinality at each level: a high-card outer terms
// (text_many_terms) ordered by a cardinality sub-agg, with a nested low-card terms
// (text_few_terms_status) also ordered by a cardinality sub-agg, plus an avg.
fn terms_many_with_single_term_2_order_by_card(index: &Index) {
let agg_req = json!({
"by_ip": {
"terms": {
"field": "text_many_terms",
"order": { "card_few_terms": "desc" }
},
"aggs": {
"card_few_terms": {
"cardinality": { "field": "text_few_terms" }
},
"nested_terms": {
"terms": {
"field": " single_term",
"order": { "distinct_path2": "desc" }
},
"aggs": {
"avg_botscore": { "avg": { "field": "score" } },
"distinct_path2": { "cardinality": { "field": "text_few_terms" } }
}
}
}
}
});
execute_agg(index, agg_req);
}
fn terms_7(index: &Index) {
let agg_req = json!({
"my_texts": { "terms": { "field": "text_few_terms_status" } },
@@ -253,6 +356,30 @@ fn terms_all_unique_with_avg_sub_agg(index: &Index) {
});
execute_agg(index, agg_req);
}
fn terms_status_with_terms_zipf_1000_sub_agg(index: &Index) {
let agg_req = json!({
"my_texts": {
"terms": { "field": "text_few_terms_status" },
"aggs": {
"nested_terms": { "terms": { "field": "text_1000_terms_zipf" } }
}
}
});
execute_agg(index, agg_req);
}
fn terms_zipf_1000_with_terms_status_sub_agg(index: &Index) {
let agg_req = json!({
"my_texts": {
"terms": { "field": "text_1000_terms_zipf" },
"aggs": {
"nested_terms": { "terms": { "field": "text_few_terms_status" } }
}
}
});
execute_agg(index, agg_req);
}
fn terms_status_with_histogram(index: &Index) {
let agg_req = json!({
"my_texts": {
@@ -265,6 +392,34 @@ fn terms_status_with_histogram(index: &Index) {
execute_agg(index, agg_req);
}
fn terms_status_with_date_histogram(index: &Index) {
let agg_req = json!({
"my_texts": {
"terms": { "field": "text_few_terms_status" },
"aggs": {
"over_time": { "date_histogram": { "field": "timestamp", "fixed_interval": "1h" } }
}
}
});
execute_agg(index, agg_req);
}
/// Same fused terms × date_histogram, but with a sibling terms aggregation next to it. The fused
/// fast path should still trigger for `my_texts` (sibling aggregations are independent top-level
/// aggregations, so they don't change its eligibility).
fn terms_status_with_date_histogram_and_sibling_terms(index: &Index) {
let agg_req = json!({
"my_texts": {
"terms": { "field": "text_few_terms_status" },
"aggs": {
"over_time": { "date_histogram": { "field": "timestamp", "fixed_interval": "1h" } }
}
},
"other_texts": { "terms": { "field": "text_few_terms" } }
});
execute_agg(index, agg_req);
}
fn terms_zipf_1000_with_histogram(index: &Index) {
let agg_req = json!({
"my_texts": {
@@ -566,7 +721,8 @@ fn get_test_index_bench(cardinality: Cardinality) -> tantivy::Result<Index> {
TextFieldIndexing::default().set_index_option(IndexRecordOption::WithFreqs),
)
.set_stored();
let text_field = schema_builder.add_text_field("text", text_fieldtype);
let text_field = schema_builder.add_text_field("text", text_fieldtype.clone());
let single_term = schema_builder.add_text_field("single_term", FAST);
let json_field = schema_builder.add_json_field("json", FAST);
let text_field_all_unique_terms =
schema_builder.add_text_field("text_all_unique_terms", STRING | FAST);
@@ -630,6 +786,8 @@ fn get_test_index_bench(cardinality: Cardinality) -> tantivy::Result<Index> {
index_writer.add_document(doc!(
json_field => json!({"mixed_type": 10.0}),
json_field => json!({"mixed_type": 10.0}),
single_term => "single_term",
single_term => "single_term",
text_field => "cool",
text_field => "cool",
text_field_all_unique_terms => "cool",
@@ -655,7 +813,9 @@ fn get_test_index_bench(cardinality: Cardinality) -> tantivy::Result<Index> {
doc_with_value /= 20;
}
let _val_max = 1_000_000.0;
for _ in 0..doc_with_value {
const SPAN_MS: i64 = 120 * 3600 * 1000; // 120 hours in ms
const NOISE_MS: i64 = 2 * 3600 * 1000; // ±2h noise
for i in 0..doc_with_value {
let val: f64 = rng.random_range(0.0..1_000_000.0);
let json = if rng.random_bool(0.1) {
// 10% are numeric values
@@ -663,7 +823,11 @@ fn get_test_index_bench(cardinality: Cardinality) -> tantivy::Result<Index> {
} else {
json!({"mixed_type": many_terms_data.choose(&mut rng).unwrap().to_string()})
};
let base_ms = (i as i64 * SPAN_MS) / doc_with_value as i64;
let noise_ms = rng.random_range(-NOISE_MS..NOISE_MS);
let ts_ms = (base_ms + noise_ms).clamp(0, SPAN_MS);
index_writer.add_document(doc!(
single_term => "single_term",
text_field => "cool",
json_field => json,
text_field_all_unique_terms => format!("unique_term_{}", rng.random::<u64>()),
@@ -674,7 +838,7 @@ fn get_test_index_bench(cardinality: Cardinality) -> tantivy::Result<Index> {
score_field => val as u64,
score_field_f64 => lg_norm.sample(&mut rng),
score_field_i64 => val as i64,
date_field => DateTime::from_timestamp_millis((val * 1_000_000.) as i64),
date_field => DateTime::from_timestamp_millis(ts_ms),
))?;
if cardinality == Cardinality::OptionalSparse {
for _ in 0..20 {

149
benches/intersection_bench.rs Normal file
View File

@@ -0,0 +1,149 @@
// Benchmarks top-K intersection of term scorers (block_wand_intersection).
//
// What's measured:
// - Conjunctive queries (+a +b, +a +b +c) with top-10 by score
// - Varying doc-frequency balance between terms (balanced, skewed, very skewed)
// - Realistic term frequencies (geometric distribution, mostly low)
// - 1M-doc single segment
//
// Run with: cargo bench --bench intersection_bench
use binggan::{black_box, BenchRunner};
use rand::prelude::*;
use rand::rngs::StdRng;
use rand::SeedableRng;
use tantivy::collector::TopDocs;
use tantivy::query::QueryParser;
use tantivy::schema::{Schema, TEXT};
use tantivy::{doc, Index, ReloadPolicy, Searcher};
const NUM_DOCS: usize = 1_000_000;
struct BenchIndex {
searcher: Searcher,
query_parser: QueryParser,
}
/// Generate term frequency from a geometric-like distribution.
/// Most values are 1, a few are 2-3, rarely higher.
/// p controls the decay: higher p → more weight on tf=1.
fn random_term_freq(rng: &mut StdRng, p: f64) -> u32 {
let mut tf = 1u32;
while tf < 10 && rng.random_bool(1.0 - p) {
tf += 1;
}
tf
}
/// Build an index with three terms (a, b, c) with given doc-frequency probabilities.
/// Each term occurrence has a realistic term frequency (geometric distribution).
/// Field length is padded with filler tokens to create varied fieldnorms.
fn build_index(p_a: f64, p_b: f64, p_c: f64) -> BenchIndex {
let mut schema_builder = Schema::builder();
let body = schema_builder.add_text_field("body", TEXT);
let schema = schema_builder.build();
let index = Index::create_in_ram(schema);
let mut rng = StdRng::from_seed([42u8; 32]);
{
let mut writer = index.writer_with_num_threads(1, 500_000_000).unwrap();
for _ in 0..NUM_DOCS {
let mut tokens: Vec<String> = Vec::new();
if rng.random_bool(p_a) {
let tf = random_term_freq(&mut rng, 0.7);
for _ in 0..tf {
tokens.push("aaa".to_string());
}
}
if rng.random_bool(p_b) {
let tf = random_term_freq(&mut rng, 0.7);
for _ in 0..tf {
tokens.push("bbb".to_string());
}
}
if rng.random_bool(p_c) {
let tf = random_term_freq(&mut rng, 0.7);
for _ in 0..tf {
tokens.push("ccc".to_string());
}
}
// Pad with filler to create varied field lengths (5-30 tokens).
let filler_count = rng.random_range(5u32..30u32);
for _ in 0..filler_count {
tokens.push("filler".to_string());
}
let text = tokens.join(" ");
writer.add_document(doc!(body => text)).unwrap();
}
writer.commit().unwrap();
}
let reader = index
.reader_builder()
.reload_policy(ReloadPolicy::Manual)
.try_into()
.unwrap();
let searcher = reader.searcher();
let query_parser = QueryParser::for_index(&index, vec![body]);
BenchIndex {
searcher,
query_parser,
}
}
fn main() {
// Scenarios: (label, p_a, p_b, p_c)
//
// "balanced": all terms ~10% → intersection ~1% of docs
// "skewed": one common (50%), one rare (2%) → intersection ~1%
// "very_skewed": one very common (80%), one very rare (0.5%) → intersection ~0.4%
// "three_balanced": three terms ~20% each → intersection ~0.8%
// "three_skewed": 50% / 10% / 2% → intersection ~0.1%
let scenarios: Vec<(&str, f64, f64, f64)> = vec![
("balanced_10%_10%", 0.10, 0.10, 0.0),
("skewed_50%_2%", 0.50, 0.02, 0.0),
("very_skewed_80%_0.5%", 0.80, 0.005, 0.0),
("three_balanced_20%_20%_20%", 0.20, 0.20, 0.20),
("three_skewed_50%_10%_2%", 0.50, 0.10, 0.02),
];
let mut runner = BenchRunner::new();
for (label, p_a, p_b, p_c) in &scenarios {
let bench_index = build_index(*p_a, *p_b, *p_c);
let mut group = runner.new_group();
group.set_name(format!("intersection — {label}"));
// Two-term intersection
if *p_a > 0.0 && *p_b > 0.0 {
let query_str = "+aaa +bbb";
let query = bench_index.query_parser.parse_query(query_str).unwrap();
let searcher = bench_index.searcher.clone();
group.register(format!("{query_str} top10"), move |_| {
let collector = TopDocs::with_limit(10).order_by_score();
black_box(searcher.search(&query, &collector).unwrap());
1usize
});
}
// Three-term intersection
if *p_c > 0.0 {
let query_str = "+aaa +bbb +ccc";
let query = bench_index.query_parser.parse_query(query_str).unwrap();
let searcher = bench_index.searcher.clone();
group.register(format!("{query_str} top10"), move |_| {
let collector = TopDocs::with_limit(10).order_by_score();
black_box(searcher.search(&query, &collector).unwrap());
1usize
});
}
group.run();
}
}

35
benches/query_parser_nested.rs Normal file
View File

@@ -0,0 +1,35 @@
// Benchmark for the query grammar parsing deeply nested queries.
//
// Regression guard for https://github.com/quickwit-oss/tantivy/issues/2498:
// at depth 20/21 the old parser took 0.87 s / 1.72 s respectively because
// `ast()` retried `occur_leaf` on backtrack, giving O(2^n) time. With the
// fix parsing is linear and completes in microseconds.
//
// Run with: `cargo bench --bench query_parser_nested`.
use binggan::{black_box, BenchRunner};
use tantivy::query_grammar::parse_query;
fn nested_query(depth: usize, leading_plus: bool) -> String {
let leading = "(".repeat(depth);
let trailing = ")".repeat(depth);
let prefix = if leading_plus { "+" } else { "" };
format!("{prefix}{leading}title:test{trailing}")
}
fn main() {
let mut runner = BenchRunner::new();
for depth in [20, 21] {
for leading_plus in [false, true] {
let query = nested_query(depth, leading_plus);
let label = format!(
"parse_nested_depth_{depth}_{}",
if leading_plus { "plus" } else { "plain" },
);
runner.bench_function(&label, move |_| {
black_box(parse_query(black_box(&query)).unwrap());
});
}
}
}

5
bitpacker/Cargo.toml
View File

@@ -18,5 +18,10 @@ homepage = "https://github.com/quickwit-oss/tantivy"
bitpacking = { version = "0.9.2", default-features = false, features = ["bitpacker1x"] }
[dev-dependencies]
binggan = "0.17.0"
rand = "0.9"
proptest = "1"
[[bench]]
name = "bench"
harness = false

153
bitpacker/benches/bench.rs
View File

@@ -1,65 +1,110 @@
#![feature(test)]
use std::cell::RefCell;
extern crate test;
use binggan::{BenchRunner, black_box};
use rand::rng;
use rand::seq::IteratorRandom;
use tantivy_bitpacker::{BitPacker, BitUnpacker, BlockedBitpacker};
#[cfg(test)]
mod tests {
use rand::rng;
use rand::seq::IteratorRandom;
use tantivy_bitpacker::{BitPacker, BitUnpacker, BlockedBitpacker};
use test::Bencher;
fn create_bitpacked_data(bit_width: u8, num_els: u32) -> Vec<u8> {
let mut bitpacker = BitPacker::new();
let mut buffer = Vec::new();
for _ in 0..num_els {
bitpacker.write(0u64, bit_width, &mut buffer).unwrap();
bitpacker.flush(&mut buffer).unwrap();
}
buffer
}
#[inline(never)]
fn create_bitpacked_data(bit_width: u8, num_els: u32) -> Vec<u8> {
let mut bitpacker = BitPacker::new();
let mut buffer = Vec::new();
for _ in 0..num_els {
// the values do not matter.
bitpacker.write(0u64, bit_width, &mut buffer).unwrap();
bitpacker.flush(&mut buffer).unwrap();
const N: usize = 100_000;
const MAX_VAL: u64 = 1_000;
const BIT_WIDTH: u8 = 10; // 2^10 = 1024 > MAX_VAL
fn create_packed_data() -> (BitUnpacker, Vec<u8>) {
let mut bitpacker = BitPacker::new();
let mut data = Vec::new();
for i in 0..N as u64 {
let val = i * MAX_VAL / N as u64;
bitpacker.write(val, BIT_WIDTH, &mut data).unwrap();
}
bitpacker.close(&mut data).unwrap();
(BitUnpacker::new(BIT_WIDTH), data)
}
fn bench_bitpacking() {
let mut runner = BenchRunner::new();
let bit_width = 3;
let num_els = 1_000_000u32;
let bit_unpacker = BitUnpacker::new(bit_width);
let data = create_bitpacked_data(bit_width, num_els);
let idxs: Vec<u32> = (0..num_els).choose_multiple(&mut rng(), 100_000);
runner.bench_function("bitpacking_read", move |_| {
let mut out = 0u64;
for &idx in &idxs {
out = out.wrapping_add(bit_unpacker.get(idx, &data[..]));
}
buffer
}
black_box(out);
});
}
#[bench]
fn bench_bitpacking_read(b: &mut Bencher) {
let bit_width = 3;
let num_els = 1_000_000u32;
let bit_unpacker = BitUnpacker::new(bit_width);
let data = create_bitpacked_data(bit_width, num_els);
let idxs: Vec<u32> = (0..num_els).choose_multiple(&mut rng(), 100_000);
b.iter(|| {
let mut out = 0u64;
for &idx in &idxs {
out = out.wrapping_add(bit_unpacker.get(idx, &data[..]));
}
out
});
fn bench_blocked_bitpacker() {
let mut runner = BenchRunner::new();
let mut blocked_bitpacker = BlockedBitpacker::new();
for val in 0..=21500 {
blocked_bitpacker.add(val * val);
}
#[bench]
fn bench_blockedbitp_read(b: &mut Bencher) {
runner.bench_function("blockedbitp_read", move |_| {
let mut out = 0u64;
for val in 0..=21500 {
out = out.wrapping_add(blocked_bitpacker.get(val));
}
black_box(out);
});
runner.bench_function("blockedbitp_create", |_| {
let mut blocked_bitpacker = BlockedBitpacker::new();
for val in 0..=21500 {
blocked_bitpacker.add(val * val);
}
b.iter(|| {
let mut out = 0u64;
for val in 0..=21500 {
out = out.wrapping_add(blocked_bitpacker.get(val));
}
out
});
}
#[bench]
fn bench_blockedbitp_create(b: &mut Bencher) {
b.iter(|| {
let mut blocked_bitpacker = BlockedBitpacker::new();
for val in 0..=21500 {
blocked_bitpacker.add(val * val);
}
blocked_bitpacker
});
}
black_box(blocked_bitpacker);
});
}
fn bench_filter_vec() {
let mut runner = BenchRunner::new();
let (unpacker, data) = create_packed_data();
let positions = RefCell::new(Vec::with_capacity(N));
runner.bench_function("filter_vec_dense", move |_| {
unpacker.get_ids_for_value_range(
250..=750,
0..N as u32,
&data,
&mut positions.borrow_mut(),
);
black_box(positions.borrow().len());
});
let (unpacker, data) = create_packed_data();
let positions = RefCell::new(Vec::with_capacity(N));
runner.bench_function("filter_vec_sparse", move |_| {
unpacker.get_ids_for_value_range(0..=50, 0..N as u32, &data, &mut positions.borrow_mut());
black_box(positions.borrow().len());
});
let (unpacker, data) = create_packed_data();
let positions = RefCell::new(Vec::with_capacity(N));
runner.bench_function("filter_vec_full", move |_| {
unpacker.get_ids_for_value_range(
0..=MAX_VAL,
0..N as u32,
&data,
&mut positions.borrow_mut(),
);
black_box(positions.borrow().len());
});
}
fn main() {
bench_bitpacking();
bench_blocked_bitpacker();
bench_filter_vec();
}

160
bitpacker/src/filter_vec/mod.rs
View File

@@ -1,8 +1,17 @@
#[cfg(all(target_arch = "aarch64", not(target_vendor = "apple")))]
use std::arch::is_aarch64_feature_detected;
use std::ops::RangeInclusive;
#[cfg(target_arch = "x86_64")]
mod avx2;
#[cfg(target_arch = "aarch64")]
mod neon;
// SVE intrinsics are not exposed on aarch64-apple-darwin.
#[cfg(all(target_arch = "aarch64", not(target_vendor = "apple")))]
mod sve;
mod scalar;
#[derive(Clone, Copy, Eq, PartialEq, Debug)]
@@ -10,6 +19,10 @@ mod scalar;
enum FilterImplPerInstructionSet {
#[cfg(target_arch = "x86_64")]
AVX2 = 0u8,
#[cfg(all(target_arch = "aarch64", not(target_vendor = "apple")))]
SVE = 3u8,
#[cfg(target_arch = "aarch64")]
Neon = 2u8,
Scalar = 1u8,
}
@@ -19,29 +32,57 @@ impl FilterImplPerInstructionSet {
match *self {
#[cfg(target_arch = "x86_64")]
FilterImplPerInstructionSet::AVX2 => is_x86_feature_detected!("avx2"),
#[cfg(all(target_arch = "aarch64", not(target_vendor = "apple")))]
FilterImplPerInstructionSet::SVE => is_aarch64_feature_detected!("sve"),
// TIL Neon is required on aarch 64.
#[cfg(target_arch = "aarch64")]
FilterImplPerInstructionSet::Neon => true,
FilterImplPerInstructionSet::Scalar => true,
}
}
}
// List of available implementation in preferred order.
// List of available implementations in preferred order.
#[cfg(target_arch = "x86_64")]
const IMPLS: [FilterImplPerInstructionSet; 2] = [
FilterImplPerInstructionSet::AVX2,
FilterImplPerInstructionSet::Scalar,
];
#[cfg(not(target_arch = "x86_64"))]
// Non-Apple aarch64: try SVE, NEON, Scalar.
#[cfg(all(target_arch = "aarch64", not(target_vendor = "apple")))]
const IMPLS: [FilterImplPerInstructionSet; 3] = [
FilterImplPerInstructionSet::SVE,
FilterImplPerInstructionSet::Neon,
FilterImplPerInstructionSet::Scalar,
];
// Apple aarch64 (M-series): SVE not available; use NEON or Scalar.
#[cfg(all(target_arch = "aarch64", target_vendor = "apple"))]
const IMPLS: [FilterImplPerInstructionSet; 2] = [
FilterImplPerInstructionSet::Neon,
FilterImplPerInstructionSet::Scalar,
];
#[cfg(not(any(target_arch = "x86_64", target_arch = "aarch64")))]
const IMPLS: [FilterImplPerInstructionSet; 1] = [FilterImplPerInstructionSet::Scalar];
impl FilterImplPerInstructionSet {
#[inline]
#[allow(unused_variables)] // on non-x86_64, code is unused.
#[allow(unused_variables)]
fn from(code: u8) -> FilterImplPerInstructionSet {
#[cfg(target_arch = "x86_64")]
if code == FilterImplPerInstructionSet::AVX2 as u8 {
return FilterImplPerInstructionSet::AVX2;
}
#[cfg(all(target_arch = "aarch64", not(target_vendor = "apple")))]
if code == FilterImplPerInstructionSet::SVE as u8 {
return FilterImplPerInstructionSet::SVE;
}
#[cfg(target_arch = "aarch64")]
if code == FilterImplPerInstructionSet::Neon as u8 {
return FilterImplPerInstructionSet::Neon;
}
FilterImplPerInstructionSet::Scalar
}
@@ -50,6 +91,13 @@ impl FilterImplPerInstructionSet {
match self {
#[cfg(target_arch = "x86_64")]
FilterImplPerInstructionSet::AVX2 => avx2::filter_vec_in_place(range, offset, output),
#[cfg(all(target_arch = "aarch64", not(target_vendor = "apple")))]
// SAFETY: SVE availability was verified by is_available() before selecting this impl.
FilterImplPerInstructionSet::SVE => unsafe {
sve::filter_vec_in_place(range, offset, output)
},
#[cfg(target_arch = "aarch64")]
FilterImplPerInstructionSet::Neon => neon::filter_vec_in_place(range, offset, output),
FilterImplPerInstructionSet::Scalar => {
scalar::filter_vec_in_place(range, offset, output)
}
@@ -57,6 +105,12 @@ impl FilterImplPerInstructionSet {
}
}
fn available_impls() -> impl Iterator<Item = FilterImplPerInstructionSet> {
IMPLS
.into_iter()
.filter(FilterImplPerInstructionSet::is_available)
}
#[inline]
fn get_best_available_instruction_set() -> FilterImplPerInstructionSet {
use std::sync::atomic::{AtomicU8, Ordering};
@@ -64,10 +118,7 @@ fn get_best_available_instruction_set() -> FilterImplPerInstructionSet {
let instruction_set_byte: u8 = INSTRUCTION_SET_BYTE.load(Ordering::Relaxed);
if instruction_set_byte == u8::MAX {
// Let's initialize the instruction set and cache it.
let instruction_set = IMPLS
.into_iter()
.find(FilterImplPerInstructionSet::is_available)
.unwrap();
let instruction_set = available_impls().next().unwrap();
INSTRUCTION_SET_BYTE.store(instruction_set as u8, Ordering::Relaxed);
return instruction_set;
}
@@ -80,12 +131,12 @@ pub fn filter_vec_in_place(range: RangeInclusive<u32>, offset: u32, output: &mut
#[cfg(test)]
mod tests {
use proptest::strategy::Strategy;
use super::*;
#[test]
fn test_get_best_available_instruction_set() {
// This does not test much unfortunately.
// We just make sure the function returns without crashing and returns the same result.
let instruction_set = get_best_available_instruction_set();
assert_eq!(get_best_available_instruction_set(), instruction_set);
}
@@ -102,6 +153,31 @@ mod tests {
}
}
#[cfg(all(target_arch = "aarch64", not(target_vendor = "apple")))]
#[test]
fn test_instruction_set_to_code_from_code() {
for instruction_set in [
FilterImplPerInstructionSet::SVE,
FilterImplPerInstructionSet::Neon,
FilterImplPerInstructionSet::Scalar,
] {
let code = instruction_set as u8;
assert_eq!(instruction_set, FilterImplPerInstructionSet::from(code));
}
}
#[cfg(all(target_arch = "aarch64", target_vendor = "apple"))]
#[test]
fn test_instruction_set_to_code_from_code() {
for instruction_set in [
FilterImplPerInstructionSet::Neon,
FilterImplPerInstructionSet::Scalar,
] {
let code = instruction_set as u8;
assert_eq!(instruction_set, FilterImplPerInstructionSet::from(code));
}
}
fn test_filter_impl_empty_aux(filter_impl: FilterImplPerInstructionSet) {
let mut output = vec![];
filter_impl.filter_vec_in_place(0..=u32::MAX, 0, &mut output);
@@ -126,11 +202,20 @@ mod tests {
assert_eq!(&output, &[1, 3, 4, 5, 6, 7, 8]);
}
fn test_filter_impl_empty_range_aux(filter_impl: FilterImplPerInstructionSet) {
// start > end: RangeInclusive::contains always returns false; output must be empty.
// The SVE path's wrapping_sub would otherwise produce a huge range_width.
let mut output = vec![3, 2, 1, 5, 11, 2, 5, 10, 2];
filter_impl.filter_vec_in_place(10..=5, 0, &mut output);
assert_eq!(&output, &[]);
}
fn test_filter_impl_test_suite(filter_impl: FilterImplPerInstructionSet) {
test_filter_impl_empty_aux(filter_impl);
test_filter_impl_simple_aux(filter_impl);
test_filter_impl_simple_aux_shifted(filter_impl);
test_filter_impl_simple_outside_i32_range(filter_impl);
test_filter_impl_empty_range_aux(filter_impl);
}
#[test]
@@ -141,25 +226,60 @@ mod tests {
}
}
#[test]
#[cfg(all(target_arch = "aarch64", not(target_vendor = "apple")))]
fn test_filter_implementation_sve() {
if FilterImplPerInstructionSet::SVE.is_available() {
test_filter_impl_test_suite(FilterImplPerInstructionSet::SVE);
}
}
#[test]
#[cfg(target_arch = "aarch64")]
fn test_filter_implementation_neon() {
test_filter_impl_test_suite(FilterImplPerInstructionSet::Neon);
}
#[test]
fn test_filter_implementation_scalar() {
test_filter_impl_test_suite(FilterImplPerInstructionSet::Scalar);
}
#[cfg(target_arch = "x86_64")]
fn max_val_strategy() -> impl proptest::strategy::Strategy<Value = u32> {
proptest::prop_oneof![
0u32..10u32,
255u32..258u32,
proptest::prelude::Just(1u32 << 25),
proptest::prelude::Just(u32::MAX - 1),
proptest::prelude::Just(u32::MAX),
]
}
fn vals_strategy() -> impl proptest::strategy::Strategy<Value = Vec<u32>> {
proptest::prop_oneof![
proptest::collection::vec(proptest::prelude::any::<u32>(), 0..300),
max_val_strategy()
.prop_flat_map(|max_val| { proptest::collection::vec(0..=max_val, 0..300) })
]
}
proptest::proptest! {
#[test]
fn test_filter_compare_scalar_and_avx2_impl_proptest(
start in proptest::prelude::any::<u32>(),
end in proptest::prelude::any::<u32>(),
fn test_filter_compare_scalar_and_impls_impl_proptest(
start in 0u32..400u32,
end in 0u32..400u32,
offset in 0u32..2u32,
mut vals in proptest::collection::vec(0..u32::MAX, 0..30)) {
if FilterImplPerInstructionSet::AVX2.is_available() {
let mut vals_clone = vals.clone();
FilterImplPerInstructionSet::AVX2.filter_vec_in_place(start..=end, offset, &mut vals);
FilterImplPerInstructionSet::Scalar.filter_vec_in_place(start..=end, offset, &mut vals_clone);
assert_eq!(&vals, &vals_clone);
}
vals in vals_strategy()) {
for implementation in available_impls() {
if implementation == FilterImplPerInstructionSet::Scalar {
continue;
}
let mut impl_output = vals.clone();
let mut scalar_output = vals.clone();
implementation.filter_vec_in_place(start..=end, offset, &mut impl_output);
FilterImplPerInstructionSet::Scalar.filter_vec_in_place(start..=end, offset, &mut scalar_output);
assert_eq!(&impl_output, &scalar_output);
}
}
}
}

118
bitpacker/src/filter_vec/neon.rs Normal file
View File

@@ -0,0 +1,118 @@
use std::arch::aarch64::*;
use std::ops::RangeInclusive;
const NUM_LANES: usize = 4;
// Compacts matching lanes to the front using a byte-level shuffle.
// `mask` is a 4-bit value: bit k=1 means lane k should appear in the output.
#[inline]
#[target_feature(enable = "neon")]
unsafe fn compact(data: uint32x4_t, mask: u8) -> uint32x4_t {
unsafe {
// SAFETY: mask is always in [0, 15] by construction (max sum of [1,2,4,8]).
// BYTE_SHUFFLE_TABLE has 16 entries, so this is always in bounds.
let shuffle = BYTE_SHUFFLE_TABLE.get_unchecked(mask as usize);
let shuffle_vec = vld1q_u8(shuffle.as_ptr());
vreinterpretq_u32_u8(vqtbl1q_u8(vreinterpretq_u8_u32(data), shuffle_vec))
}
}
// Safe (not unsafe) because NEON is mandatory on aarch64: no runtime feature check needed.
#[inline(never)]
pub fn filter_vec_in_place(range: RangeInclusive<u32>, offset: u32, output: &mut Vec<u32>) {
let num_words = output.len() / NUM_LANES;
let mut output_len = unsafe {
filter_vec_neon_aux(
output.as_ptr(),
range.clone(),
output.as_mut_ptr(),
offset,
num_words,
)
};
let remainder_start = num_words * NUM_LANES;
for i in remainder_start..output.len() {
let val = output[i];
output[output_len] = offset + i as u32;
output_len += if range.contains(&val) { 1 } else { 0 };
}
output.truncate(output_len);
}
#[target_feature(enable = "neon")]
unsafe fn filter_vec_neon_aux(
input: *const u32,
range: RangeInclusive<u32>,
output: *mut u32,
offset: u32,
num_words: usize,
) -> usize {
unsafe {
let mut input = input;
let mut output_tail = output;
let range_start_simd = vdupq_n_u32(*range.start());
let range_end_simd = vdupq_n_u32(*range.end());
let mut ids = vld1q_u32([offset, offset + 1, offset + 2, offset + 3].as_ptr());
let shift = vdupq_n_u32(NUM_LANES as u32);
let bit_weights = vld1q_u32([1u32, 2, 4, 8].as_ptr());
for _ in 0..num_words {
let word = vld1q_u32(input);
// Unsigned compares: CMHS (compare higher or same) tests `word >= start`
// and `end >= word`. ANDing both gives the inside-range mask directly,
// which is cheaper than computing `outside` and then negating.
let ge_start = vcgeq_u32(word, range_start_simd);
let le_end = vcleq_u32(word, range_end_simd);
// inside[k] = 0xFFFFFFFF if val[k] is in range, 0 otherwise.
let inside = vandq_u32(ge_start, le_end);
// Build the 4-bit mask: AND bit_weights with the inside lane mask, so each
// inside lane contributes its bit_weight (1, 2, 4, or 8). Summing yields the
// 4-bit mask in one addv.
let inside_bits = vandq_u32(bit_weights, inside);
let mask = vaddvq_u32(inside_bits) as u8;
// mask is mathematically bounded: max value is 1+2+4+8=15 (all lanes match)
debug_assert!(mask <= 15, "mask must fit in 4 bits: {}", mask);
// Count of matching lanes = popcount(mask). Derives the count directly from
// the mask instead of running a parallel SIMD reduction over `outside`.
let added_len = mask.count_ones() as usize;
// Safe because mask is guaranteed to be in [0, 15]
let filtered_ids = compact(ids, mask);
vst1q_u32(output_tail, filtered_ids);
output_tail = output_tail.add(added_len);
ids = vaddq_u32(ids, shift);
input = input.add(NUM_LANES);
}
output_tail.offset_from(output) as usize
}
}
// Byte shuffle patterns to compact matching lanes to the front of the vector.
// Index is a 4-bit mask: bit k=1 means lane k (bytes 4k..4k+3) is in-range.
// The j-th set bit determines which input lane goes to output position j.
const BYTE_SHUFFLE_TABLE: [[u8; 16]; 16] = [
[
16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16,
], // 0b0000: none
[0, 1, 2, 3, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16], // 0b0001: lane 0
[4, 5, 6, 7, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16], // 0b0010: lane 1
[0, 1, 2, 3, 4, 5, 6, 7, 16, 16, 16, 16, 16, 16, 16, 16], // 0b0011: lanes 0,1
[8, 9, 10, 11, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16], // 0b0100: lane 2
[0, 1, 2, 3, 8, 9, 10, 11, 16, 16, 16, 16, 16, 16, 16, 16], // 0b0101: lanes 0,2
[4, 5, 6, 7, 8, 9, 10, 11, 16, 16, 16, 16, 16, 16, 16, 16], // 0b0110: lanes 1,2
[0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 16, 16, 16, 16], // 0b0111: lanes 0,1,2
[
12, 13, 14, 15, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16,
], // 0b1000: lane 3
[0, 1, 2, 3, 12, 13, 14, 15, 16, 16, 16, 16, 16, 16, 16, 16], // 0b1001: lanes 0,3
[4, 5, 6, 7, 12, 13, 14, 15, 16, 16, 16, 16, 16, 16, 16, 16], // 0b1010: lanes 1,3
[0, 1, 2, 3, 4, 5, 6, 7, 12, 13, 14, 15, 16, 16, 16, 16], // 0b1011: lanes 0,1,3
[8, 9, 10, 11, 12, 13, 14, 15, 16, 16, 16, 16, 16, 16, 16, 16], // 0b1100: lanes 2,3
[0, 1, 2, 3, 8, 9, 10, 11, 12, 13, 14, 15, 16, 16, 16, 16], // 0b1101: lanes 0,2,3
[4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 16, 16, 16], // 0b1110: lanes 1,2,3
[0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15], // 0b1111: all lanes
];

260
bitpacker/src/filter_vec/sve.rs Normal file
View File

@@ -0,0 +1,260 @@
use std::ops::RangeInclusive;
// SVE vector length (in u32 lanes) is not a compile-time constant; query at runtime.
// Safe to call only when SVE is confirmed available via is_aarch64_feature_detected!("sve").
#[target_feature(enable = "sve")]
unsafe fn num_lanes() -> usize {
let vl: usize;
unsafe {
core::arch::asm!(
"cntw {vl}",
vl = out(reg) vl,
options(nostack, nomem, preserves_flags),
);
}
vl
}
// SAFETY: caller must ensure SVE is available (checked via is_aarch64_feature_detected!("sve")).
// Unlike NEON, SVE is optional on aarch64 and not guaranteed by the target architecture.
pub unsafe fn filter_vec_in_place(range: RangeInclusive<u32>, offset: u32, output: &mut Vec<u32>) {
if range.start() > range.end() {
output.clear();
return;
}
let vl = unsafe { num_lanes() };
let num_words = output.len() / vl;
let range_start = *range.start();
// Unsigned subtraction trick: val ∈ [lo, hi] ↔ (val - lo) ≤ᵤ (hi - lo).
// Values below lo wrap around to large u32, so the single unsigned ≤ excludes them.
let range_width = range.end().wrapping_sub(range_start);
let mut output_len = unsafe {
filter_vec_sve_aux(
output.as_ptr(),
range_start,
range_width,
output.as_mut_ptr(),
offset,
num_words,
vl,
)
};
let remainder_start = num_words * vl;
for i in remainder_start..output.len() {
let val = output[i];
output[output_len] = offset + i as u32;
output_len += if range.contains(&val) { 1 } else { 0 };
}
output.truncate(output_len);
}
// Register allocation for the asm! blocks:
// z0 ids_a (index vector for first half of each pair, advances by step2 each iter)
// z1 range_width broadcast
// z2 range_start broadcast
// z3 step2 broadcast (2 * vl)
// z4 ids_b (index vector for second half, = ids_a + step, advances by step2)
// z5 scratch: loaded word_a, then compacted_a
// z6 scratch: loaded word_b, then compacted_b
// p0 all-true predicate (ptrue p0.s)
// p1 in-range mask for word_a
// p2 in-range mask for word_b
#[target_feature(enable = "sve")]
unsafe fn filter_vec_sve_aux(
input: *const u32,
range_start: u32,
range_width: u32,
output: *mut u32,
offset: u32,
num_words: usize,
vl: usize,
) -> usize {
let num_pairs = num_words / 2;
let mut input_ptr = input;
let mut output_tail = output;
if num_pairs > 0 {
unsafe {
// We rely on asm! because the SVE intrinsics are not available in stable Rust.
// The code that follows was generated by Rustc nightly based on the intrinsics version
// at the bottom of this file.
core::arch::asm!(
// --- Setup ---
// All-true predicate for 32-bit lanes.
"ptrue p0.s",
// ids_a = [offset, offset+1, offset+2, ...]
"index z0.s, {offset:w}, #1",
// Broadcast scalars into SVE vectors.
"mov z1.s, {range_width:w}",
"mov z2.s, {range_start:w}",
// vl_gpr = number of 32-bit lanes (cntw).
"cntw {vl_gpr}",
// step2_bytes will first hold 2*vl (for the step2 vector), then 2*VL in bytes.
"lsl {step2_bytes}, {vl_gpr}, #1",
// z4 = step = [vl, vl, ...]; will become ids_b after the add below.
"mov z4.s, {vl_gpr:w}",
// z3 = step2 = [2*vl, 2*vl, ...], used to advance both id vectors each iter.
"mov z3.s, {step2_bytes:w}",
// Repurpose step2_bytes to hold the byte stride for advancing the input pointer
// by two full SVE vectors per iteration.
"rdvl {step2_bytes}, #2",
// ids_b = ids_a + step = [offset+vl, offset+vl+1, ...]
"add z4.s, z0.s, z4.s",
// --- Main loop: process two SVE vectors (ids_a and ids_b) per iteration ---
"0:",
// Load two consecutive SVE vectors from input.
"ld1w {{z5.s}}, p0/z, [{input}]",
"ld1w {{z6.s}}, p0/z, [{input}, #1, mul vl]",
// Advance input pointer by 2 * VL bytes.
"add {input}, {input}, {step2_bytes}",
// Unsigned shift: subtract range_start so in-range check becomes a single cmpu ≤.
"sub z5.s, z5.s, z2.s",
"sub z6.s, z6.s, z2.s",
// in_range: shifted value ≤ range_width (unsigned, so values below lo also fail).
"cmphs p1.s, p0/z, z1.s, z5.s",
"cmphs p2.s, p0/z, z1.s, z6.s",
// Count matching lanes; both cntp calls have independent inputs for OOO parallelism.
"cntp {cnt_a}, p0, p1.s",
"compact z5.s, p1, z0.s",
"compact z6.s, p2, z4.s",
"cntp {cnt_b}, p0, p2.s",
// Advance id vectors for the next iteration.
"add z0.s, z0.s, z3.s",
"add z4.s, z4.s, z3.s",
// Store compacted ids. Only the first cnt_a / cnt_b slots are valid; the rest
// will be overwritten by subsequent iterations before the final truncate.
"str z5, [{out}]",
"st1w {{z6.s}}, p0, [{out}, {cnt_a}, lsl #2]",
"add {out}, {out}, {cnt_a}, lsl #2",
"add {out}, {out}, {cnt_b}, lsl #2",
"subs {pairs}, {pairs}, #1",
"b.ne 0b",
// --- Operands ---
input = inout(reg) input_ptr,
out = inout(reg) output_tail,
pairs = inout(reg) num_pairs => _,
offset = in(reg) offset,
range_start = in(reg) range_start,
range_width = in(reg) range_width,
vl_gpr = out(reg) _,
step2_bytes = out(reg) _,
cnt_a = out(reg) _,
cnt_b = out(reg) _,
out("p0") _, out("p1") _, out("p2") _,
out("v0") _, out("v1") _, out("v2") _, out("v3") _,
out("v4") _, out("v5") _, out("v6") _,
options(nostack),
);
}
}
// Handle an odd trailing vector.
if num_words % 2 == 1 {
// ids_a for the odd word starts at offset + num_pairs * 2 * vl.
// input_ptr was advanced by the main loop and now points at the odd word.
let odd_offset =
offset.wrapping_add((num_pairs as u32).wrapping_mul(2).wrapping_mul(vl as u32));
unsafe {
core::arch::asm!(
"ptrue p0.s",
"index z0.s, {odd_offset:w}, #1",
"mov z1.s, {range_width:w}",
"mov z2.s, {range_start:w}",
"ld1w {{z3.s}}, p0/z, [{input}]",
"sub z3.s, z3.s, z2.s",
"cmphs p1.s, p0/z, z1.s, z3.s",
"cntp {cnt}, p0, p1.s",
"compact z0.s, p1, z0.s",
"str z0, [{out}]",
"add {out}, {out}, {cnt}, lsl #2",
odd_offset = in(reg) odd_offset,
range_width = in(reg) range_width,
range_start = in(reg) range_start,
input = in(reg) input_ptr,
out = inout(reg) output_tail,
cnt = out(reg) _,
out("p0") _, out("p1") _,
out("v0") _, out("v1") _, out("v2") _, out("v3") _,
options(nostack),
);
}
}
unsafe { output_tail.offset_from(output) as usize }
}
// SVE implements with intrinsics.
//
// #[target_feature(enable = "sve")]
// unsafe fn filter_vec_sve_aux(
// input: *const u32,
// range_start: u32,
// range_width: u32,
// output: *mut u32,
// offset: u32,
// num_words: usize,
// vl: usize,
// ) -> usize {
// unsafe {
// let all_true = svptrue_b32();
// let range_start_simd = svdup_n_u32(range_start);
// let range_width_simd = svdup_n_u32(range_width);
// // ids_a covers [offset .. offset+vl), ids_b covers the next vl ids.
// // Keeping them separate breaks the loop-carried dependency through ids so
// // both compact/cntp chains are fully independent within each unrolled body.
// let mut ids_a = svindex_u32(offset, 1);
// let step = svdup_n_u32(vl as u32);
// let step2 = svdup_n_u32(2 * vl as u32);
// let mut ids_b = svadd_u32_x(all_true, ids_a, step);
// let mut input = input;
// let mut output_tail = output;
// // Unrolled ×2: both cntp calls have independent inputs and execute in parallel.
// // The two output_tail updates are sequential but together cost 4+1+1=6 cy per
// // pair vs 5+5=10 cy for two scalar iterations, breaking the cntp latency chain.
// let num_pairs = num_words / 2;
// for _ in 0..num_pairs {
// let word_a = svld1_u32(all_true, input);
// let word_b = svld1_u32(all_true, input.add(vl));
// let shifted_a = svsub_u32_x(all_true, word_a, range_start_simd);
// let shifted_b = svsub_u32_x(all_true, word_b, range_start_simd);
// let in_range_a = svcmple_u32(all_true, shifted_a, range_width_simd);
// let in_range_b = svcmple_u32(all_true, shifted_b, range_width_simd);
// let compacted_a = svcompact_u32(in_range_a, ids_a);
// let compacted_b = svcompact_u32(in_range_b, ids_b);
// // cntp_a and cntp_b have independent inputs: OOO engine issues them in parallel.
// let added_len_a = svcntp_b32(all_true, in_range_a) as usize;
// let added_len_b = svcntp_b32(all_true, in_range_b) as usize;
// // Write the full vector — only the first added_len slots are valid.
// // Subsequent iterations overwrite the trailing zeros before truncate.
// svst1_u32(all_true, output_tail, compacted_a);
// output_tail = output_tail.add(added_len_a);
// svst1_u32(all_true, output_tail, compacted_b);
// output_tail = output_tail.add(added_len_b);
// ids_a = svadd_u32_x(all_true, ids_a, step2);
// ids_b = svadd_u32_x(all_true, ids_b, step2);
// input = input.add(2 * vl);
// }
// // Handle an odd trailing word.
// if num_words % 2 == 1 {
// let word = svld1_u32(all_true, input);
// let shifted = svsub_u32_x(all_true, word, range_start_simd);
// let in_range = svcmple_u32(all_true, shifted, range_width_simd);
// let added_len = svcntp_b32(all_true, in_range) as usize;
// let compacted_ids = svcompact_u32(in_range, ids_a);
// svst1_u32(all_true, output_tail, compacted_ids);
// output_tail = output_tail.add(added_len);
// }
// output_tail.offset_from(output) as usize
// }
// }

2
columnar/Cargo.toml
View File

@@ -23,7 +23,7 @@ downcast-rs = "2.0.1"
proptest = "1"
more-asserts = "0.3.1"
rand = "0.9"
binggan = "0.15.3"
binggan = "0.17.0"
[[bench]]
name = "bench_merge"

76
columnar/src/block_accessor.rs
View File

@@ -17,7 +17,18 @@ impl<T: PartialOrd + Copy + std::fmt::Debug + Send + Sync + 'static + Default>
pub fn fetch_block<'a>(&'a mut self, docs: &'a [u32], accessor: &Column<T>) {
if accessor.index.get_cardinality().is_full() {
self.val_cache.resize(docs.len(), T::default());
accessor.values.get_vals(docs, &mut self.val_cache);
// When the docs form a contiguous ascending run we can fetch the values
// as a single range. This lets codecs (e.g. bitpacked) bulk-decode the
// slice instead of gathering value-by-value, and avoids per-value dynamic
// dispatch. `docs` is always sorted ascending and free of duplicates here,
// so comparing the endpoints is enough to detect contiguity.
if is_contiguous(docs) {
accessor
.values
.get_range(docs[0] as u64, &mut self.val_cache);
} else {
accessor.values.get_vals(docs, &mut self.val_cache);
}
} else {
self.docid_cache.clear();
self.row_id_cache.clear();
@@ -33,14 +44,14 @@ impl<T: PartialOrd + Copy + std::fmt::Debug + Send + Sync + 'static + Default>
&mut self,
docs: &[u32],
accessor: &Column<T>,
missing: Option<T>,
missing_opt: Option<T>,
) {
self.fetch_block(docs, accessor);
// no missing values
if accessor.index.get_cardinality().is_full() {
return;
}
let Some(missing) = missing else {
let Some(missing) = missing_opt else {
return;
};
@@ -158,6 +169,22 @@ impl<T: PartialOrd + Copy + std::fmt::Debug + Send + Sync + 'static + Default>
}
}
/// Returns true if `docs` is a contiguous ascending run `[d, d + 1, ..., d + n - 1]`.
///
/// Assumes `docs` is sorted ascending and free of duplicates (the invariant for the
/// doc blocks passed to `fetch_block`), so comparing the endpoints is sufficient.
#[inline]
fn is_contiguous(docs: &[u32]) -> bool {
let (Some(&first), Some(&last)) = (docs.first(), docs.last()) else {
return false;
};
debug_assert!(
docs.windows(2).all(|w| w[0] < w[1]),
"fetch_block requires docs sorted ascending without duplicates"
);
(last - first) as usize + 1 == docs.len()
}
/// Given two sorted lists of docids `docs` and `hits`, hits is a subset of `docs`.
/// Return all docs that are not in `hits`.
fn find_missing_docs<F>(docs: &[u32], hits: &[u32], mut callback: F)
@@ -191,6 +218,7 @@ where F: FnMut(u32) {
}
#[cfg(test)]
#[allow(clippy::field_reassign_with_default)]
mod tests {
use super::*;
@@ -287,4 +315,46 @@ mod tests {
assert_eq!(accessor.docid_cache, vec![0]);
assert_eq!(accessor.val_cache, vec![1]);
}
#[test]
fn test_is_contiguous() {
assert!(!is_contiguous(&[]));
assert!(is_contiguous(&[5]));
assert!(is_contiguous(&[5, 6, 7, 8]));
assert!(is_contiguous(&[0, 1, 2]));
assert!(!is_contiguous(&[5, 7, 8]));
assert!(!is_contiguous(&[0, 1, 3]));
}
#[test]
fn test_fetch_block_contiguous_and_gather_match() {
use crate::column_index::ColumnIndex;
use crate::column_values::{
ALL_U64_CODEC_TYPES, serialize_and_load_u64_based_column_values,
};
let vals: Vec<u64> = (0..200u64).map(|i| i * 7 + 3).collect();
let values =
serialize_and_load_u64_based_column_values::<u64>(&&vals[..], &ALL_U64_CODEC_TYPES);
let column = Column {
index: ColumnIndex::Full,
values,
};
let check = |accessor: &mut ColumnBlockAccessor<u64>, docs: &[u32]| {
accessor.fetch_block(docs, &column);
let got: Vec<(u32, u64)> = accessor.iter_docid_vals(docs, &column).collect();
let expected: Vec<(u32, u64)> = docs.iter().map(|&d| (d, vals[d as usize])).collect();
assert_eq!(got, expected);
};
let mut accessor = ColumnBlockAccessor::<u64>::default();
// Contiguous block -> get_range fast path.
check(&mut accessor, &(10..74).collect::<Vec<u32>>());
// Non-contiguous block -> get_vals gather path.
check(&mut accessor, &[0, 5, 9, 100, 199]);
// Single doc and full span.
check(&mut accessor, &[42]);
check(&mut accessor, &(0..200).collect::<Vec<u32>>());
}
}

12
columnar/src/column_values/mod.rs
View File

@@ -119,8 +119,18 @@ pub trait ColumnValues<T: PartialOrd = u64>: Send + Sync + DowncastSync {
/// the segment's `maxdoc`.
#[inline(always)]
fn get_range(&self, start: u64, output: &mut [T]) {
for (out, idx) in output.iter_mut().zip(start..) {
let mut out_chunks = output.chunks_exact_mut(4);
let mut idx = start;
for out_x4 in out_chunks.by_ref() {
out_x4[0] = self.get_val(idx as u32);
out_x4[1] = self.get_val((idx + 1) as u32);
out_x4[2] = self.get_val((idx + 2) as u32);
out_x4[3] = self.get_val((idx + 3) as u32);
idx += 4;
}
for out in out_chunks.into_remainder() {
*out = self.get_val(idx as u32);
idx += 1;
}
}

16
columnar/src/column_values/u64_based/tests.rs
View File

@@ -121,6 +121,22 @@ pub(crate) fn create_and_validate<TColumnCodec: ColumnCodec>(
reader.get_vals(&all_docs, &mut buffer);
assert_eq!(vals, buffer);
// Validate `get_range` over the full column and a sub-range. The sub-range starts
// at a non-zero offset to exercise the entrance-ramp alignment of the batch decode.
buffer.resize(all_docs.len(), 0);
reader.get_range(0, &mut buffer);
assert_eq!(vals, buffer, "get_range (full) mismatch in data set {name}");
if vals.len() >= 2 {
let start = 1usize;
buffer.resize(vals.len() - start, 0);
reader.get_range(start as u64, &mut buffer);
assert_eq!(
&vals[start..],
&buffer[..],
"get_range (sub-range) mismatch in data set {name}"
);
}
if !vals.is_empty() {
let test_rand_idx = rand::rng().random_range(0..=vals.len() - 1);
let expected_positions: Vec<u32> = vals

2
common/Cargo.toml
View File

@@ -19,6 +19,6 @@ time = { version = "0.3.47", features = ["serde-well-known"] }
serde = { version = "1.0.136", features = ["derive"] }
[dev-dependencies]
binggan = "0.15.3"
binggan = "0.17.0"
proptest = "1.0.0"
rand = "0.9"

3
common/src/bitset.rs
View File

@@ -47,6 +47,9 @@ impl TinySet {
TinySet(val)
}
/// An empty `TinySet` constant.
pub const EMPTY: TinySet = TinySet(0u64);
/// Returns an empty `TinySet`.
#[inline]
pub fn empty() -> TinySet {

2
common/src/file_slice.rs
View File

@@ -121,7 +121,7 @@ pub struct FileSlice {
impl fmt::Debug for FileSlice {
fn fmt(&self, f: &mut fmt::Formatter) -> fmt::Result {
write!(f, "FileSlice({:?}, {:?})", &self.data, self.range)
write!(f, "FileSlice({:?}, {:?})", self.data, self.range)
}
}

68
query-grammar/src/query_grammar.rs
View File

@@ -1045,18 +1045,43 @@ fn operand_leaf(inp: &str) -> IResult<&str, (Option<BinaryOperand>, Option<Occur
}
fn ast(inp: &str) -> IResult<&str, UserInputAst> {
let boolean_expr = map_res(
separated_pair(occur_leaf, multispace1, many1(operand_leaf)),
|(left, right)| aggregate_binary_expressions(left, right),
);
let single_leaf = map(occur_leaf, |(occur, ast)| {
if occur == Some(Occur::MustNot) {
ast.unary(Occur::MustNot)
} else {
ast
}
});
delimited(multispace0, alt((boolean_expr, single_leaf)), multispace0)(inp)
// Parse `occur_leaf` once, then conditionally extend into a boolean
// expression. The previous implementation used `alt((boolean_expr,
// single_leaf))` which, when the input was a single leaf with no
// following operand, would parse `occur_leaf` once for `boolean_expr`,
// fail at `multispace1`, backtrack, then re-parse `occur_leaf` for
// `single_leaf`. With recursively-nested groups like `(+(+(+a)))`, that
// doubling at every level produced O(2^n) parse time. Parsing once and
// peeking ahead for the operand keeps it O(n).
delimited(
multispace0,
|inp| {
let (rest, first) = occur_leaf(inp)?;
// Only fall back on `Err::Error` (recoverable), mirroring
// `alt`'s behaviour. `Err::Failure` and `Err::Incomplete`
// must propagate so cut points and streaming needs are not
// accidentally swallowed if they are ever introduced in the
// operand parsers.
match preceded(multispace1, many1(operand_leaf))(rest) {
Ok((rest, more)) => {
let combined = aggregate_binary_expressions(first, more)
.map_err(|_| nom::Err::Error(Error::new(inp, ErrorKind::MapRes)))?;
Ok((rest, combined))
}
Err(nom::Err::Error(_)) => {
let (occur, ast) = first;
let single = if occur == Some(Occur::MustNot) {
ast.unary(Occur::MustNot)
} else {
ast
};
Ok((rest, single))
}
Err(e) => Err(e),
}
},
multispace0,
)(inp)
}
fn ast_infallible(inp: &str) -> JResult<&str, UserInputAst> {
@@ -1891,4 +1916,23 @@ mod test {
r#"(+"field":'happy tax payer' +"other_field":1)"#,
);
}
// Regression test for https://github.com/quickwit-oss/tantivy/issues/2498:
// deeply nested parenthesized queries used to take O(2^n) time because the
// top-level `ast()` parser tried `boolean_expr` first and re-parsed the
// inner `occur_leaf` when it backtracked to `single_leaf`. Depth 60 would
// take ~10^18 operations under the regression; with the fix it parses
// instantly. We use `test_parse_query_to_ast_helper` so this test would
// never finish if the regression returned.
#[test]
fn test_parse_deeply_nested_query() {
let depth = 60;
let leading: String = "(".repeat(depth);
let trailing: String = ")".repeat(depth);
let query = format!("{leading}title:test{trailing}");
test_parse_query_to_ast_helper(&query, r#""title":test"#);
let query_with_plus = format!("+{leading}title:test{trailing}");
test_parse_query_to_ast_helper(&query_with_plus, r#""title":test"#);
}
}

297
src/aggregation/agg_data.rs
View File

@@ -10,18 +10,18 @@ use crate::aggregation::accessor_helpers::{
};
use crate::aggregation::agg_req::{Aggregation, AggregationVariants, Aggregations};
use crate::aggregation::bucket::{
build_segment_filter_collector, build_segment_range_collector, CompositeAggReqData,
CompositeAggregation, CompositeSourceAccessors, FilterAggReqData, HistogramAggReqData,
HistogramBounds, IncludeExcludeParam, MissingTermAggReqData, RangeAggReqData,
SegmentHistogramCollector, TermMissingAgg, TermsAggReqData, TermsAggregation,
TermsAggregationInternal,
build_segment_filter_collector, build_segment_histogram_collector,
build_segment_range_collector, CompositeAggReqData, CompositeAggregation,
CompositeSourceAccessors, FilterAggReqData, HistogramAggReqData, HistogramBounds,
IncludeExcludeParam, MissingTermAggReqData, RangeAggReqData, TermMissingAgg, TermsAggReqData,
TermsAggregation, TermsAggregationInternal,
};
use crate::aggregation::metric::{
build_segment_stats_collector, AverageAggregation, CardinalityAggReqData,
CardinalityAggregationReq, CountAggregation, ExtendedStatsAggregation, MaxAggregation,
MetricAggReqData, MinAggregation, SegmentCardinalityCollector, SegmentExtendedStatsCollector,
SegmentPercentilesCollector, StatsAggregation, StatsType, SumAggregation, TopHitsAggReqData,
TopHitsSegmentCollector,
SegmentPercentilesCollector, StatsAggregation, StatsType, SumAggregation, TermOrdSet,
TopHitsAggReqData, TopHitsSegmentCollector, BITSET_MAX_TERM_ORD,
};
use crate::aggregation::segment_agg_result::{
GenericSegmentAggregationResultsCollector, SegmentAggregationCollector,
@@ -41,7 +41,7 @@ pub struct AggregationsSegmentCtx {
impl AggregationsSegmentCtx {
pub(crate) fn push_term_req_data(&mut self, data: TermsAggReqData) -> usize {
self.per_request.term_req_data.push(Some(Box::new(data)));
self.per_request.term_req_data.push(data);
self.per_request.term_req_data.len() - 1
}
pub(crate) fn push_cardinality_req_data(&mut self, data: CardinalityAggReqData) -> usize {
@@ -61,31 +61,25 @@ impl AggregationsSegmentCtx {
self.per_request.missing_term_req_data.len() - 1
}
pub(crate) fn push_histogram_req_data(&mut self, data: HistogramAggReqData) -> usize {
self.per_request
.histogram_req_data
.push(Some(Box::new(data)));
self.per_request.histogram_req_data.push(data);
self.per_request.histogram_req_data.len() - 1
}
pub(crate) fn push_range_req_data(&mut self, data: RangeAggReqData) -> usize {
self.per_request.range_req_data.push(Some(Box::new(data)));
self.per_request.range_req_data.push(data);
self.per_request.range_req_data.len() - 1
}
pub(crate) fn push_filter_req_data(&mut self, data: FilterAggReqData) -> usize {
self.per_request.filter_req_data.push(Some(Box::new(data)));
self.per_request.filter_req_data.push(data);
self.per_request.filter_req_data.len() - 1
}
pub(crate) fn push_composite_req_data(&mut self, data: CompositeAggReqData) -> usize {
self.per_request
.composite_req_data
.push(Some(Box::new(data)));
self.per_request.composite_req_data.push(data);
self.per_request.composite_req_data.len() - 1
}
#[inline]
pub(crate) fn get_term_req_data(&self, idx: usize) -> &TermsAggReqData {
self.per_request.term_req_data[idx]
.as_deref()
.expect("term_req_data slot is empty (taken)")
&self.per_request.term_req_data[idx]
}
#[inline]
pub(crate) fn get_cardinality_req_data(&self, idx: usize) -> &CardinalityAggReqData {
@@ -103,116 +97,6 @@ impl AggregationsSegmentCtx {
pub(crate) fn get_missing_term_req_data(&self, idx: usize) -> &MissingTermAggReqData {
&self.per_request.missing_term_req_data[idx]
}
#[inline]
pub(crate) fn get_histogram_req_data(&self, idx: usize) -> &HistogramAggReqData {
self.per_request.histogram_req_data[idx]
.as_deref()
.expect("histogram_req_data slot is empty (taken)")
}
#[inline]
pub(crate) fn get_range_req_data(&self, idx: usize) -> &RangeAggReqData {
self.per_request.range_req_data[idx]
.as_deref()
.expect("range_req_data slot is empty (taken)")
}
#[inline]
pub(crate) fn get_composite_req_data(&self, idx: usize) -> &CompositeAggReqData {
self.per_request.composite_req_data[idx]
.as_deref()
.expect("composite_req_data slot is empty (taken)")
}
// ---------- mutable getters ----------
#[inline]
pub(crate) fn get_metric_req_data_mut(&mut self, idx: usize) -> &mut MetricAggReqData {
&mut self.per_request.stats_metric_req_data[idx]
}
#[inline]
pub(crate) fn get_cardinality_req_data_mut(
&mut self,
idx: usize,
) -> &mut CardinalityAggReqData {
&mut self.per_request.cardinality_req_data[idx]
}
#[inline]
pub(crate) fn get_histogram_req_data_mut(&mut self, idx: usize) -> &mut HistogramAggReqData {
self.per_request.histogram_req_data[idx]
.as_deref_mut()
.expect("histogram_req_data slot is empty (taken)")
}
// ---------- take / put (terms, histogram, range) ----------
/// Move out the boxed Histogram request at `idx`, leaving `None`.
#[inline]
pub(crate) fn take_histogram_req_data(&mut self, idx: usize) -> Box<HistogramAggReqData> {
self.per_request.histogram_req_data[idx]
.take()
.expect("histogram_req_data slot is empty (taken)")
}
/// Put back a Histogram request into an empty slot at `idx`.
#[inline]
pub(crate) fn put_back_histogram_req_data(
&mut self,
idx: usize,
value: Box<HistogramAggReqData>,
) {
debug_assert!(self.per_request.histogram_req_data[idx].is_none());
self.per_request.histogram_req_data[idx] = Some(value);
}
/// Move out the boxed Range request at `idx`, leaving `None`.
#[inline]
pub(crate) fn take_range_req_data(&mut self, idx: usize) -> Box<RangeAggReqData> {
self.per_request.range_req_data[idx]
.take()
.expect("range_req_data slot is empty (taken)")
}
/// Put back a Range request into an empty slot at `idx`.
#[inline]
pub(crate) fn put_back_range_req_data(&mut self, idx: usize, value: Box<RangeAggReqData>) {
debug_assert!(self.per_request.range_req_data[idx].is_none());
self.per_request.range_req_data[idx] = Some(value);
}
/// Move out the boxed Filter request at `idx`, leaving `None`.
#[inline]
pub(crate) fn take_filter_req_data(&mut self, idx: usize) -> Box<FilterAggReqData> {
self.per_request.filter_req_data[idx]
.take()
.expect("filter_req_data slot is empty (taken)")
}
/// Put back a Filter request into an empty slot at `idx`.
#[inline]
pub(crate) fn put_back_filter_req_data(&mut self, idx: usize, value: Box<FilterAggReqData>) {
debug_assert!(self.per_request.filter_req_data[idx].is_none());
self.per_request.filter_req_data[idx] = Some(value);
}
/// Move out the Composite request at `idx`.
#[inline]
pub(crate) fn take_composite_req_data(&mut self, idx: usize) -> Box<CompositeAggReqData> {
self.per_request.composite_req_data[idx]
.take()
.expect("composite_req_data slot is empty (taken)")
}
/// Put back a Composite request into an empty slot at `idx`.
#[inline]
pub(crate) fn put_back_composite_req_data(
&mut self,
idx: usize,
value: Box<CompositeAggReqData>,
) {
debug_assert!(self.per_request.composite_req_data[idx].is_none());
self.per_request.composite_req_data[idx] = Some(value);
}
}
/// Each type of aggregation has its own request data struct. This struct holds
@@ -223,15 +107,14 @@ impl AggregationsSegmentCtx {
/// for a node with [AggKind::Terms]).
#[derive(Default)]
pub struct PerRequestAggSegCtx {
// Box for cheap take/put - Only necessary for bucket aggs that have sub-aggregations
/// TermsAggReqData contains the request data for a terms aggregation.
pub term_req_data: Vec<Option<Box<TermsAggReqData>>>,
pub term_req_data: Vec<TermsAggReqData>,
/// HistogramAggReqData contains the request data for a histogram aggregation.
pub histogram_req_data: Vec<Option<Box<HistogramAggReqData>>>,
pub histogram_req_data: Vec<HistogramAggReqData>,
/// RangeAggReqData contains the request data for a range aggregation.
pub range_req_data: Vec<Option<Box<RangeAggReqData>>>,
pub range_req_data: Vec<RangeAggReqData>,
/// FilterAggReqData contains the request data for a filter aggregation.
pub filter_req_data: Vec<Option<Box<FilterAggReqData>>>,
pub filter_req_data: Vec<FilterAggReqData>,
/// Shared by avg, min, max, sum, stats, extended_stats, count
pub stats_metric_req_data: Vec<MetricAggReqData>,
/// CardinalityAggReqData contains the request data for a cardinality aggregation.
@@ -241,7 +124,7 @@ pub struct PerRequestAggSegCtx {
/// MissingTermAggReqData contains the request data for a missing term aggregation.
pub missing_term_req_data: Vec<MissingTermAggReqData>,
/// CompositeAggReqData contains the request data for a composite aggregation.
pub composite_req_data: Vec<Option<Box<CompositeAggReqData>>>,
pub composite_req_data: Vec<CompositeAggReqData>,
/// Request tree used to build collectors.
pub agg_tree: Vec<AggRefNode>,
@@ -252,22 +135,22 @@ impl PerRequestAggSegCtx {
fn get_memory_consumption(&self) -> usize {
self.term_req_data
.iter()
.map(|b| b.as_ref().unwrap().get_memory_consumption())
.map(|t| t.get_memory_consumption())
.sum::<usize>()
+ self
.histogram_req_data
.iter()
.map(|b| b.as_ref().unwrap().get_memory_consumption())
.map(|t| t.get_memory_consumption())
.sum::<usize>()
+ self
.range_req_data
.iter()
.map(|b| b.as_ref().unwrap().get_memory_consumption())
.map(|t| t.get_memory_consumption())
.sum::<usize>()
+ self
.filter_req_data
.iter()
.map(|b| b.as_ref().unwrap().get_memory_consumption())
.map(|t| t.get_memory_consumption())
.sum::<usize>()
+ self
.stats_metric_req_data
@@ -292,7 +175,7 @@ impl PerRequestAggSegCtx {
+ self
.composite_req_data
.iter()
.map(|b| b.as_ref().map(|d| d.get_memory_consumption()).unwrap_or(0))
.map(|t| t.get_memory_consumption())
.sum::<usize>()
+ self.agg_tree.len() * std::mem::size_of::<AggRefNode>()
}
@@ -301,40 +184,16 @@ impl PerRequestAggSegCtx {
let idx = node.idx_in_req_data;
let kind = node.kind;
match kind {
AggKind::Terms => self.term_req_data[idx]
.as_deref()
.expect("term_req_data slot is empty (taken)")
.name
.as_str(),
AggKind::Terms => self.term_req_data[idx].name.as_str(),
AggKind::Cardinality => &self.cardinality_req_data[idx].name,
AggKind::StatsKind(_) => &self.stats_metric_req_data[idx].name,
AggKind::TopHits => &self.top_hits_req_data[idx].name,
AggKind::MissingTerm => &self.missing_term_req_data[idx].name,
AggKind::Histogram => self.histogram_req_data[idx]
.as_deref()
.expect("histogram_req_data slot is empty (taken)")
.name
.as_str(),
AggKind::DateHistogram => self.histogram_req_data[idx]
.as_deref()
.expect("histogram_req_data slot is empty (taken)")
.name
.as_str(),
AggKind::Range => self.range_req_data[idx]
.as_deref()
.expect("range_req_data slot is empty (taken)")
.name
.as_str(),
AggKind::Filter => self.filter_req_data[idx]
.as_deref()
.expect("filter_req_data slot is empty (taken)")
.name
.as_str(),
AggKind::Composite => self.composite_req_data[idx]
.as_deref()
.expect("composite_req_data slot is empty (taken)")
.name
.as_str(),
AggKind::Histogram => self.histogram_req_data[idx].name.as_str(),
AggKind::DateHistogram => self.histogram_req_data[idx].name.as_str(),
AggKind::Range => self.range_req_data[idx].name.as_str(),
AggKind::Filter => self.filter_req_data[idx].name.as_str(),
AggKind::Composite => self.composite_req_data[idx].name.as_str(),
}
}
@@ -412,13 +271,39 @@ pub(crate) fn build_segment_agg_collector(
Ok(Box::new(TermMissingAgg::new(req, node)?))
}
AggKind::Cardinality => {
let req_data = &mut req.get_cardinality_req_data_mut(node.idx_in_req_data);
Ok(Box::new(SegmentCardinalityCollector::from_req(
req_data.column_type,
node.idx_in_req_data,
req_data.accessor.clone(),
req_data.missing_value_for_accessor,
)))
let req_data = req.get_cardinality_req_data(node.idx_in_req_data);
// For str columns, choose the per-bucket entries representation
// based on the segment's column.max_value():
// * small (< BITSET_MAX_TERM_ORD): `BitSet`, pre-allocated, no promotion machinery.
// * large: `TermOrdSet` (sparse FxHashSet that promotes to a paged bitset).
// For non-str columns the `entries` field is unused (values go
// straight into the HLL sketch); we still pick `TermOrdSet`
// because its empty Sparse(FxHashSet) costs nothing.
let is_str = req_data.column_type == ColumnType::Str;
let max_term_ord_inclusive = if is_str {
req_data.accessor.max_value()
} else {
0
};
let collector: Box<dyn SegmentAggregationCollector> =
if is_str && max_term_ord_inclusive < BITSET_MAX_TERM_ORD {
Box::new(SegmentCardinalityCollector::<BitSet>::from_req(
req_data.column_type,
node.idx_in_req_data,
req_data.accessor.clone(),
req_data.missing_value_for_accessor,
max_term_ord_inclusive,
))
} else {
Box::new(SegmentCardinalityCollector::<TermOrdSet>::from_req(
req_data.column_type,
node.idx_in_req_data,
req_data.accessor.clone(),
req_data.missing_value_for_accessor,
max_term_ord_inclusive,
))
};
Ok(collector)
}
AggKind::StatsKind(stats_type) => {
let req_data = &mut req.per_request.stats_metric_req_data[node.idx_in_req_data];
@@ -433,7 +318,7 @@ pub(crate) fn build_segment_agg_collector(
SegmentExtendedStatsCollector::from_req(req_data, sigma),
)),
StatsType::Percentiles => {
let req_data = req.get_metric_req_data_mut(node.idx_in_req_data);
let req_data = req.get_metric_req_data(node.idx_in_req_data);
Ok(Box::new(
SegmentPercentilesCollector::from_req_and_validate(
req_data.field_type,
@@ -453,12 +338,8 @@ pub(crate) fn build_segment_agg_collector(
req_data.segment_ordinal,
)))
}
AggKind::Histogram => Ok(Box::new(SegmentHistogramCollector::from_req_and_validate(
req, node,
)?)),
AggKind::DateHistogram => Ok(Box::new(SegmentHistogramCollector::from_req_and_validate(
req, node,
)?)),
AggKind::Histogram => build_segment_histogram_collector(req, node),
AggKind::DateHistogram => build_segment_histogram_collector(req, node),
AggKind::Range => Ok(build_segment_range_collector(req, node)?),
AggKind::Filter => build_segment_filter_collector(req, node),
AggKind::Composite => Ok(Box::new(
@@ -773,23 +654,18 @@ fn build_nodes(
let schema = reader.schema();
let tokenizers = &data.context.tokenizers;
let query = filter_req.parse_query(schema, tokenizers)?;
let evaluator = crate::aggregation::bucket::DocumentQueryEvaluator::new(
query,
schema.clone(),
reader,
)?;
// Pre-allocate buffer for batch filtering
let max_doc = reader.max_doc();
let buffer_capacity = crate::docset::COLLECT_BLOCK_BUFFER_LEN.min(max_doc as usize);
let matching_docs_buffer = Vec::with_capacity(buffer_capacity);
let evaluator =
std::rc::Rc::new(crate::aggregation::bucket::DocumentQueryEvaluator::new(
query,
schema.clone(),
reader,
)?);
let idx_in_req_data = data.push_filter_req_data(FilterAggReqData {
name: agg_name.to_string(),
req: filter_req.clone(),
segment_reader: reader.clone(),
evaluator,
matching_docs_buffer,
is_top_level,
});
let children = build_children(&req.sub_aggregation, reader, segment_ordinal, data)?;
@@ -985,8 +861,12 @@ fn build_terms_or_cardinality_nodes(
let str_col = str_dict_column
.as_ref()
.expect("str_dict_column must exist for string column");
allowed_term_ids =
build_allowed_term_ids_for_str(str_col, &req.include, &req.exclude)?;
allowed_term_ids = build_allowed_term_ids_for_str(
str_col,
&req.include,
&req.exclude,
missing.is_some(),
)?;
};
let idx_in_req_data = data.push_term_req_data(TermsAggReqData {
accessor,
@@ -1002,10 +882,20 @@ fn build_terms_or_cardinality_nodes(
(idx_in_req_data, AggKind::Terms)
}
TermsOrCardinalityRequest::Cardinality(ref req) => {
// `str_dict_column` is computed once per field; for JSON paths
// with mixed types it's `Some` even on the numeric req_data.
// Cardinality only consults it for the str column path, so
// gate by column_type to avoid driving non-str collectors
// through the coupon-cache path.
let str_dict_column_for_req = if column_type == ColumnType::Str {
str_dict_column.clone()
} else {
None
};
let idx_in_req_data = data.push_cardinality_req_data(CardinalityAggReqData {
accessor,
column_type,
str_dict_column: str_dict_column.clone(),
str_dict_column: str_dict_column_for_req,
missing_value_for_accessor,
name: agg_name.to_string(),
req: req.clone(),
@@ -1025,16 +915,21 @@ fn build_terms_or_cardinality_nodes(
/// Builds a single BitSet of allowed term ordinals for a string dictionary column according to
/// include/exclude parameters.
///
/// When `reserve_missing_sentinel` is true, the bitset will have 1 additional slot for the missing
/// term ordinal
fn build_allowed_term_ids_for_str(
str_col: &StrColumn,
include: &Option<IncludeExcludeParam>,
exclude: &Option<IncludeExcludeParam>,
reserve_missing_sentinel: bool,
) -> crate::Result<Option<BitSet>> {
let mut allowed: Option<BitSet> = None;
let num_terms = str_col.dictionary().num_terms() as u32;
let missing_sentinel_adjustment = if reserve_missing_sentinel { 1 } else { 0 };
let allowed_capacity = str_col.dictionary().num_terms() as u32 + missing_sentinel_adjustment;
if let Some(include) = include {
// add matches
allowed = Some(BitSet::with_max_value(num_terms));
allowed = Some(BitSet::with_max_value(allowed_capacity));
let allowed = allowed.as_mut().unwrap();
for_each_matching_term_ord(str_col, include, |ord| allowed.insert(ord))?;
};
@@ -1042,7 +937,7 @@ fn build_allowed_term_ids_for_str(
if let Some(exclude) = exclude {
if allowed.is_none() {
// Start with all terms allowed
allowed = Some(BitSet::with_max_value_and_full(num_terms));
allowed = Some(BitSet::with_max_value_and_full(allowed_capacity));
}
let allowed = allowed.as_mut().unwrap();
for_each_matching_term_ord(str_col, exclude, |ord| allowed.remove(ord))?;

65
src/aggregation/agg_req.rs
View File

@@ -115,6 +115,71 @@ pub fn get_fast_field_names(aggs: &Aggregations) -> HashSet<String> {
fast_field_names
}
/// Validates that all fields referenced in the aggregation request exist in the schema
/// and are configured as fast fields.
///
/// This is a convenience function for upfront validation before executing aggregations.
/// Returns an error if any field doesn't exist or is not a fast field.
///
/// Validation is intentionally opt-in rather than baked into aggregation execution: the
/// default lenient behavior (returning empty results for missing fields) supports
/// schema evolution and federated queries where the same request runs against segments
/// or indices with different schemas.
///
/// # Example
/// ```
/// use tantivy::aggregation::agg_req::{Aggregations, validate_aggregation_fields_exist};
/// use tantivy::schema::{Schema, FAST};
/// use tantivy::Index;
///
/// # fn main() -> tantivy::Result<()> {
/// // Create a simple index
/// let mut schema_builder = Schema::builder();
/// schema_builder.add_f64_field("price", FAST);
/// let schema = schema_builder.build();
/// let index = Index::create_in_ram(schema);
///
/// // Parse aggregation request
/// let agg_req: Aggregations = serde_json::from_str(r#"{
/// "avg_price": { "avg": { "field": "price" } }
/// }"#)?;
///
/// let reader = index.reader()?;
/// let searcher = reader.searcher();
///
/// // Validate fields before executing
/// for segment_reader in searcher.segment_readers() {
/// validate_aggregation_fields_exist(&agg_req, segment_reader)?;
/// }
/// # Ok(())
/// # }
/// ```
pub fn validate_aggregation_fields_exist(
aggs: &Aggregations,
reader: &crate::SegmentReader,
) -> crate::Result<()> {
let field_names = get_fast_field_names(aggs);
let schema = reader.schema();
for field_name in field_names {
// Check if the field is either directly in the schema or could be part of a json field
// present in the schema, and verify it's a fast field.
if let Some((field, _path)) = schema.find_field(&field_name) {
let field_type = schema.get_field_entry(field).field_type();
if !field_type.is_fast() {
return Err(crate::TantivyError::SchemaError(format!(
"Field '{}' is not a fast field. Aggregations require fast fields.",
field_name
)));
}
} else {
return Err(crate::TantivyError::FieldNotFound(field_name));
}
}
Ok(())
}
#[derive(Clone, Debug, PartialEq, Serialize, Deserialize)]
/// All aggregation types.
pub enum AggregationVariants {

3
src/aggregation/agg_result.rs
View File

@@ -208,7 +208,8 @@ pub enum BucketEntries<T> {
}
impl<T> BucketEntries<T> {
fn iter<'a>(&'a self) -> Box<dyn Iterator<Item = &'a T> + 'a> {
/// Iterate over all bucket entries.
pub fn iter<'a>(&'a self) -> Box<dyn Iterator<Item = &'a T> + 'a> {
match self {
BucketEntries::Vec(vec) => Box::new(vec.iter()),
BucketEntries::HashMap(map) => Box::new(map.values()),

43
src/aggregation/agg_tests.rs
View File

@@ -1436,3 +1436,46 @@ fn test_aggregation_on_json_object_mixed_numerical_segments() {
)
);
}
#[test]
fn test_aggregation_field_validation_helper() {
// Test the standalone validation helper function for field validation
let index = get_test_index_2_segments(false).unwrap();
let reader = index.reader().unwrap();
let searcher = reader.searcher();
let segment_reader = searcher.segment_reader(0);
// Test with invalid field
let agg_req: Aggregations = serde_json::from_str(
r#"{
"avg_test": {
"avg": { "field": "nonexistent_field" }
}
}"#,
)
.unwrap();
let result =
crate::aggregation::agg_req::validate_aggregation_fields_exist(&agg_req, segment_reader);
assert!(result.is_err());
match result {
Err(crate::TantivyError::FieldNotFound(field_name)) => {
assert_eq!(field_name, "nonexistent_field");
}
_ => panic!("Expected FieldNotFound error, got: {:?}", result),
}
// Test with valid field
let agg_req: Aggregations = serde_json::from_str(
r#"{
"avg_test": {
"avg": { "field": "score" }
}
}"#,
)
.unwrap();
let result =
crate::aggregation::agg_req::validate_aggregation_fields_exist(&agg_req, segment_reader);
assert!(result.is_ok());
}

5
src/aggregation/bucket/composite/accessors.rs
View File

@@ -16,6 +16,7 @@ use crate::{SegmentReader, TantivyError};
/// Contains all information required by the SegmentCompositeCollector to perform the
/// composite aggregation on a segment.
#[derive(Debug, Clone)]
pub struct CompositeAggReqData {
/// The name of the aggregation.
pub name: String,
@@ -34,6 +35,7 @@ impl CompositeAggReqData {
}
/// Accessors for a single column in a composite source.
#[derive(Debug, Clone)]
pub struct CompositeAccessor {
/// The fast field column
pub column: Column<u64>,
@@ -48,6 +50,7 @@ pub struct CompositeAccessor {
}
/// Accessors to all the columns that belong to the field of a composite source.
#[derive(Debug, Clone)]
pub struct CompositeSourceAccessors {
/// The accessors for this source
pub accessors: Vec<CompositeAccessor>,
@@ -358,7 +361,7 @@ impl PrecomputedDateInterval {
///
/// Some column types (term, IP) might not have an exact representation of the
/// specified after key
#[derive(Debug)]
#[derive(Debug, Clone)]
pub enum PrecomputedAfterKey {
/// The after key could be exactly represented in the column space.
Exact(u64),

48
src/aggregation/bucket/composite/collector.rs
View File

@@ -21,7 +21,7 @@ use crate::aggregation::bucket::composite::map::{DynArrayHeapMap, MAX_DYN_ARRAY_
use crate::aggregation::bucket::{
CalendarInterval, CompositeAggregationSource, MissingOrder, Order,
};
use crate::aggregation::cached_sub_aggs::{CachedSubAggs, HighCardSubAggCache};
use crate::aggregation::buffered_sub_aggs::{BufferedSubAggs, HighCardSubAggBuffer};
use crate::aggregation::intermediate_agg_result::{
CompositeIntermediateKey, IntermediateAggregationResult, IntermediateAggregationResults,
IntermediateBucketResult, IntermediateCompositeBucketEntry, IntermediateCompositeBucketResult,
@@ -118,8 +118,8 @@ impl InternalValueRepr {
pub struct SegmentCompositeCollector {
/// One DynArrayHeapMap per parent bucket.
parent_buckets: Vec<DynArrayHeapMap<InternalValueRepr, CompositeBucketCollector>>,
accessor_idx: usize,
sub_agg: Option<CachedSubAggs<HighCardSubAggCache>>,
req_data: CompositeAggReqData,
sub_agg: Option<BufferedSubAggs<HighCardSubAggBuffer>>,
bucket_id_provider: BucketIdProvider,
/// Number of sources, needed when creating new DynArrayHeapMaps.
num_sources: usize,
@@ -132,10 +132,7 @@ impl SegmentAggregationCollector for SegmentCompositeCollector {
results: &mut IntermediateAggregationResults,
parent_bucket_id: BucketId,
) -> crate::Result<()> {
let name = agg_data
.get_composite_req_data(self.accessor_idx)
.name
.clone();
let name = self.req_data.name.clone();
let buckets = self.add_intermediate_bucket_result(agg_data, parent_bucket_id)?;
results.push(
@@ -153,12 +150,11 @@ impl SegmentAggregationCollector for SegmentCompositeCollector {
agg_data: &mut AggregationsSegmentCtx,
) -> crate::Result<()> {
let mem_pre = self.get_memory_consumption(parent_bucket_id);
let composite_agg_data = agg_data.take_composite_req_data(self.accessor_idx);
for doc in docs {
let mut visitor = CompositeKeyVisitor {
doc_id: *doc,
composite_agg_data: &composite_agg_data,
composite_agg_data: &self.req_data,
buckets: &mut self.parent_buckets[parent_bucket_id as usize],
sub_agg: &mut self.sub_agg,
bucket_id_provider: &mut self.bucket_id_provider,
@@ -166,7 +162,6 @@ impl SegmentAggregationCollector for SegmentCompositeCollector {
};
visitor.visit(0, true)?;
}
agg_data.put_back_composite_req_data(self.accessor_idx, composite_agg_data);
if let Some(sub_agg) = &mut self.sub_agg {
sub_agg.check_flush_local(agg_data)?;
@@ -199,6 +194,17 @@ impl SegmentAggregationCollector for SegmentCompositeCollector {
}
Ok(())
}
fn compute_metric_value(
&self,
_bucket_id: BucketId,
_sub_agg_name: &str,
_sub_agg_property: &str,
_agg_data: &AggregationsSegmentCtx,
) -> Option<f64> {
// Composite is a multi-bucket agg with no single value to extract.
None
}
}
impl SegmentCompositeCollector {
@@ -210,22 +216,27 @@ impl SegmentCompositeCollector {
req_data: &mut AggregationsSegmentCtx,
node: &AggRefNode,
) -> crate::Result<Self> {
validate_req(req_data, node.idx_in_req_data)?;
let composite_req_data =
req_data.per_request.composite_req_data[node.idx_in_req_data].clone();
validate_req(&composite_req_data)?;
req_data
.context
.limits
.add_memory_consumed(composite_req_data.get_memory_consumption() as u64)?;
let has_sub_aggregations = !node.children.is_empty();
let sub_agg = if has_sub_aggregations {
let sub_agg_collector = build_segment_agg_collectors(req_data, &node.children)?;
Some(CachedSubAggs::new(sub_agg_collector))
Some(BufferedSubAggs::new(sub_agg_collector))
} else {
None
};
let composite_req_data = req_data.get_composite_req_data(node.idx_in_req_data);
let num_sources = composite_req_data.req.sources.len();
Ok(SegmentCompositeCollector {
parent_buckets: vec![DynArrayHeapMap::try_new(num_sources)?],
accessor_idx: node.idx_in_req_data,
req_data: composite_req_data,
sub_agg,
bucket_id_provider: BucketIdProvider::default(),
num_sources,
@@ -247,7 +258,7 @@ impl SegmentCompositeCollector {
let mut dict: FxHashMap<Vec<CompositeIntermediateKey>, IntermediateCompositeBucketEntry> =
Default::default();
dict.reserve(heap_map.size());
let composite_data = agg_data.get_composite_req_data(self.accessor_idx);
let composite_data = &self.req_data;
for (key_internal_repr, agg) in heap_map.into_iter() {
let key = resolve_key(&key_internal_repr, composite_data)?;
let mut sub_aggregation_res = IntermediateAggregationResults::default();
@@ -287,8 +298,7 @@ impl SegmentCompositeCollector {
}
}
fn validate_req(req_data: &mut AggregationsSegmentCtx, accessor_idx: usize) -> crate::Result<()> {
let composite_data = req_data.get_composite_req_data(accessor_idx);
fn validate_req(composite_data: &CompositeAggReqData) -> crate::Result<()> {
let req = &composite_data.req;
if req.sources.is_empty() {
return Err(TantivyError::InvalidArgument(
@@ -329,7 +339,7 @@ fn collect_bucket_with_limit(
limit_num_buckets: usize,
buckets: &mut DynArrayHeapMap<InternalValueRepr, CompositeBucketCollector>,
key: &[InternalValueRepr],
sub_agg: &mut Option<CachedSubAggs<HighCardSubAggCache>>,
sub_agg: &mut Option<BufferedSubAggs<HighCardSubAggBuffer>>,
bucket_id_provider: &mut BucketIdProvider,
) {
let mut record_in_bucket = |bucket: &mut CompositeBucketCollector| {
@@ -485,7 +495,7 @@ struct CompositeKeyVisitor<'a> {
doc_id: crate::DocId,
composite_agg_data: &'a CompositeAggReqData,
buckets: &'a mut DynArrayHeapMap<InternalValueRepr, CompositeBucketCollector>,
sub_agg: &'a mut Option<CachedSubAggs<HighCardSubAggCache>>,
sub_agg: &'a mut Option<BufferedSubAggs<HighCardSubAggBuffer>>,
bucket_id_provider: &'a mut BucketIdProvider,
sub_level_values: SmallVec<[InternalValueRepr; MAX_DYN_ARRAY_SIZE]>,
}

92
src/aggregation/bucket/composite/mod.rs
View File

@@ -511,14 +511,14 @@ mod tests {
fn datetime_from_iso_str(date_str: &str) -> common::DateTime {
let dt = OffsetDateTime::parse(date_str, &Rfc3339)
.expect(&format!("Failed to parse date: {}", date_str));
.unwrap_or_else(|_| panic!("Failed to parse date: {}", date_str));
let timestamp_secs = dt.unix_timestamp_nanos();
common::DateTime::from_timestamp_nanos(timestamp_secs as i64)
}
fn ms_timestamp_from_iso_str(date_str: &str) -> i64 {
let dt = OffsetDateTime::parse(date_str, &Rfc3339)
.expect(&format!("Failed to parse date: {}", date_str));
.unwrap_or_else(|_| panic!("Failed to parse date: {}", date_str));
(dt.unix_timestamp_nanos() / 1_000_000) as i64
}
@@ -548,7 +548,7 @@ mod tests {
agg_req_json["my_composite"]["composite"]["after"] = after_key.take().unwrap();
}
let agg_req: Aggregations = serde_json::from_value(agg_req_json).unwrap();
let res = exec_request(agg_req.clone(), &index).unwrap();
let res = exec_request(agg_req.clone(), index).unwrap();
let expected_page_buckets = &expected_buckets_vec[page_idx * page_size
..std::cmp::min((page_idx + 1) * page_size, expected_buckets_vec.len())];
assert_eq!(
@@ -559,34 +559,30 @@ mod tests {
page_size,
agg_req,
);
if page_idx + 1 < page_count {
assert!(
res["my_composite"].get("after_key").is_some(),
"expected after_key on all but last page"
);
after_key = Some(res["my_composite"]["after_key"].clone());
} else if res["my_composite"].get("after_key").is_some() {
// currently we sometime have an after_key on the last page,
// check that the next "page" is empty
let agg_req_json = json!({
"my_composite": {
"composite": {
"sources": composite_agg_sources,
"size": page_size,
"after": res["my_composite"]["after_key"].clone(),
}
}
});
let agg_req: Aggregations = serde_json::from_value(agg_req_json).unwrap();
let res = exec_request(agg_req.clone(), &index).unwrap();
assert_eq!(
res["my_composite"]["buckets"],
json!([]),
"expected no buckets when using after_key from last page, query: {:?}",
agg_req
);
}
assert!(
res["my_composite"].get("after_key").is_some(),
"expected after_key on every non-empty page"
);
after_key = Some(res["my_composite"]["after_key"].clone());
}
// Using the after_key from the last page must yield an empty page.
let agg_req_json = json!({
"my_composite": {
"composite": {
"sources": composite_agg_sources,
"size": page_size,
"after": after_key,
}
}
});
let agg_req: Aggregations = serde_json::from_value(agg_req_json).unwrap();
let res = exec_request(agg_req.clone(), index).unwrap();
assert_eq!(
res["my_composite"]["buckets"],
json!([]),
"expected no buckets when using after_key from last page, query: {:?}",
agg_req
);
}
}
@@ -711,8 +707,28 @@ mod tests {
{"key": {"myterm": "terme"}, "doc_count": 1}
])
);
assert!(res["my_composite"].get("after_key").is_none());
// paginating past last page should be empty
let agg_req_json = json!({
"my_composite": {
"composite": {
"sources": [
{"myterm": {"terms": {"field": "string_id"}}}
],
"size": 3,
"after": &res["my_composite"]["after_key"]
}
}
});
let agg_req: Aggregations = serde_json::from_value(agg_req_json).unwrap();
let res = exec_request(agg_req.clone(), &index).unwrap();
assert!(res["my_composite"].get("after_key").is_none());
assert_eq!(
res["my_composite"]["buckets"],
json!([]),
"expected no buckets when using after_key from last page, query: {:?}",
agg_req
);
Ok(())
}
@@ -820,7 +836,10 @@ mod tests {
{"key": {"myterm": "apple"}, "doc_count": 1}
])
);
assert!(res["fruity_aggreg"].get("after_key").is_none());
assert_eq!(
res["fruity_aggreg"]["after_key"],
json!({"myterm": "str:apple"})
);
Ok(())
}
@@ -1792,7 +1811,14 @@ mod tests {
{"key": {"month": ms_timestamp_from_iso_str("2021-02-01T00:00:00Z"), "category": "books"}, "doc_count": 1},
]),
);
assert!(res["my_composite"].get("after_key").is_none());
let feb_2021_ns = ms_timestamp_from_iso_str("2021-02-01T00:00:00Z") * 1_000_000;
assert_eq!(
res["my_composite"]["after_key"],
json!({
"month": format!("dt:{}", feb_2021_ns),
"category": "str:books"
})
);
Ok(())
}

96
src/aggregation/bucket/filter.rs
View File

@@ -1,4 +1,5 @@
use std::fmt::Debug;
use std::rc::Rc;
use common::BitSet;
use serde::{Deserialize, Deserializer, Serialize, Serializer};
@@ -6,8 +7,8 @@ use serde::{Deserialize, Deserializer, Serialize, Serializer};
use crate::aggregation::agg_data::{
build_segment_agg_collectors, AggRefNode, AggregationsSegmentCtx,
};
use crate::aggregation::cached_sub_aggs::{
CachedSubAggs, HighCardSubAggCache, LowCardSubAggCache, SubAggCache,
use crate::aggregation::buffered_sub_aggs::{
BufferedSubAggs, HighCardSubAggBuffer, LowCardSubAggBuffer, SubAggBuffer,
};
use crate::aggregation::intermediate_agg_result::{
IntermediateAggregationResult, IntermediateAggregationResults, IntermediateBucketResult,
@@ -396,6 +397,7 @@ impl PartialEq for FilterAggregation {
/// Request data for filter aggregation
/// This struct holds the per-segment data needed to execute a filter aggregation
#[derive(Clone)]
pub struct FilterAggReqData {
/// The name of the filter aggregation
pub name: String,
@@ -403,22 +405,20 @@ pub struct FilterAggReqData {
pub req: FilterAggregation,
/// The segment reader
pub segment_reader: SegmentReader,
/// Document evaluator for the filter query (precomputed BitSet)
/// This is built once when the request data is created
pub evaluator: DocumentQueryEvaluator,
/// Reusable buffer for matching documents to minimize allocations during collection
pub matching_docs_buffer: Vec<DocId>,
/// Document evaluator for the filter query (precomputed BitSet).
/// Wrapped in `Rc` so cloning the request data does not duplicate the (potentially large)
/// underlying BitSet.
pub evaluator: Rc<DocumentQueryEvaluator>,
/// True if this filter aggregation is at the top level of the aggregation tree (not nested).
pub is_top_level: bool,
}
impl FilterAggReqData {
pub(crate) fn get_memory_consumption(&self) -> usize {
// Estimate: name + segment reader reference + bitset + buffer capacity
// Estimate: name + segment reader reference + bitset
self.name.len()
+ std::mem::size_of::<SegmentReader>()
+ self.evaluator.bitset.len() / 8 // BitSet memory (bits to bytes)
+ self.matching_docs_buffer.capacity() * std::mem::size_of::<DocId>()
+ std::mem::size_of::<bool>()
}
}
@@ -503,21 +503,24 @@ struct DocCount {
}
/// Segment collector for filter aggregation
pub struct SegmentFilterCollector<C: SubAggCache> {
pub struct SegmentFilterCollector<B: SubAggBuffer> {
/// Document counts per parent bucket
parent_buckets: Vec<DocCount>,
/// Sub-aggregation collectors
sub_aggregations: Option<CachedSubAggs<C>>,
sub_aggregations: Option<BufferedSubAggs<B>>,
bucket_id_provider: BucketIdProvider,
/// Accessor index for this filter aggregation (to access FilterAggReqData)
accessor_idx: usize,
/// Per-segment filter request data, owned by this collector.
req_data: FilterAggReqData,
/// Reusable buffer for matching documents to minimize allocations during collection.
matching_docs_buffer: Vec<DocId>,
}
impl<C: SubAggCache> SegmentFilterCollector<C> {
impl<B: SubAggBuffer> SegmentFilterCollector<B> {
/// Create a new filter segment collector following the new agg_data pattern
pub(crate) fn from_req_and_validate(
req: &mut AggregationsSegmentCtx,
node: &AggRefNode,
req_data: FilterAggReqData,
) -> crate::Result<Self> {
// Build sub-aggregation collectors if any
let sub_agg_collector = if !node.children.is_empty() {
@@ -525,13 +528,17 @@ impl<C: SubAggCache> SegmentFilterCollector<C> {
} else {
None
};
let sub_agg_collector = sub_agg_collector.map(CachedSubAggs::new);
let sub_agg_collector = sub_agg_collector.map(BufferedSubAggs::new);
let max_doc = req_data.segment_reader.max_doc();
let buffer_capacity = crate::docset::COLLECT_BLOCK_BUFFER_LEN.min(max_doc as usize);
Ok(SegmentFilterCollector {
parent_buckets: Vec::new(),
sub_aggregations: sub_agg_collector,
accessor_idx: node.idx_in_req_data,
req_data,
bucket_id_provider: BucketIdProvider::default(),
matching_docs_buffer: Vec::with_capacity(buffer_capacity),
})
}
}
@@ -540,33 +547,38 @@ pub(crate) fn build_segment_filter_collector(
req: &mut AggregationsSegmentCtx,
node: &AggRefNode,
) -> crate::Result<Box<dyn SegmentAggregationCollector>> {
let is_top_level = req.per_request.filter_req_data[node.idx_in_req_data]
.as_ref()
.expect("filter_req_data slot is empty")
.is_top_level;
let req_data = req.per_request.filter_req_data[node.idx_in_req_data].clone();
req.context
.limits
.add_memory_consumed(req_data.get_memory_consumption() as u64)?;
let is_top_level = req_data.is_top_level;
if is_top_level {
Ok(Box::new(
SegmentFilterCollector::<LowCardSubAggCache>::from_req_and_validate(req, node)?,
SegmentFilterCollector::<LowCardSubAggBuffer>::from_req_and_validate(
req, node, req_data,
)?,
))
} else {
Ok(Box::new(
SegmentFilterCollector::<HighCardSubAggCache>::from_req_and_validate(req, node)?,
SegmentFilterCollector::<HighCardSubAggBuffer>::from_req_and_validate(
req, node, req_data,
)?,
))
}
}
impl<C: SubAggCache> Debug for SegmentFilterCollector<C> {
impl<B: SubAggBuffer> Debug for SegmentFilterCollector<B> {
fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result {
f.debug_struct("SegmentFilterCollector")
.field("buckets", &self.parent_buckets)
.field("has_sub_aggs", &self.sub_aggregations.is_some())
.field("accessor_idx", &self.accessor_idx)
.field("name", &self.req_data.name)
.finish()
}
}
impl<C: SubAggCache> SegmentAggregationCollector for SegmentFilterCollector<C> {
impl<B: SubAggBuffer> SegmentAggregationCollector for SegmentFilterCollector<B> {
fn add_intermediate_aggregation_result(
&mut self,
agg_data: &AggregationsSegmentCtx,
@@ -598,11 +610,7 @@ impl<C: SubAggCache> SegmentAggregationCollector for SegmentFilterCollector<C> {
};
// Get the name of this filter aggregation
let name = agg_data.per_request.filter_req_data[self.accessor_idx]
.as_ref()
.expect("filter_req_data slot is empty")
.name
.clone();
let name = self.req_data.name.clone();
results.push(
name,
@@ -623,27 +631,24 @@ impl<C: SubAggCache> SegmentAggregationCollector for SegmentFilterCollector<C> {
}
let mut bucket = self.parent_buckets[parent_bucket_id as usize];
// Take the request data to avoid borrow checker issues with sub-aggregations
let mut req = agg_data.take_filter_req_data(self.accessor_idx);
// Use batch filtering with O(1) BitSet lookups
req.matching_docs_buffer.clear();
req.evaluator
.filter_batch(docs, &mut req.matching_docs_buffer);
self.matching_docs_buffer.clear();
self.req_data
.evaluator
.filter_batch(docs, &mut self.matching_docs_buffer);
bucket.doc_count += req.matching_docs_buffer.len() as u64;
bucket.doc_count += self.matching_docs_buffer.len() as u64;
// Batch process sub-aggregations if we have matches
if !req.matching_docs_buffer.is_empty() {
if !self.matching_docs_buffer.is_empty() {
if let Some(sub_aggs) = &mut self.sub_aggregations {
for &doc_id in &req.matching_docs_buffer {
for &doc_id in &self.matching_docs_buffer {
sub_aggs.push(bucket.bucket_id, doc_id);
}
}
}
// Put the request data back
agg_data.put_back_filter_req_data(self.accessor_idx, req);
if let Some(sub_aggs) = &mut self.sub_aggregations {
sub_aggs.check_flush_local(agg_data)?;
}
@@ -674,6 +679,17 @@ impl<C: SubAggCache> SegmentAggregationCollector for SegmentFilterCollector<C> {
}
Ok(())
}
fn compute_metric_value(
&self,
_bucket_id: BucketId,
_sub_agg_name: &str,
_sub_agg_property: &str,
_agg_data: &AggregationsSegmentCtx,
) -> Option<f64> {
// TODO: forward into the inner `sub_agg` for nested order paths (`filter.metric`).
None
}
}
/// Intermediate result for filter aggregation

507
src/aggregation/bucket/histogram/histogram.rs
View File

@@ -10,7 +10,7 @@ use crate::aggregation::agg_data::{
};
use crate::aggregation::agg_req::Aggregations;
use crate::aggregation::agg_result::BucketEntry;
use crate::aggregation::cached_sub_aggs::{CachedSubAggs, HighCardCachedSubAggs};
use crate::aggregation::buffered_sub_aggs::{BufferedSubAggs, HighCardBufferedSubAggs};
use crate::aggregation::intermediate_agg_result::{
IntermediateAggregationResult, IntermediateAggregationResults, IntermediateBucketResult,
IntermediateHistogramBucketEntry,
@@ -21,6 +21,7 @@ use crate::TantivyError;
/// Contains all information required by the SegmentHistogramCollector to perform the
/// histogram or date_histogram aggregation on a segment.
#[derive(Debug, Clone)]
pub struct HistogramAggReqData {
/// The column accessor to access the fast field values.
pub accessor: Column<u64>,
@@ -243,22 +244,55 @@ impl Display for HistogramBounds {
}
impl HistogramBounds {
fn contains(&self, val: f64) -> bool {
pub(crate) fn contains(&self, val: f64) -> bool {
val >= self.min && val <= self.max
}
}
#[derive(Default, Clone, Debug, PartialEq)]
pub(crate) struct SegmentHistogramBucketEntry {
pub key: f64,
pub doc_count: u64,
pub bucket_id: BucketId,
/// The per-bucket identifier stored in a [`SegmentHistogramBucketEntry`].
///
/// It is [`BucketId`] when the histogram has sub aggregations (which key their state by it), and
/// the zero-sized `()` when it does not. Without sub aggregations the id is never read, so storing
/// `()` drops 8 bytes per bucket (24 -> 16) and turns id assignment into a no-op.
pub trait BucketIdSlot: Copy + Default + std::fmt::Debug + PartialEq {
/// Assigns the next id from the provider, called once when a bucket is first filled.
fn assign(provider: &mut BucketIdProvider) -> Self;
/// Resolves to the `BucketId` for sub-aggregation bookkeeping.
///
/// Only ever called for the [`BucketId`] slot: the `()` slot is used exactly when there are no
/// sub aggregations, so every call site is guarded by `sub_agg.is_some()` and is dead for `()`.
fn to_bucket_id(self) -> BucketId;
}
impl BucketIdSlot for BucketId {
#[inline(always)]
fn assign(provider: &mut BucketIdProvider) -> Self {
provider.next_bucket_id()
}
#[inline(always)]
fn to_bucket_id(self) -> BucketId {
self
}
}
impl BucketIdSlot for () {
#[inline(always)]
fn assign(_provider: &mut BucketIdProvider) -> Self {}
#[inline(always)]
fn to_bucket_id(self) -> BucketId {
unreachable!("bucket ids are only resolved when sub aggregations are present")
}
}
impl SegmentHistogramBucketEntry {
#[derive(Default, Clone, Debug, PartialEq)]
pub(crate) struct SegmentHistogramBucketEntry<B> {
pub key: f64,
pub doc_count: u64,
pub bucket_id: B,
}
impl<B: BucketIdSlot> SegmentHistogramBucketEntry<B> {
pub(crate) fn into_intermediate_bucket_entry(
self,
sub_aggregation: &mut Option<HighCardCachedSubAggs>,
sub_aggregation: &mut Option<HighCardBufferedSubAggs>,
agg_data: &AggregationsSegmentCtx,
) -> crate::Result<IntermediateHistogramBucketEntry> {
let mut sub_aggregation_res = IntermediateAggregationResults::default();
@@ -268,7 +302,7 @@ impl SegmentHistogramBucketEntry {
.add_intermediate_aggregation_result(
agg_data,
&mut sub_aggregation_res,
self.bucket_id,
self.bucket_id.to_bucket_id(),
)?;
}
Ok(IntermediateHistogramBucketEntry {
@@ -279,34 +313,147 @@ impl SegmentHistogramBucketEntry {
}
}
#[derive(Clone, Debug, Default)]
struct HistogramBuckets {
pub buckets: FxHashMap<i64, SegmentHistogramBucketEntry>,
/// The contiguous bucket range a histogram can span, derived from the column min/max (clamped to
/// the histogram bounds). Buckets in `[base_pos, base_pos + len)` can be stored in a flat `Vec`
/// indexed by `bucket_pos - base_pos`, avoiding the hash map on the hot path.
#[derive(Clone, Copy, Debug)]
pub(crate) struct DenseRange {
/// `bucket_pos` mapped to index 0 of the dense `Vec`.
pub(crate) base_pos: i64,
/// Number of bucket positions in the range.
pub(crate) len: usize,
}
/// Storage for the histogram buckets of a single parent bucket.
///
/// Starts out sparse (a hash map keyed by `bucket_pos`). Once enough distinct buckets have been
/// filled that we are clearly going to cover most of the column's theoretical range, it switches
/// to a dense `Vec` indexed by `bucket_pos - base_pos`, which removes hashing from the hot loop.
#[derive(Clone, Debug)]
enum HistogramBuckets<B> {
Sparse(FxHashMap<i64, SegmentHistogramBucketEntry<B>>),
Dense {
base_pos: i64,
/// One slot per bucket position; a slot with `doc_count == 0` has not been hit yet.
buckets: Vec<SegmentHistogramBucketEntry<B>>,
},
}
impl<B> Default for HistogramBuckets<B> {
fn default() -> Self {
HistogramBuckets::Sparse(FxHashMap::default())
}
}
impl<B: BucketIdSlot> HistogramBuckets<B> {
fn memory_consumption(&self) -> u64 {
let num_slots = match self {
HistogramBuckets::Sparse(map) => map.capacity(),
HistogramBuckets::Dense { buckets, .. } => buckets.capacity(),
};
num_slots as u64 * std::mem::size_of::<SegmentHistogramBucketEntry<B>>() as u64
}
/// Switches from sparse to dense storage once the dense `Vec` would use no more memory than the
/// hash map does now, so the switch never increases memory. Called at block boundaries.
///
/// The `Vec` holds one `Entry` per bucket position in the range. The map additionally stores
/// the key and a control byte per slot, at a load factor of 7/16..7/8, so for a dense histogram
/// its footprint grows past the `Vec` well before full coverage. And since the `Vec` never
/// grows afterwards while the map would keep growing, dense only gets relatively cheaper — so
/// no upper bound on the range is needed: a large but sparse range simply never crosses over.
#[inline]
fn maybe_densify(&mut self, dense_range: Option<DenseRange>) {
let Some(range) = dense_range else { return };
let HistogramBuckets::Sparse(map) = self else {
return;
};
let dense_bytes = range
.len
.saturating_mul(std::mem::size_of::<SegmentHistogramBucketEntry<B>>());
let sparse_bytes = map
.capacity()
.saturating_mul(std::mem::size_of::<(i64, SegmentHistogramBucketEntry<B>)>() + 1);
if dense_bytes > sparse_bytes {
return;
}
let map = std::mem::take(map);
let mut buckets = vec![SegmentHistogramBucketEntry::<B>::default(); range.len];
for (bucket_pos, entry) in map {
buckets[(bucket_pos - range.base_pos) as usize] = entry;
}
*self = HistogramBuckets::Dense {
base_pos: range.base_pos,
buckets,
};
}
/// Returns the bucket entry for `bucket_pos`, setting its key (and `bucket_id`, when `B` is
/// [`BucketId`]) on first use.
///
/// For the dense variant `bucket_pos` is guaranteed to be inside the range, since it is
/// derived from the column min/max that bounds every value (see [`compute_dense_range`]).
#[inline]
fn get_or_create(
&mut self,
bucket_pos: i64,
bucket_id_provider: &mut BucketIdProvider,
key_from_pos: impl FnOnce(i64) -> f64,
) -> &mut SegmentHistogramBucketEntry<B> {
match self {
HistogramBuckets::Sparse(map) => {
map.entry(bucket_pos)
.or_insert_with(|| SegmentHistogramBucketEntry {
key: key_from_pos(bucket_pos),
doc_count: 0,
bucket_id: B::assign(bucket_id_provider),
})
}
HistogramBuckets::Dense { base_pos, buckets } => {
let idx = (bucket_pos - *base_pos) as usize;
debug_assert!(idx < buckets.len(), "bucket_pos outside the dense range");
let entry = &mut buckets[idx];
if entry.doc_count == 0 {
entry.key = key_from_pos(bucket_pos);
entry.bucket_id = B::assign(bucket_id_provider);
}
entry
}
}
}
/// Consumes the storage, yielding all non-empty bucket entries.
fn into_filled_entries(self) -> Vec<SegmentHistogramBucketEntry<B>> {
match self {
HistogramBuckets::Sparse(map) => map.into_values().collect(),
HistogramBuckets::Dense { buckets, .. } => {
buckets.into_iter().filter(|b| b.doc_count > 0).collect()
}
}
}
}
/// The collector puts values from the fast field into the correct buckets and does a conversion to
/// the correct datatype.
#[derive(Debug)]
pub struct SegmentHistogramCollector {
pub struct SegmentHistogramCollector<B> {
/// The buckets containing the aggregation data.
/// One Histogram bucket per parent bucket id.
parent_buckets: Vec<HistogramBuckets>,
sub_agg: Option<HighCardCachedSubAggs>,
accessor_idx: usize,
parent_buckets: Vec<HistogramBuckets<B>>,
sub_agg: Option<HighCardBufferedSubAggs>,
req_data: HistogramAggReqData,
bucket_id_provider: BucketIdProvider,
/// Theoretical bucket range derived from the column min/max, if dense `Vec` storage is
/// viable. `None` keeps every parent bucket in the sparse hash map.
dense_range: Option<DenseRange>,
}
impl SegmentAggregationCollector for SegmentHistogramCollector {
impl<B: BucketIdSlot> SegmentAggregationCollector for SegmentHistogramCollector<B> {
fn add_intermediate_aggregation_result(
&mut self,
agg_data: &AggregationsSegmentCtx,
results: &mut IntermediateAggregationResults,
parent_bucket_id: BucketId,
) -> crate::Result<()> {
let name = agg_data
.get_histogram_req_data(self.accessor_idx)
.name
.clone();
let name = self.req_data.name.clone();
// TODO: avoid prepare_max_bucket here and handle empty buckets.
self.prepare_max_bucket(parent_bucket_id, agg_data)?;
let histogram = std::mem::take(&mut self.parent_buckets[parent_bucket_id as usize]);
@@ -323,10 +470,13 @@ impl SegmentAggregationCollector for SegmentHistogramCollector {
docs: &[crate::DocId],
agg_data: &mut AggregationsSegmentCtx,
) -> crate::Result<()> {
let req = agg_data.take_histogram_req_data(self.accessor_idx);
let mem_pre = self.get_memory_consumption();
let buckets = &mut self.parent_buckets[parent_bucket_id as usize].buckets;
let mem_pre = self.get_memory_consumption(parent_bucket_id);
let dense_range = self.dense_range;
let store = &mut self.parent_buckets[parent_bucket_id as usize];
// Upgrade to dense storage before processing the block if the buckets are dense enough.
store.maybe_densify(dense_range);
let req = &self.req_data;
let bounds = req.bounds;
let interval = req.req.interval;
let offset = req.offset;
@@ -335,35 +485,43 @@ impl SegmentAggregationCollector for SegmentHistogramCollector {
agg_data
.column_block_accessor
.fetch_block(docs, &req.accessor);
for (doc, val) in agg_data
.column_block_accessor
.iter_docid_vals(docs, &req.accessor)
{
let val = f64_from_fastfield_u64(val, req.field_type);
let bucket_pos = get_bucket_pos(val);
if bounds.contains(val) {
let bucket = buckets.entry(bucket_pos).or_insert_with(|| {
let key = get_bucket_key_from_pos(bucket_pos as f64, interval, offset);
SegmentHistogramBucketEntry {
key,
doc_count: 0,
bucket_id: self.bucket_id_provider.next_bucket_id(),
}
});
bucket.doc_count += 1;
if let Some(sub_agg) = &mut self.sub_agg {
sub_agg.push(bucket.bucket_id, doc);
// special path for nested buckets
if let Some(sub_agg) = &mut self.sub_agg {
for (doc, val) in agg_data
.column_block_accessor
.iter_docid_vals(docs, &req.accessor)
{
let val = f64_from_fastfield_u64(val, req.field_type);
if bounds.contains(val) {
let bucket = store.get_or_create(
get_bucket_pos(val),
&mut self.bucket_id_provider,
|pos| get_bucket_key_from_pos(pos as f64, interval, offset),
);
bucket.doc_count += 1;
sub_agg.push(bucket.bucket_id.to_bucket_id(), doc);
}
}
} else {
for val in agg_data.column_block_accessor.iter_vals() {
let val = f64_from_fastfield_u64(val, req.field_type);
if bounds.contains(val) {
let bucket = store.get_or_create(
get_bucket_pos(val),
&mut self.bucket_id_provider,
|pos| get_bucket_key_from_pos(pos as f64, interval, offset),
);
bucket.doc_count += 1;
}
}
}
agg_data.put_back_histogram_req_data(self.accessor_idx, req);
let mem_delta = self.get_memory_consumption() - mem_pre;
if mem_delta > 0 {
agg_data
.context
.limits
.add_memory_consumed(mem_delta as u64)?;
// `checked_sub` is `None` when densifying shrank the accounted memory; only account growth.
if let Some(mem_delta) = self
.get_memory_consumption(parent_bucket_id)
.checked_sub(mem_pre)
{
agg_data.context.limits.add_memory_consumed(mem_delta)?;
}
if let Some(sub_agg) = &mut self.sub_agg {
@@ -386,39 +544,45 @@ impl SegmentAggregationCollector for SegmentHistogramCollector {
_agg_data: &AggregationsSegmentCtx,
) -> crate::Result<()> {
while self.parent_buckets.len() <= max_bucket as usize {
self.parent_buckets.push(HistogramBuckets {
buckets: FxHashMap::default(),
});
self.parent_buckets.push(HistogramBuckets::default());
}
Ok(())
}
fn compute_metric_value(
&self,
_bucket_id: BucketId,
_sub_agg_name: &str,
_sub_agg_property: &str,
_agg_data: &AggregationsSegmentCtx,
) -> Option<f64> {
// Histogram is a multi-bucket agg with no single value to extract.
None
}
}
impl SegmentHistogramCollector {
fn get_memory_consumption(&self) -> usize {
let self_mem = std::mem::size_of::<Self>();
let buckets_mem = self.parent_buckets.len() * std::mem::size_of::<HistogramBuckets>();
self_mem + buckets_mem
impl<B: BucketIdSlot> SegmentHistogramCollector<B> {
fn get_memory_consumption(&self, parent_bucket_id: BucketId) -> u64 {
self.parent_buckets[parent_bucket_id as usize].memory_consumption()
}
/// Converts the collector result into a intermediate bucket result.
fn add_intermediate_bucket_result(
&mut self,
agg_data: &AggregationsSegmentCtx,
histogram: HistogramBuckets,
histogram: HistogramBuckets<B>,
) -> crate::Result<IntermediateBucketResult> {
let mut buckets = Vec::with_capacity(histogram.buckets.len());
let filled = histogram.into_filled_entries();
let mut buckets = Vec::with_capacity(filled.len());
for bucket in histogram.buckets.into_values() {
for bucket in filled {
let bucket_res = bucket.into_intermediate_bucket_entry(&mut self.sub_agg, agg_data);
buckets.push(bucket_res?);
}
buckets.sort_unstable_by(|b1, b2| b1.key.total_cmp(&b2.key));
let is_date_agg = agg_data
.get_histogram_req_data(self.accessor_idx)
.field_type
== ColumnType::DateTime;
let is_date_agg = self.req_data.field_type == ColumnType::DateTime;
Ok(IntermediateBucketResult::Histogram {
buckets,
is_date_agg,
@@ -434,32 +598,163 @@ impl SegmentHistogramCollector {
} else {
None
};
let req_data = agg_data.get_histogram_req_data_mut(node.idx_in_req_data);
req_data.req.validate()?;
if req_data.field_type == ColumnType::DateTime && !req_data.is_date_histogram {
req_data.req.normalize_date_time();
}
req_data.bounds = req_data.req.hard_bounds.unwrap_or(HistogramBounds {
min: f64::MIN,
max: f64::MAX,
});
req_data.offset = req_data.req.offset.unwrap_or(0.0);
let sub_agg = sub_agg.map(CachedSubAggs::new);
let mut req_data = agg_data.per_request.histogram_req_data[node.idx_in_req_data].clone();
normalize_histogram_req(&mut req_data)?;
agg_data
.context
.limits
.add_memory_consumed(req_data.get_memory_consumption() as u64)?;
let dense_range = compute_dense_range(
&req_data.accessor,
req_data.field_type,
req_data.req.interval,
req_data.offset,
req_data.bounds,
);
let sub_agg = sub_agg.map(BufferedSubAggs::new);
Ok(Self {
parent_buckets: Default::default(),
sub_agg,
accessor_idx: node.idx_in_req_data,
req_data,
bucket_id_provider: BucketIdProvider::default(),
dense_range,
})
}
}
impl SegmentHistogramCollector<()> {
/// Builds a histogram collector whose parent `t` is a dense histogram filled from
/// `counts[t * num_time_buckets .. (t + 1) * num_time_buckets]` (row-major). Used by the fused
/// terms×histogram collector to turn its flat 2D counters into the regular intermediate result,
/// so cross-segment merging is shared with the general path.
pub(crate) fn from_dense_rows(
req_data: HistogramAggReqData,
base_pos: i64,
num_time_buckets: usize,
counts: &[u32],
) -> Self {
let interval = req_data.req.interval;
let offset = req_data.offset;
let num_parents = if num_time_buckets == 0 {
0
} else {
counts.len() / num_time_buckets
};
let parent_buckets = (0..num_parents)
.map(|t| {
let row = &counts[t * num_time_buckets..(t + 1) * num_time_buckets];
let buckets = row
.iter()
.enumerate()
.map(|(b, &doc_count)| SegmentHistogramBucketEntry {
key: get_bucket_key_from_pos(
(base_pos + b as i64) as f64,
interval,
offset,
),
doc_count: doc_count as u64,
bucket_id: (),
})
.collect();
HistogramBuckets::Dense { base_pos, buckets }
})
.collect();
Self {
parent_buckets,
sub_agg: None,
req_data,
bucket_id_provider: BucketIdProvider::default(),
dense_range: None,
}
}
}
/// Validates and normalizes a histogram request in place: applies date ns-normalization (for a
/// `histogram` on a date column) and resolves `bounds`/`offset` from the request.
fn normalize_histogram_req(req_data: &mut HistogramAggReqData) -> crate::Result<()> {
req_data.req.validate()?;
if req_data.field_type == ColumnType::DateTime && !req_data.is_date_histogram {
req_data.req.normalize_date_time();
}
req_data.bounds = req_data.req.hard_bounds.unwrap_or(HistogramBounds {
min: f64::MIN,
max: f64::MAX,
});
req_data.offset = req_data.req.offset.unwrap_or(0.0);
Ok(())
}
/// Clones and normalizes (resolving interval/offset/bounds) the histogram request at `node`, and
/// returns it together with its dense bucket range — or `None` if the column has no usable range.
/// Used by the fused terms×histogram collector, which then owns the normalized request.
pub(crate) fn prepare_histogram_dense_range(
agg_data: &AggregationsSegmentCtx,
node: &AggRefNode,
) -> crate::Result<Option<(HistogramAggReqData, DenseRange)>> {
let mut req_data = agg_data.per_request.histogram_req_data[node.idx_in_req_data].clone();
normalize_histogram_req(&mut req_data)?;
let dense_range = compute_dense_range(
&req_data.accessor,
req_data.field_type,
req_data.req.interval,
req_data.offset,
req_data.bounds,
);
Ok(dense_range.map(|range| (req_data, range)))
}
/// Builds a boxed histogram (or date histogram) segment collector, picking the bucket-id storage
/// based on whether there are sub aggregations: `()` (no id stored) when there are none, otherwise
/// [`BucketId`].
pub(crate) fn build_segment_histogram_collector(
agg_data: &mut AggregationsSegmentCtx,
node: &AggRefNode,
) -> crate::Result<Box<dyn SegmentAggregationCollector>> {
if node.children.is_empty() {
Ok(Box::new(
SegmentHistogramCollector::<()>::from_req_and_validate(agg_data, node)?,
))
} else {
Ok(Box::new(
SegmentHistogramCollector::<BucketId>::from_req_and_validate(agg_data, node)?,
))
}
}
#[inline]
fn get_bucket_pos_f64(val: f64, interval: f64, offset: f64) -> f64 {
pub(crate) fn get_bucket_pos_f64(val: f64, interval: f64, offset: f64) -> f64 {
((val - offset) / interval).floor()
}
/// Computes the dense bucket range for a column from its min/max value (clamped to the histogram
/// bounds), or `None` if there are no values within bounds (or the range overflows `usize`).
///
/// There is no upper bound on the range: whether dense storage is actually used is decided later,
/// per parent bucket, by [`HistogramBuckets::maybe_densify`] based on the memory it would save.
///
/// The column min/max bound every value the collector can see, so a `Vec` sized to this range can
/// be indexed by `bucket_pos - base_pos` without any out-of-bounds check on the hot path.
fn compute_dense_range(
accessor: &Column<u64>,
field_type: ColumnType,
interval: f64,
offset: f64,
bounds: HistogramBounds,
) -> Option<DenseRange> {
let col_min = f64_from_fastfield_u64(accessor.min_value(), field_type);
let col_max = f64_from_fastfield_u64(accessor.max_value(), field_type);
let lo = col_min.max(bounds.min);
let hi = col_max.min(bounds.max);
if lo > hi {
return None;
}
let base_pos = get_bucket_pos_f64(lo, interval, offset) as i64;
let top_pos = get_bucket_pos_f64(hi, interval, offset) as i64;
let len = usize::try_from(top_pos.checked_sub(base_pos)?.checked_add(1)?).ok()?;
(len > 0).then_some(DenseRange { base_pos, len })
}
#[inline]
fn get_bucket_key_from_pos(bucket_pos: f64, interval: f64, offset: f64) -> f64 {
bucket_pos * interval + offset
@@ -764,6 +1059,62 @@ mod tests {
Ok(())
}
#[test]
fn histogram_dense_storage_test() -> crate::Result<()> {
histogram_dense_storage_test_with_opt(false)?;
histogram_dense_storage_test_with_opt(true)?;
Ok(())
}
/// Exercises the switch from sparse hash map to dense `Vec` storage. The switch happens at a
/// block boundary (a block is `COLLECT_BLOCK_BUFFER_LEN` = 64 docs), so we need many docs in a
/// single segment, densely covering the bucket range. `with_sub_agg` toggles the `iter_vals`
/// fast path vs. the `iter_docid_vals` path used when there is a sub aggregation.
fn histogram_dense_storage_test_with_opt(with_sub_agg: bool) -> crate::Result<()> {
let num_buckets = 50usize;
let docs_per_bucket = 10usize;
// Value `k` repeated `docs_per_bucket` times for each bucket `k`, so every value in bucket
// `k` equals `k` and the per-bucket average is exactly `k`.
let values: Vec<f64> = (0..num_buckets * docs_per_bucket)
.map(|i| (i % num_buckets) as f64)
.collect();
// `merge_segments = true` collapses the per-value segments into a single segment with all
// the docs, which is collected in 64-doc blocks and therefore switches to dense storage.
let index = get_test_index_from_values(true, &values)?;
let agg_req: Aggregations = serde_json::from_value(if with_sub_agg {
json!({
"histogram": {
"histogram": { "field": "score_f64", "interval": 1.0 },
"aggs": { "avg": { "avg": { "field": "score_f64" } } }
}
})
} else {
json!({
"histogram": {
"histogram": { "field": "score_f64", "interval": 1.0 }
}
})
})
.unwrap();
let res = exec_request(agg_req, &index)?;
for k in 0..num_buckets {
assert_eq!(res["histogram"]["buckets"][k]["key"], k as f64);
assert_eq!(
res["histogram"]["buckets"][k]["doc_count"],
docs_per_bucket as u64
);
if with_sub_agg {
assert_eq!(res["histogram"]["buckets"][k]["avg"]["value"], k as f64);
}
}
assert_eq!(res["histogram"]["buckets"][num_buckets], Value::Null);
Ok(())
}
#[test]
fn histogram_memory_limit() -> crate::Result<()> {
let index = get_test_index_with_num_docs(true, 100)?;

137
src/aggregation/bucket/range.rs
View File

@@ -9,8 +9,9 @@ use crate::aggregation::agg_data::{
build_segment_agg_collectors, AggRefNode, AggregationsSegmentCtx,
};
use crate::aggregation::agg_limits::AggregationLimitsGuard;
use crate::aggregation::cached_sub_aggs::{
CachedSubAggs, HighCardSubAggCache, LowCardCachedSubAggs, LowCardSubAggCache, SubAggCache,
use crate::aggregation::buffered_sub_aggs::{
BufferedSubAggs, HighCardSubAggBuffer, LowCardBufferedSubAggs, LowCardSubAggBuffer,
SubAggBuffer,
};
use crate::aggregation::intermediate_agg_result::{
IntermediateAggregationResult, IntermediateAggregationResults, IntermediateBucketResult,
@@ -22,6 +23,7 @@ use crate::TantivyError;
/// Contains all information required by the SegmentRangeCollector to perform the
/// range aggregation on a segment.
#[derive(Debug, Clone)]
pub struct RangeAggReqData {
/// The column accessor to access the fast field values.
pub accessor: Column<u64>,
@@ -155,13 +157,13 @@ pub(crate) struct SegmentRangeAndBucketEntry {
/// The collector puts values from the fast field into the correct buckets and does a conversion to
/// the correct datatype.
pub struct SegmentRangeCollector<C: SubAggCache> {
pub struct SegmentRangeCollector<B: SubAggBuffer> {
/// The buckets containing the aggregation data.
/// One for each ParentBucketId
parent_buckets: Vec<Vec<SegmentRangeAndBucketEntry>>,
column_type: ColumnType,
pub(crate) accessor_idx: usize,
sub_agg: Option<CachedSubAggs<C>>,
pub(crate) req_data: RangeAggReqData,
sub_agg: Option<BufferedSubAggs<B>>,
/// Here things get a bit weird. We need to assign unique bucket ids across all
/// parent buckets. So we keep track of the next available bucket id here.
/// This allows a kind of flattening of the bucket ids across all parent buckets.
@@ -178,12 +180,12 @@ pub struct SegmentRangeCollector<C: SubAggCache> {
limits: AggregationLimitsGuard,
}
impl<C: SubAggCache> Debug for SegmentRangeCollector<C> {
impl<B: SubAggBuffer> Debug for SegmentRangeCollector<B> {
fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result {
f.debug_struct("SegmentRangeCollector")
.field("parent_buckets_len", &self.parent_buckets.len())
.field("column_type", &self.column_type)
.field("accessor_idx", &self.accessor_idx)
.field("name", &self.req_data.name)
.field("has_sub_agg", &self.sub_agg.is_some())
.finish()
}
@@ -229,7 +231,7 @@ impl SegmentRangeBucketEntry {
}
}
impl<C: SubAggCache> SegmentAggregationCollector for SegmentRangeCollector<C> {
impl<B: SubAggBuffer> SegmentAggregationCollector for SegmentRangeCollector<B> {
fn add_intermediate_aggregation_result(
&mut self,
agg_data: &AggregationsSegmentCtx,
@@ -238,10 +240,7 @@ impl<C: SubAggCache> SegmentAggregationCollector for SegmentRangeCollector<C> {
) -> crate::Result<()> {
self.prepare_max_bucket(parent_bucket_id, agg_data)?;
let field_type = self.column_type;
let name = agg_data
.get_range_req_data(self.accessor_idx)
.name
.to_string();
let name = self.req_data.name.to_string();
let buckets = std::mem::take(&mut self.parent_buckets[parent_bucket_id as usize]);
@@ -280,17 +279,15 @@ impl<C: SubAggCache> SegmentAggregationCollector for SegmentRangeCollector<C> {
docs: &[crate::DocId],
agg_data: &mut AggregationsSegmentCtx,
) -> crate::Result<()> {
let req = agg_data.take_range_req_data(self.accessor_idx);
agg_data
.column_block_accessor
.fetch_block(docs, &req.accessor);
.fetch_block(docs, &self.req_data.accessor);
let buckets = &mut self.parent_buckets[parent_bucket_id as usize];
for (doc, val) in agg_data
.column_block_accessor
.iter_docid_vals(docs, &req.accessor)
.iter_docid_vals(docs, &self.req_data.accessor)
{
let bucket_pos = get_bucket_pos(val, buckets);
let bucket = &mut buckets[bucket_pos];
@@ -300,7 +297,6 @@ impl<C: SubAggCache> SegmentAggregationCollector for SegmentRangeCollector<C> {
}
}
agg_data.put_back_range_req_data(self.accessor_idx, req);
if let Some(sub_agg) = self.sub_agg.as_mut() {
sub_agg.check_flush_local(agg_data)?;
}
@@ -318,15 +314,26 @@ impl<C: SubAggCache> SegmentAggregationCollector for SegmentRangeCollector<C> {
fn prepare_max_bucket(
&mut self,
max_bucket: BucketId,
agg_data: &AggregationsSegmentCtx,
_agg_data: &AggregationsSegmentCtx,
) -> crate::Result<()> {
while self.parent_buckets.len() <= max_bucket as usize {
let new_buckets = self.create_new_buckets(agg_data)?;
let new_buckets = self.create_new_buckets()?;
self.parent_buckets.push(new_buckets);
}
Ok(())
}
fn compute_metric_value(
&self,
_bucket_id: BucketId,
_sub_agg_name: &str,
_sub_agg_property: &str,
_agg_data: &AggregationsSegmentCtx,
) -> Option<f64> {
// Range is a multi-bucket agg with no single value to extract.
None
}
}
/// Build a concrete `SegmentRangeCollector` with either a Vec- or HashMap-backed
/// bucket storage, depending on the column type and aggregation level.
@@ -334,8 +341,11 @@ pub(crate) fn build_segment_range_collector(
agg_data: &mut AggregationsSegmentCtx,
node: &AggRefNode,
) -> crate::Result<Box<dyn SegmentAggregationCollector>> {
let accessor_idx = node.idx_in_req_data;
let req_data = agg_data.get_range_req_data(node.idx_in_req_data);
let req_data = agg_data.per_request.range_req_data[node.idx_in_req_data].clone();
agg_data
.context
.limits
.add_memory_consumed(req_data.get_memory_consumption() as u64)?;
let field_type = req_data.field_type;
// TODO: A better metric instead of is_top_level would be the number of buckets expected.
@@ -350,19 +360,19 @@ pub(crate) fn build_segment_range_collector(
};
if is_low_card {
Ok(Box::new(SegmentRangeCollector::<LowCardSubAggCache> {
sub_agg: sub_agg.map(LowCardCachedSubAggs::new),
Ok(Box::new(SegmentRangeCollector::<LowCardSubAggBuffer> {
sub_agg: sub_agg.map(LowCardBufferedSubAggs::new),
column_type: field_type,
accessor_idx,
req_data,
parent_buckets: Vec::new(),
bucket_id_provider: BucketIdProvider::default(),
limits: agg_data.context.limits.clone(),
}))
} else {
Ok(Box::new(SegmentRangeCollector::<HighCardSubAggCache> {
sub_agg: sub_agg.map(CachedSubAggs::new),
Ok(Box::new(SegmentRangeCollector::<HighCardSubAggBuffer> {
sub_agg: sub_agg.map(BufferedSubAggs::new),
column_type: field_type,
accessor_idx,
req_data,
parent_buckets: Vec::new(),
bucket_id_provider: BucketIdProvider::default(),
limits: agg_data.context.limits.clone(),
@@ -370,13 +380,10 @@ pub(crate) fn build_segment_range_collector(
}
}
impl<C: SubAggCache> SegmentRangeCollector<C> {
pub(crate) fn create_new_buckets(
&mut self,
agg_data: &AggregationsSegmentCtx,
) -> crate::Result<Vec<SegmentRangeAndBucketEntry>> {
impl<B: SubAggBuffer> SegmentRangeCollector<B> {
pub(crate) fn create_new_buckets(&mut self) -> crate::Result<Vec<SegmentRangeAndBucketEntry>> {
let field_type = self.column_type;
let req_data = agg_data.get_range_req_data(self.accessor_idx);
let req_data = &self.req_data;
// The range input on the request is f64.
// We need to convert to u64 ranges, because we read the values as u64.
// The mapping from the conversion is monotonic so ordering is preserved.
@@ -551,17 +558,16 @@ mod tests {
get_test_index_with_num_docs,
};
pub fn get_collector_from_ranges(
ranges: Vec<RangeAggregationRange>,
pub fn build_test_buckets(
ranges: &[RangeAggregationRange],
field_type: ColumnType,
) -> SegmentRangeCollector<HighCardSubAggCache> {
) -> Vec<SegmentRangeAndBucketEntry> {
let req = RangeAggregation {
field: "dummy".to_string(),
ranges,
ranges: ranges.to_vec(),
..Default::default()
};
// Build buckets directly as in from_req_and_validate without AggregationsData
let buckets: Vec<_> = extend_validate_ranges(&req.ranges, &field_type)
extend_validate_ranges(&req.ranges, &field_type)
.expect("unexpected error in extend_validate_ranges")
.iter()
.map(|range| {
@@ -592,16 +598,7 @@ mod tests {
},
}
})
.collect();
SegmentRangeCollector {
parent_buckets: vec![buckets],
column_type: field_type,
accessor_idx: 0,
sub_agg: None,
bucket_id_provider: Default::default(),
limits: AggregationLimitsGuard::default(),
}
.collect()
}
#[test]
@@ -844,10 +841,10 @@ mod tests {
#[test]
fn bucket_test_extend_range_hole() {
let buckets = vec![(10f64..20f64).into(), (30f64..40f64).into()];
let collector = get_collector_from_ranges(buckets, ColumnType::F64);
let buckets = [(10f64..20f64).into(), (30f64..40f64).into()];
let parent_buckets = [build_test_buckets(&buckets, ColumnType::F64)];
let buckets = collector.parent_buckets[0].clone();
let buckets = parent_buckets[0].clone();
assert_eq!(buckets[0].range.start, u64::MIN);
assert_eq!(buckets[0].range.end, 10f64.to_u64());
assert_eq!(buckets[1].range.start, 10f64.to_u64());
@@ -863,14 +860,14 @@ mod tests {
fn bucket_test_range_conversion_special_case() {
// the monotonic conversion between f64 and u64, does not map f64::MIN.to_u64() ==
// u64::MIN, but the into trait converts f64::MIN/MAX to None
let buckets = vec![
let buckets = [
(f64::MIN..10f64).into(),
(10f64..20f64).into(),
(20f64..f64::MAX).into(),
];
let collector = get_collector_from_ranges(buckets, ColumnType::F64);
let parent_buckets = [build_test_buckets(&buckets, ColumnType::F64)];
let buckets = collector.parent_buckets[0].clone();
let buckets = parent_buckets[0].clone();
assert_eq!(buckets[0].range.start, u64::MIN);
assert_eq!(buckets[0].range.end, 10f64.to_u64());
assert_eq!(buckets[1].range.start, 10f64.to_u64());
@@ -882,28 +879,28 @@ mod tests {
#[test]
fn bucket_range_test_negative_vals() {
let buckets = vec![(-10f64..-1f64).into()];
let collector = get_collector_from_ranges(buckets, ColumnType::F64);
let buckets = [(-10f64..-1f64).into()];
let parent_buckets = [build_test_buckets(&buckets, ColumnType::F64)];
let buckets = collector.parent_buckets[0].clone();
let buckets = parent_buckets[0].clone();
assert_eq!(&buckets[0].bucket.key.to_string(), "*--10");
assert_eq!(&buckets[buckets.len() - 1].bucket.key.to_string(), "-1-*");
}
#[test]
fn bucket_range_test_positive_vals() {
let buckets = vec![(0f64..10f64).into()];
let collector = get_collector_from_ranges(buckets, ColumnType::F64);
let buckets = [(0f64..10f64).into()];
let parent_buckets = [build_test_buckets(&buckets, ColumnType::F64)];
let buckets = collector.parent_buckets[0].clone();
let buckets = parent_buckets[0].clone();
assert_eq!(&buckets[0].bucket.key.to_string(), "*-0");
assert_eq!(&buckets[buckets.len() - 1].bucket.key.to_string(), "10-*");
}
#[test]
fn range_binary_search_test_u64() {
let check_ranges = |ranges: Vec<RangeAggregationRange>| {
let collector = get_collector_from_ranges(ranges, ColumnType::U64);
let search = |val: u64| get_bucket_pos(val, &collector.parent_buckets[0]);
let check_ranges = |ranges: &[RangeAggregationRange]| {
let parent_buckets = [build_test_buckets(ranges, ColumnType::U64)];
let search = |val: u64| get_bucket_pos(val, &parent_buckets[0]);
assert_eq!(search(u64::MIN), 0);
assert_eq!(search(9), 0);
@@ -916,7 +913,7 @@ mod tests {
};
let ranges = vec![(10.0..100.0).into()];
check_ranges(ranges);
check_ranges(&ranges);
let ranges = vec![
RangeAggregationRange {
@@ -926,7 +923,7 @@ mod tests {
},
(10.0..100.0).into(),
];
check_ranges(ranges);
check_ranges(&ranges);
let ranges = vec![
RangeAggregationRange {
@@ -941,15 +938,15 @@ mod tests {
from: Some(100.0),
},
];
check_ranges(ranges);
check_ranges(&ranges);
}
#[test]
fn range_binary_search_test_f64() {
let ranges = vec![(10.0..100.0).into()];
let ranges = [(10.0..100.0).into()];
let collector = get_collector_from_ranges(ranges, ColumnType::F64);
let search = |val: u64| get_bucket_pos(val, &collector.parent_buckets[0]);
let parent_buckets = [build_test_buckets(&ranges, ColumnType::F64)];
let search = |val: u64| get_bucket_pos(val, &parent_buckets[0]);
assert_eq!(search(u64::MIN), 0);
assert_eq!(search(9f64.to_u64()), 0);

600
src/aggregation/bucket/term_agg.rs → src/aggregation/bucket/term_agg/mod.rs
View File

@@ -1,5 +1,4 @@
use std::fmt::Debug;
use std::io;
use std::net::Ipv6Addr;
use columnar::column_values::CompactSpaceU64Accessor;
@@ -17,8 +16,9 @@ use crate::aggregation::agg_data::{
};
use crate::aggregation::agg_limits::MemoryConsumption;
use crate::aggregation::agg_req::Aggregations;
use crate::aggregation::cached_sub_aggs::{
CachedSubAggs, HighCardSubAggCache, LowCardCachedSubAggs, LowCardSubAggCache, SubAggCache,
use crate::aggregation::buffered_sub_aggs::{
BufferedSubAggs, HighCardSubAggBuffer, LowCardBufferedSubAggs, LowCardSubAggBuffer,
SubAggBuffer,
};
use crate::aggregation::intermediate_agg_result::{
IntermediateAggregationResult, IntermediateAggregationResults, IntermediateBucketResult,
@@ -29,6 +29,8 @@ use crate::aggregation::{format_date, BucketId, Key};
use crate::error::DataCorruption;
use crate::TantivyError;
mod term_histogram;
/// Contains all information required by the SegmentTermCollector to perform the
/// terms aggregation on a segment.
#[derive(Debug, Clone)]
@@ -352,19 +354,15 @@ pub(crate) fn build_segment_term_collector(
)));
}
// Validate sub aggregation exists when ordering by sub-aggregation.
{
if let OrderTarget::SubAggregation(sub_agg_name) = &terms_req_data.req.order.target {
let (agg_name, _agg_property) = get_agg_name_and_property(sub_agg_name);
node.get_sub_agg(agg_name, &req_data.per_request)
.ok_or_else(|| {
TantivyError::InvalidArgument(format!(
"could not find aggregation with name {agg_name} in metric \
sub_aggregations"
))
})?;
}
// Validate that the referenced sub-aggregation exists when ordering by one.
if let OrderTarget::SubAggregation(sub_agg_name) = &terms_req_data.req.order.target {
let (agg_name, _agg_property) = get_agg_name_and_property(sub_agg_name);
node.get_sub_agg(agg_name, &req_data.per_request)
.ok_or_else(|| {
TantivyError::InvalidArgument(format!(
"could not find aggregation with name {agg_name} in metric sub_aggregations"
))
})?;
}
// Build sub-aggregation blueprint if there are children.
@@ -381,6 +379,18 @@ pub(crate) fn build_segment_term_collector(
let max_term_id: u64 =
col_max_value.max(terms_req_data.missing_value_for_accessor.unwrap_or(0u64));
// Fused fast path: low-cardinality terms × a single `histogram`/`date_histogram` leaf over full
// columns with a small enough bucket grid. Anything else falls through to the general path.
if let Some(collector) = term_histogram::maybe_build_collector(
req_data,
node,
&terms_req_data,
max_term_id,
is_top_level,
)? {
return Ok(collector);
}
let sub_agg_collector = if has_sub_aggregations {
Some(build_segment_agg_collectors(req_data, &node.children)?)
} else {
@@ -391,7 +401,7 @@ pub(crate) fn build_segment_term_collector(
// Decide which bucket storage is best suited for this aggregation.
if is_top_level && max_term_id < MAX_NUM_TERMS_FOR_VEC && !has_sub_aggregations {
let term_buckets = VecTermBucketsNoAgg::new(max_term_id + 1, &mut bucket_id_provider);
let collector: SegmentTermCollector<_, HighCardSubAggCache> = SegmentTermCollector {
let collector: SegmentTermCollector<_, HighCardSubAggBuffer> = SegmentTermCollector {
parent_buckets: vec![term_buckets],
sub_agg: None,
bucket_id_provider,
@@ -401,8 +411,8 @@ pub(crate) fn build_segment_term_collector(
Ok(Box::new(collector))
} else if is_top_level && max_term_id < MAX_NUM_TERMS_FOR_VEC {
let term_buckets = VecTermBuckets::new(max_term_id + 1, &mut bucket_id_provider);
let sub_agg = sub_agg_collector.map(LowCardCachedSubAggs::new);
let collector: SegmentTermCollector<_, LowCardSubAggCache> = SegmentTermCollector {
let sub_agg = sub_agg_collector.map(LowCardBufferedSubAggs::new);
let collector: SegmentTermCollector<_, LowCardSubAggBuffer> = SegmentTermCollector {
parent_buckets: vec![term_buckets],
sub_agg,
bucket_id_provider,
@@ -414,8 +424,8 @@ pub(crate) fn build_segment_term_collector(
let term_buckets: PagedTermMap =
PagedTermMap::new(max_term_id + 1, &mut bucket_id_provider);
// Build sub-aggregation blueprint (flat pairs)
let sub_agg = sub_agg_collector.map(CachedSubAggs::new);
let collector: SegmentTermCollector<PagedTermMap, HighCardSubAggCache> =
let sub_agg = sub_agg_collector.map(BufferedSubAggs::new);
let collector: SegmentTermCollector<PagedTermMap, HighCardSubAggBuffer> =
SegmentTermCollector {
parent_buckets: vec![term_buckets],
sub_agg,
@@ -427,8 +437,8 @@ pub(crate) fn build_segment_term_collector(
} else {
let term_buckets: HashMapTermBuckets = HashMapTermBuckets::default();
// Build sub-aggregation blueprint (flat pairs)
let sub_agg = sub_agg_collector.map(CachedSubAggs::new);
let collector: SegmentTermCollector<HashMapTermBuckets, HighCardSubAggCache> =
let sub_agg = sub_agg_collector.map(BufferedSubAggs::new);
let collector: SegmentTermCollector<HashMapTermBuckets, HighCardSubAggBuffer> =
SegmentTermCollector {
parent_buckets: vec![term_buckets],
sub_agg,
@@ -758,10 +768,10 @@ impl TermAggregationMap for VecTermBuckets {
/// The collector puts values from the fast field into the correct buckets and does a conversion to
/// the correct datatype.
#[derive(Debug)]
struct SegmentTermCollector<TermMap: TermAggregationMap, C: SubAggCache> {
struct SegmentTermCollector<TermMap: TermAggregationMap, B: SubAggBuffer> {
/// The buckets containing the aggregation data.
parent_buckets: Vec<TermMap>,
sub_agg: Option<CachedSubAggs<C>>,
sub_agg: Option<BufferedSubAggs<B>>,
bucket_id_provider: BucketIdProvider,
max_term_id: u64,
terms_req_data: TermsAggReqData,
@@ -772,8 +782,8 @@ pub(crate) fn get_agg_name_and_property(name: &str) -> (&str, &str) {
(agg_name, agg_property)
}
impl<TermMap: TermAggregationMap, C: SubAggCache> SegmentAggregationCollector
for SegmentTermCollector<TermMap, C>
impl<TermMap: TermAggregationMap, B: SubAggBuffer> SegmentAggregationCollector
for SegmentTermCollector<TermMap, B>
{
fn add_intermediate_aggregation_result(
&mut self,
@@ -790,8 +800,14 @@ impl<TermMap: TermAggregationMap, C: SubAggCache> SegmentAggregationCollector
let term_req = &self.terms_req_data;
let name = term_req.name.clone();
let bucket =
Self::into_intermediate_bucket_result(term_req, &mut self.sub_agg, bucket, agg_data)?;
let bucket = Self::into_intermediate_bucket_result(
term_req,
self.sub_agg
.as_mut()
.map(BufferedSubAggs::get_sub_agg_collector),
bucket,
agg_data,
)?;
results.push(name, IntermediateAggregationResult::Bucket(bucket))?;
Ok(())
}
@@ -881,6 +897,17 @@ impl<TermMap: TermAggregationMap, C: SubAggCache> SegmentAggregationCollector
}
Ok(())
}
fn compute_metric_value(
&self,
_bucket_id: BucketId,
_sub_agg_name: &str,
_sub_agg_property: &str,
_agg_data: &AggregationsSegmentCtx,
) -> Option<f64> {
// Terms is a multi-bucket agg with no single value to extract.
None
}
}
/// Missing value are represented as a sentinel value in the column.
@@ -907,10 +934,38 @@ fn extract_missing_value<T>(
Some((key, bucket))
}
impl<TermMap, C> SegmentTermCollector<TermMap, C>
fn reborrow_opt_collector<'a>(
opt: &'a mut Option<&mut dyn SegmentAggregationCollector>,
) -> Option<&'a mut dyn SegmentAggregationCollector> {
match opt {
Some(inner) => Some(*inner),
None => None,
}
}
fn into_intermediate_bucket_entry(
bucket: Bucket,
sub_agg_collector: Option<&mut dyn SegmentAggregationCollector>,
agg_data: &AggregationsSegmentCtx,
) -> crate::Result<IntermediateTermBucketEntry> {
let mut sub_aggregation_res = IntermediateAggregationResults::default();
if let Some(sub_agg_collector) = sub_agg_collector {
sub_agg_collector.add_intermediate_aggregation_result(
agg_data,
&mut sub_aggregation_res,
bucket.bucket_id,
)?;
}
Ok(IntermediateTermBucketEntry {
doc_count: bucket.count,
sub_aggregation: sub_aggregation_res,
})
}
impl<TermMap, B> SegmentTermCollector<TermMap, B>
where
TermMap: TermAggregationMap,
C: SubAggCache,
B: SubAggBuffer,
{
#[inline]
fn get_memory_consumption(&self, parent_bucket_id: BucketId) -> usize {
@@ -920,15 +975,12 @@ where
#[inline]
pub(crate) fn into_intermediate_bucket_result(
term_req: &TermsAggReqData,
sub_agg: &mut Option<CachedSubAggs<C>>,
mut sub_agg_collector: Option<&mut dyn SegmentAggregationCollector>,
term_buckets: TermMap,
agg_data: &AggregationsSegmentCtx,
) -> crate::Result<IntermediateBucketResult> {
let mut entries: Vec<(u64, Bucket)> = term_buckets.into_vec();
let order_by_sub_aggregation =
matches!(term_req.req.order.target, OrderTarget::SubAggregation(_));
match &term_req.req.order.target {
OrderTarget::Key => {
// We rely on the fact, that term ordinals match the order of the strings
@@ -940,10 +992,37 @@ where
entries.sort_unstable_by_key(|bucket| bucket.0);
}
}
OrderTarget::SubAggregation(_name) => {
// don't sort and cut off since it's hard to make assumptions on the quality of the
// results when cutting off du to unknown nature of the sub_aggregation (possible
// to check).
OrderTarget::SubAggregation(sub_agg_path) => {
// Peek segment-level metric values, sort, then fall through to
// `cut_off_buckets`. Like Elasticsearch, we always cut off when ordering
// by a sub-agg: top-K results are approximate and may differ from the
// global ordering, especially for non-monotonic metrics like avg/min.
let coll = sub_agg_collector.as_deref().ok_or_else(|| {
TantivyError::InvalidArgument(format!(
"Could not find sub-aggregation collector for path {sub_agg_path}"
))
})?;
let (agg_name, agg_prop) = get_agg_name_and_property(sub_agg_path);
// Fetch values up-front; otherwise sort would re-compute per comparison
let mut keyed: Vec<(f64, (u64, Bucket))> = entries
.into_iter()
.map(|bucket| {
let metric_value = coll
.compute_metric_value(bucket.1.bucket_id, agg_name, agg_prop, agg_data)
.unwrap_or(0.0);
(metric_value, bucket)
})
.collect();
if term_req.req.order.order == Order::Desc {
keyed.sort_unstable_by(|a, b| {
b.0.partial_cmp(&a.0).unwrap_or(std::cmp::Ordering::Equal)
});
} else {
keyed.sort_unstable_by(|a, b| {
a.0.partial_cmp(&b.0).unwrap_or(std::cmp::Ordering::Equal)
});
}
entries = keyed.into_iter().map(|(_, e)| e).collect();
}
OrderTarget::Count => {
if term_req.req.order.order == Order::Desc {
@@ -954,40 +1033,12 @@ where
}
}
let (term_doc_count_before_cutoff, sum_other_doc_count) = if order_by_sub_aggregation {
(0, 0)
} else {
cut_off_buckets(&mut entries, term_req.req.segment_size as usize)
};
let (term_doc_count_before_cutoff, sum_other_doc_count) =
cut_off_buckets(&mut entries, term_req.req.segment_size as usize);
let mut dict: FxHashMap<IntermediateKey, IntermediateTermBucketEntry> = Default::default();
dict.reserve(entries.len());
let into_intermediate_bucket_entry =
|bucket: Bucket,
sub_agg: &mut Option<CachedSubAggs<C>>|
-> crate::Result<IntermediateTermBucketEntry> {
if let Some(sub_agg) = sub_agg {
let mut sub_aggregation_res = IntermediateAggregationResults::default();
sub_agg
.get_sub_agg_collector()
.add_intermediate_aggregation_result(
agg_data,
&mut sub_aggregation_res,
bucket.bucket_id,
)?;
Ok(IntermediateTermBucketEntry {
doc_count: bucket.count,
sub_aggregation: sub_aggregation_res,
})
} else {
Ok(IntermediateTermBucketEntry {
doc_count: bucket.count,
sub_aggregation: Default::default(),
})
}
};
if term_req.column_type == ColumnType::Str {
let fallback_dict = Dictionary::empty();
let term_dict = term_req
@@ -998,7 +1049,11 @@ where
if let Some((intermediate_key, bucket)) = extract_missing_value(&mut entries, term_req)
{
let intermediate_entry = into_intermediate_bucket_entry(bucket, sub_agg)?;
let intermediate_entry = into_intermediate_bucket_entry(
bucket,
reborrow_opt_collector(&mut sub_agg_collector),
agg_data,
)?;
dict.insert(intermediate_key, intermediate_entry);
}
@@ -1006,19 +1061,28 @@ where
entries.sort_unstable_by_key(|bucket| bucket.0);
let (term_ids, buckets): (Vec<u64>, Vec<Bucket>) = entries.into_iter().unzip();
let mut buckets_it = buckets.into_iter();
term_dict.sorted_ords_to_term_cb(term_ids.into_iter(), |term| {
let bucket = buckets_it.next().unwrap();
let intermediate_entry =
into_intermediate_bucket_entry(bucket, sub_agg).map_err(io::Error::other)?;
let intermediate_entries: Vec<IntermediateTermBucketEntry> = buckets
.into_iter()
.map(|bucket| {
into_intermediate_bucket_entry(
bucket,
reborrow_opt_collector(&mut sub_agg_collector),
agg_data,
)
})
.collect::<crate::Result<_>>()?;
let mut intermediate_entry_it = intermediate_entries.into_iter();
term_dict.sorted_ords_to_term_cb(&term_ids[..], |term| {
let intermediate_entry = intermediate_entry_it.next().unwrap();
dict.insert(
IntermediateKey::Str(
String::from_utf8(term.to_vec()).expect("could not convert to String"),
),
intermediate_entry,
);
Ok(())
})?;
if term_req.req.min_doc_count == 0 {
@@ -1053,14 +1117,22 @@ where
}
} else if term_req.column_type == ColumnType::DateTime {
for (val, doc_count) in entries {
let intermediate_entry = into_intermediate_bucket_entry(doc_count, sub_agg)?;
let intermediate_entry = into_intermediate_bucket_entry(
doc_count,
reborrow_opt_collector(&mut sub_agg_collector),
agg_data,
)?;
let val = i64::from_u64(val);
let date = format_date(val)?;
dict.insert(IntermediateKey::Str(date), intermediate_entry);
}
} else if term_req.column_type == ColumnType::Bool {
for (val, doc_count) in entries {
let intermediate_entry = into_intermediate_bucket_entry(doc_count, sub_agg)?;
let intermediate_entry = into_intermediate_bucket_entry(
doc_count,
reborrow_opt_collector(&mut sub_agg_collector),
agg_data,
)?;
let val = bool::from_u64(val);
dict.insert(IntermediateKey::Bool(val), intermediate_entry);
}
@@ -1080,14 +1152,22 @@ where
})?;
for (val, doc_count) in entries {
let intermediate_entry = into_intermediate_bucket_entry(doc_count, sub_agg)?;
let intermediate_entry = into_intermediate_bucket_entry(
doc_count,
reborrow_opt_collector(&mut sub_agg_collector),
agg_data,
)?;
let val: u128 = compact_space_accessor.compact_to_u128(val as u32);
let val = Ipv6Addr::from_u128(val);
dict.insert(IntermediateKey::IpAddr(val), intermediate_entry);
}
} else {
for (val, doc_count) in entries {
let intermediate_entry = into_intermediate_bucket_entry(doc_count, sub_agg)?;
let intermediate_entry = into_intermediate_bucket_entry(
doc_count,
reborrow_opt_collector(&mut sub_agg_collector),
agg_data,
)?;
if term_req.column_type == ColumnType::U64 {
dict.insert(IntermediateKey::U64(val), intermediate_entry);
} else if term_req.column_type == ColumnType::I64 {
@@ -1121,13 +1201,13 @@ where
}
}
impl<TermMap: TermAggregationMap, C: SubAggCache> SegmentTermCollector<TermMap, C> {
impl<TermMap: TermAggregationMap, B: SubAggBuffer> SegmentTermCollector<TermMap, B> {
#[inline]
fn collect_terms_with_docs(
iter: impl Iterator<Item = (crate::DocId, u64)>,
term_buckets: &mut TermMap,
bucket_id_provider: &mut BucketIdProvider,
sub_agg: &mut CachedSubAggs<C>,
sub_agg: &mut BufferedSubAggs<B>,
) {
for (doc, term_id) in iter {
let bucket_id = term_buckets.term_entry(term_id, bucket_id_provider);
@@ -1200,7 +1280,7 @@ mod tests {
use crate::aggregation::{AggregationLimitsGuard, DistributedAggregationCollector};
use crate::indexer::NoMergePolicy;
use crate::query::AllQuery;
use crate::schema::{IntoIpv6Addr, Schema, FAST, STRING};
use crate::schema::{IntoIpv6Addr, Schema, FAST, INDEXED, STRING, TEXT};
use crate::{Index, IndexWriter};
#[test]
@@ -1729,6 +1809,263 @@ mod tests {
Ok(())
}
#[test]
fn terms_aggregation_order_by_cardinality_desc_single_segment() -> crate::Result<()> {
terms_aggregation_order_by_cardinality_desc(true)
}
#[test]
fn terms_aggregation_order_by_cardinality_desc_multi_segment() -> crate::Result<()> {
terms_aggregation_order_by_cardinality_desc(false)
}
fn terms_aggregation_order_by_cardinality_desc(merge_segments: bool) -> crate::Result<()> {
// Distinct score values per bucket key: A→5, B→1, C→3.
// Order by cardinality desc must yield A, C, B.
let segment_and_terms = vec![vec![
(1.0, "A".to_string()),
(2.0, "A".to_string()),
(3.0, "A".to_string()),
(4.0, "A".to_string()),
(5.0, "A".to_string()),
(1.0, "B".to_string()),
(1.0, "B".to_string()),
(1.0, "B".to_string()),
(1.0, "C".to_string()),
(2.0, "C".to_string()),
(3.0, "C".to_string()),
]];
let index = get_test_index_from_values_and_terms(merge_segments, &segment_and_terms)?;
let agg_req: Aggregations = serde_json::from_value(json!({
"my_texts": {
"terms": {
"field": "string_id",
"order": { "card": "desc" }
},
"aggs": {
"card": { "cardinality": { "field": "score" } }
}
}
}))
.unwrap();
let res = exec_request(agg_req, &index)?;
assert_eq!(res["my_texts"]["buckets"][0]["key"], "A");
assert_eq!(res["my_texts"]["buckets"][0]["card"]["value"], 5.0);
assert_eq!(res["my_texts"]["buckets"][1]["key"], "C");
assert_eq!(res["my_texts"]["buckets"][1]["card"]["value"], 3.0);
assert_eq!(res["my_texts"]["buckets"][2]["key"], "B");
assert_eq!(res["my_texts"]["buckets"][2]["card"]["value"], 1.0);
// Asc engages the segment-cutoff path too (monotonic-safe: discarded buckets had
// local card >= cutoff, so merged card >= cutoff and they cannot be globally smallest).
let agg_req: Aggregations = serde_json::from_value(json!({
"my_texts": {
"terms": {
"field": "string_id",
"order": { "card": "asc" }
},
"aggs": {
"card": { "cardinality": { "field": "score" } }
}
}
}))
.unwrap();
let res = exec_request(agg_req, &index)?;
assert_eq!(res["my_texts"]["buckets"][0]["key"], "B");
assert_eq!(res["my_texts"]["buckets"][1]["key"], "C");
assert_eq!(res["my_texts"]["buckets"][2]["key"], "A");
// size=2 with desc engages the segment cutoff: must keep top-2 by cardinality (A, C),
// and `sum_other_doc_count` reflects the dropped B (3 docs).
let agg_req: Aggregations = serde_json::from_value(json!({
"my_texts": {
"terms": {
"field": "string_id",
"size": 2,
"order": { "card": "desc" }
},
"aggs": {
"card": { "cardinality": { "field": "score" } }
}
}
}))
.unwrap();
let res = exec_request(agg_req, &index)?;
assert_eq!(res["my_texts"]["buckets"][0]["key"], "A");
assert_eq!(res["my_texts"]["buckets"][1]["key"], "C");
assert_eq!(res["my_texts"]["buckets"].as_array().unwrap().len(), 2);
// size=2 with asc engages the segment cutoff: must keep bottom-2 by cardinality (B, C).
let agg_req: Aggregations = serde_json::from_value(json!({
"my_texts": {
"terms": {
"field": "string_id",
"size": 2,
"order": { "card": "asc" }
},
"aggs": {
"card": { "cardinality": { "field": "score" } }
}
}
}))
.unwrap();
let res = exec_request(agg_req, &index)?;
assert_eq!(res["my_texts"]["buckets"][0]["key"], "B");
assert_eq!(res["my_texts"]["buckets"][1]["key"], "C");
assert_eq!(res["my_texts"]["buckets"].as_array().unwrap().len(), 2);
Ok(())
}
#[test]
fn terms_aggregation_order_by_sum_single_segment() -> crate::Result<()> {
terms_aggregation_order_by_sum(true)
}
#[test]
fn terms_aggregation_order_by_sum_multi_segment() -> crate::Result<()> {
terms_aggregation_order_by_sum(false)
}
fn terms_aggregation_order_by_sum(merge_segments: bool) -> crate::Result<()> {
// Per-bucket sums on the U64 `score` column (non-negative => sum is monotonic):
// A → 1+2+3+4+5 = 15, B → 1+1+1 = 3, C → 1+2+3 = 6.
let segment_and_terms = vec![
vec![
(1.0, "A".to_string()),
(2.0, "A".to_string()),
(3.0, "A".to_string()),
(1.0, "B".to_string()),
(1.0, "C".to_string()),
],
vec![
(4.0, "A".to_string()),
(5.0, "A".to_string()),
(1.0, "B".to_string()),
(1.0, "B".to_string()),
(2.0, "C".to_string()),
(3.0, "C".to_string()),
],
];
let index = get_test_index_from_values_and_terms(merge_segments, &segment_and_terms)?;
// Desc on a Sum metric engages the fast path (column is U64).
let agg_req: Aggregations = serde_json::from_value(json!({
"my_texts": {
"terms": {
"field": "string_id",
"order": { "total": "desc" }
},
"aggs": {
"total": { "sum": { "field": "score" } }
}
}
}))
.unwrap();
let res = exec_request(agg_req, &index)?;
assert_eq!(res["my_texts"]["buckets"][0]["key"], "A");
assert_eq!(res["my_texts"]["buckets"][0]["total"]["value"], 15.0);
assert_eq!(res["my_texts"]["buckets"][1]["key"], "C");
assert_eq!(res["my_texts"]["buckets"][1]["total"]["value"], 6.0);
assert_eq!(res["my_texts"]["buckets"][2]["key"], "B");
assert_eq!(res["my_texts"]["buckets"][2]["total"]["value"], 3.0);
// Asc engages the fast path too — discarded buckets had local sum >= cutoff,
// and merged sum >= local (non-negative addends), so they cannot be globally smallest.
let agg_req: Aggregations = serde_json::from_value(json!({
"my_texts": {
"terms": {
"field": "string_id",
"order": { "total": "asc" }
},
"aggs": {
"total": { "sum": { "field": "score" } }
}
}
}))
.unwrap();
let res = exec_request(agg_req, &index)?;
assert_eq!(res["my_texts"]["buckets"][0]["key"], "B");
assert_eq!(res["my_texts"]["buckets"][1]["key"], "C");
assert_eq!(res["my_texts"]["buckets"][2]["key"], "A");
// size=2 desc with cutoff: top-2 by sum (A, C).
let agg_req: Aggregations = serde_json::from_value(json!({
"my_texts": {
"terms": {
"field": "string_id",
"size": 2,
"order": { "total": "desc" }
},
"aggs": {
"total": { "sum": { "field": "score" } }
}
}
}))
.unwrap();
let res = exec_request(agg_req, &index)?;
assert_eq!(res["my_texts"]["buckets"][0]["key"], "A");
assert_eq!(res["my_texts"]["buckets"][1]["key"], "C");
assert_eq!(res["my_texts"]["buckets"].as_array().unwrap().len(), 2);
// Stats sub-property: ordering by `mystats.sum` on a U64 column also engages.
let agg_req: Aggregations = serde_json::from_value(json!({
"my_texts": {
"terms": {
"field": "string_id",
"order": { "mystats.sum": "desc" }
},
"aggs": {
"mystats": { "stats": { "field": "score" } }
}
}
}))
.unwrap();
let res = exec_request(agg_req, &index)?;
assert_eq!(res["my_texts"]["buckets"][0]["key"], "A");
assert_eq!(res["my_texts"]["buckets"][1]["key"], "C");
assert_eq!(res["my_texts"]["buckets"][2]["key"], "B");
// Sum on a signed column (I64) takes the same cutoff path. Results may be
// approximate near the boundary on adversarial data, but for this dataset the
// top-K is unambiguous.
let agg_req: Aggregations = serde_json::from_value(json!({
"my_texts": {
"terms": {
"field": "string_id",
"order": { "total": "desc" }
},
"aggs": {
"total": { "sum": { "field": "score_i64" } }
}
}
}))
.unwrap();
let res = exec_request(agg_req, &index)?;
assert_eq!(res["my_texts"]["buckets"][0]["key"], "A");
assert_eq!(res["my_texts"]["buckets"][1]["key"], "C");
assert_eq!(res["my_texts"]["buckets"][2]["key"], "B");
// Order by extended_stats sub-property exercises compute_metric_value on the
// ExtendedStats collector. A→max=5, B→max=1, C→max=3, so desc by max → A, C, B.
let agg_req: Aggregations = serde_json::from_value(json!({
"my_texts": {
"terms": {
"field": "string_id",
"order": { "ext.max": "desc" }
},
"aggs": {
"ext": { "extended_stats": { "field": "score" } }
}
}
}))
.unwrap();
let res = exec_request(agg_req, &index)?;
assert_eq!(res["my_texts"]["buckets"][0]["key"], "A");
assert_eq!(res["my_texts"]["buckets"][1]["key"], "C");
assert_eq!(res["my_texts"]["buckets"][2]["key"], "B");
Ok(())
}
#[test]
fn terms_aggregation_test_order_key_single_segment() -> crate::Result<()> {
terms_aggregation_test_order_key_merge_segment(true)
@@ -2894,4 +3231,101 @@ mod tests {
Ok(())
}
fn prep_index_with_n_unique_terms_plus_one_null(n: u64) -> crate::Result<Index> {
let mut schema_builder = Schema::builder();
let id_field = schema_builder.add_u64_field("id", INDEXED);
let title_field = schema_builder.add_text_field("title", TEXT | FAST);
let schema = schema_builder.build();
let index = Index::create_in_ram(schema.clone());
// set to one thread to guarantee all docs end up in the same segment
let mut writer = index.writer_with_num_threads(1, 50_000_000)?;
writer.add_document(doc!(
id_field => 0u64,
))?;
for i in 1u64..=n {
let title = format!("foo{i}");
writer.add_document(doc!(
id_field => i,
title_field => title,
))?;
}
writer.commit()?;
Ok(index)
}
#[test]
fn null_bitset_bounds_check_regression() -> crate::Result<()> {
// include cases
for i in 0..=4 {
let index = prep_index_with_n_unique_terms_plus_one_null(i * 64)?;
let normal_req: Aggregations = serde_json::from_value(json!({
"my_bool": {
"terms": {
"field": "title",
"missing": "__NULL__",
"size": 1000,
}
}
}))?;
let include_req: Aggregations = serde_json::from_value(json!({
"my_bool": {
"terms": {
"field": "title",
"include": "foo(.*)",
"missing": "__NULL__",
"size": 1000,
}
}
}))?;
let exclude_req: Aggregations = serde_json::from_value(json!({
"my_bool": {
"terms": {
"field": "title",
"exclude": "foo(.*)",
"missing": "__NULL__",
"size": 1000,
}
}
}))?;
let normal_res = exec_request(normal_req, &index)?;
let normal_buckets = normal_res["my_bool"]["buckets"].as_array().unwrap();
assert_eq!(
normal_buckets.len(),
(i * 64) as usize + 1,
"The normal request should return all 'foo' buckets, plus the missing term bucket",
);
let include_res = exec_request(include_req, &index)?;
eprintln!("include_res: {include_res:?}");
let include_buckets = include_res["my_bool"]["buckets"].as_array().unwrap();
assert_eq!(
include_buckets.len(),
(i * 64) as usize,
"The include request should return all 'foo' buckets, and not the missing term \
bucket",
);
assert!(include_buckets
.iter()
.all(|b| b["key"].as_str().unwrap().starts_with("foo")));
let exclude_res = exec_request(exclude_req, &index)?;
let exclude_buckets = exclude_res["my_bool"]["buckets"].as_array().unwrap();
if i != 0 {
// TODO: Remove this if after fixing exclude + missing bug
assert_eq!(
exclude_buckets.len(),
1,
"The exclude request should exclude all 'foo' buckets, and only the missing \
term bucket",
);
assert_eq!(exclude_buckets[0]["key"], "__NULL__");
}
}
Ok(())
}
}

479
src/aggregation/bucket/term_agg/term_histogram.rs Normal file
View File

@@ -0,0 +1,479 @@
//! Fused collector for the very common shape `terms` (low cardinality) × a single
//! `histogram`/`date_histogram` sub-aggregation with nothing nested below it.
//!
//! See [`SegmentTermHistogramCollector`] for the approach and [`maybe_build_collector`] for the
//! conditions under which it is used.
use columnar::ColumnBlockAccessor;
use super::{Bucket, SegmentTermCollector, TermsAggReqData, VecTermBuckets};
use crate::aggregation::agg_data::{AggKind, AggRefNode, AggregationsSegmentCtx};
use crate::aggregation::bucket::{
get_bucket_pos_f64, prepare_histogram_dense_range, HistogramAggReqData,
SegmentHistogramCollector,
};
use crate::aggregation::buffered_sub_aggs::LowCardSubAggBuffer;
use crate::aggregation::intermediate_agg_result::{
IntermediateAggregationResult, IntermediateAggregationResults,
};
use crate::aggregation::segment_agg_result::{BucketIdProvider, SegmentAggregationCollector};
use crate::aggregation::{f64_from_fastfield_u64, BucketId};
/// Maximum number of cells (`num_terms × num_time_buckets`) in the fused flat 2D grid. Above this
/// the grid would be too large/cache-unfriendly, so we fall back to the general buffered path.
/// `1 << 14` cells = 128 KB of `u64` counters, comfortably L2-resident.
const MAX_FUSED_GRID_BUCKETS: usize = 1 << 14;
/// Fused collector for `terms` (low cardinality) × a single `histogram`/`date_histogram` leaf with
/// nothing nested below it, when the resulting `num_terms × num_time_buckets` grid is small (see
/// [`MAX_FUSED_GRID_BUCKETS`]).
///
/// It keeps a flat, fully dense 2D counter grid (`counts[term * num_time_buckets + bucket]`) and a
/// per-term total. A single pass reads both the term and histogram columns in document order and
/// bumps the counters directly — no doc-id buffering, no per-term scattered re-fetch, no dynamic
/// dispatch on flush, no per-bucket key/id storage during collection (keys are derived from the
/// index at the end).
///
/// At result time the flat grid is expanded back into the regular term map + histogram storage and
/// handed to the shared intermediate-result builders, so cross-segment merging is identical to the
/// general path.
#[derive(Debug)]
pub(crate) struct SegmentTermHistogramCollector {
/// `[num_terms]` total doc count per term bucket (independent of the histogram bounds).
/// `u32` is enough: a per-segment count can't exceed the segment's doc count (`DocId` is
/// `u32`); the fused path is only taken when `num_docs < u32::MAX` (see
/// `maybe_build_collector`).
term_counts: Vec<u32>,
/// Flat row-major `[num_terms * num_time_buckets]` histogram counters (`u32`, see
/// `term_counts`).
counts: Vec<u32>,
/// Histogram buckets per term (the dense time-range length).
num_time_buckets: usize,
/// `bucket_pos` mapped to time-bucket index 0.
base_pos: i64,
terms_req_data: TermsAggReqData,
/// The (cloned, normalized) histogram request: its column + interval/offset/bounds.
hist_req_data: HistogramAggReqData,
/// Private block accessors for both columns. We read them together, so each needs its own
/// (the shared `agg_data` scratch accessor only holds one block at a time). Owning them keeps
/// `collect` independent of `agg_data`.
term_block: ColumnBlockAccessor<u64>,
hist_block: ColumnBlockAccessor<u64>,
/// When true, `term_counts` is left untouched during `collect` and derived from the grid
/// row-sums at result time. Valid only when there are no hard bounds: then every doc lands in
/// the dense range, so the per-term total equals the sum over its histogram buckets, and we
/// can drop both the per-doc `term_counts` increment and the (always-true) bounds check.
///
/// This is a real hot-loop win (~17% on `terms_status_with_(date_)histogram`): with low term
/// cardinality, `term_counts[term_id] += 1` repeatedly hits the same few addresses, so the
/// read-modify-write serializes across iterations on store-to-load forwarding latency. Summing
/// the grid once at the end avoids that loop-carried dependency entirely.
derive_term_counts_from_histogram: bool,
}
impl SegmentAggregationCollector for SegmentTermHistogramCollector {
fn add_intermediate_aggregation_result(
&mut self,
agg_data: &AggregationsSegmentCtx,
results: &mut IntermediateAggregationResults,
parent_bucket_id: BucketId,
) -> crate::Result<()> {
debug_assert_eq!(
parent_bucket_id, 0,
"fused term-histogram collector is top-level only"
);
// Expand the flat grid back into the regular structures and reuse the shared builders, so
// ordering/cut-off/dict handling and cross-segment merging match the general path exactly.
if self.derive_term_counts_from_histogram {
// `collect` skipped the per-doc total; recover it as the grid row-sum (every doc landed
// in a bucket, since there are no hard bounds).
let nb = self.num_time_buckets;
let counts = &self.counts;
for (t, slot) in self.term_counts.iter_mut().enumerate() {
*slot = counts[t * nb..(t + 1) * nb].iter().sum();
}
}
let mut bucket_id_provider = BucketIdProvider::default();
let term_buckets = VecTermBuckets {
buckets: self
.term_counts
.iter()
.map(|&count| Bucket {
count,
bucket_id: bucket_id_provider.next_bucket_id(),
})
.collect(),
};
let mut histogram = SegmentHistogramCollector::<()>::from_dense_rows(
self.hist_req_data.clone(),
self.base_pos,
self.num_time_buckets,
&self.counts,
);
let name = self.terms_req_data.name.clone();
let bucket = SegmentTermCollector::<VecTermBuckets, LowCardSubAggBuffer>::into_intermediate_bucket_result(
&self.terms_req_data,
Some(&mut histogram as &mut dyn SegmentAggregationCollector),
term_buckets,
agg_data,
)?;
results.push(name, IntermediateAggregationResult::Bucket(bucket))?;
Ok(())
}
#[inline]
fn collect(
&mut self,
parent_bucket_id: BucketId,
docs: &[crate::DocId],
_agg_data: &mut AggregationsSegmentCtx,
) -> crate::Result<()> {
debug_assert_eq!(
parent_bucket_id, 0,
"fused term-histogram collector is top-level only"
);
// Fetch both columns into our own accessors (we read them together, so they can't share the
// single `agg_data` scratch accessor). The collector owns all its inputs, so `collect`
// doesn't touch `agg_data`.
self.term_block
.fetch_block(docs, &self.terms_req_data.accessor);
self.hist_block
.fetch_block(docs, &self.hist_req_data.accessor);
let field_type = self.hist_req_data.field_type;
let bounds = self.hist_req_data.bounds;
let interval = self.hist_req_data.req.interval;
let offset = self.hist_req_data.offset;
let base_pos = self.base_pos;
let num_time_buckets = self.num_time_buckets;
let term_counts = &mut self.term_counts;
let counts = &mut self.counts;
// Both columns are full (checked at construction), so values align with `docs` positionally
// and are read together in one pass.
if self.derive_term_counts_from_histogram {
// No hard bounds: every doc lands in a bucket, so we skip the per-doc `term_counts`
// increment (derived from the grid at flush) and the always-true bounds check.
for (term_id, hist_raw) in self.term_block.iter_vals().zip(self.hist_block.iter_vals())
{
let term_id = term_id as usize;
let val = f64_from_fastfield_u64(hist_raw, field_type);
let bucket = (get_bucket_pos_f64(val, interval, offset) as i64 - base_pos) as usize;
debug_assert!(
bucket < num_time_buckets,
"histogram bucket outside dense range"
);
counts[term_id * num_time_buckets + bucket] += 1;
}
} else {
for (term_id, hist_raw) in self.term_block.iter_vals().zip(self.hist_block.iter_vals())
{
let term_id = term_id as usize;
term_counts[term_id] += 1;
let val = f64_from_fastfield_u64(hist_raw, field_type);
if bounds.contains(val) {
let bucket =
(get_bucket_pos_f64(val, interval, offset) as i64 - base_pos) as usize;
debug_assert!(
bucket < num_time_buckets,
"histogram bucket outside dense range"
);
counts[term_id * num_time_buckets + bucket] += 1;
}
}
}
Ok(())
}
fn flush(&mut self, _agg_data: &mut AggregationsSegmentCtx) -> crate::Result<()> {
// Nothing is buffered: `collect` writes the flat grid directly.
Ok(())
}
fn prepare_max_bucket(
&mut self,
_max_bucket: BucketId,
_agg_data: &AggregationsSegmentCtx,
) -> crate::Result<()> {
// Top-level: the flat grid is allocated up front.
Ok(())
}
fn compute_metric_value(
&self,
_bucket_id: BucketId,
_sub_agg_name: &str,
_sub_agg_property: &str,
_agg_data: &AggregationsSegmentCtx,
) -> Option<f64> {
None
}
}
/// Builds the fused terms×histogram collector for a single top-level parent, when the shape is
/// eligible. Returns `Ok(None)` to fall back to the general buffered terms path.
///
/// Eligibility: top-level, low-cardinality terms over a full column with no missing/include-exclude
/// handling; a single `histogram`/`date_histogram` leaf (no nesting below it) over a full column;
/// and a `num_terms × num_time_buckets` grid no larger than [`MAX_FUSED_GRID_BUCKETS`].
pub(super) fn maybe_build_collector(
agg_data: &mut AggregationsSegmentCtx,
node: &AggRefNode,
terms_req_data: &TermsAggReqData,
max_term_id: u64,
is_top_level: bool,
) -> crate::Result<Option<Box<dyn SegmentAggregationCollector>>> {
// Both columns must be full (one value per doc) so their values align positionally with `docs`
// and we can zip them. Requiring full columns also makes the terms agg's `missing` config a
// no-op (`fetch_block_with_missing` early-returns on full columns), so we needn't check for it.
//
// We don't cap the term cardinality here: the flat grid is bounded by the total cell count
// (`num_terms * num_time_buckets <= MAX_FUSED_GRID_BUCKETS`) checked below, which subsumes it.
let fuseable = is_top_level
&& terms_req_data.allowed_term_ids.is_none()
&& terms_req_data.accessor.get_cardinality().is_full()
// The flat counters are `u32`; a per-segment count can't exceed the doc count, so this
// guarantees no overflow (essentially always true, as `DocId` is `u32`).
&& terms_req_data.accessor.num_docs() < u32::MAX
&& node.children.len() == 1
&& matches!(
node.children[0].kind,
AggKind::Histogram | AggKind::DateHistogram
)
&& node.children[0].children.is_empty()
&& agg_data.per_request.histogram_req_data[node.children[0].idx_in_req_data]
.accessor
.get_cardinality()
.is_full();
if !fuseable {
return Ok(None);
}
// Clone + normalize the histogram request and get its dense bucket range; only take the fused
// path when the flat `num_terms × num_time_buckets` grid is small enough.
let Some((hist_req_data, range)) = prepare_histogram_dense_range(agg_data, &node.children[0])?
else {
return Ok(None);
};
let num_terms = (max_term_id + 1) as usize;
if num_terms.saturating_mul(range.len) > MAX_FUSED_GRID_BUCKETS {
return Ok(None);
}
let counts = vec![0u32; num_terms * range.len];
agg_data
.context
.limits
.add_memory_consumed((counts.len() * std::mem::size_of::<u32>()) as u64)?;
// With no hard bounds every doc lands in the dense range, so the per-term totals can be derived
// from the grid at result time instead of incremented per doc.
let derive_term_counts_from_histogram =
hist_req_data.bounds.min == f64::MIN && hist_req_data.bounds.max == f64::MAX;
Ok(Some(Box::new(SegmentTermHistogramCollector {
term_counts: vec![0u32; num_terms],
counts,
num_time_buckets: range.len,
base_pos: range.base_pos,
terms_req_data: terms_req_data.clone(),
hist_req_data,
term_block: ColumnBlockAccessor::default(),
hist_block: ColumnBlockAccessor::default(),
derive_term_counts_from_histogram,
})))
}
#[cfg(test)]
mod tests {
use crate::aggregation::agg_req::Aggregations;
use crate::aggregation::tests::{exec_request, get_test_index_from_values_and_terms};
/// Hand-computed correctness check for the fused terms×histogram fast path
/// ([`super::SegmentTermHistogramCollector`]): low-cardinality terms × a histogram leaf over
/// full columns, exercised single- and multi-segment.
#[test]
fn fused_term_histogram_test() -> crate::Result<()> {
fused_term_histogram_with_opt(false)?;
fused_term_histogram_with_opt(true)?;
Ok(())
}
fn fused_term_histogram_with_opt(merge_segments: bool) -> crate::Result<()> {
// 300 docs: term = {a, b, c} by i % 3, histogram value = i % 20 (interval 1 => buckets
// 0..19). gcd(3, 20) = 1, so every (term, bucket) pair occurs exactly 300 / 60 = 5 times.
let docs: Vec<(f64, String)> = (0..300u64)
.map(|i| {
(
(i % 20) as f64,
["a", "b", "c"][(i % 3) as usize].to_string(),
)
})
.collect();
// Two segments, to also exercise cross-segment merging of the fused per-term histograms.
let segments = vec![docs[..150].to_vec(), docs[150..].to_vec()];
let index = get_test_index_from_values_and_terms(merge_segments, &segments)?;
let agg_req: Aggregations = serde_json::from_value(serde_json::json!({
"by_term": {
"terms": { "field": "string_id", "order": { "_key": "asc" } },
"aggs": {
"histo": { "histogram": { "field": "score_f64", "interval": 1.0 } }
}
}
}))
.unwrap();
let res = exec_request(agg_req, &index)?;
for (term_idx, term) in ["a", "b", "c"].iter().enumerate() {
assert_eq!(res["by_term"]["buckets"][term_idx]["key"], *term);
assert_eq!(res["by_term"]["buckets"][term_idx]["doc_count"], 100);
let histo = &res["by_term"]["buckets"][term_idx]["histo"]["buckets"];
for b in 0..20usize {
assert_eq!(histo[b]["key"], b as f64, "term {term} bucket {b}");
assert_eq!(histo[b]["doc_count"], 5, "term {term} bucket {b}");
}
assert_eq!(histo[20], serde_json::Value::Null);
}
assert_eq!(res["by_term"]["buckets"][3], serde_json::Value::Null);
Ok(())
}
/// A `missing` config on a *full* term column still takes the fused path (the string sentinel
/// is just `col_max + 1`, so the column stays low-cardinality). Since no doc is missing, the
/// real term buckets must be exactly as without `missing`.
#[test]
fn fused_term_histogram_with_missing_on_full_column() -> crate::Result<()> {
let docs: Vec<(f64, String)> = (0..300u64)
.map(|i| {
(
(i % 20) as f64,
["a", "b", "c"][(i % 3) as usize].to_string(),
)
})
.collect();
let index = get_test_index_from_values_and_terms(true, &[docs])?;
let agg_req: Aggregations = serde_json::from_value(serde_json::json!({
"by_term": {
"terms": { "field": "string_id", "missing": "MISSING", "order": { "_key": "asc" } },
"aggs": {
"histo": { "histogram": { "field": "score_f64", "interval": 1.0 } }
}
}
}))
.unwrap();
let res = exec_request(agg_req, &index)?;
// Column is full, so "MISSING" never applies: a, b, c are unchanged (100 docs, 5 per
// bucket).
for (term_idx, term) in ["a", "b", "c"].iter().enumerate() {
assert_eq!(res["by_term"]["buckets"][term_idx]["key"], *term);
assert_eq!(res["by_term"]["buckets"][term_idx]["doc_count"], 100);
let histo = &res["by_term"]["buckets"][term_idx]["histo"]["buckets"];
for b in 0..20usize {
assert_eq!(histo[b]["doc_count"], 5, "term {term} bucket {b}");
}
}
Ok(())
}
/// Term cardinality above the general path's `MAX_NUM_TERMS_FOR_VEC` (100) still fuses: the
/// flat grid is bounded by the total cell count (`num_terms * num_time_buckets`), not the
/// term count.
#[test]
fn fused_term_histogram_many_terms() -> crate::Result<()> {
let num_terms = 150usize;
let docs_per_term = 2usize;
// All docs share histogram value 0 (a single bucket), so the grid is 150 x 1 = 150 cells.
let docs: Vec<(f64, String)> = (0..num_terms * docs_per_term)
.map(|i| (0.0, format!("t{:03}", i % num_terms)))
.collect();
let index = get_test_index_from_values_and_terms(true, &[docs])?;
let agg_req: Aggregations = serde_json::from_value(serde_json::json!({
"by_term": {
"terms": { "field": "string_id", "size": 1000, "order": { "_key": "asc" } },
"aggs": {
"histo": { "histogram": { "field": "score_f64", "interval": 1.0 } }
}
}
}))
.unwrap();
let res = exec_request(agg_req, &index)?;
let buckets = res["by_term"]["buckets"].as_array().unwrap();
assert_eq!(buckets.len(), num_terms);
for (i, bucket) in buckets.iter().enumerate() {
assert_eq!(bucket["key"], format!("t{i:03}"));
assert_eq!(bucket["doc_count"], docs_per_term as u64);
assert_eq!(bucket["histo"]["buckets"][0]["key"], 0.0);
assert_eq!(
bucket["histo"]["buckets"][0]["doc_count"],
docs_per_term as u64
);
}
Ok(())
}
/// `hard_bounds` exercises the non-derived `term_counts` branch: a term's `doc_count` must
/// count *every* doc with that term, including docs whose histogram value is outside the
/// bounds (those are excluded from the histogram buckets but still counted for the term). This
/// is the case where the per-doc `term_counts` increment cannot be replaced by the grid
/// row-sum.
#[test]
fn fused_term_histogram_with_hard_bounds() -> crate::Result<()> {
// 300 docs: term = {a, b, c} by i % 3, value = i % 20. Per term: 100 docs, each value in
// 0..=19 occurring 5 times.
let docs: Vec<(f64, String)> = (0..300u64)
.map(|i| {
(
(i % 20) as f64,
["a", "b", "c"][(i % 3) as usize].to_string(),
)
})
.collect();
let index = get_test_index_from_values_and_terms(true, &[docs])?;
// hard_bounds [5, 14] (inclusive) keeps only values 5..=14 in the histogram (10 buckets);
// values 0..=4 and 15..=19 are out of bounds.
let agg_req: Aggregations = serde_json::from_value(serde_json::json!({
"by_term": {
"terms": { "field": "string_id", "order": { "_key": "asc" } },
"aggs": {
"histo": {
"histogram": {
"field": "score_f64",
"interval": 1.0,
"hard_bounds": { "min": 5.0, "max": 14.0 }
}
}
}
}
}))
.unwrap();
let res = exec_request(agg_req, &index)?;
for (term_idx, term) in ["a", "b", "c"].iter().enumerate() {
assert_eq!(res["by_term"]["buckets"][term_idx]["key"], *term);
// doc_count includes the 50 per-term docs whose value is outside [5, 14].
assert_eq!(res["by_term"]["buckets"][term_idx]["doc_count"], 100);
let histo = &res["by_term"]["buckets"][term_idx]["histo"]["buckets"];
for b in 0..10usize {
let key = 5 + b;
assert_eq!(histo[b]["key"], key as f64, "term {term} bucket key {key}");
assert_eq!(histo[b]["doc_count"], 5, "term {term} bucket {key}");
}
// Only the 10 in-bounds buckets exist.
assert_eq!(histo[10], serde_json::Value::Null);
}
Ok(())
}
}

17
src/aggregation/bucket/term_missing_agg.rs
View File

@@ -5,7 +5,7 @@ use crate::aggregation::agg_data::{
build_segment_agg_collectors, AggRefNode, AggregationsSegmentCtx,
};
use crate::aggregation::bucket::term_agg::TermsAggregation;
use crate::aggregation::cached_sub_aggs::{CachedSubAggs, HighCardCachedSubAggs};
use crate::aggregation::buffered_sub_aggs::{BufferedSubAggs, HighCardBufferedSubAggs};
use crate::aggregation::intermediate_agg_result::{
IntermediateAggregationResult, IntermediateAggregationResults, IntermediateBucketResult,
IntermediateKey, IntermediateTermBucketEntry, IntermediateTermBucketResult,
@@ -47,7 +47,7 @@ struct MissingCount {
#[derive(Default, Debug)]
pub struct TermMissingAgg {
accessor_idx: usize,
sub_agg: Option<HighCardCachedSubAggs>,
sub_agg: Option<HighCardBufferedSubAggs>,
/// Idx = parent bucket id, Value = missing count for that bucket
missing_count_per_bucket: Vec<MissingCount>,
bucket_id_provider: BucketIdProvider,
@@ -66,7 +66,7 @@ impl TermMissingAgg {
None
};
let sub_agg = sub_agg.map(CachedSubAggs::new);
let sub_agg = sub_agg.map(BufferedSubAggs::new);
let bucket_id_provider = BucketIdProvider::default();
Ok(Self {
@@ -177,6 +177,17 @@ impl SegmentAggregationCollector for TermMissingAgg {
}
Ok(())
}
fn compute_metric_value(
&self,
_bucket_id: BucketId,
_sub_agg_name: &str,
_sub_agg_property: &str,
_agg_data: &AggregationsSegmentCtx,
) -> Option<f64> {
// TODO: forward to `sub_agg` for nested order paths (`missing_agg>metric`).
None
}
}
#[cfg(test)]

46
src/aggregation/cached_sub_aggs.rs → src/aggregation/buffered_sub_aggs.rs
View File

@@ -6,7 +6,7 @@ use crate::aggregation::bucket::MAX_NUM_TERMS_FOR_VEC;
use crate::aggregation::BucketId;
use crate::DocId;
/// A cache for sub-aggregations, storing doc ids per bucket id.
/// A buffer for sub-aggregations, storing doc ids per bucket id.
/// Depending on the cardinality of the parent aggregation, we use different
/// storage strategies.
///
@@ -24,21 +24,21 @@ use crate::DocId;
/// aggregations.
/// What this datastructure does in general is to group docs by bucket id.
#[derive(Debug)]
pub(crate) struct CachedSubAggs<C: SubAggCache> {
cache: C,
pub(crate) struct BufferedSubAggs<B: SubAggBuffer> {
buffer: B,
sub_agg_collector: Box<dyn SegmentAggregationCollector>,
num_docs: usize,
}
pub type LowCardCachedSubAggs = CachedSubAggs<LowCardSubAggCache>;
pub type HighCardCachedSubAggs = CachedSubAggs<HighCardSubAggCache>;
pub type LowCardBufferedSubAggs = BufferedSubAggs<LowCardSubAggBuffer>;
pub type HighCardBufferedSubAggs = BufferedSubAggs<HighCardSubAggBuffer>;
const FLUSH_THRESHOLD: usize = 2048;
/// A trait for caching sub-aggregation doc ids per bucket id.
/// A trait for buffering sub-aggregation doc ids per bucket id.
/// Different implementations can be used depending on the cardinality
/// of the parent aggregation.
pub trait SubAggCache: Debug {
pub trait SubAggBuffer: Debug {
fn new() -> Self;
fn push(&mut self, bucket_id: BucketId, doc_id: DocId);
fn flush_local(
@@ -49,22 +49,22 @@ pub trait SubAggCache: Debug {
) -> crate::Result<()>;
}
impl<Backend: SubAggCache + Debug> CachedSubAggs<Backend> {
impl<Backend: SubAggBuffer + Debug> BufferedSubAggs<Backend> {
pub fn new(sub_agg: Box<dyn SegmentAggregationCollector>) -> Self {
Self {
cache: Backend::new(),
buffer: Backend::new(),
sub_agg_collector: sub_agg,
num_docs: 0,
}
}
pub fn get_sub_agg_collector(&mut self) -> &mut Box<dyn SegmentAggregationCollector> {
&mut self.sub_agg_collector
pub fn get_sub_agg_collector(&mut self) -> &mut dyn SegmentAggregationCollector {
&mut *self.sub_agg_collector
}
#[inline]
pub fn push(&mut self, bucket_id: BucketId, doc_id: DocId) {
self.cache.push(bucket_id, doc_id);
self.buffer.push(bucket_id, doc_id);
self.num_docs += 1;
}
@@ -75,7 +75,7 @@ impl<Backend: SubAggCache + Debug> CachedSubAggs<Backend> {
agg_data: &mut AggregationsSegmentCtx,
) -> crate::Result<()> {
if self.num_docs >= FLUSH_THRESHOLD {
self.cache
self.buffer
.flush_local(&mut self.sub_agg_collector, agg_data, false)?;
self.num_docs = 0;
}
@@ -85,7 +85,7 @@ impl<Backend: SubAggCache + Debug> CachedSubAggs<Backend> {
/// Note: this _does_ flush the sub aggregations.
pub fn flush(&mut self, agg_data: &mut AggregationsSegmentCtx) -> crate::Result<()> {
if self.num_docs != 0 {
self.cache
self.buffer
.flush_local(&mut self.sub_agg_collector, agg_data, true)?;
self.num_docs = 0;
}
@@ -94,11 +94,11 @@ impl<Backend: SubAggCache + Debug> CachedSubAggs<Backend> {
}
}
/// Number of partitions for high cardinality sub-aggregation cache.
/// Number of partitions for high cardinality sub-aggregation buffer.
const NUM_PARTITIONS: usize = 16;
#[derive(Debug)]
pub(crate) struct HighCardSubAggCache {
pub(crate) struct HighCardSubAggBuffer {
/// This weird partitioning is used to do some cheap grouping on the bucket ids.
/// bucket ids are dense, e.g. when we don't detect the cardinality as low cardinality,
/// but there are just 16 bucket ids, each bucket id will go to its own partition.
@@ -108,7 +108,7 @@ pub(crate) struct HighCardSubAggCache {
partitions: Box<[PartitionEntry; NUM_PARTITIONS]>,
}
impl HighCardSubAggCache {
impl HighCardSubAggBuffer {
#[inline]
fn clear(&mut self) {
for partition in self.partitions.iter_mut() {
@@ -131,13 +131,14 @@ impl PartitionEntry {
}
}
impl SubAggCache for HighCardSubAggCache {
impl SubAggBuffer for HighCardSubAggBuffer {
fn new() -> Self {
Self {
partitions: Box::new(core::array::from_fn(|_| PartitionEntry::default())),
}
}
#[inline]
fn push(&mut self, bucket_id: BucketId, doc_id: DocId) {
let idx = bucket_id % NUM_PARTITIONS as u32;
let slot = &mut self.partitions[idx as usize];
@@ -173,14 +174,14 @@ impl SubAggCache for HighCardSubAggCache {
}
#[derive(Debug)]
pub(crate) struct LowCardSubAggCache {
/// Cache doc ids per bucket for sub-aggregations.
pub(crate) struct LowCardSubAggBuffer {
/// Buffer doc ids per bucket for sub-aggregations.
///
/// The outer Vec is indexed by BucketId.
per_bucket_docs: Vec<Vec<DocId>>,
}
impl LowCardSubAggCache {
impl LowCardSubAggBuffer {
#[inline]
fn clear(&mut self) {
for v in &mut self.per_bucket_docs {
@@ -189,13 +190,14 @@ impl LowCardSubAggCache {
}
}
impl SubAggCache for LowCardSubAggCache {
impl SubAggBuffer for LowCardSubAggBuffer {
fn new() -> Self {
Self {
per_bucket_docs: Vec::new(),
}
}
#[inline]
fn push(&mut self, bucket_id: BucketId, doc_id: DocId) {
let idx = bucket_id as usize;
if self.per_bucket_docs.len() <= idx {

6
src/aggregation/collector.rs
View File

@@ -1,6 +1,6 @@
use super::agg_req::Aggregations;
use super::agg_result::AggregationResults;
use super::cached_sub_aggs::LowCardCachedSubAggs;
use super::buffered_sub_aggs::LowCardBufferedSubAggs;
use super::intermediate_agg_result::IntermediateAggregationResults;
use super::AggContextParams;
// group buffering strategy is chosen explicitly by callers; no need to hash-group on the fly.
@@ -136,7 +136,7 @@ fn merge_fruits(
/// `AggregationSegmentCollector` does the aggregation collection on a segment.
pub struct AggregationSegmentCollector {
aggs_with_accessor: AggregationsSegmentCtx,
agg_collector: LowCardCachedSubAggs,
agg_collector: LowCardBufferedSubAggs,
error: Option<TantivyError>,
}
@@ -152,7 +152,7 @@ impl AggregationSegmentCollector {
let mut agg_data =
build_aggregations_data_from_req(agg, reader, segment_ordinal, context.clone())?;
let mut result =
LowCardCachedSubAggs::new(build_segment_agg_collectors_root(&mut agg_data)?);
LowCardBufferedSubAggs::new(build_segment_agg_collectors_root(&mut agg_data)?);
result
.get_sub_agg_collector()
.prepare_max_bucket(0, &agg_data)?; // prepare for bucket zero

32
src/aggregation/intermediate_agg_result.rs
View File

@@ -1004,24 +1004,20 @@ impl IntermediateCompositeBucketResult {
) -> crate::Result<BucketResult> {
let trimmed_entry_vec =
trim_composite_buckets(self.entries, &self.orders, self.target_size)?;
let after_key = if trimmed_entry_vec.len() == req.size as usize {
trimmed_entry_vec
.last()
.map(|bucket| {
let (intermediate_key, _entry) = bucket;
intermediate_key
.iter()
.enumerate()
.map(|(idx, intermediate_key)| {
let source = &req.sources[idx];
(source.name().to_string(), intermediate_key.clone().into())
})
.collect()
})
.unwrap()
} else {
FxHashMap::default()
};
let after_key = trimmed_entry_vec
.last()
.map(|bucket| {
let (intermediate_key, _entry) = bucket;
intermediate_key
.iter()
.enumerate()
.map(|(idx, intermediate_key)| {
let source = &req.sources[idx];
(source.name().to_string(), intermediate_key.clone().into())
})
.collect()
})
.unwrap_or_default();
let buckets = trimmed_entry_vec
.into_iter()

970
src/aggregation/metric/cardinality.rs
View File

File diff suppressed because it is too large Load Diff

20
src/aggregation/metric/extended_stats.rs
View File

@@ -399,6 +399,26 @@ impl SegmentAggregationCollector for SegmentExtendedStatsCollector {
}
Ok(())
}
fn compute_metric_value(
&self,
bucket_id: BucketId,
sub_agg_name: &str,
sub_agg_property: &str,
_agg_data: &AggregationsSegmentCtx,
) -> Option<f64> {
if self.name != sub_agg_name {
return None;
}
let extended = self.buckets.get(bucket_id as usize)?;
// Finalize is a pure read of accumulators — calling it here for the cutoff sort
// doesn't disturb the eventual intermediate result.
extended
.finalize()
.get_value(sub_agg_property)
.ok()
.flatten()
}
}
#[cfg(test)]

5
src/aggregation/metric/mod.rs
View File

@@ -107,10 +107,9 @@ pub enum PercentileValues {
#[derive(Clone, Debug, PartialEq, Serialize, Deserialize)]
/// The entry when requesting percentiles with keyed: false
pub struct PercentileValuesVecEntry {
/// Percentile
/// The percentile key (e.g. 1.0, 5.0, 25.0).
pub key: f64,
/// Value at the percentile
/// The percentile value. `NaN` when there are no values.
pub value: f64,
}

20
src/aggregation/metric/percentiles.rs
View File

@@ -312,6 +312,26 @@ impl SegmentAggregationCollector for SegmentPercentilesCollector {
}
Ok(())
}
fn compute_metric_value(
&self,
bucket_id: BucketId,
sub_agg_name: &str,
sub_agg_property: &str,
agg_data: &AggregationsSegmentCtx,
) -> Option<f64> {
if agg_data.get_metric_req_data(self.accessor_idx).name != sub_agg_name {
return None;
}
let percentile: f64 = sub_agg_property.parse().ok()?;
if !(0.0..=100.0).contains(&percentile) {
return None;
}
let bucket = self.buckets.get(bucket_id as usize)?;
// DDSketch.quantile is a pure read; calling it here for the cutoff sort does
// not affect the intermediate state used for the final result.
bucket.sketch.quantile(percentile / 100.0).ok().flatten()
}
}
#[cfg(test)]

34
src/aggregation/metric/stats.rs
View File

@@ -321,6 +321,40 @@ impl<const COLUMN_TYPE_ID: u8> SegmentAggregationCollector
}
Ok(())
}
fn compute_metric_value(
&self,
bucket_id: BucketId,
sub_agg_name: &str,
sub_agg_property: &str,
_agg_data: &AggregationsSegmentCtx,
) -> Option<f64> {
if self.name != sub_agg_name {
return None;
}
let stats = self.buckets.get(bucket_id as usize)?;
// The property depends on what we're collecting:
// - StatsType::Stats exposes count/sum/min/max/avg via dotted property.
// - Single-value kinds (Sum/Count/Min/Max/Average) expect an empty property and return
// the value they were configured to collect.
let prop = match self.collecting_for {
StatsType::Stats if !sub_agg_property.is_empty() => sub_agg_property,
StatsType::Sum if sub_agg_property.is_empty() => "sum",
StatsType::Count if sub_agg_property.is_empty() => "count",
StatsType::Max if sub_agg_property.is_empty() => "max",
StatsType::Min if sub_agg_property.is_empty() => "min",
StatsType::Average if sub_agg_property.is_empty() => "avg",
_ => return None,
};
match prop {
"count" => Some(stats.count as f64),
"sum" => Some(stats.sum),
"min" if stats.count > 0 => Some(stats.min),
"max" if stats.count > 0 => Some(stats.max),
"avg" if stats.count > 0 => Some(stats.sum / stats.count as f64),
_ => None,
}
}
}
#[inline]

11
src/aggregation/metric/top_hits.rs
View File

@@ -644,6 +644,17 @@ impl SegmentAggregationCollector for TopHitsSegmentCollector {
);
Ok(())
}
fn compute_metric_value(
&self,
_bucket_id: BucketId,
_sub_agg_name: &str,
_sub_agg_property: &str,
_agg_data: &AggregationsSegmentCtx,
) -> Option<f64> {
// top_hits is not a numeric metric and cannot be used as an order target.
None
}
}
#[cfg(test)]

2
src/aggregation/mod.rs
View File

@@ -133,7 +133,7 @@ mod agg_limits;
pub mod agg_req;
pub mod agg_result;
pub mod bucket;
pub(crate) mod cached_sub_aggs;
pub(crate) mod buffered_sub_aggs;
mod collector;
mod date;
mod error;

42
src/aggregation/segment_agg_result.rs
View File

@@ -76,6 +76,31 @@ pub trait SegmentAggregationCollector: Debug {
fn flush(&mut self, _agg_data: &mut AggregationsSegmentCtx) -> crate::Result<()> {
Ok(())
}
/// Compute the segment-level metric value of the named direct-child metric for `bucket_id`.
///
/// Used by parent term aggs that order by a sub-aggregation: the parent sorts on
/// this value and cuts off at segment time, matching the approximation tradeoff
/// Elasticsearch makes for any sub-agg ordering.
///
/// `sub_agg_property` is the dotted suffix (e.g. `"sum"` in `mystats.sum`); empty when
/// the metric is a single-value kind such as cardinality.
///
/// Returns `None` only on name mismatch, unknown property, or empty bucket. Implementations
/// may finalize their per-bucket state (e.g. compute a percentile from a sketch); calls
/// must be idempotent so the final intermediate result is unaffected.
///
/// No default impl on purpose: every collector must decide explicitly whether it
/// produces a metric value, forwards into children (single-bucket aggs), or rejects
/// the lookup. A silent `None` default would let a parent term agg's cutoff sort all
/// buckets to the same key and drop arbitrary winners.
fn compute_metric_value(
&self,
bucket_id: BucketId,
sub_agg_name: &str,
sub_agg_property: &str,
agg_data: &AggregationsSegmentCtx,
) -> Option<f64>;
}
#[derive(Default)]
@@ -137,4 +162,21 @@ impl SegmentAggregationCollector for GenericSegmentAggregationResultsCollector {
}
Ok(())
}
fn compute_metric_value(
&self,
bucket_id: BucketId,
sub_agg_name: &str,
sub_agg_property: &str,
agg_data: &AggregationsSegmentCtx,
) -> Option<f64> {
for agg in &self.aggs {
if let Some(value) =
agg.compute_metric_value(bucket_id, sub_agg_name, sub_agg_property, agg_data)
{
return Some(value);
}
}
None
}
}

10
src/collector/count_collector.rs
View File

@@ -1,5 +1,6 @@
use super::Collector;
use crate::collector::SegmentCollector;
use crate::query::Weight;
use crate::{DocId, Score, SegmentOrdinal, SegmentReader};
/// `CountCollector` collector only counts how many
@@ -55,6 +56,15 @@ impl Collector for Count {
fn merge_fruits(&self, segment_counts: Vec<usize>) -> crate::Result<usize> {
Ok(segment_counts.into_iter().sum())
}
fn collect_segment(
&self,
weight: &dyn Weight,
_segment_ord: u32,
reader: &SegmentReader,
) -> crate::Result<usize> {
Ok(weight.count(reader)? as usize)
}
}
#[derive(Default)]

64
src/collector/facet_collector.rs
View File

@@ -389,6 +389,13 @@ impl SegmentCollector for FacetSegmentCollector {
}
let mut facet = vec![];
let (facet_ord, facet_depth) = self.unique_facet_ords[collapsed_facet_ord];
// u64::MAX is used as a sentinel for unmapped ordinals (e.g. when a
// document has the exact registered facet, not a child of it).
// Passing it to ord_to_term would resolve to the last dictionary
// entry and produce a spurious facet from an unrelated branch.
if facet_ord == u64::MAX {
continue;
}
// TODO handle errors.
if facet_dict.ord_to_term(facet_ord, &mut facet).is_ok() {
if let Some((end_collapsed_facet, _)) = facet
@@ -814,6 +821,63 @@ mod tests {
assert!(!super::is_child_facet(&b"foo\0bar"[..], &b"foo"[..]));
assert!(!super::is_child_facet(&b"foo"[..], &b"foobar\0baz"[..]));
}
// Regression test for https://github.com/quickwit-oss/tantivy/issues/2494
// When a document has the exact registered facet path (not just a child),
// harvest() must not turn the unmapped sentinel into a spurious root entry.
#[test]
fn test_facet_collector_wrong_root() -> crate::Result<()> {
let mut schema_builder = Schema::builder();
let facet_field = schema_builder.add_facet_field("facet", FacetOptions::default());
let schema = schema_builder.build();
let index = Index::create_in_ram(schema);
let mut index_writer: IndexWriter = index.writer_for_tests()?;
let facets: Vec<&str> = vec![
"/science-fiction/asimov",
"/science-fiction/clarke",
"/science-fiction/dick",
"/science-fiction/herbert",
"/science-fiction/orwell",
// This exact match on the registered facet is the bug trigger:
// its ordinal maps to the sentinel (u64::MAX, 0) in the collapse
// mapping, which without the fix resolves to an unrelated term.
"/fantasy/epic-fantasy",
"/fantasy/epic-fantasy/tolkien",
"/fantasy/epic-fantasy/martin",
];
for facet_str in &facets {
index_writer.add_document(doc!(
facet_field => Facet::from(*facet_str)
))?;
}
index_writer.commit()?;
let reader = index.reader()?;
let searcher = reader.searcher();
let term = Term::from_facet(facet_field, &Facet::from("/fantasy/epic-fantasy"));
let query = TermQuery::new(term, IndexRecordOption::Basic);
let mut facet_collector = FacetCollector::for_field("facet");
facet_collector.add_facet("/fantasy/epic-fantasy");
let counts: FacetCounts = searcher.search(&query, &facet_collector)?;
let result: Vec<(String, u64)> = counts
.get("/")
.map(|(facet, count)| (facet.to_string(), count))
.collect();
// Only children of /fantasy/epic-fantasy should appear, not /science-fiction
assert_eq!(
result,
vec![
("/fantasy/epic-fantasy/martin".to_string(), 1),
("/fantasy/epic-fantasy/tolkien".to_string(), 1),
]
);
Ok(())
}
}
#[cfg(all(test, feature = "unstable"))]

216
src/collector/sort_key/sort_by_score.rs
View File

@@ -1,5 +1,8 @@
use std::cmp::{Ordering, Reverse};
use std::collections::BinaryHeap;
use crate::collector::sort_key::NaturalComparator;
use crate::collector::{SegmentSortKeyComputer, SortKeyComputer, TopNComputer};
use crate::collector::{SegmentSortKeyComputer, SortKeyComputer};
use crate::{DocAddress, DocId, Score};
/// Sort by similarity score.
@@ -25,6 +28,10 @@ impl SortKeyComputer for SortBySimilarityScore {
}
// Sorting by score is special in that it allows for the Block-Wand optimization.
//
// We use a BinaryHeap (TopNHeap) instead of TopNComputer here so that the
// threshold is always the exact K-th best score. TopNComputer only updates its
// threshold every K docs (at truncation), giving Block-WAND a stale bound.
fn collect_segment_top_k(
&self,
k: usize,
@@ -32,12 +39,10 @@ impl SortKeyComputer for SortBySimilarityScore {
reader: &crate::SegmentReader,
segment_ord: u32,
) -> crate::Result<Vec<(Self::SortKey, DocAddress)>> {
let mut top_n: TopNComputer<Score, DocId, Self::Comparator> =
TopNComputer::new_with_comparator(k, self.comparator());
let mut top_n = TopNHeap::new(k);
if let Some(alive_bitset) = reader.alive_bitset() {
let mut threshold = Score::MIN;
top_n.threshold = Some(threshold);
weight.for_each_pruning(Score::MIN, reader, &mut |doc, score| {
if alive_bitset.is_deleted(doc) {
return threshold;
@@ -56,7 +61,7 @@ impl SortKeyComputer for SortBySimilarityScore {
Ok(top_n
.into_vec()
.into_iter()
.map(|cid| (cid.sort_key, DocAddress::new(segment_ord, cid.doc)))
.map(|(score, doc)| (score, DocAddress::new(segment_ord, doc)))
.collect())
}
}
@@ -75,3 +80,204 @@ impl SegmentSortKeyComputer for SortBySimilarityScore {
score
}
}
/// Min-heap entry: higher score = greater, lower doc wins ties.
struct ScoreHeapEntry {
score: Score,
doc: DocId,
}
impl Eq for ScoreHeapEntry {}
impl PartialEq for ScoreHeapEntry {
fn eq(&self, other: &Self) -> bool {
self.cmp(other) == Ordering::Equal
}
}
impl PartialOrd for ScoreHeapEntry {
fn partial_cmp(&self, other: &Self) -> Option<Ordering> {
Some(self.cmp(other))
}
}
impl Ord for ScoreHeapEntry {
fn cmp(&self, other: &Self) -> Ordering {
self.score
.partial_cmp(&other.score)
.unwrap_or(Ordering::Equal)
.then_with(|| other.doc.cmp(&self.doc))
}
}
/// Heap-based top-K for score collection. O(log K) per insert, but the threshold
/// is always tight, so Block-WAND prunes better than with [`TopNComputer`]'s
/// buffer/median approach.
///
/// Like [`TopNComputer`], items must arrive in ascending doc order, and equal
/// scores are rejected (strict `>`) so that lower doc IDs win ties.
///
/// [`TopNComputer`]: crate::collector::TopNComputer
struct TopNHeap {
heap: BinaryHeap<Reverse<ScoreHeapEntry>>,
top_n: usize,
threshold: Option<Score>,
}
impl TopNHeap {
fn new(top_n: usize) -> Self {
TopNHeap {
heap: BinaryHeap::with_capacity(top_n),
top_n,
threshold: None,
}
}
#[inline]
fn push(&mut self, score: Score, doc: DocId) {
if self.heap.len() < self.top_n {
self.heap.push(Reverse(ScoreHeapEntry { score, doc }));
if self.heap.len() == self.top_n {
self.threshold = self.heap.peek().map(|Reverse(entry)| entry.score);
}
} else if let Some(threshold) = self.threshold {
if score > threshold {
// peek_mut + assign is a single sift-down, vs pop + push = two sifts.
if let Some(mut min) = self.heap.peek_mut() {
*min = Reverse(ScoreHeapEntry { score, doc });
}
self.threshold = self.heap.peek().map(|Reverse(entry)| entry.score);
}
}
}
fn into_vec(self) -> Vec<(Score, DocId)> {
self.heap
.into_vec()
.into_iter()
.map(|Reverse(entry)| (entry.score, entry.doc))
.collect()
}
}
#[cfg(test)]
mod tests {
use proptest::prelude::*;
use super::*;
use crate::collector::sort_key::NaturalComparator;
use crate::collector::TopNComputer;
#[test]
fn test_top_n_heap_zero_capacity() {
let mut heap = TopNHeap::new(0);
heap.push(1.0, 0);
heap.push(2.0, 1);
assert!(heap.into_vec().is_empty());
}
#[test]
fn test_top_n_heap_basic() {
let mut heap = TopNHeap::new(2);
heap.push(1.0, 0);
heap.push(3.0, 1);
heap.push(2.0, 2);
let mut results = heap.into_vec();
results.sort_by(|a, b| b.0.partial_cmp(&a.0).unwrap().then_with(|| a.1.cmp(&b.1)));
assert_eq!(results, vec![(3.0, 1), (2.0, 2)]);
}
#[test]
fn test_top_n_heap_threshold_always_accurate() {
let mut heap = TopNHeap::new(2);
assert_eq!(heap.threshold, None);
heap.push(1.0, 0);
assert_eq!(heap.threshold, None);
heap.push(3.0, 1);
assert_eq!(heap.threshold, Some(1.0));
heap.push(2.0, 2); // evicts 1.0
assert_eq!(heap.threshold, Some(2.0));
heap.push(4.0, 3); // evicts 2.0
assert_eq!(heap.threshold, Some(3.0));
}
#[test]
fn test_top_n_heap_tiebreaking_lower_doc_wins() {
let mut heap = TopNHeap::new(2);
heap.push(5.0, 0);
heap.push(5.0, 1);
heap.push(5.0, 2); // rejected: not strictly > threshold
let mut results = heap.into_vec();
results.sort_by_key(|&(_, doc)| doc);
assert_eq!(results, vec![(5.0, 0), (5.0, 1)]);
}
#[test]
fn test_top_n_heap_single_element() {
let mut heap = TopNHeap::new(1);
heap.push(1.0, 0);
assert_eq!(heap.threshold, Some(1.0));
heap.push(0.5, 1); // rejected
heap.push(2.0, 2); // accepted
assert_eq!(heap.threshold, Some(2.0));
let results = heap.into_vec();
assert_eq!(results, vec![(2.0, 2)]);
}
#[test]
fn test_top_n_heap_under_capacity() {
let mut heap = TopNHeap::new(5);
heap.push(3.0, 0);
heap.push(1.0, 1);
heap.push(2.0, 2);
// Only 3 elements, capacity is 5 — all should be kept
assert_eq!(heap.threshold, None);
let mut results = heap.into_vec();
results.sort_by(|a, b| b.0.partial_cmp(&a.0).unwrap().then_with(|| a.1.cmp(&b.1)));
assert_eq!(results, vec![(3.0, 0), (2.0, 2), (1.0, 1)]);
}
proptest! {
#[test]
fn test_top_n_heap_matches_top_n_computer(
limit in 0..20_usize,
mut docs in proptest::collection::vec((0..1000_u32, 0..1000_u32), 0..200_usize),
) {
// Both require ascending doc order.
docs.sort_by_key(|(_, doc_id)| *doc_id);
docs.dedup_by_key(|(_, doc_id)| *doc_id);
let mut heap = TopNHeap::new(limit);
let mut computer: TopNComputer<Score, DocId, NaturalComparator> =
TopNComputer::new_with_comparator(limit, NaturalComparator);
for &(score_u32, doc) in &docs {
let score = score_u32 as Score;
heap.push(score, doc);
computer.push(score, doc);
}
let mut heap_results = heap.into_vec();
heap_results.sort_by(|a, b| {
b.0.partial_cmp(&a.0).unwrap().then_with(|| a.1.cmp(&b.1))
});
let computer_results: Vec<(Score, DocId)> = computer
.into_sorted_vec()
.into_iter()
.map(|cd| (cd.sort_key, cd.doc))
.collect();
prop_assert_eq!(heap_results, computer_results);
}
}
}

2
src/collector/sort_key/sort_by_static_fast_value.rs
View File

@@ -52,7 +52,7 @@ impl<T: FastValue> SortKeyComputer for SortByStaticFastValue<T> {
if schema_type != T::to_type() {
return Err(crate::TantivyError::SchemaError(format!(
"Field `{}` is of type {schema_type:?}, not of the type {:?}.",
&self.field,
self.field,
T::to_type()
)));
}

4
src/collector/top_score_collector.rs
View File

@@ -513,7 +513,9 @@ pub struct TopNComputer<Score, D, C> {
/// The buffer reverses sort order to get top-semantics instead of bottom-semantics
buffer: Vec<ComparableDoc<Score, D>>,
top_n: usize,
pub(crate) threshold: Option<Score>,
/// The current threshold for pruning. Documents with scores at or below
/// this value are skipped by `push()`. Updated when the buffer is truncated.
pub threshold: Option<Score>,
comparator: C,
}

54
src/docset.rs
View File

@@ -1,5 +1,7 @@
use std::borrow::{Borrow, BorrowMut};
use common::TinySet;
use crate::fastfield::AliveBitSet;
use crate::DocId;
@@ -14,6 +16,12 @@ pub const TERMINATED: DocId = i32::MAX as u32;
/// exactly this size as long as we can fill the buffer.
pub const COLLECT_BLOCK_BUFFER_LEN: usize = 64;
/// Number of `TinySet` (64-bit) buckets in a block used by [`DocSet::fill_bitset_block`].
pub const BLOCK_NUM_TINYBITSETS: usize = 16;
/// Number of doc IDs covered by one block: `BLOCK_NUM_TINYBITSETS * 64 = 1024`.
pub const BLOCK_WINDOW: u32 = BLOCK_NUM_TINYBITSETS as u32 * 64;
/// Represents an iterable set of sorted doc ids.
pub trait DocSet: Send {
/// Goes to the next element.
@@ -160,6 +168,31 @@ pub trait DocSet: Send {
self.size_hint() as u64
}
/// Fills a bitmask representing which documents in `[min_doc, min_doc + BLOCK_WINDOW)` are
/// present in this docset.
///
/// The window is divided into `BLOCK_NUM_TINYBITSETS` buckets of 64 docs each.
/// Returns the next doc `>= min_doc + BLOCK_WINDOW`, or `TERMINATED` if exhausted.
fn fill_bitset_block(
&mut self,
min_doc: DocId,
mask: &mut [TinySet; BLOCK_NUM_TINYBITSETS],
) -> DocId {
self.seek(min_doc);
let horizon = min_doc + BLOCK_WINDOW;
loop {
let doc = self.doc();
if doc >= horizon {
return doc;
}
let delta = doc - min_doc;
mask[(delta / 64) as usize].insert_mut(delta % 64);
if self.advance() == TERMINATED {
return TERMINATED;
}
}
}
/// Returns the number documents matching.
/// Calling this method consumes the `DocSet`.
fn count(&mut self, alive_bitset: &AliveBitSet) -> u32 {
@@ -214,6 +247,18 @@ impl DocSet for &mut dyn DocSet {
(**self).seek_danger(target)
}
fn fill_buffer(&mut self, buffer: &mut [DocId; COLLECT_BLOCK_BUFFER_LEN]) -> usize {
(**self).fill_buffer(buffer)
}
fn fill_bitset_block(
&mut self,
min_doc: DocId,
mask: &mut [TinySet; BLOCK_NUM_TINYBITSETS],
) -> DocId {
(**self).fill_bitset_block(min_doc, mask)
}
fn doc(&self) -> u32 {
(**self).doc()
}
@@ -256,6 +301,15 @@ impl<TDocSet: DocSet + ?Sized> DocSet for Box<TDocSet> {
unboxed.fill_buffer(buffer)
}
fn fill_bitset_block(
&mut self,
min_doc: DocId,
mask: &mut [TinySet; BLOCK_NUM_TINYBITSETS],
) -> DocId {
let unboxed: &mut TDocSet = self.borrow_mut();
unboxed.fill_bitset_block(min_doc, mask)
}
fn doc(&self) -> DocId {
let unboxed: &TDocSet = self.borrow();
unboxed.doc()

13
src/index/segment_reader.rs
View File

@@ -6,6 +6,7 @@ use common::{ByteCount, HasLen};
use fnv::FnvHashMap;
use itertools::Itertools;
use crate::directory::error::OpenReadError;
use crate::directory::{CompositeFile, FileSlice};
use crate::error::DataCorruption;
use crate::fastfield::{intersect_alive_bitsets, AliveBitSet, FacetReader, FastFieldReaders};
@@ -159,12 +160,10 @@ impl SegmentReader {
let postings_file = segment.open_read(SegmentComponent::Postings)?;
let postings_composite = CompositeFile::open(&postings_file)?;
let positions_composite = {
if let Ok(positions_file) = segment.open_read(SegmentComponent::Positions) {
CompositeFile::open(&positions_file)?
} else {
CompositeFile::empty()
}
let positions_composite = match segment.open_read(SegmentComponent::Positions) {
Ok(positions_file) => CompositeFile::open(&positions_file)?,
Err(OpenReadError::FileDoesNotExist(_)) => CompositeFile::empty(),
Err(open_read_error) => return Err(open_read_error.into()),
};
let schema = segment.schema();
@@ -323,7 +322,7 @@ impl SegmentReader {
// Without expand dots enabled dots need to be escaped.
let escaped_json_path = json_path.replace('.', "\\.");
let full_path = format!("{field_name}.{escaped_json_path}");
let full_path_unescaped = format!("{}.{}", field_name, &json_path);
let full_path_unescaped = format!("{}.{}", field_name, json_path);
map_to_canonical.insert(full_path_unescaped, full_path.to_string());
full_path
} else {

11
src/postings/block_segment_postings.rs
View File

@@ -249,6 +249,12 @@ impl BlockSegmentPostings {
/// Returns the length of the current block.
///
/// Returns the decoded term-frequency buffer for the current block.
#[inline]
pub(crate) fn freq_output_array(&self) -> &[u32] {
self.freq_decoder.output_array()
}
/// All blocks have a length of `NUM_DOCS_PER_BLOCK`,
/// except the last block that may have a length
/// of any number between 1 and `NUM_DOCS_PER_BLOCK - 1`
@@ -298,6 +304,11 @@ impl BlockSegmentPostings {
}
}
#[inline]
pub(crate) fn has_remaining_docs(&self) -> bool {
self.skip_reader.has_remaining_docs()
}
pub(crate) fn block_is_loaded(&self) -> bool {
self.block_loaded
}

97
src/postings/recorder.rs
View File

@@ -275,8 +275,9 @@ impl Recorder for TfAndPositionRecorder {
mod tests {
use common::write_u32_vint;
use stacker::MemoryArena;
use super::{BufferLender, VInt32Reader};
use super::{BufferLender, Recorder, TermFrequencyRecorder, VInt32Reader};
#[test]
fn test_buffer_lender() {
@@ -314,4 +315,98 @@ mod tests {
let res: Vec<u32> = VInt32Reader::new(&buffer[..]).collect();
assert_eq!(&res[..], &vals[..]);
}
// ── TermFrequencyRecorder ─────────────────────────────────────────────────
#[test]
fn term_frequency_recorder_has_term_freq() {
let rec = TermFrequencyRecorder::default();
assert!(
rec.has_term_freq(),
"TermFrequencyRecorder must advertise term-frequency support"
);
}
#[test]
fn term_frequency_recorder_term_doc_freq_single_doc() {
let mut arena = MemoryArena::default();
let mut rec = TermFrequencyRecorder::default();
// Record one document with two term occurrences.
rec.new_doc(0, &mut arena);
rec.record_position(0, &mut arena);
rec.record_position(1, &mut arena);
rec.close_doc(&mut arena);
assert_eq!(
rec.term_doc_freq(),
Some(1),
"term_doc_freq should be 1 after recording one document"
);
}
#[test]
fn term_frequency_recorder_term_doc_freq_multiple_docs() {
let mut arena = MemoryArena::default();
let mut rec = TermFrequencyRecorder::default();
// Three documents with 1, 3, and 2 occurrences respectively.
for (doc, tf) in [(0u32, 1u32), (5, 3), (10, 2)] {
rec.new_doc(doc, &mut arena);
for pos in 0..tf {
rec.record_position(pos, &mut arena);
}
rec.close_doc(&mut arena);
}
assert_eq!(
rec.term_doc_freq(),
Some(3),
"term_doc_freq should equal the number of documents recorded"
);
}
#[test]
fn term_frequency_recorder_zero_docs() {
let rec = TermFrequencyRecorder::default();
assert_eq!(
rec.term_doc_freq(),
Some(0),
"term_doc_freq should be 0 before any document is recorded"
);
}
#[test]
fn term_frequency_recorder_single_occurrence_per_doc() {
let mut arena = MemoryArena::default();
let mut rec = TermFrequencyRecorder::default();
// Each document has exactly one occurrence — the minimum non-trivial case.
for doc in [1u32, 2, 100] {
rec.new_doc(doc, &mut arena);
rec.record_position(0, &mut arena);
rec.close_doc(&mut arena);
}
assert_eq!(rec.term_doc_freq(), Some(3));
}
#[test]
fn term_frequency_recorder_high_frequency_doc() {
let mut arena = MemoryArena::default();
let mut rec = TermFrequencyRecorder::default();
// A document where the term appears many times.
rec.new_doc(42, &mut arena);
for pos in 0..1000 {
rec.record_position(pos, &mut arena);
}
rec.close_doc(&mut arena);
assert_eq!(
rec.term_doc_freq(),
Some(1),
"term_doc_freq counts documents, not occurrences"
);
}
}

5
src/postings/skip.rs
View File

@@ -146,6 +146,11 @@ impl SkipReader {
skip_reader
}
#[inline(always)]
pub fn has_remaining_docs(&self) -> bool {
self.remaining_docs != 0
}
pub fn reset(&mut self, data: OwnedBytes, doc_freq: u32) {
self.last_doc_in_block = if doc_freq >= COMPRESSION_BLOCK_SIZE as u32 {
0

464
src/query/boolean_query/block_wand_intersection.rs Normal file
View File

@@ -0,0 +1,464 @@
use crate::postings::compression::COMPRESSION_BLOCK_SIZE;
use crate::query::term_query::TermScorer;
use crate::query::Scorer;
use crate::{DocId, DocSet, Score, TERMINATED};
/// Block-max pruning for top-K over intersection of term scorers.
///
/// Uses the least-frequent term as "leader" to define 128-doc processing windows.
/// For each window, the sum of block_max_scores is compared to the current threshold;
/// if the block can't beat it, the entire block is skipped.
///
/// Within non-skipped blocks, individual documents are pruned by checking whether
/// leader_score + sum(secondary block_max_scores) can exceed the threshold before
/// performing the expensive intersection membership check (seeking into secondary scorers).
///
/// # Preconditions
/// - `scorers` has at least 2 elements
/// - All scorers read frequencies (`FreqReadingOption::ReadFreq`)
pub(crate) fn block_wand_intersection(
mut scorers: Vec<TermScorer>,
mut threshold: Score,
callback: &mut dyn FnMut(DocId, Score) -> Score,
) {
assert!(scorers.len() >= 2);
// Sort by cost (ascending). scorers[0] becomes the "leader" (rarest term).
scorers.sort_by_key(TermScorer::size_hint);
let (leader, secondaries) = scorers.split_first_mut().unwrap();
// Precompute global max scores for early termination checks.
let leader_max_score: Score = leader.max_score();
let secondaries_global_max_sum: Score = secondaries.iter().map(TermScorer::max_score).sum();
// Early exit: no document can possibly beat the threshold.
if leader_max_score + secondaries_global_max_sum <= threshold {
return;
}
// Borrow fieldnorm reader and BM25 weight before the main loop.
// These are immutable references to disjoint fields from block_cursor,
// but Rust's borrow checker can't see through method calls, so we
// extract them once upfront.
let fieldnorm_reader = leader.fieldnorm_reader().clone();
let bm25_weight = leader.bm25_weight().clone();
let mut doc = leader.doc();
let mut secondary_block_max_scores: Box<[f32]> =
vec![0.0f32; secondaries.len()].into_boxed_slice();
let mut secondary_suffix_block_max: Box<[f32]> =
vec![0.0f32; secondaries.len()].into_boxed_slice();
while doc < TERMINATED {
// --- Phase 1: Block-level pruning ---
//
// Position all skip readers on the block containing `doc`.
// seek_block is cheap: it only advances the skip reader, no block decompression.
leader.seek_block(doc);
let leader_block_max: Score = leader.block_max_score();
// Compute the window end as the minimum last_doc_in_block across all scorers.
// This ensures the block_max values are valid for all docs in [doc, window_end].
// Different scorers have independently aligned blocks, so we must use the
// smallest window where all block_max values hold.
let mut window_end: DocId = leader.last_doc_in_block();
let mut secondary_block_max_sum: Score = 0.0;
let num_secondaries = secondaries.len();
for (idx, secondary) in secondaries.iter_mut().enumerate() {
secondary.block_cursor().seek_block(doc);
if !secondary.block_cursor().has_remaining_docs() {
return;
}
window_end = window_end.min(secondary.last_doc_in_block());
let bms = secondary.block_max_score();
secondary_block_max_scores[idx] = bms;
secondary_block_max_sum += bms;
}
if leader_block_max + secondary_block_max_sum <= threshold {
// The entire window cannot beat the threshold. Skip past it.
doc = window_end + 1;
continue;
}
// --- Phase 2: Batch processing within the window ---
//
// Score-first approach: decode the leader's block, filter by threshold,
// then check intersection membership only for survivors. This avoids expensive
// secondary seeks for docs that can't beat the threshold.
let block_cursor = leader.block_cursor();
// seek loads the block and returns the in-block index of the first doc >= `doc`.
let start_idx = block_cursor.seek(doc);
// Use the branchless binary search on the doc decoder to find the first
// index past window_end.
let end_idx = block_cursor
.doc_decoder
.seek_within_block(window_end + 1)
.min(block_cursor.block_len());
let block_docs = &block_cursor.doc_decoder.output_array()[start_idx..end_idx];
let block_freqs = &block_cursor.freq_output_array()[start_idx..end_idx];
// Pass 1: Batch-compute leader BM25 scores and branchlessly filter
// candidates that can't beat the threshold.
//
// The trick: always write to the buffer at `num_candidates`, then
// conditionally advance the count. The compiler can turn this into
// a cmov instead of a branch, avoiding misprediction costs.
let score_threshold = threshold - secondary_block_max_sum;
let mut candidate_doc_ids = [0u32; COMPRESSION_BLOCK_SIZE];
let mut candidate_scores = [0.0f32; COMPRESSION_BLOCK_SIZE];
let mut num_candidates = 0usize;
for (candidate_doc, term_freq) in
block_docs.iter().copied().zip(block_freqs.iter().copied())
{
let fieldnorm_id = fieldnorm_reader.fieldnorm_id(candidate_doc);
let leader_score = bm25_weight.score(fieldnorm_id, term_freq);
candidate_doc_ids[num_candidates] = candidate_doc;
candidate_scores[num_candidates] = leader_score;
num_candidates += (leader_score > score_threshold) as usize;
}
// Precompute suffix sums: suffix[i] = sum of block_max for secondaries[i+1..].
// Used in Phase 2 to prune candidates that can't beat threshold even with
// remaining secondaries contributing their block_max.
if num_candidates == 0 {
doc = window_end + 1;
continue;
}
let mut running = 0.0f32;
for idx in (0..num_secondaries).rev() {
secondary_suffix_block_max[idx] = running;
running += secondary_block_max_scores[idx];
}
// Pass 2: Check intersection membership only for survivors.
// score_threshold may be stale (threshold can increase from callbacks),
// but that's conservative — we may check a few extra candidates, never miss one.
'next_candidate: for candidate_idx in 0..num_candidates {
let candidate_doc = candidate_doc_ids[candidate_idx];
let mut total_score: Score = candidate_scores[candidate_idx];
for (secondary_idx, secondary) in secondaries.iter_mut().enumerate() {
// If a previous candidate already advanced this secondary past
// candidate_doc, the candidate can't be in the intersection.
if secondary.doc() > candidate_doc {
continue 'next_candidate;
}
let seek_result = secondary.seek(candidate_doc);
if seek_result != candidate_doc {
continue 'next_candidate;
}
total_score += secondary.score();
// Prune: even if all remaining secondaries score at their block max,
// can we still beat the threshold?
if total_score + secondary_suffix_block_max[secondary_idx] <= threshold {
continue 'next_candidate;
}
}
// All secondaries matched.
if total_score > threshold {
threshold = callback(candidate_doc, total_score);
if leader_max_score + secondaries_global_max_sum <= threshold {
return;
}
}
}
doc = window_end + 1;
}
}
#[cfg(test)]
mod tests {
use std::cmp::Ordering;
use std::collections::BinaryHeap;
use proptest::prelude::*;
use crate::query::term_query::TermScorer;
use crate::query::{Bm25Weight, Scorer};
use crate::{DocId, DocSet, Score, TERMINATED};
struct Float(Score);
impl Eq for Float {}
impl PartialEq for Float {
fn eq(&self, other: &Self) -> bool {
self.cmp(other) == Ordering::Equal
}
}
impl PartialOrd for Float {
fn partial_cmp(&self, other: &Self) -> Option<Ordering> {
Some(self.cmp(other))
}
}
impl Ord for Float {
fn cmp(&self, other: &Self) -> Ordering {
other.0.partial_cmp(&self.0).unwrap_or(Ordering::Equal)
}
}
fn nearly_equals(left: Score, right: Score) -> bool {
(left - right).abs() < 0.0001 * (left + right).abs()
}
/// Run block_wand_intersection and collect (doc, score) pairs above threshold.
fn compute_checkpoints_block_wand_intersection(
term_scorers: Vec<TermScorer>,
top_k: usize,
) -> Vec<(DocId, Score)> {
let mut heap: BinaryHeap<Float> = BinaryHeap::with_capacity(top_k);
let mut checkpoints: Vec<(DocId, Score)> = Vec::new();
let mut limit: Score = 0.0;
let callback = &mut |doc, score| {
heap.push(Float(score));
if heap.len() > top_k {
heap.pop().unwrap();
}
if heap.len() == top_k {
limit = heap.peek().unwrap().0;
}
if !nearly_equals(score, limit) {
checkpoints.push((doc, score));
}
limit
};
super::block_wand_intersection(term_scorers, Score::MIN, callback);
checkpoints
}
/// Naive baseline: intersect by iterating all docs.
fn compute_checkpoints_naive_intersection(
mut term_scorers: Vec<TermScorer>,
top_k: usize,
) -> Vec<(DocId, Score)> {
let mut heap: BinaryHeap<Float> = BinaryHeap::with_capacity(top_k);
let mut checkpoints: Vec<(DocId, Score)> = Vec::new();
let mut limit = Score::MIN;
// Sort by cost to use the cheapest as driver.
term_scorers.sort_by_key(|s| s.cost());
let (leader, secondaries) = term_scorers.split_first_mut().unwrap();
let mut doc = leader.doc();
while doc != TERMINATED {
let mut all_match = true;
for secondary in secondaries.iter_mut() {
let secondary_doc = secondary.doc();
let seek_result = if secondary_doc <= doc {
secondary.seek(doc)
} else {
secondary_doc
};
if seek_result != doc {
all_match = false;
break;
}
}
if all_match {
let score: Score =
leader.score() + secondaries.iter_mut().map(|s| s.score()).sum::<Score>();
if score > limit {
heap.push(Float(score));
if heap.len() > top_k {
heap.pop().unwrap();
}
if heap.len() == top_k {
limit = heap.peek().unwrap().0;
}
if !nearly_equals(score, limit) {
checkpoints.push((doc, score));
}
}
}
doc = leader.advance();
}
checkpoints
}
const MAX_TERM_FREQ: u32 = 100u32;
fn posting_list(max_doc: u32) -> BoxedStrategy<Vec<(DocId, u32)>> {
(1..max_doc + 1)
.prop_flat_map(move |doc_freq| {
(
proptest::bits::bitset::sampled(doc_freq as usize, 0..max_doc as usize),
proptest::collection::vec(1u32..MAX_TERM_FREQ, doc_freq as usize),
)
})
.prop_map(|(docset, term_freqs)| {
docset
.iter()
.map(|doc| doc as u32)
.zip(term_freqs.iter().cloned())
.collect::<Vec<_>>()
})
.boxed()
}
#[expect(clippy::type_complexity)]
fn gen_term_scorers(num_scorers: usize) -> BoxedStrategy<(Vec<Vec<(DocId, u32)>>, Vec<u32>)> {
(1u32..100u32)
.prop_flat_map(move |max_doc: u32| {
(
proptest::collection::vec(posting_list(max_doc), num_scorers),
proptest::collection::vec(2u32..10u32 * MAX_TERM_FREQ, max_doc as usize),
)
})
.boxed()
}
fn test_block_wand_intersection_aux(posting_lists: &[Vec<(DocId, u32)>], fieldnorms: &[u32]) {
// Repeat docs 64 times to create multi-block scenarios, matching block_wand.rs test
// strategy.
const REPEAT: usize = 64;
let fieldnorms_expanded: Vec<u32> = fieldnorms
.iter()
.cloned()
.flat_map(|fieldnorm| std::iter::repeat_n(fieldnorm, REPEAT))
.collect();
let postings_lists_expanded: Vec<Vec<(DocId, u32)>> = posting_lists
.iter()
.map(|posting_list| {
posting_list
.iter()
.cloned()
.flat_map(|(doc, term_freq)| {
(0_u32..REPEAT as u32).map(move |offset| {
(
doc * (REPEAT as u32) + offset,
if offset == 0 { term_freq } else { 1 },
)
})
})
.collect::<Vec<(DocId, u32)>>()
})
.collect();
let total_fieldnorms: u64 = fieldnorms_expanded
.iter()
.cloned()
.map(|fieldnorm| fieldnorm as u64)
.sum();
let average_fieldnorm = (total_fieldnorms as Score) / (fieldnorms_expanded.len() as Score);
let max_doc = fieldnorms_expanded.len();
let make_scorers = || -> Vec<TermScorer> {
postings_lists_expanded
.iter()
.map(|postings| {
let bm25_weight = Bm25Weight::for_one_term(
postings.len() as u64,
max_doc as u64,
average_fieldnorm,
);
TermScorer::create_for_test(postings, &fieldnorms_expanded[..], bm25_weight)
})
.collect()
};
for top_k in 1..4 {
let checkpoints_optimized =
compute_checkpoints_block_wand_intersection(make_scorers(), top_k);
let checkpoints_naive = compute_checkpoints_naive_intersection(make_scorers(), top_k);
assert_eq!(
checkpoints_optimized.len(),
checkpoints_naive.len(),
"Mismatch in checkpoint count for top_k={top_k}"
);
for (&(left_doc, left_score), &(right_doc, right_score)) in
checkpoints_optimized.iter().zip(checkpoints_naive.iter())
{
assert_eq!(left_doc, right_doc);
assert!(
nearly_equals(left_score, right_score),
"Score mismatch for doc {left_doc}: {left_score} vs {right_score}"
);
}
}
}
proptest! {
#![proptest_config(ProptestConfig::with_cases(500))]
#[test]
fn test_block_wand_intersection_two_scorers(
(posting_lists, fieldnorms) in gen_term_scorers(2)
) {
test_block_wand_intersection_aux(&posting_lists[..], &fieldnorms[..]);
}
}
proptest! {
#![proptest_config(ProptestConfig::with_cases(500))]
#[test]
fn test_block_wand_intersection_three_scorers(
(posting_lists, fieldnorms) in gen_term_scorers(3)
) {
test_block_wand_intersection_aux(&posting_lists[..], &fieldnorms[..]);
}
}
#[test]
fn test_block_wand_intersection_disjoint() {
// Two posting lists with no overlap — intersection is empty.
let fieldnorms: Vec<u32> = vec![10; 200];
let average_fieldnorm = 10.0;
let postings_a: Vec<(DocId, u32)> = (0..100).map(|d| (d, 1)).collect();
let postings_b: Vec<(DocId, u32)> = (100..200).map(|d| (d, 1)).collect();
let scorer_a = TermScorer::create_for_test(
&postings_a,
&fieldnorms,
Bm25Weight::for_one_term(100, 200, average_fieldnorm),
);
let scorer_b = TermScorer::create_for_test(
&postings_b,
&fieldnorms,
Bm25Weight::for_one_term(100, 200, average_fieldnorm),
);
let checkpoints = compute_checkpoints_block_wand_intersection(vec![scorer_a, scorer_b], 10);
assert!(checkpoints.is_empty());
}
#[test]
fn test_block_wand_intersection_all_overlap() {
// Two posting lists with full overlap.
let fieldnorms: Vec<u32> = vec![10; 50];
let average_fieldnorm = 10.0;
let postings: Vec<(DocId, u32)> = (0..50).map(|d| (d, 3)).collect();
let make_scorer = || {
TermScorer::create_for_test(
&postings,
&fieldnorms,
Bm25Weight::for_one_term(50, 50, average_fieldnorm),
)
};
let checkpoints_opt =
compute_checkpoints_block_wand_intersection(vec![make_scorer(), make_scorer()], 5);
let checkpoints_naive =
compute_checkpoints_naive_intersection(vec![make_scorer(), make_scorer()], 5);
assert_eq!(checkpoints_opt.len(), checkpoints_naive.len());
}
}

30
src/query/boolean_query/block_wand.rs → src/query/boolean_query/block_wand_union.rs
View File

@@ -50,7 +50,7 @@ fn block_max_was_too_low_advance_one_scorer(
scorers: &mut [TermScorerWithMaxScore],
pivot_len: usize,
) {
debug_assert!(is_sorted(scorers.iter().map(|scorer| scorer.doc())));
debug_assert!(scorers.iter().map(|scorer| scorer.doc()).is_sorted());
let mut scorer_to_seek = pivot_len - 1;
let mut global_max_score = scorers[scorer_to_seek].max_score;
let mut doc_to_seek_after = scorers[scorer_to_seek].last_doc_in_block();
@@ -76,7 +76,7 @@ fn block_max_was_too_low_advance_one_scorer(
scorers[scorer_to_seek].seek(doc_to_seek_after);
restore_ordering(scorers, scorer_to_seek);
debug_assert!(is_sorted(scorers.iter().map(|scorer| scorer.doc())));
debug_assert!(scorers.iter().map(|scorer| scorer.doc()).is_sorted());
}
// Given a list of term_scorers and a `ord` and assuming that `term_scorers[ord]` is sorted
@@ -90,7 +90,7 @@ fn restore_ordering(term_scorers: &mut [TermScorerWithMaxScore], ord: usize) {
}
term_scorers.swap(i, i - 1);
}
debug_assert!(is_sorted(term_scorers.iter().map(|scorer| scorer.doc())));
debug_assert!(term_scorers.iter().map(|scorer| scorer.doc()).is_sorted());
}
// Attempts to advance all term_scorers between `&term_scorers[0..before_len]` to the pivot.
@@ -150,17 +150,21 @@ pub fn block_wand(
mut threshold: Score,
callback: &mut dyn FnMut(u32, Score) -> Score,
) {
scorers.retain(|scorer| scorer.doc() < TERMINATED);
if scorers.len() == 1 {
let scorer = scorers.pop().unwrap();
return block_wand_single_scorer(scorer, threshold, callback);
}
let mut scorers: Vec<TermScorerWithMaxScore> = scorers
.iter_mut()
.map(TermScorerWithMaxScore::from)
.collect();
scorers.sort_by_key(|scorer| scorer.doc());
// At this point we need to ensure that the scorers are sorted!
debug_assert!(is_sorted(scorers.iter().map(|scorer| scorer.doc())));
scorers.sort_by_key(|scorer| scorer.doc());
while let Some((before_pivot_len, pivot_len, pivot_doc)) =
find_pivot_doc(&scorers[..], threshold)
{
debug_assert!(is_sorted(scorers.iter().map(|scorer| scorer.doc())));
debug_assert!(scorers.iter().map(|scorer| scorer.doc()).is_sorted());
debug_assert_ne!(pivot_doc, TERMINATED);
debug_assert!(before_pivot_len < pivot_len);
@@ -228,7 +232,7 @@ pub fn block_wand_single_scorer(
loop {
// We position the scorer on a block that can reach
// the threshold.
while scorer.block_max_score() < threshold {
while scorer.block_max_score() <= threshold {
let last_doc_in_block = scorer.last_doc_in_block();
if last_doc_in_block == TERMINATED {
return;
@@ -286,18 +290,6 @@ impl DerefMut for TermScorerWithMaxScore<'_> {
}
}
fn is_sorted<I: Iterator<Item = DocId>>(mut it: I) -> bool {
if let Some(first) = it.next() {
let mut prev = first;
for doc in it {
if doc < prev {
return false;
}
prev = doc;
}
}
true
}
#[cfg(test)]
mod tests {
use std::cmp::Ordering;

94
src/query/boolean_query/boolean_weight.rs
View File

@@ -16,6 +16,7 @@ use crate::{DocId, Score};
enum SpecializedScorer {
TermUnion(Vec<TermScorer>),
TermIntersection(Vec<TermScorer>),
Other(Box<dyn Scorer>),
}
@@ -49,10 +50,9 @@ where
TScoreCombiner: ScoreCombiner,
{
assert!(!scorers.is_empty());
if scorers.len() == 1 {
if scorers.len() == 1 && !scorers[0].is::<TermScorer>() {
return SpecializedScorer::Other(scorers.into_iter().next().unwrap()); //< we checked the size beforehand
}
{
let is_all_term_queries = scorers.iter().all(|scorer| scorer.is::<TermScorer>());
if is_all_term_queries {
@@ -66,6 +66,9 @@ where
{
// Block wand is only available if we read frequencies.
return SpecializedScorer::TermUnion(scorers);
} else if scorers.len() == 1 {
// Single TermScorer without freq reading — unwrap directly.
return SpecializedScorer::Other(Box::new(scorers.into_iter().next().unwrap()));
} else {
return SpecializedScorer::Other(Box::new(BufferedUnionScorer::build(
scorers,
@@ -93,6 +96,13 @@ fn into_box_scorer<TScoreCombiner: ScoreCombiner>(
BufferedUnionScorer::build(term_scorers, score_combiner_fn, num_docs);
Box::new(union_scorer)
}
SpecializedScorer::TermIntersection(term_scorers) => {
let boxed_scorers: Vec<Box<dyn Scorer>> = term_scorers
.into_iter()
.map(|s| Box::new(s) as Box<dyn Scorer>)
.collect();
intersect_scorers(boxed_scorers, num_docs)
}
SpecializedScorer::Other(scorer) => scorer,
}
}
@@ -297,14 +307,43 @@ impl<TScoreCombiner: ScoreCombiner> BooleanWeight<TScoreCombiner> {
// Result depends entirely on MUST + any removed AllScorers.
let combined_all_scorer_count = must_special_scorer_counts.num_all_scorers
+ should_special_scorer_counts.num_all_scorers;
let boxed_scorer: Box<dyn Scorer> = effective_must_scorer(
must_scorers,
combined_all_scorer_count,
reader.max_doc(),
num_docs,
)
.unwrap_or_else(|| Box::new(EmptyScorer));
SpecializedScorer::Other(boxed_scorer)
// Try to detect a pure TermScorer intersection for block-max optimization.
// Preconditions: no removed AllScorers, at least 2 scorers, all TermScorer
// with frequency reading enabled.
if combined_all_scorer_count == 0
&& must_scorers.len() >= 2
&& must_scorers.iter().all(|s| s.is::<TermScorer>())
{
let term_scorers: Vec<TermScorer> = must_scorers
.into_iter()
.map(|s| *(s.downcast::<TermScorer>().map_err(|_| ()).unwrap()))
.collect();
if term_scorers
.iter()
.all(|s| s.freq_reading_option() == FreqReadingOption::ReadFreq)
{
SpecializedScorer::TermIntersection(term_scorers)
} else {
let must_scorers: Vec<Box<dyn Scorer>> = term_scorers
.into_iter()
.map(|s| Box::new(s) as Box<dyn Scorer>)
.collect();
let boxed_scorer: Box<dyn Scorer> =
effective_must_scorer(must_scorers, 0, reader.max_doc(), num_docs)
.unwrap_or_else(|| Box::new(EmptyScorer));
SpecializedScorer::Other(boxed_scorer)
}
} else {
let boxed_scorer: Box<dyn Scorer> = effective_must_scorer(
must_scorers,
combined_all_scorer_count,
reader.max_doc(),
num_docs,
)
.unwrap_or_else(|| Box::new(EmptyScorer));
SpecializedScorer::Other(boxed_scorer)
}
}
(ShouldScorersCombinationMethod::Optional(should_scorer), must_scorers) => {
// Optional SHOULD: contributes to scoring but not required for matching.
@@ -463,15 +502,21 @@ impl<TScoreCombiner: ScoreCombiner + Sync> Weight for BooleanWeight<TScoreCombin
callback: &mut dyn FnMut(DocId, Score),
) -> crate::Result<()> {
let scorer = self.complex_scorer(reader, 1.0, &self.score_combiner_fn)?;
let num_docs = reader.num_docs();
match scorer {
SpecializedScorer::TermUnion(term_scorers) => {
let mut union_scorer = BufferedUnionScorer::build(
term_scorers,
&self.score_combiner_fn,
reader.num_docs(),
);
let mut union_scorer =
BufferedUnionScorer::build(term_scorers, &self.score_combiner_fn, num_docs);
for_each_scorer(&mut union_scorer, callback);
}
SpecializedScorer::TermIntersection(term_scorers) => {
let boxed_scorers: Vec<Box<dyn Scorer>> = term_scorers
.into_iter()
.map(|term_scorer| Box::new(term_scorer) as Box<dyn Scorer>)
.collect();
let mut intersection = intersect_scorers(boxed_scorers, num_docs);
for_each_scorer(intersection.as_mut(), callback);
}
SpecializedScorer::Other(mut scorer) => {
for_each_scorer(scorer.as_mut(), callback);
}
@@ -485,17 +530,23 @@ impl<TScoreCombiner: ScoreCombiner + Sync> Weight for BooleanWeight<TScoreCombin
callback: &mut dyn FnMut(&[DocId]),
) -> crate::Result<()> {
let scorer = self.complex_scorer(reader, 1.0, || DoNothingCombiner)?;
let num_docs = reader.num_docs();
let mut buffer = [0u32; COLLECT_BLOCK_BUFFER_LEN];
match scorer {
SpecializedScorer::TermUnion(term_scorers) => {
let mut union_scorer = BufferedUnionScorer::build(
term_scorers,
&self.score_combiner_fn,
reader.num_docs(),
);
let mut union_scorer =
BufferedUnionScorer::build(term_scorers, &self.score_combiner_fn, num_docs);
for_each_docset_buffered(&mut union_scorer, &mut buffer, callback);
}
SpecializedScorer::TermIntersection(term_scorers) => {
let boxed_scorers: Vec<Box<dyn Scorer>> = term_scorers
.into_iter()
.map(|term_scorer| Box::new(term_scorer) as Box<dyn Scorer>)
.collect();
let mut intersection = intersect_scorers(boxed_scorers, num_docs);
for_each_docset_buffered(intersection.as_mut(), &mut buffer, callback);
}
SpecializedScorer::Other(mut scorer) => {
for_each_docset_buffered(scorer.as_mut(), &mut buffer, callback);
}
@@ -524,6 +575,9 @@ impl<TScoreCombiner: ScoreCombiner + Sync> Weight for BooleanWeight<TScoreCombin
SpecializedScorer::TermUnion(term_scorers) => {
super::block_wand(term_scorers, threshold, callback);
}
SpecializedScorer::TermIntersection(term_scorers) => {
super::block_wand_intersection(term_scorers, threshold, callback);
}
SpecializedScorer::Other(mut scorer) => {
for_each_pruning_scorer(scorer.as_mut(), threshold, callback);
}

6
src/query/boolean_query/mod.rs
View File

@@ -1,8 +1,10 @@
mod block_wand;
mod block_wand_intersection;
mod block_wand_union;
mod boolean_query;
mod boolean_weight;
pub(crate) use self::block_wand::{block_wand, block_wand_single_scorer};
pub(crate) use self::block_wand_intersection::block_wand_intersection;
pub(crate) use self::block_wand_union::{block_wand, block_wand_single_scorer};
pub use self::boolean_query::BooleanQuery;
pub use self::boolean_weight::BooleanWeight;

182
src/query/intersection.rs
View File

@@ -1,5 +1,7 @@
use common::TinySet;
use super::size_hint::estimate_intersection;
use crate::docset::{DocSet, SeekDangerResult, TERMINATED};
use crate::docset::{DocSet, SeekDangerResult, BLOCK_NUM_TINYBITSETS, TERMINATED};
use crate::query::term_query::TermScorer;
use crate::query::{EmptyScorer, Scorer};
use crate::{DocId, Score};
@@ -17,7 +19,7 @@ use crate::{DocId, Score};
/// `size_hint` of the intersection.
pub fn intersect_scorers(
mut scorers: Vec<Box<dyn Scorer>>,
num_docs_segment: u32,
segment_num_docs: u32,
) -> Box<dyn Scorer> {
if scorers.is_empty() {
return Box::new(EmptyScorer);
@@ -42,14 +44,14 @@ pub fn intersect_scorers(
left: *(left.downcast::<TermScorer>().map_err(|_| ()).unwrap()),
right: *(right.downcast::<TermScorer>().map_err(|_| ()).unwrap()),
others: scorers,
num_docs: num_docs_segment,
segment_num_docs,
});
}
Box::new(Intersection {
left,
right,
others: scorers,
num_docs: num_docs_segment,
segment_num_docs,
})
}
@@ -58,7 +60,7 @@ pub struct Intersection<TDocSet: DocSet, TOtherDocSet: DocSet = Box<dyn Scorer>>
left: TDocSet,
right: TDocSet,
others: Vec<TOtherDocSet>,
num_docs: u32,
segment_num_docs: u32,
}
fn go_to_first_doc<TDocSet: DocSet>(docsets: &mut [TDocSet]) -> DocId {
@@ -78,7 +80,10 @@ fn go_to_first_doc<TDocSet: DocSet>(docsets: &mut [TDocSet]) -> DocId {
impl<TDocSet: DocSet> Intersection<TDocSet, TDocSet> {
/// num_docs is the number of documents in the segment.
pub(crate) fn new(mut docsets: Vec<TDocSet>, num_docs: u32) -> Intersection<TDocSet, TDocSet> {
pub(crate) fn new(
mut docsets: Vec<TDocSet>,
segment_num_docs: u32,
) -> Intersection<TDocSet, TDocSet> {
let num_docsets = docsets.len();
assert!(num_docsets >= 2);
docsets.sort_by_key(|docset| docset.cost());
@@ -97,7 +102,7 @@ impl<TDocSet: DocSet> Intersection<TDocSet, TDocSet> {
left,
right,
others: docsets,
num_docs,
segment_num_docs,
}
}
}
@@ -214,7 +219,7 @@ impl<TDocSet: DocSet, TOtherDocSet: DocSet> DocSet for Intersection<TDocSet, TOt
[self.left.size_hint(), self.right.size_hint()]
.into_iter()
.chain(self.others.iter().map(DocSet::size_hint)),
self.num_docs,
self.segment_num_docs,
)
}
@@ -224,6 +229,91 @@ impl<TDocSet: DocSet, TOtherDocSet: DocSet> DocSet for Intersection<TDocSet, TOt
// If there are docsets that are bad at skipping, they should also influence the cost.
self.left.cost()
}
fn count_including_deleted(&mut self) -> u32 {
const DENSITY_THRESHOLD_INVERSE: u32 = 32;
if self
.left
.size_hint()
.saturating_mul(DENSITY_THRESHOLD_INVERSE)
< self.segment_num_docs
{
// Sparse path: if the lead iterator covers less than ~3% of docs,
// the block approach wastes time on mostly-empty blocks.
self.count_including_deleted_sparse()
} else {
// Dense approach. We push documents into a block bitset to then
// perform count using popcount.
self.count_including_deleted_dense()
}
}
}
const EMPTY_BLOCK: [TinySet; BLOCK_NUM_TINYBITSETS] = [TinySet::EMPTY; BLOCK_NUM_TINYBITSETS];
/// ANDs `other` into `mask` in-place. Returns `true` if the result is all zeros.
#[inline]
fn and_blocks_and_return_is_empty(
mask: &mut [TinySet; BLOCK_NUM_TINYBITSETS],
update: &[TinySet; BLOCK_NUM_TINYBITSETS],
) -> bool {
let mut all_empty = true;
for (mask_tinyset, update_tinyset) in mask.iter_mut().zip(update.iter()) {
*mask_tinyset = mask_tinyset.intersect(*update_tinyset);
all_empty &= mask_tinyset.is_empty();
}
all_empty
}
impl<TDocSet: DocSet, TOtherDocSet: DocSet> Intersection<TDocSet, TOtherDocSet> {
fn count_including_deleted_sparse(&mut self) -> u32 {
let mut count = 0u32;
let mut doc = self.doc();
while doc != TERMINATED {
count += 1;
doc = self.advance();
}
count
}
/// Dense block-wise bitmask intersection count.
///
/// Fills a 1024-doc window from each iterator, ANDs the bitmasks together,
/// and popcounts the result. `fill_bitset_block` handles seeking tails forward
/// when they lag behind the current block.
fn count_including_deleted_dense(&mut self) -> u32 {
let mut count = 0u32;
let mut next_base = self.left.doc();
while next_base < TERMINATED {
let base = next_base;
// Fill lead bitmask.
let mut mask = EMPTY_BLOCK;
next_base = next_base.max(self.left.fill_bitset_block(base, &mut mask));
let mut tail_mask = EMPTY_BLOCK;
next_base = next_base.max(self.right.fill_bitset_block(base, &mut tail_mask));
if and_blocks_and_return_is_empty(&mut mask, &tail_mask) {
continue;
}
// AND with each additional tail.
for other in &mut self.others {
let mut other_mask = EMPTY_BLOCK;
next_base = next_base.max(other.fill_bitset_block(base, &mut other_mask));
if and_blocks_and_return_is_empty(&mut mask, &other_mask) {
continue;
}
}
for tinyset in &mask {
count += tinyset.len();
}
}
count
}
}
impl<TScorer, TOtherScorer> Scorer for Intersection<TScorer, TOtherScorer>
@@ -421,6 +511,82 @@ mod tests {
}
}
proptest! {
#[test]
fn prop_test_count_including_deleted_matches_default(
a in sorted_deduped_vec(1200, 400),
b in sorted_deduped_vec(1200, 400),
c in sorted_deduped_vec(1200, 400),
num_docs in 1200u32..2000u32,
) {
// Compute expected count via set intersection.
let expected: u32 = a.iter()
.filter(|doc| b.contains(doc) && c.contains(doc))
.count() as u32;
// Test count_including_deleted (dense path).
let make_intersection = || {
Intersection::new(
vec![
VecDocSet::from(a.clone()),
VecDocSet::from(b.clone()),
VecDocSet::from(c.clone()),
],
num_docs,
)
};
let mut intersection = make_intersection();
let count = intersection.count_including_deleted();
prop_assert_eq!(count, expected,
"count_including_deleted mismatch: a={:?}, b={:?}, c={:?}", a, b, c);
}
}
#[test]
fn test_count_including_deleted_two_way() {
let left = VecDocSet::from(vec![1, 3, 9]);
let right = VecDocSet::from(vec![3, 4, 9, 18]);
let mut intersection = Intersection::new(vec![left, right], 100);
assert_eq!(intersection.count_including_deleted(), 2);
}
#[test]
fn test_count_including_deleted_empty() {
let a = VecDocSet::from(vec![1, 3]);
let b = VecDocSet::from(vec![1, 4]);
let c = VecDocSet::from(vec![3, 9]);
let mut intersection = Intersection::new(vec![a, b, c], 100);
assert_eq!(intersection.count_including_deleted(), 0);
}
/// Test with enough documents to exercise the dense path (>= num_docs/32).
#[test]
fn test_count_including_deleted_dense_path() {
// Create dense docsets: many docs relative to segment size.
let docs_a: Vec<u32> = (0..2000).step_by(2).collect(); // even numbers 0..2000
let docs_b: Vec<u32> = (0..2000).step_by(3).collect(); // multiples of 3
let expected = docs_a.iter().filter(|d| *d % 3 == 0).count() as u32;
let a = VecDocSet::from(docs_a);
let b = VecDocSet::from(docs_b);
let mut intersection = Intersection::new(vec![a, b], 2000);
assert_eq!(intersection.count_including_deleted(), expected);
}
/// Test that spans multiple blocks (>1024 docs).
#[test]
fn test_count_including_deleted_multi_block() {
let docs_a: Vec<u32> = (0..5000).collect();
let docs_b: Vec<u32> = (0..5000).step_by(7).collect();
let expected = docs_b.len() as u32; // all of b is in a
let a = VecDocSet::from(docs_a);
let b = VecDocSet::from(docs_b);
let mut intersection = Intersection::new(vec![a, b], 5000);
assert_eq!(intersection.count_including_deleted(), expected);
}
#[test]
fn test_bug_2811_intersection_candidate_should_increase() {
let mut schema_builder = Schema::builder();

23
src/query/term_query/term_scorer.rs
View File

@@ -1,6 +1,6 @@
use crate::docset::DocSet;
use crate::fieldnorm::FieldNormReader;
use crate::postings::{FreqReadingOption, Postings, SegmentPostings};
use crate::postings::{BlockSegmentPostings, FreqReadingOption, Postings, SegmentPostings};
use crate::query::bm25::Bm25Weight;
use crate::query::{Explanation, Scorer};
use crate::{DocId, Score};
@@ -95,6 +95,21 @@ impl TermScorer {
pub fn last_doc_in_block(&self) -> DocId {
self.postings.block_cursor.skip_reader().last_doc_in_block()
}
/// Returns a mutable reference to the underlying block cursor.
pub(crate) fn block_cursor(&mut self) -> &mut BlockSegmentPostings {
&mut self.postings.block_cursor
}
/// Returns a reference to the fieldnorm reader for batch lookups.
pub(crate) fn fieldnorm_reader(&self) -> &FieldNormReader {
&self.fieldnorm_reader
}
/// Returns a reference to the BM25 weight for batch score computation.
pub(crate) fn bm25_weight(&self) -> &Bm25Weight {
&self.similarity_weight
}
}
impl DocSet for TermScorer {
@@ -117,6 +132,12 @@ impl DocSet for TermScorer {
fn size_hint(&self) -> u32 {
self.postings.size_hint()
}
// TODO
// It is probably possible to optimize fill_bitset_block for TermScorer,
// working directly with the blocks, enabling vectorization.
// I did not manage to get a performance improvement on Mac ARM,
// and do not have access to x86 to investigate.
}
impl Scorer for TermScorer {

8
src/store/index/skip_index.rs
View File

@@ -94,13 +94,7 @@ impl SkipIndex {
byte_range: 0..first_layer_len,
};
for layer in &self.layers {
if let Some(checkpoint) =
layer.seek_start_at_offset(target, cur_checkpoint.byte_range.start)
{
cur_checkpoint = checkpoint;
} else {
return None;
}
cur_checkpoint = layer.seek_start_at_offset(target, cur_checkpoint.byte_range.start)?;
}
Some(cur_checkpoint)
}

2
sstable/Cargo.toml
View File

@@ -23,7 +23,7 @@ zstd-compression = ["zstd"]
[dev-dependencies]
proptest = "1"
criterion = { version = "0.5", default-features = false }
criterion = { version = "0.8", default-features = false }
names = "0.14"
rand = "0.9"

152
sstable/src/dictionary.rs
View File

@@ -14,11 +14,8 @@ use itertools::Itertools;
use tantivy_fst::Automaton;
use tantivy_fst::automaton::AlwaysMatch;
use crate::sstable_index_v3::SSTableIndexV3Empty;
use crate::streamer::{Streamer, StreamerBuilder};
use crate::{
BlockAddr, DeltaReader, Reader, SSTable, SSTableIndex, SSTableIndexV3, TermOrdinal, VoidSSTable,
};
use crate::{BlockAddr, DeltaReader, Reader, SSTable, SSTableIndex, TermOrdinal, VoidSSTable};
/// An SSTable is a sorted map that associates sorted `&[u8]` keys
/// to any kind of typed values.
@@ -288,33 +285,7 @@ impl<TSSTable: SSTable> Dictionary<TSSTable> {
let (sstable_slice, index_slice) = main_slice.split(index_offset as usize);
let sstable_index_bytes = index_slice.read_bytes()?;
let sstable_index = match version {
2 => SSTableIndex::V2(
crate::sstable_index_v2::SSTableIndex::load(sstable_index_bytes).map_err(|_| {
io::Error::new(io::ErrorKind::InvalidData, "SSTable corruption")
})?,
),
3 => {
let (sstable_index_bytes, mut footerv3_len_bytes) = sstable_index_bytes.rsplit(8);
let store_offset = u64::deserialize(&mut footerv3_len_bytes)?;
if store_offset != 0 {
SSTableIndex::V3(
SSTableIndexV3::load(sstable_index_bytes, store_offset).map_err(|_| {
io::Error::new(io::ErrorKind::InvalidData, "SSTable corruption")
})?,
)
} else {
// if store_offset is zero, there is no index, so we build a pseudo-index
// assuming a single block of sstable covering everything.
SSTableIndex::V3Empty(SSTableIndexV3Empty::load(index_offset as usize))
}
}
_ => {
return Err(io::Error::other(format!(
"Unsupported sstable version, expected one of [2, 3], found {version}"
)));
}
};
let sstable_index = SSTableIndex::open(version, index_offset, sstable_index_bytes)?;
Ok(Dictionary {
sstable_slice,
@@ -512,21 +483,28 @@ impl<TSSTable: SSTable> Dictionary<TSSTable> {
/// Returns the terms for a _sorted_ list of term ordinals.
///
/// Returns true if and only if all terms have been found.
pub fn sorted_ords_to_term_cb<F: FnMut(&[u8]) -> io::Result<()>>(
pub fn sorted_ords_to_term_cb(
&self,
mut ords: impl Iterator<Item = TermOrdinal>,
mut cb: F,
ords: &[TermOrdinal],
mut cb: impl FnMut(&[u8]),
) -> io::Result<bool> {
assert!(ords.is_sorted());
let mut ords = ords.iter().copied();
let Some(mut ord) = ords.next() else {
return Ok(true);
};
// Open the block for the first ordinal.
let mut bytes = Vec::new();
let mut current_block_addr = self.sstable_index.get_block_with_ord(ord);
let (mut current_block_addr, block_id) = self.sstable_index.get_and_locate_with_ord(ord);
let mut current_sstable_delta_reader =
self.sstable_delta_reader_block(current_block_addr.clone())?;
let mut current_block_ordinal = current_block_addr.first_ordinal;
let mut current_block_end_bound = self
.sstable_index
.get_block(block_id + 1)
.map(|block_addr| block_addr.first_ordinal)
.unwrap_or(u64::MAX);
loop {
// move to the ord inside the current block
@@ -538,33 +516,38 @@ impl<TSSTable: SSTable> Dictionary<TSSTable> {
bytes.extend_from_slice(current_sstable_delta_reader.suffix());
current_block_ordinal += 1;
}
cb(&bytes)?;
cb(&bytes);
// fetch the next ordinal
let Some(next_ord) = ords.next() else {
return Ok(true);
let next_ord = loop {
let Some(next_ord) = ords.next() else {
return Ok(true);
};
if next_ord == ord {
// This is the same ordinal, let's just call the callback directly.
cb(&bytes);
} else {
// we checked it was sorted beforehands
debug_assert!(next_ord > ord);
break next_ord;
}
};
// advance forward if the new ord is different than the one we just processed
//
// this allows the input TermOrdinal iterator to contain duplicates, so long as it's
// still sorted
if next_ord < ord {
panic!("Ordinals were not sorted: received {next_ord} after {ord}");
} else if next_ord > ord {
// check if block changed for new term_ord
let new_block_addr = self.sstable_index.get_block_with_ord(next_ord);
if new_block_addr != current_block_addr {
current_block_addr = new_block_addr;
current_block_ordinal = current_block_addr.first_ordinal;
current_sstable_delta_reader =
self.sstable_delta_reader_block(current_block_addr.clone())?;
bytes.clear();
}
ord = next_ord;
} else {
// The next ord is equal to the previous ord: no need to seek or advance.
if next_ord >= current_block_end_bound {
let (new_block_addr, block_id) =
self.sstable_index.get_and_locate_with_ord(next_ord);
current_block_addr = new_block_addr;
current_block_ordinal = current_block_addr.first_ordinal;
current_sstable_delta_reader =
self.sstable_delta_reader_block(current_block_addr.clone())?;
bytes.clear();
current_block_end_bound = self
.sstable_index
.get_block(block_id + 1)
.map(|block_addr| block_addr.first_ordinal)
.unwrap_or(u64::MAX)
}
ord = next_ord;
}
}
@@ -671,8 +654,8 @@ mod tests {
use common::OwnedBytes;
use super::Dictionary;
use crate::MonotonicU64SSTable;
use crate::dictionary::TermOrdHit;
use crate::{MonotonicU64SSTable, TermOrdinal};
#[derive(Debug)]
struct PermissionedHandle {
@@ -935,25 +918,24 @@ mod tests {
}
#[test]
fn test_ords_term() {
fn test_sorted_ords_to_term() {
let (dic, _slice) = make_test_sstable();
// Single term
let mut terms = Vec::new();
assert!(
dic.sorted_ords_to_term_cb(100_000..100_001, |term| {
dic.sorted_ords_to_term_cb(&[100_000], |term| {
terms.push(term.to_vec());
Ok(())
})
.unwrap()
);
assert_eq!(terms, vec![format!("{:05X}", 100_000).into_bytes(),]);
// Single term
let mut terms = Vec::new();
let ords: Vec<TermOrdinal> = (100_001..100_002).collect();
assert!(
dic.sorted_ords_to_term_cb(100_001..100_002, |term| {
dic.sorted_ords_to_term_cb(&ords, |term| {
terms.push(term.to_vec());
Ok(())
})
.unwrap()
);
@@ -961,9 +943,8 @@ mod tests {
// both terms
let mut terms = Vec::new();
assert!(
dic.sorted_ords_to_term_cb(100_000..100_002, |term| {
dic.sorted_ords_to_term_cb(&[100_000, 100_001], |term| {
terms.push(term.to_vec());
Ok(())
})
.unwrap()
);
@@ -976,10 +957,10 @@ mod tests {
);
// Test cross block
let mut terms = Vec::new();
let ords: Vec<TermOrdinal> = (98653..=98655).collect();
assert!(
dic.sorted_ords_to_term_cb(98653..=98655, |term| {
dic.sorted_ords_to_term_cb(&ords, |term| {
terms.push(term.to_vec());
Ok(())
})
.unwrap()
);
@@ -991,6 +972,43 @@ mod tests {
format!("{:05X}", 98655).into_bytes(),
]
);
// redundant
let mut terms = Vec::new();
let ords: Vec<TermOrdinal> = vec![1, 1, 2];
assert!(
dic.sorted_ords_to_term_cb(&ords, |term| {
terms.push(term.to_vec());
})
.unwrap()
);
assert_eq!(
terms,
vec![
format!("{:05X}", 1).into_bytes(),
format!("{:05X}", 1).into_bytes(),
format!("{:05X}", 2).into_bytes(),
]
);
// redundant cross block
let mut terms = Vec::new();
let ords: Vec<TermOrdinal> = vec![98653, 98653, 98654, 98654, 98655, 98655];
assert!(
dic.sorted_ords_to_term_cb(&ords, |term| {
terms.push(term.to_vec());
})
.unwrap()
);
assert_eq!(
terms,
vec![
format!("{:05X}", 98_653).into_bytes(),
format!("{:05X}", 98_653).into_bytes(),
format!("{:05X}", 98_654).into_bytes(),
format!("{:05X}", 98_654).into_bytes(),
format!("{:05X}", 98_655).into_bytes(),
format!("{:05X}", 98_655).into_bytes(),
]
);
}
#[test]

319
sstable/src/index/mod.rs Normal file
View File

@@ -0,0 +1,319 @@
pub(crate) mod v2;
pub(crate) mod v3;
use std::io::{self, Read, Write};
use std::ops::Range;
use common::{BinarySerializable, FixedSize, OwnedBytes};
use tantivy_fst::{Automaton, MapBuilder};
use crate::{TermOrdinal, common_prefix_len};
#[derive(Debug, Clone)]
pub enum SSTableIndex {
V2(v2::SSTableIndex),
V3(v3::SSTableIndexV3),
V3Empty(v3::SSTableIndexV3Empty),
}
impl SSTableIndex {
pub(crate) fn open(
version: u32,
index_offset: u64,
index_bytes: OwnedBytes,
) -> io::Result<Self> {
let index = match version {
2 => {
SSTableIndex::V2(v2::SSTableIndex::load(index_bytes).map_err(|_| {
io::Error::new(io::ErrorKind::InvalidData, "SSTable corruption")
})?)
}
3 => {
let (index_bytes, mut footerv3_len_bytes) = index_bytes.rsplit(8);
let store_offset = u64::deserialize(&mut footerv3_len_bytes)?;
if store_offset != 0 {
SSTableIndex::V3(v3::SSTableIndexV3::load(index_bytes, store_offset).map_err(
|_| io::Error::new(io::ErrorKind::InvalidData, "SSTable corruption"),
)?)
} else {
// if store_offset is zero, there is no index, so we build a pseudo-index
// assuming a single block of sstable covering everything.
SSTableIndex::V3Empty(v3::SSTableIndexV3Empty::load(index_offset as usize))
}
}
_ => {
return Err(io::Error::other(format!(
"Unsupported sstable version, expected one of [2, 3], found {version}"
)));
}
};
Ok(index)
}
/// Get the [`BlockAddr`] of the requested block.
pub(crate) fn get_block(&self, block_id: u64) -> Option<BlockAddr> {
match self {
SSTableIndex::V2(v2_index) => v2_index.get_block(block_id as usize),
SSTableIndex::V3(v3_index) => v3_index.get_block(block_id),
SSTableIndex::V3Empty(v3_empty) => v3_empty.get_block(block_id),
}
}
/// Get the block id of the block that would contain `key`.
///
/// Returns None if `key` is lexicographically after the last key recorded.
pub(crate) fn locate_with_key(&self, key: &[u8]) -> Option<u64> {
match self {
SSTableIndex::V2(v2_index) => v2_index.locate_with_key(key).map(|i| i as u64),
SSTableIndex::V3(v3_index) => v3_index.locate_with_key(key),
SSTableIndex::V3Empty(v3_empty) => v3_empty.locate_with_key(key),
}
}
/// Get the [`BlockAddr`] of the block that would contain `key`.
///
/// Returns None if `key` is lexicographically after the last key recorded.
pub fn get_block_with_key(&self, key: &[u8]) -> Option<BlockAddr> {
match self {
SSTableIndex::V2(v2_index) => v2_index.get_block_with_key(key),
SSTableIndex::V3(v3_index) => v3_index.get_block_with_key(key),
SSTableIndex::V3Empty(v3_empty) => v3_empty.get_block_with_key(key),
}
}
pub(crate) fn locate_with_ord(&self, ord: TermOrdinal) -> u64 {
match self {
SSTableIndex::V2(v2_index) => v2_index.locate_with_ord(ord) as u64,
SSTableIndex::V3(v3_index) => v3_index.locate_with_ord(ord),
SSTableIndex::V3Empty(v3_empty) => v3_empty.locate_with_ord(ord),
}
}
/// Get the [`BlockAddr`] of the block containing the `ord`-th term.
pub(crate) fn get_block_with_ord(&self, ord: TermOrdinal) -> BlockAddr {
match self {
SSTableIndex::V2(v2_index) => v2_index.get_block_with_ord(ord),
SSTableIndex::V3(v3_index) => v3_index.get_block_with_ord(ord),
SSTableIndex::V3Empty(v3_empty) => v3_empty.get_block_with_ord(ord),
}
}
pub(crate) fn get_and_locate_with_ord(&self, ord: TermOrdinal) -> (BlockAddr, u64) {
match self {
SSTableIndex::V2(v2_index) => v2_index.get_and_locate_with_ord(ord),
SSTableIndex::V3(v3_index) => v3_index.get_and_locate_with_ord(ord),
SSTableIndex::V3Empty(v3_empty) => v3_empty.get_and_locate_with_ord(ord),
}
}
pub fn get_block_for_automaton<'a>(
&'a self,
automaton: &'a impl Automaton,
) -> impl Iterator<Item = (u64, BlockAddr)> + 'a {
match self {
SSTableIndex::V2(v2_index) => {
BlockIter::V2(v2_index.get_block_for_automaton(automaton))
}
SSTableIndex::V3(v3_index) => {
BlockIter::V3(v3_index.get_block_for_automaton(automaton))
}
SSTableIndex::V3Empty(v3_empty) => {
BlockIter::V3Empty(std::iter::once((0, v3_empty.block_addr.clone())))
}
}
}
}
enum BlockIter<V2, V3, T> {
V2(V2),
V3(V3),
V3Empty(std::iter::Once<T>),
}
impl<V2: Iterator<Item = T>, V3: Iterator<Item = T>, T> Iterator for BlockIter<V2, V3, T> {
type Item = T;
fn next(&mut self) -> Option<Self::Item> {
match self {
BlockIter::V2(v2) => v2.next(),
BlockIter::V3(v3) => v3.next(),
BlockIter::V3Empty(once) => once.next(),
}
}
}
#[derive(Clone, Eq, PartialEq, Debug)]
pub struct BlockAddr {
pub first_ordinal: u64,
pub byte_range: Range<usize>,
}
impl BlockAddr {
fn to_block_start(&self) -> BlockStartAddr {
BlockStartAddr {
first_ordinal: self.first_ordinal,
byte_range_start: self.byte_range.start,
}
}
}
#[derive(Debug, Clone, PartialEq, Eq)]
struct BlockStartAddr {
first_ordinal: u64,
byte_range_start: usize,
}
impl BlockStartAddr {
fn to_block_addr(&self, byte_range_end: usize) -> BlockAddr {
BlockAddr {
first_ordinal: self.first_ordinal,
byte_range: self.byte_range_start..byte_range_end,
}
}
}
#[derive(Debug, Clone)]
pub(crate) struct BlockMeta {
/// Any byte string that is lexicographically greater or equal to
/// the last key in the block,
/// and yet strictly smaller than the first key in the next block.
pub last_key_or_greater: Vec<u8>,
pub block_addr: BlockAddr,
}
impl BinarySerializable for BlockStartAddr {
fn serialize<W: Write + ?Sized>(&self, writer: &mut W) -> io::Result<()> {
let start = self.byte_range_start as u64;
start.serialize(writer)?;
self.first_ordinal.serialize(writer)
}
fn deserialize<R: Read>(reader: &mut R) -> io::Result<Self> {
let byte_range_start = u64::deserialize(reader)? as usize;
let first_ordinal = u64::deserialize(reader)?;
Ok(BlockStartAddr {
first_ordinal,
byte_range_start,
})
}
// Provided method
fn num_bytes(&self) -> u64 {
BlockStartAddr::SIZE_IN_BYTES as u64
}
}
impl FixedSize for BlockStartAddr {
const SIZE_IN_BYTES: usize = 2 * u64::SIZE_IN_BYTES;
}
/// Given that left < right,
/// mutates `left into a shorter byte string left'` that
/// matches `left <= left' < right`.
fn find_shorter_str_in_between(left: &mut Vec<u8>, right: &[u8]) {
assert!(&left[..] < right);
let common_len = common_prefix_len(left, right);
if left.len() == common_len {
return;
}
// It is possible to do one character shorter in some case,
// but it is not worth the extra complexity
for pos in (common_len + 1)..left.len() {
if left[pos] != u8::MAX {
left[pos] += 1;
left.truncate(pos + 1);
return;
}
}
}
#[derive(Default)]
pub struct SSTableIndexBuilder {
blocks: Vec<BlockMeta>,
}
impl SSTableIndexBuilder {
/// In order to make the index as light as possible, we
/// try to find a shorter alternative to the last key of the last block
/// that is still smaller than the next key.
pub(crate) fn shorten_last_block_key_given_next_key(&mut self, next_key: &[u8]) {
if let Some(last_block) = self.blocks.last_mut() {
find_shorter_str_in_between(&mut last_block.last_key_or_greater, next_key);
}
}
pub fn add_block(&mut self, last_key: &[u8], byte_range: Range<usize>, first_ordinal: u64) {
self.blocks.push(BlockMeta {
last_key_or_greater: last_key.to_vec(),
block_addr: BlockAddr {
byte_range,
first_ordinal,
},
})
}
pub fn serialize<W: std::io::Write>(&self, wrt: W) -> io::Result<u64> {
if self.blocks.len() <= 1 {
return Ok(0);
}
let counting_writer = common::CountingWriter::wrap(wrt);
let mut map_builder = MapBuilder::new(counting_writer).map_err(fst_error_to_io_error)?;
for (i, block) in self.blocks.iter().enumerate() {
map_builder
.insert(&block.last_key_or_greater, i as u64)
.map_err(fst_error_to_io_error)?;
}
let counting_writer = map_builder.into_inner().map_err(fst_error_to_io_error)?;
let written_bytes = counting_writer.written_bytes();
let mut wrt = counting_writer.finish();
let mut block_store_writer = v3::BlockAddrStoreWriter::new();
for block in &self.blocks {
block_store_writer.write_block_meta(block.block_addr.clone())?;
}
block_store_writer.serialize(&mut wrt)?;
Ok(written_bytes)
}
}
fn fst_error_to_io_error(error: tantivy_fst::Error) -> io::Error {
match error {
tantivy_fst::Error::Fst(fst_error) => io::Error::other(fst_error),
tantivy_fst::Error::Io(ioerror) => ioerror,
}
}
#[cfg(test)]
mod tests {
#[track_caller]
fn test_find_shorter_str_in_between_aux(left: &[u8], right: &[u8]) {
let mut left_buf = left.to_vec();
super::find_shorter_str_in_between(&mut left_buf, right);
assert!(left_buf.len() <= left.len());
assert!(left <= &left_buf);
assert!(&left_buf[..] < right);
}
#[test]
fn test_find_shorter_str_in_between() {
test_find_shorter_str_in_between_aux(b"", b"hello");
test_find_shorter_str_in_between_aux(b"abc", b"abcd");
test_find_shorter_str_in_between_aux(b"abcd", b"abd");
test_find_shorter_str_in_between_aux(&[0, 0, 0], &[1]);
test_find_shorter_str_in_between_aux(&[0, 0, 0], &[0, 0, 1]);
test_find_shorter_str_in_between_aux(&[0, 0, 255, 255, 255, 0u8], &[0, 1]);
}
use proptest::prelude::*;
proptest! {
#![proptest_config(ProptestConfig::with_cases(100))]
#[test]
fn test_proptest_find_shorter_str(left in any::<Vec<u8>>(), right in any::<Vec<u8>>()) {
if left < right {
test_find_shorter_str_in_between_aux(&left, &right);
}
}
}
}

7
sstable/src/sstable_index_v2.rs → sstable/src/index/v2.rs
View File

@@ -77,6 +77,13 @@ impl SSTableIndex {
self.get_block(self.locate_with_ord(ord)).unwrap()
}
pub(crate) fn get_and_locate_with_ord(&self, ord: TermOrdinal) -> (BlockAddr, u64) {
let location = self.locate_with_ord(ord);
// locate_with_ord always returns an index within range
let block_addr = self.get_block(location).unwrap();
(block_addr, location as u64)
}
pub(crate) fn get_block_for_automaton<'a>(
&'a self,
automaton: &'a impl Automaton,

294
sstable/src/sstable_index_v3.rs → sstable/src/index/v3.rs
View File

@@ -1,106 +1,14 @@
use std::io::{self, Read, Write};
use std::ops::Range;
use std::sync::Arc;
use common::{BinarySerializable, FixedSize, OwnedBytes};
use tantivy_bitpacker::{BitPacker, compute_num_bits};
use tantivy_fst::raw::Fst;
use tantivy_fst::{Automaton, IntoStreamer, Map, MapBuilder, Streamer};
use tantivy_fst::{Automaton, IntoStreamer, Map, Streamer};
use super::{BlockAddr, BlockStartAddr};
use crate::block_match_automaton::can_block_match_automaton;
use crate::{SSTableDataCorruption, TermOrdinal, common_prefix_len};
#[derive(Debug, Clone)]
pub enum SSTableIndex {
V2(crate::sstable_index_v2::SSTableIndex),
V3(SSTableIndexV3),
V3Empty(SSTableIndexV3Empty),
}
impl SSTableIndex {
/// Get the [`BlockAddr`] of the requested block.
pub(crate) fn get_block(&self, block_id: u64) -> Option<BlockAddr> {
match self {
SSTableIndex::V2(v2_index) => v2_index.get_block(block_id as usize),
SSTableIndex::V3(v3_index) => v3_index.get_block(block_id),
SSTableIndex::V3Empty(v3_empty) => v3_empty.get_block(block_id),
}
}
/// Get the block id of the block that would contain `key`.
///
/// Returns None if `key` is lexicographically after the last key recorded.
pub(crate) fn locate_with_key(&self, key: &[u8]) -> Option<u64> {
match self {
SSTableIndex::V2(v2_index) => v2_index.locate_with_key(key).map(|i| i as u64),
SSTableIndex::V3(v3_index) => v3_index.locate_with_key(key),
SSTableIndex::V3Empty(v3_empty) => v3_empty.locate_with_key(key),
}
}
/// Get the [`BlockAddr`] of the block that would contain `key`.
///
/// Returns None if `key` is lexicographically after the last key recorded.
pub fn get_block_with_key(&self, key: &[u8]) -> Option<BlockAddr> {
match self {
SSTableIndex::V2(v2_index) => v2_index.get_block_with_key(key),
SSTableIndex::V3(v3_index) => v3_index.get_block_with_key(key),
SSTableIndex::V3Empty(v3_empty) => v3_empty.get_block_with_key(key),
}
}
pub(crate) fn locate_with_ord(&self, ord: TermOrdinal) -> u64 {
match self {
SSTableIndex::V2(v2_index) => v2_index.locate_with_ord(ord) as u64,
SSTableIndex::V3(v3_index) => v3_index.locate_with_ord(ord),
SSTableIndex::V3Empty(v3_empty) => v3_empty.locate_with_ord(ord),
}
}
/// Get the [`BlockAddr`] of the block containing the `ord`-th term.
pub(crate) fn get_block_with_ord(&self, ord: TermOrdinal) -> BlockAddr {
match self {
SSTableIndex::V2(v2_index) => v2_index.get_block_with_ord(ord),
SSTableIndex::V3(v3_index) => v3_index.get_block_with_ord(ord),
SSTableIndex::V3Empty(v3_empty) => v3_empty.get_block_with_ord(ord),
}
}
pub fn get_block_for_automaton<'a>(
&'a self,
automaton: &'a impl Automaton,
) -> impl Iterator<Item = (u64, BlockAddr)> + 'a {
match self {
SSTableIndex::V2(v2_index) => {
BlockIter::V2(v2_index.get_block_for_automaton(automaton))
}
SSTableIndex::V3(v3_index) => {
BlockIter::V3(v3_index.get_block_for_automaton(automaton))
}
SSTableIndex::V3Empty(v3_empty) => {
BlockIter::V3Empty(std::iter::once((0, v3_empty.block_addr.clone())))
}
}
}
}
enum BlockIter<V2, V3, T> {
V2(V2),
V3(V3),
V3Empty(std::iter::Once<T>),
}
impl<V2: Iterator<Item = T>, V3: Iterator<Item = T>, T> Iterator for BlockIter<V2, V3, T> {
type Item = T;
fn next(&mut self) -> Option<Self::Item> {
match self {
BlockIter::V2(v2) => v2.next(),
BlockIter::V3(v3) => v3.next(),
BlockIter::V3Empty(once) => once.next(),
}
}
}
use crate::{SSTableDataCorruption, TermOrdinal};
#[derive(Debug, Clone)]
pub struct SSTableIndexV3 {
@@ -160,6 +68,11 @@ impl SSTableIndexV3 {
self.block_addr_store.binary_search_ord(ord).1
}
pub(crate) fn get_and_locate_with_ord(&self, ord: TermOrdinal) -> (BlockAddr, u64) {
let (location, block_addr) = self.block_addr_store.binary_search_ord(ord);
(block_addr, location)
}
pub(crate) fn get_block_for_automaton<'a>(
&'a self,
automaton: &'a impl Automaton,
@@ -216,7 +129,7 @@ impl<A: Automaton> Iterator for GetBlockForAutomaton<'_, A> {
#[derive(Debug, Clone)]
pub struct SSTableIndexV3Empty {
block_addr: BlockAddr,
pub block_addr: BlockAddr,
}
impl SSTableIndexV3Empty {
@@ -230,8 +143,8 @@ impl SSTableIndexV3Empty {
}
/// Get the [`BlockAddr`] of the requested block.
pub(crate) fn get_block(&self, _block_id: u64) -> Option<BlockAddr> {
Some(self.block_addr.clone())
pub(crate) fn get_block(&self, block_id: u64) -> Option<BlockAddr> {
(block_id == 0).then(|| self.block_addr.clone())
}
/// Get the block id of the block that would contain `key`.
@@ -256,146 +169,9 @@ impl SSTableIndexV3Empty {
pub(crate) fn get_block_with_ord(&self, _ord: TermOrdinal) -> BlockAddr {
self.block_addr.clone()
}
}
#[derive(Clone, Eq, PartialEq, Debug)]
pub struct BlockAddr {
pub first_ordinal: u64,
pub byte_range: Range<usize>,
}
impl BlockAddr {
fn to_block_start(&self) -> BlockStartAddr {
BlockStartAddr {
first_ordinal: self.first_ordinal,
byte_range_start: self.byte_range.start,
}
}
}
#[derive(Debug, Clone, PartialEq, Eq)]
struct BlockStartAddr {
first_ordinal: u64,
byte_range_start: usize,
}
impl BlockStartAddr {
fn to_block_addr(&self, byte_range_end: usize) -> BlockAddr {
BlockAddr {
first_ordinal: self.first_ordinal,
byte_range: self.byte_range_start..byte_range_end,
}
}
}
#[derive(Debug, Clone)]
pub(crate) struct BlockMeta {
/// Any byte string that is lexicographically greater or equal to
/// the last key in the block,
/// and yet strictly smaller than the first key in the next block.
pub last_key_or_greater: Vec<u8>,
pub block_addr: BlockAddr,
}
impl BinarySerializable for BlockStartAddr {
fn serialize<W: Write + ?Sized>(&self, writer: &mut W) -> io::Result<()> {
let start = self.byte_range_start as u64;
start.serialize(writer)?;
self.first_ordinal.serialize(writer)
}
fn deserialize<R: Read>(reader: &mut R) -> io::Result<Self> {
let byte_range_start = u64::deserialize(reader)? as usize;
let first_ordinal = u64::deserialize(reader)?;
Ok(BlockStartAddr {
first_ordinal,
byte_range_start,
})
}
// Provided method
fn num_bytes(&self) -> u64 {
BlockStartAddr::SIZE_IN_BYTES as u64
}
}
impl FixedSize for BlockStartAddr {
const SIZE_IN_BYTES: usize = 2 * u64::SIZE_IN_BYTES;
}
/// Given that left < right,
/// mutates `left into a shorter byte string left'` that
/// matches `left <= left' < right`.
fn find_shorter_str_in_between(left: &mut Vec<u8>, right: &[u8]) {
assert!(&left[..] < right);
let common_len = common_prefix_len(left, right);
if left.len() == common_len {
return;
}
// It is possible to do one character shorter in some case,
// but it is not worth the extra complexity
for pos in (common_len + 1)..left.len() {
if left[pos] != u8::MAX {
left[pos] += 1;
left.truncate(pos + 1);
return;
}
}
}
#[derive(Default)]
pub struct SSTableIndexBuilder {
blocks: Vec<BlockMeta>,
}
impl SSTableIndexBuilder {
/// In order to make the index as light as possible, we
/// try to find a shorter alternative to the last key of the last block
/// that is still smaller than the next key.
pub(crate) fn shorten_last_block_key_given_next_key(&mut self, next_key: &[u8]) {
if let Some(last_block) = self.blocks.last_mut() {
find_shorter_str_in_between(&mut last_block.last_key_or_greater, next_key);
}
}
pub fn add_block(&mut self, last_key: &[u8], byte_range: Range<usize>, first_ordinal: u64) {
self.blocks.push(BlockMeta {
last_key_or_greater: last_key.to_vec(),
block_addr: BlockAddr {
byte_range,
first_ordinal,
},
})
}
pub fn serialize<W: std::io::Write>(&self, wrt: W) -> io::Result<u64> {
if self.blocks.len() <= 1 {
return Ok(0);
}
let counting_writer = common::CountingWriter::wrap(wrt);
let mut map_builder = MapBuilder::new(counting_writer).map_err(fst_error_to_io_error)?;
for (i, block) in self.blocks.iter().enumerate() {
map_builder
.insert(&block.last_key_or_greater, i as u64)
.map_err(fst_error_to_io_error)?;
}
let counting_writer = map_builder.into_inner().map_err(fst_error_to_io_error)?;
let written_bytes = counting_writer.written_bytes();
let mut wrt = counting_writer.finish();
let mut block_store_writer = BlockAddrStoreWriter::new();
for block in &self.blocks {
block_store_writer.write_block_meta(block.block_addr.clone())?;
}
block_store_writer.serialize(&mut wrt)?;
Ok(written_bytes)
}
}
fn fst_error_to_io_error(error: tantivy_fst::Error) -> io::Error {
match error {
tantivy_fst::Error::Fst(fst_error) => io::Error::other(fst_error),
tantivy_fst::Error::Io(ioerror) => ioerror,
pub(crate) fn get_and_locate_with_ord(&self, _ord: TermOrdinal) -> (BlockAddr, u64) {
(self.block_addr.clone(), 0)
}
}
@@ -647,14 +423,14 @@ fn binary_search(max: u64, cmp_fn: impl Fn(u64) -> std::cmp::Ordering) -> Result
Err(left)
}
struct BlockAddrStoreWriter {
pub(crate) struct BlockAddrStoreWriter {
buffer_block_metas: Vec<u8>,
buffer_addrs: Vec<u8>,
block_addrs: Vec<BlockAddr>,
}
impl BlockAddrStoreWriter {
fn new() -> Self {
pub(crate) fn new() -> Self {
BlockAddrStoreWriter {
buffer_block_metas: Vec::new(),
buffer_addrs: Vec::new(),
@@ -662,7 +438,7 @@ impl BlockAddrStoreWriter {
}
}
fn flush_block(&mut self) -> io::Result<()> {
pub(crate) fn flush_block(&mut self) -> io::Result<()> {
if self.block_addrs.is_empty() {
return Ok(());
}
@@ -741,7 +517,7 @@ impl BlockAddrStoreWriter {
Ok(())
}
fn write_block_meta(&mut self, block_addr: BlockAddr) -> io::Result<()> {
pub(crate) fn write_block_meta(&mut self, block_addr: BlockAddr) -> io::Result<()> {
self.block_addrs.push(block_addr);
if self.block_addrs.len() >= STORE_BLOCK_LEN {
self.flush_block()?;
@@ -749,7 +525,7 @@ impl BlockAddrStoreWriter {
Ok(())
}
fn serialize<W: std::io::Write>(&mut self, wrt: &mut W) -> io::Result<()> {
pub(crate) fn serialize<W: std::io::Write>(&mut self, wrt: &mut W) -> io::Result<()> {
self.flush_block()?;
let len = self.buffer_block_metas.len() as u64;
len.serialize(wrt)?;
@@ -824,8 +600,9 @@ mod tests {
use common::OwnedBytes;
use super::*;
use crate::SSTableDataCorruption;
use crate::block_match_automaton::tests::EqBuffer;
use crate::index::BlockMeta;
use crate::{SSTableDataCorruption, SSTableIndexBuilder};
#[test]
fn test_sstable_index() {
@@ -874,36 +651,7 @@ mod tests {
assert!(matches!(data_corruption_err, SSTableDataCorruption));
}
#[track_caller]
fn test_find_shorter_str_in_between_aux(left: &[u8], right: &[u8]) {
let mut left_buf = left.to_vec();
super::find_shorter_str_in_between(&mut left_buf, right);
assert!(left_buf.len() <= left.len());
assert!(left <= &left_buf);
assert!(&left_buf[..] < right);
}
#[test]
fn test_find_shorter_str_in_between() {
test_find_shorter_str_in_between_aux(b"", b"hello");
test_find_shorter_str_in_between_aux(b"abc", b"abcd");
test_find_shorter_str_in_between_aux(b"abcd", b"abd");
test_find_shorter_str_in_between_aux(&[0, 0, 0], &[1]);
test_find_shorter_str_in_between_aux(&[0, 0, 0], &[0, 0, 1]);
test_find_shorter_str_in_between_aux(&[0, 0, 255, 255, 255, 0u8], &[0, 1]);
}
use proptest::prelude::*;
proptest! {
#![proptest_config(ProptestConfig::with_cases(100))]
#[test]
fn test_proptest_find_shorter_str(left in any::<Vec<u8>>(), right in any::<Vec<u8>>()) {
if left < right {
test_find_shorter_str_in_between_aux(&left, &right);
}
}
}
// use proptest::prelude::*;
#[test]
fn test_find_best_slop() {

5
sstable/src/lib.rs
View File

@@ -47,9 +47,8 @@ pub mod merge;
mod streamer;
pub mod value;
mod sstable_index_v3;
pub use sstable_index_v3::{BlockAddr, SSTableIndex, SSTableIndexBuilder, SSTableIndexV3};
mod sstable_index_v2;
mod index;
pub use index::{BlockAddr, SSTableIndex, SSTableIndexBuilder};
pub(crate) mod vint;
pub use dictionary::{Dictionary, TermOrdHit};
pub use streamer::{Streamer, StreamerBuilder};

4
stacker/Cargo.toml
View File

@@ -8,7 +8,7 @@ repository = "https://github.com/quickwit-oss/tantivy"
description = "term hashmap used for indexing"
[dependencies]
murmurhash32 = "0.3"
murmurhash32 = "0.4"
common = { version = "0.11", path = "../common/", package = "tantivy-common" }
ahash = { version = "0.8.11", default-features = false, optional = true }
@@ -27,7 +27,7 @@ rand = "0.9"
zipf = "7.0.0"
rustc-hash = "2.1.0"
proptest = "1.2.0"
binggan = { version = "0.15.3" }
binggan = { version = "0.17.0" }
rand_distr = "0.5"
[features]