rust/tantivy
1
0
Fork 0
mirror of https://github.com/quickwit-oss/tantivy.git synced 2025-12-28 04:52:55 +00:00
Code Issues Packages Projects Releases Wiki Activity

Compare commits

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

94 Commits
0.22.0 ... test_parse

Author SHA1 Message Date
Pascal Seitz
4a9262cd2c accept * as field name 2024-12-06 09:42:18 +01:00
PSeitz
876a579e5d queryparser: add field respecification test (#2550) 2024-12-02 14:17:12 +01:00
PSeitz
4c52499622 clippy (#2549) 2024-11-29 16:08:21 +08:00
PSeitz
52d4e81e70 update CHANGELOG (#2546) 2024-11-27 20:49:35 +08:00
dependabot[bot]
c71ea7b2ef Update thiserror requirement from 1.0.30 to 2.0.1 (#2542) 
Updates the requirements on [thiserror](https://github.com/dtolnay/thiserror) to permit the latest version.
- [Release notes](https://github.com/dtolnay/thiserror/releases)
- [Commits](https://github.com/dtolnay/thiserror/compare/1.0.30...2.0.1)

---
updated-dependencies:
- dependency-name: thiserror
  dependency-type: direct:production
...

Signed-off-by: dependabot[bot] <support@github.com>
Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com>
 2024-11-09 08:08:34 +08:00
Paul Masurel
c35a782747 Updating rustc-hash and clippy fixes (#2532) 
* Updating rustc-hash and clippy fixes

* fix terms_aggregation_min_doc_count_special_case

---------

Co-authored-by: Pascal Seitz <pascal.seitz@gmail.com>
 2024-11-01 13:46:26 +08:00
dependabot[bot]
c66af2c0a9 Update binggan requirement from 0.12.0 to 0.14.0 (#2530) 
* Update binggan requirement from 0.12.0 to 0.14.0

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

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

* fix build

---------

Signed-off-by: dependabot[bot] <support@github.com>
Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com>
Co-authored-by: Pascal Seitz <pascal.seitz@gmail.com>
 2024-10-24 09:41:35 +08:00
Joan Antoni RE
f9ac055847 Fix some links in architecture docs (#2528) 2024-10-23 21:06:54 +09:00
PSeitz
21d057059e clippy (#2527) 
* clippy

* clippy

* clippy

* clippy

* convert allow to expect and remove unused

* cargo fmt

* cleanup

* export sample

* clippy
 2024-10-22 09:26:54 +08:00
PSeitz
dca508b4ca remove read_postings_no_deletes (#2526) 
closes #2525
 2024-10-22 09:52:43 +09:00
PSeitz
aebae9965d add RegexPhraseQuery (#2516) 
* add RegexPhraseQuery

RegexPhraseQuery supports phrase queries with regex. It supports regex
and wildcards. E.g. a query with wildcards:
"b* b* wolf" matches "big bad wolf"
Slop is supported as well:
"b* wolf"~2 matches "big bad wolf"

Regex queries may match a lot of terms where we still need to
keep track which term hit to load the positions.
The phrase query algorithm groups terms by their frequency
together in the union to prefilter groups early.

This PR comes with some new datastructures:

SimpleUnion - A union docset for a list of docsets. It doesn't do any
caching and is therefore well suited for datasets with lots of skipping.
(phrase search, but intersections in general)

LoadedPostings - Like SegmentPostings, but all docs and positions are loaded in
memory. SegmentPostings uses 1840 bytes per instance with its caches,
which is equivalent to 460 docids.
LoadedPostings is used for terms which have less than 100 docs.
LoadedPostings is only used to reduce memory consumption.

BitSetPostingUnion - Creates a `Posting` that uses the bitset for docid
hits and the docsets for positions. The BitSet is the precalculated
union of the docsets
In the RegexPhraseQuery there is a size limit of 512 docsets per PreAggregatedUnion,
before creating a new one.

Renamed Union to BufferedUnionScorer
Added proptests to test different union types.

* cleanup

* use Box instead of Vec

* use RefCell instead of term_freq(&mut)

* remove wildcard mode

* move RefCell to outer

* clippy
 2024-10-21 18:29:17 +08:00
Marvin
e7e3e3f44c make casing in docs more consistent (#2524) 
* make casing in docs more consistent

* more

* lowercase tantivy
 2024-10-21 17:59:41 +09:00
PSeitz
2f2db16ec1 store DateTime as nanoseconds in doc store (#2486) 
* store DateTime as nanoseconds in doc store

The doc store DateTime was truncated to microseconds previously. This
removes this truncation, while still keeping backwards compatibility.

This is done by adding the trait `ConfigurableBinarySerializable`, which
works like `BinarySerializable`, but with a config that allows de/serialize
as different date time precision currently.

bump version format to 7.
add compat test to check the date time truncation.

* remove configurable binary serialize, add enum for doc store version

* test doc store version ord
 2024-10-18 10:50:20 +08:00
Paul Masurel
d152e29687 Fixed citation (#2523) 2024-10-17 10:19:50 +09:00
Paul Masurel
285bcc25c9 Added citation.cff (#2522) 2024-10-17 09:43:35 +09:00
PSeitz
7b65ad922d use binggan for stacker bench (#2492) 
* use binggan for stacker bench

```
alice (num terms: 174693)
hashmap                    Memory: 1.3 MB     Avg: 367.19 MiB/s (-1.34%)    Median: 368.10 MiB/s (-1.34%)    [378.75 MiB/s .. 352.81 MiB/s]
hasmap with postings       Memory: 2.4 MB     Avg: 237.29 MiB/s (-2.19%)    Median: 240.22 MiB/s (-1.61%)    [248.26 MiB/s .. 210.66 MiB/s]
fxhashmap ref postings     Memory: 2.9 MB     Avg: 171.94 MiB/s (-3.22%)    Median: 174.13 MiB/s (-2.69%)    [185.94 MiB/s .. 152.43 MiB/s]
fxhasmap owned postings    Memory: 3.5 MB     Avg: 96.993 MiB/s (-4.20%)    Median: 97.410 MiB/s (-4.48%)    [102.78 MiB/s .. 82.745 MiB/s]
numbers unique 100k
hashmap                 Memory: 5.2 MB     Avg: 334.17 MiB/s (-3.06%)    Median: 352.61 MiB/s (+0.77%)    [362.60 MiB/s .. 213.03 MiB/s]
hasmap with postings    Memory: 6.3 MB     Avg: 316.96 MiB/s (-0.02%)    Median: 325.16 MiB/s (-0.04%)    [338.36 MiB/s .. 218.60 MiB/s]
zipfs numbers 100k
hashmap                 Memory: 1.3 MB     Avg: 1.2342 GiB/s (+2.87%)    Median: 1.2677 GiB/s (+4.66%)    [1.3130 GiB/s .. 915.93 MiB/s]
hasmap with postings    Memory: 2.4 MB     Avg: 485.16 MiB/s (+2.68%)    Median: 494.70 MiB/s (+4.42%)    [505.31 MiB/s .. 413.14 MiB/s]
numbers unique 1mio
hashmap                 Memory: 35.7 MB     Avg: 169.68 MiB/s (-1.08%)    Median: 166.80 MiB/s (-3.87%)    [201.33 MiB/s .. 154.26 MiB/s]
hasmap with postings    Memory: 39.8 MB     Avg: 149.49 MiB/s (-3.07%)    Median: 150.85 MiB/s (-1.45%)    [160.76 MiB/s .. 130.94 MiB/s]
zipfs numbers 1mio
hashmap                 Memory: 1.3 MB     Avg: 1.2185 GiB/s (-2.33%)     Median: 1.2291 GiB/s (-2.33%)     [1.2905 GiB/s .. 1.0742 GiB/s]
hasmap with postings    Memory: 5.5 MB     Avg: 358.43 MiB/s (-11.63%)    Median: 356.95 MiB/s (-12.85%)    [444.94 MiB/s .. 302.46 MiB/s]
numbers unique 2mio
hashmap                 Memory: 70.3 MB     Avg: 163.65 MiB/s (+8.37%)    Median: 162.83 MiB/s (+8.80%)    [190.20 MiB/s .. 144.70 MiB/s]
hasmap with postings    Memory: 78.6 MB     Avg: 148.00 MiB/s (+7.75%)    Median: 151.53 MiB/s (+9.11%)    [166.92 MiB/s .. 120.09 MiB/s]
zipfs numbers 2mio
hashmap                 Memory: 1.3 MB     Avg: 1.2535 GiB/s (+2.59%)    Median: 1.2654 GiB/s (+0.36%)    [1.2938 GiB/s .. 1.0592 GiB/s]
hasmap with postings    Memory: 9.7 MB     Avg: 377.96 MiB/s (-4.94%)    Median: 381.82 MiB/s (-3.67%)    [426.14 MiB/s .. 335.66 MiB/s]
numbers unique 5mio
hashmap                 Memory: 277.9 MB     Avg: 121.30 MiB/s (+2.00%)    Median: 121.99 MiB/s (+2.99%)    [132.51 MiB/s .. 110.32 MiB/s]
hasmap with postings    Memory: 295.7 MB     Avg: 114.23 MiB/s (+2.13%)    Median: 115.26 MiB/s (+2.94%)    [124.08 MiB/s .. 103.38 MiB/s]
zipfs numbers 5mio
hashmap                 Memory: 1.3 MB      Avg: 1.2326 GiB/s (+0.63%)    Median: 1.2400 GiB/s (+0.71%)    [1.2755 GiB/s .. 1.0923 GiB/s]
hasmap with postings    Memory: 25.4 MB     Avg: 360.49 MiB/s (+1.07%)    Median: 363.44 MiB/s (+1.27%)    [404.88 MiB/s .. 300.38 MiB/s]
```

* rename bench

* update binggan

* rename to HASHMAP_CAPACITY
 2024-10-16 11:41:33 +08:00
dependabot[bot]
99be20cedd Update binggan requirement from 0.10.0 to 0.12.0 (#2519) 
* Update binggan requirement from 0.10.0 to 0.12.0

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

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

* fix build

---------

Signed-off-by: dependabot[bot] <support@github.com>
Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com>
Co-authored-by: Pascal Seitz <pascal.seitz@gmail.com>
 2024-10-16 11:36:04 +08:00
Bruce Mitchener
5f026901b8 Update MSRV to 1.75 (#2515) 
This is required by the `fs4` dependency. There are other
things that need something later than 1.66.

Both quickwit and the Python binding already require something
newer.
 2024-10-16 10:32:16 +08:00
baishen
6dfa2df06f fix OwnedBytes debug panic (#2512) 2024-10-16 10:31:40 +08:00
Bruce Mitchener
c17e513377 Reduce typo count. (#2510) 2024-10-10 09:55:37 +08:00
PSeitz
2f5a269c70 update packages (#2500) 
fixes some warnings
 2024-09-25 17:46:18 +08:00
PSeitz
50532260e3 update changelog (#2496) 2024-09-25 10:28:53 +08:00
Tri
8bd6eb06e6 feat: make SegmentMeta.with_max_doc public (#2499) 
* chore: add container

* feat: make max doc editable externally

* chore: expose another method

* chore: remove comments

* remove unused devcontainer

* chore: manually match nightly format

* chore: change weird formating

* revert format change

* fix: format with nightly
 2024-09-23 12:39:36 +08:00
PSeitz
55b0b52457 Fix AggregationLimits (#2495) 
* change AggregationLimits behavior

This fixes an issue encountered with the current behaviour of
AggregationLimits.
Previously we had AggregationLimits and RessourceLimitGuard, which both
track the memory, but only RessourceLimitGuard released memory when
dropped, while AggregationLimits did not.

This PR changes AggregationLimits to be a guard itself and removes the
RessourceLimitGuard.

* rename AggregationLimits to AggregationLimitsGuard
 2024-09-17 14:25:47 +08:00
dependabot[bot]
56fc56c5b9 Update binggan requirement from 0.8.0 to 0.10.0 (#2493) 
* Update binggan requirement from 0.8.0 to 0.10.0

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

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

* update PR

---------

Signed-off-by: dependabot[bot] <support@github.com>
Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com>
Co-authored-by: Pascal Seitz <pascal.seitz@gmail.com>
 2024-09-10 14:26:06 +08:00
trinity-1686a
85395d942a fix clippy lints from 1.80-1.81 (#2488) 
* fix some clippy lints

* fix clippy::doc_lazy_continuation

* fix some lints for 1.82
 2024-09-05 14:33:05 +02:00
PSeitz
a206c3ccd3 add compat tests (#2485) 2024-09-04 18:26:57 +08:00
Chaya
dc5d31c116 grammar and misspellings (#2483) 
* grammar

* grammar

* misspelling
 2024-09-04 12:45:31 +08:00
gezihuzi
95a4ddea3e Fix: Improve collapse_overlapped_ranges function (#2474) 
* Fix: Improve collapse_overlapped_ranges function

- Refactor into separate sort_and_deduplicate_ranges and merge_overlapping_ranges functions
- Enhance sorting to consider both start and end of ranges
- Optimize merging logic to handle adjacent ranges
- Add comprehensive examples in function documentation
- Ensure proper handling of duplicate and unsorted input ranges
- Improve overall efficiency and readability of range collapsing algorithm

* move debug_assert

---------

Co-authored-by: PSeitz <PSeitz@users.noreply.github.com>
 2024-09-04 12:39:13 +08:00
trinity-1686a
ab5125d3dc remove unused trait bounds and outdated doc comment (#2478) 2024-09-03 16:31:51 +02:00
trinity-1686a
9f81d59ecd make find_field_with_default return json fields without path (#2476) 
* make find_field_with_default return json fields without path

* add tests for find_field_with_default
 2024-08-19 15:25:29 +02:00
PSeitz
c71ec8086d add FastFieldRangeQuery, rename (#2477) 
* add FastFieldRangeQuery, rename

* remove Query impl
 2024-08-19 09:02:00 +02:00
PSeitz
27be6aed91 lift clauses in LogicalAst (#2449) 
(a OR b) OR (c OR d) can be simplified to (a OR b OR c OR d)
(a AND b) AND (c AND d) can be simplified to (a AND b AND c AND d)

This directly affects how queries are executed

remove unused SumWithCoordsCombiner
the number of fields is unused and private
 2024-08-14 19:21:26 +02:00
PSeitz
3d1c4b313a support ff range queries on json fields (#2456) 
* support ff range queries on json fields

* fix term date truncation

* use inverted index range query for phrase prefix queries

* rename to InvertedIndexRangeQuery

* fix column filter, add mixed column test
 2024-08-02 00:06:50 +08:00
PSeitz
0d4e319965 add Key::I64 and Key::U64 variants in aggregation (#2468) 
* add Key::I64 and Key::U64 variants in aggregation

Currently all `Key` numerical values are returned as f64. This causes problems in some
cases with the precision and the way f64 is serialized.

This PR adds `Key::I64` and `Key::U64` variants and uses them in the term
aggregation.

* add clarification comment
 2024-07-31 20:29:32 +08:00
PSeitz
75dc3eb298 extend custom order deserialization (#2451) 
allow arrays
improve validation
closes https://github.com/quickwit-oss/tantivy/issues/2435
 2024-07-30 18:36:08 +08:00
PSeitz
3f6d225086 fix potential endless loop in merge (#2457) 
avoid single segments lists without deletes as merge candidates, as they will be moved
to a merge operation and filtered for merging in the next
consider_merge_options call. In rare cases this may end up in a endless
merge loop where only single segments where nothing is to be done are
merged.
 2024-07-30 16:37:20 +08:00
PSeitz
d8843c608c make FastFieldRangeWeight::new pub (#2460) 2024-07-29 10:39:27 +08:00
PSeitz
7ebcc15b17 add support for str fast field range query (#2453) 
* add support for str fast field range query

Add support for range queries on fast fields, by converting term bounds to
term ordinals bounds.

closes https://github.com/quickwit-oss/tantivy/issues/2023

* extend tests, rename

* update comment

* update comment
 2024-07-17 09:31:42 +08:00
PSeitz
1b4076691f refactor fast field query (#2452) 
As preparation of #2023 and #1709

* Use Term to pass parameters
* merge u64 and ip fast field range query

Side note: I did not rename range_query_u64_fastfield, because then git can't track the changes.
 2024-07-15 18:08:05 +08:00
Robert Caulk
eab660873a doc: fix typo in readme (#2450) 2024-07-09 15:12:22 +08:00
PSeitz
232f37126e fix coverage (#2448) 2024-07-05 12:04:18 +08:00
PSeitz
13e9885dfd faster term aggregation fetch terms (#2447) 
big impact for term aggregations with large `size` parameter (e.g. 1000)
add top 1000 term agg bench

full
terms_few                                      Memory: 27.3 KB (+79.09%)    Avg: 3.8058ms (+2.40%)      Median: 3.7192ms (+3.47%)       [3.6224ms .. 4.3721ms]
terms_many                                     Memory: 6.9 MB               Avg: 12.6102ms (-4.70%)     Median: 12.1389ms (-6.58%)      [10.2847ms .. 15.4857ms]
terms_many_top_1000                            Memory: 6.9 MB               Avg: 15.8216ms (-83.19%)    Median: 15.4899ms (-83.46%)     [13.4250ms .. 20.6897ms]
terms_many_order_by_term                       Memory: 6.9 MB               Avg: 14.7820ms (-3.95%)     Median: 14.2236ms (-4.28%)      [12.6669ms .. 21.0968ms]
terms_many_with_top_hits                       Memory: 58.2 MB              Avg: 551.6218ms (+7.18%)    Median: 549.8826ms (+11.01%)    [496.7371ms .. 592.1299ms]
terms_many_with_avg_sub_agg                    Memory: 27.8 MB              Avg: 197.7029ms (+2.66%)    Median: 190.1564ms (+0.64%)     [167.9226ms .. 245.6651ms]
terms_many_json_mixed_type_with_avg_sub_agg    Memory: 42.0 MB (+0.00%)     Avg: 242.0121ms (+0.92%)    Median: 237.7084ms (-2.85%)     [201.9959ms .. 302.2136ms]
terms_few_with_cardinality_agg                 Memory: 10.6 MB              Avg: 122.6036ms (+1.21%)    Median: 119.0033ms (+2.60%)     [109.2859ms .. 161.5858ms]
range_agg_with_term_agg_few                    Memory: 45.4 KB (+39.75%)    Avg: 24.5454ms (+2.14%)     Median: 24.2861ms (+2.44%)      [23.5109ms .. 27.8406ms]
range_agg_with_term_agg_many                   Memory: 6.9 MB               Avg: 56.8049ms (+3.01%)     Median: 50.9706ms (+1.52%)      [41.4517ms .. 90.3934ms]
dense
terms_few                                      Memory: 28.8 KB (+81.74%)    Avg: 8.9092ms (-2.24%)      Median: 8.7143ms (-1.31%)      [8.6148ms .. 10.3868ms]
terms_many                                     Memory: 6.9 MB (-0.00%)      Avg: 17.9604ms (-10.18%)    Median: 17.1552ms (-11.93%)    [14.8979ms .. 26.2779ms]
terms_many_top_1000                            Memory: 6.9 MB               Avg: 21.4963ms (-78.90%)    Median: 21.2924ms (-78.98%)    [18.2033ms .. 28.0087ms]
terms_many_order_by_term                       Memory: 6.9 MB               Avg: 20.4167ms (-9.13%)     Median: 19.5596ms (-11.37%)    [17.5153ms .. 29.5987ms]
terms_many_with_top_hits                       Memory: 58.2 MB              Avg: 518.4474ms (-6.41%)    Median: 514.9180ms (-9.44%)    [471.5550ms .. 579.0220ms]
terms_many_with_avg_sub_agg                    Memory: 27.8 MB              Avg: 263.6702ms (-2.78%)    Median: 260.8775ms (-2.55%)    [239.5754ms .. 304.6669ms]
terms_many_json_mixed_type_with_avg_sub_agg    Memory: 42.0 MB              Avg: 299.9791ms (-2.01%)    Median: 302.2180ms (-3.08%)    [239.2080ms .. 346.3649ms]
terms_few_with_cardinality_agg                 Memory: 10.6 MB              Avg: 136.3303ms (-3.12%)    Median: 132.3831ms (-2.88%)    [123.7564ms .. 164.7914ms]
range_agg_with_term_agg_few                    Memory: 47.1 KB (+37.81%)    Avg: 35.4538ms (+0.66%)     Median: 34.8754ms (-0.56%)     [34.2287ms .. 40.0884ms]
range_agg_with_term_agg_many                   Memory: 6.9 MB               Avg: 72.2269ms (-4.38%)     Median: 66.1174ms (-4.98%)     [55.5125ms .. 124.1622ms]
sparse
terms_few                                      Memory: 27.3 KB (+69.68%)    Avg: 19.6053ms (-1.15%)     Median: 19.4543ms (-0.38%)     [19.3056ms .. 24.0547ms]
terms_many                                     Memory: 1.8 MB               Avg: 21.2886ms (-6.28%)     Median: 21.1287ms (-6.65%)     [20.6640ms .. 24.6144ms]
terms_many_top_1000                            Memory: 2.6 MB               Avg: 23.4869ms (-85.53%)    Median: 23.3393ms (-85.61%)    [22.7789ms .. 25.0896ms]
terms_many_order_by_term                       Memory: 1.8 MB               Avg: 21.7437ms (-7.78%)     Median: 21.6272ms (-7.66%)     [21.0409ms .. 23.6517ms]
terms_many_with_top_hits                       Memory: 13.1 MB              Avg: 43.7926ms (-2.76%)     Median: 44.3602ms (+0.01%)     [37.8039ms .. 51.0451ms]
terms_many_with_avg_sub_agg                    Memory: 7.5 MB               Avg: 34.6307ms (+3.72%)     Median: 33.4522ms (+1.16%)     [32.4418ms .. 41.4196ms]
terms_many_json_mixed_type_with_avg_sub_agg    Memory: 7.4 MB               Avg: 46.4318ms (+1.16%)     Median: 46.4050ms (+2.03%)     [44.5986ms .. 48.5142ms]
terms_few_with_cardinality_agg                 Memory: 680.0 KB (-0.04%)    Avg: 35.4410ms (+2.05%)     Median: 35.1384ms (+1.19%)     [34.4402ms .. 39.1082ms]
range_agg_with_term_agg_few                    Memory: 45.7 KB (+39.44%)    Avg: 22.7760ms (+0.44%)     Median: 22.5152ms (-0.35%)     [22.3078ms .. 26.1567ms]
range_agg_with_term_agg_many                   Memory: 1.8 MB               Avg: 25.7696ms (-4.45%)     Median: 25.4009ms (-5.61%)     [24.7874ms .. 29.6434ms]
multivalue
terms_few                                      Memory: 244.4 KB            Avg: 15.1253ms (-2.85%)     Median: 15.0988ms (-0.54%)     [14.8790ms .. 15.8193ms]
terms_many                                     Memory: 6.9 MB (-0.00%)     Avg: 26.3019ms (-6.24%)     Median: 26.3662ms (-4.94%)     [21.3553ms .. 31.0564ms]
terms_many_top_1000                            Memory: 6.9 MB              Avg: 29.5212ms (-72.90%)    Median: 29.4257ms (-72.84%)    [24.2645ms .. 35.1607ms]
terms_many_order_by_term                       Memory: 6.9 MB              Avg: 28.6076ms (-4.93%)     Median: 28.1059ms (-6.64%)     [24.0845ms .. 34.1493ms]
terms_many_with_top_hits                       Memory: 58.3 MB             Avg: 570.1548ms (+1.52%)    Median: 572.7759ms (+0.53%)    [525.9567ms .. 617.0862ms]
terms_many_with_avg_sub_agg                    Memory: 27.8 MB             Avg: 305.5207ms (+0.24%)    Median: 296.0101ms (-0.22%)    [277.8579ms .. 373.5914ms]
terms_many_json_mixed_type_with_avg_sub_agg    Memory: 42.0 MB (-0.00%)    Avg: 324.7342ms (-2.51%)    Median: 319.0025ms (-2.58%)    [298.7122ms .. 368.6144ms]
terms_few_with_cardinality_agg                 Memory: 10.8 MB             Avg: 151.6126ms (-2.54%)    Median: 149.0616ms (-0.32%)    [136.5592ms .. 181.8942ms]
range_agg_with_term_agg_few                    Memory: 248.2 KB            Avg: 49.5225ms (+3.11%)     Median: 48.3994ms (+3.18%)     [46.4134ms .. 60.5989ms]
range_agg_with_term_agg_many                   Memory: 6.9 MB              Avg: 85.9824ms (-3.66%)     Median: 78.4266ms (-3.85%)     [64.1231ms .. 128.5279ms]
 2024-07-03 12:42:59 +08:00
PSeitz
56d79cb203 fix cardinality aggregation performance (#2446) 
* fix cardinality aggregation performance

fix cardinality performance by fetching multiple terms at once. This
avoids decompressing the same block and keeps the buffer state between
terms.

add cardinality aggregation benchmark

bump rust version to 1.66

Performance comparison to before (AllQuery)
```
full
cardinality_agg                   Memory: 3.5 MB (-0.00%)    Avg: 21.2256ms (-97.78%)    Median: 21.0042ms (-97.82%)    [20.4717ms .. 23.6206ms]
terms_few_with_cardinality_agg    Memory: 10.6 MB            Avg: 81.9293ms (-97.37%)    Median: 81.5526ms (-97.38%)    [79.7564ms .. 88.0374ms]
dense
cardinality_agg                   Memory: 3.6 MB (-0.00%)    Avg: 25.9372ms (-97.24%)    Median: 25.7744ms (-97.25%)    [24.7241ms .. 27.8793ms]
terms_few_with_cardinality_agg    Memory: 10.6 MB            Avg: 93.9897ms (-96.91%)    Median: 92.7821ms (-96.94%)    [90.3312ms .. 117.4076ms]
sparse
cardinality_agg                   Memory: 895.4 KB (-0.00%)    Avg: 22.5113ms (-95.01%)    Median: 22.5629ms (-94.99%)    [22.1628ms .. 22.9436ms]
terms_few_with_cardinality_agg    Memory: 680.2 KB             Avg: 26.4250ms (-94.85%)    Median: 26.4135ms (-94.86%)    [26.3210ms .. 26.6774ms]
```

* clippy

* assert for sorted ordinals
 2024-07-02 15:29:00 +08:00
Paul Masurel
0f4c2e27cf Fixes bug that causes out-of-order sstable key. (#2445) 
The previous way to address the problem was to replace \u{0000}
with 0 in different places.

This logic had several flaws:
Done on the serializer side (like it was for the columnar), there was
a collision problem.

If a document in the segment contained a json field with a \0 and
antoher doc contained the same json field but `0` then we were sending
the same field path twice to the serializer.

Another option would have been to normalizes all values on the writer
side.

This PR simplifies the logic and simply ignore json path containing a
\0, both in the columnar and the inverted index.

Closes #2442
 2024-07-01 15:40:07 +08:00
落叶乌龟
f9ae295507 feat(query): Make BooleanQuery supports minimum_number_should_match (#2405) 
* feat(query): Make `BooleanQuery` supports `minimum_number_should_match`. see issue #2398

In this commit, a novel scorer named DisjunctionScorer is introduced, which performs the union of inverted chains with the minimal required elements. BTW, it's implemented via a min-heap. Necessary modifications on `BooleanQuery` and `BooleanWeight` are performed as well.

* fixup! fix test

* fixup!: refactor code.

1. More meaningful names.
2. Add Cache for `Disjunction`'s scorers, and fix bug.
3. Optimize `BooleanWeight::complex_scorer`

Thanks
 Paul Masurel <paul@quickwit.io>

* squash!: come up with better variable naming.

* squash!: fix naming issues.

* squash!: fix typo.

* squash!: Remove CombinationMethod::FullIntersection
 2024-07-01 15:39:41 +08:00
Raphael Coeffic
d9db5302d9 feat: cardinality aggregation (#2337) 
* WiP: cardinality aggregation

* Collect unique entries first, then insert into HyperLogLog

* Handle `missing`

* Hybrid approach

* Review changes

- insert `missing` value at most once
- `term_id` -> `term_ord`
- iterate directly over entries without collecting first

* Use salted hasher to include column type

* fix: formatting

* More review fixes

* Add cardinality to test_aggregation_flushing

* Formatting
 2024-07-01 07:49:42 +08:00
Paul Masurel
e453848134 Recycling buffer in PrefixPhraseScorer (#2443) 2024-06-24 17:11:53 +09:00
PSeitz
59084143ef use optional index in multivalued index (#2439) 
* use optional index in multivalued index

For mostly empty multivalued indices there was a large overhead during
creation when iterating all docids. This is alleviated by placing an
optional index in the multivalued index to mark documents that have values.

There's some performance overhead when accessing values in a multivalued
index. The accessing cost is now optional index + multivalue index. The
sparse codec performs relatively bad with the binary_search when accessing
data. This is reflected in the benchmarks below.

This changes the format of columnar to v2, but code is added to handle the v1
formats.

```
     Running benches/bench_access.rs (/home/pascal/Development/tantivy/optional_multivalues/target/release/deps/bench_access-ea323c028db88db4)
multi sparse 1/13
access_values_for_doc        Avg: 42.8946ms (+241.80%)    Median: 42.8869ms (+244.10%)    [42.7484ms .. 43.1074ms]
access_first_vals            Avg: 42.8022ms (+421.93%)    Median: 42.7553ms (+439.84%)    [42.6794ms .. 43.7404ms]
multi 2x
access_values_for_doc        Avg: 31.1244ms (+24.17%)    Median: 30.8339ms (+23.46%)    [30.7192ms .. 33.6059ms]
access_first_vals            Avg: 24.3070ms (+70.92%)    Median: 24.0966ms (+70.18%)    [23.9328ms .. 26.4851ms]
sparse 1/13
access_values_for_doc        Avg: 42.2490ms (+0.61%)    Median: 42.2346ms (+2.28%)    [41.8988ms .. 43.7821ms]
access_first_vals            Avg: 43.6272ms (+0.23%)    Median: 43.6197ms (+1.78%)    [43.4920ms .. 43.9009ms]
dense 1/12
access_values_for_doc        Avg: 8.6184ms (+23.18%)    Median: 8.6126ms (+23.78%)    [8.5843ms .. 8.7527ms]
access_first_vals            Avg: 6.8112ms (+4.47%)     Median: 6.8002ms (+4.55%)     [6.7887ms .. 6.8991ms]
full
access_values_for_doc        Avg: 9.4073ms (-5.09%)    Median: 9.4023ms (-2.23%)    [9.3694ms .. 9.4568ms]
access_first_vals            Avg: 4.9531ms (+6.24%)    Median: 4.9502ms (+7.85%)    [4.9423ms .. 4.9718ms]
```

```
     Running benches/bench_merge.rs (/home/pascal/Development/tantivy/optional_multivalues/target/release/deps/bench_merge-475697dfceb3639f)
merge_multi 2x_and_multi 2x                          Avg: 20.2280ms (+34.33%)    Median: 20.1829ms (+35.33%)    [19.9933ms .. 20.8806ms]
merge_multi sparse 1/13_and_multi sparse 1/13        Avg: 0.8961ms (-78.04%)     Median: 0.8943ms (-77.61%)     [0.8899ms .. 0.9272ms]
merge_dense 1/12_and_dense 1/12                      Avg: 0.6619ms (-1.26%)      Median: 0.6616ms (+2.20%)      [0.6473ms .. 0.6837ms]
merge_sparse 1/13_and_sparse 1/13                    Avg: 0.5508ms (-0.85%)      Median: 0.5508ms (+2.80%)      [0.5420ms .. 0.5634ms]
merge_sparse 1/13_and_dense 1/12                     Avg: 0.6046ms (-4.64%)      Median: 0.6038ms (+2.80%)      [0.5939ms .. 0.6296ms]
merge_multi sparse 1/13_and_dense 1/12               Avg: 0.9111ms (-83.48%)     Median: 0.9063ms (-83.50%)     [0.9047ms .. 0.9663ms]
merge_multi sparse 1/13_and_sparse 1/13              Avg: 0.8451ms (-89.49%)     Median: 0.8428ms (-89.43%)     [0.8411ms .. 0.8563ms]
merge_multi 2x_and_dense 1/12                        Avg: 10.6624ms (-4.82%)     Median: 10.6568ms (-4.49%)     [10.5738ms .. 10.8353ms]
merge_multi 2x_and_sparse 1/13                       Avg: 10.6336ms (-22.95%)    Median: 10.5925ms (-22.33%)    [10.5149ms .. 11.5657ms]
```

* Update columnar/src/columnar/format_version.rs

Co-authored-by: Paul Masurel <paul@quickwit.io>

* Update columnar/src/column_index/mod.rs

Co-authored-by: Paul Masurel <paul@quickwit.io>

---------

Co-authored-by: Paul Masurel <paul@quickwit.io>
 2024-06-19 14:54:12 +08:00
PSeitz
511b027350 update columnar bench (#2438) 
* update columnar bench

* fix compile
 2024-06-14 10:42:35 +08:00
Philippe Noël
322f47eb47 Add ParadeDB to Companies List (#1) (#2437) 
* Add ParadeDB logo
 2024-06-14 09:12:58 +09:00
PSeitz
72f61ff89c remove index sorting (#2434) 
closes https://github.com/quickwit-oss/tantivy/issues/2352
 2024-06-13 15:51:53 +08:00
PSeitz
a141c3ec59 add columnar format compatibiliy tests (#2433) 
* add columnar format compatibiliy tests

* always try to write current format
 2024-06-13 15:04:52 +08:00
PSeitz
e90e7a25ae add access benchmark for columnar (#2432) 2024-06-12 14:29:15 +08:00
PSeitz
c3b92a5412 fix compiler warning, cleanup (#2393) 
fix compiler warning for missing feature flag
remove unused variables
cleanup unused methods
 2024-06-11 16:03:50 +08:00
PSeitz
2f55511064 extend indexwriter proptests (#2342) 
* index random values in proptest

* add proptest with multiple docs
 2024-06-11 16:02:57 +08:00
trinity-1686a
08b9fc0b31 fix de-escaping too much in query parser (#2427) 
* fix de-escaping too much in query parser
 2024-06-10 11:19:01 +02:00
PSeitz
714f363d43 add bench & test for columnar merging (#2428) 
* add merge columnar proptest

* add columnar merge benchmark
 2024-06-10 16:26:16 +08:00
PSeitz
93ff7365b0 reduce top hits aggregation memory consumption (#2426) 
move request structure out of top hits aggregation collector and use from the
passed structure instead

full
terms_many_with_top_hits    Memory: 58.2 MB (-43.64%)    Avg: 425.9680ms (-21.38%)    Median: 415.1097ms (-23.56%)    [395.5303ms .. 484.6325ms]
dense
terms_many_with_top_hits    Memory: 58.2 MB (-43.64%)    Avg: 440.0817ms (-19.68%)    Median: 432.2286ms (-21.10%)    [403.5632ms .. 497.7541ms]
sparse
terms_many_with_top_hits    Memory: 13.1 MB (-49.31%)    Avg: 33.3568ms (-32.19%)    Median: 33.0834ms (-31.86%)    [32.5126ms .. 35.7397ms]
multivalue
terms_many_with_top_hits    Memory: 58.2 MB (-43.64%)    Avg: 414.2340ms (-25.44%)    Median: 413.4144ms (-25.64%)    [403.9919ms .. 430.3170ms]
 2024-06-06 22:32:58 +08:00
Adam Reichold
8151925068 Panicking in spawned Rayon tasks will abort the process by default. (#2409) 2024-06-04 17:04:30 +09:00
dependabot[bot]
b960e40bc8 Update sketches-ddsketch requirement from 0.2.1 to 0.3.0 (#2423) 
Updates the requirements on [sketches-ddsketch](https://github.com/mheffner/rust-sketches-ddsketch) to permit the latest version.
- [Release notes](https://github.com/mheffner/rust-sketches-ddsketch/releases)
- [Commits](https://github.com/mheffner/rust-sketches-ddsketch/compare/v0.2.1...v0.3.0)

---
updated-dependencies:
- dependency-name: sketches-ddsketch
  dependency-type: direct:production
...

Signed-off-by: dependabot[bot] <support@github.com>
Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com>
 2024-06-04 15:50:23 +08:00
giovannicuccu
1095c9b073 Issue 1787 extended stats (#2247) 
* first version of extended stats along with its tests

* using IntermediateExtendStats instead of IntermediateStats with all tests passing

* Created struct for request and response

* first test with extended_stats

* kahan summation and tests with approximate equality

* version ready for merge

* removed approx dependency

* refactor for using ExtendedStats only when needed

* interim version

* refined version with code formatted

* refactored a struct

* cosmetic refactor

* fix after merge

* fix format

* added extended_stat bench

* merge and new benchmark for extended stats

* split stat segment collectors

* wrapped intermediate extended stat with a box to limit memory usage

* Revert "wrapped intermediate extended stat with a box to limit memory usage"

This reverts commit 5b4aa9f393.

* some code reformat, commented kahan summation

* refactor after review

* refactor after code review

* fix after incorrectly restoring kahan summation

* modifications for code review + bug fix in merge_fruit

* refactor assert_nearly_equals macro

* update after code review

---------

Co-authored-by: Giovanni Cuccu <gcuccu@imolainformatica.it>
 2024-06-04 14:25:17 +08:00
PSeitz
c0686515a9 update one_shot (#2420) 2024-05-31 11:07:35 +08:00
trinity-1686a
455156f51c improve query parser (#2416) 
* support escape sequence in more place

and fix bug with singlequoted strings

* add query parser test for range query on default field
 2024-05-30 17:29:27 +02:00
Meng Zhang
4143d31865 chore: fix build as the rev is gone (#2417) 2024-05-29 09:49:16 +08:00
Hamir Mahal
0c634adbe1 style: simplify strings with string interpolation (#2412) 
* style: simplify strings with string interpolation

* fix: formatting
 2024-05-27 09:16:47 +02:00
PSeitz
2e3641c2ae return CompactDocValue instead of trait (#2410) 
The CompactDocValue is easier to handle than the trait in some cases like comparison
and conversion
 2024-05-27 07:33:50 +02:00
Paul Masurel
b806122c81 Fixing flaky test (#2407) 2024-05-22 10:10:55 +09:00
PSeitz
e1679f3fb9 compact doc (#2402) 
* compact doc

* add any value type

* pass references when building CompactDoc

* remove OwnedValue from API

* clippy

* clippy

* fail on large documents

* fmt

* cleanup

* cleanup

* implement Value for different types

fix serde_json date Value implementation

* fmt

* cleanup

* fmt

* cleanup

* store positions instead of pos+len

* remove nodes array

* remove mediumvec

* cleanup

* infallible serialize into vec

* remove positions indirection

* remove 24MB limitation in document

use u32 for Addr
Remove the 3 byte addressing limitation and use VInt instead

* cleanup

* extend test

* cleanup, add comments

* rename, remove pub
 2024-05-21 10:16:08 +02:00
dependabot[bot]
5a80420b10 --- (#2406) 
updated-dependencies:
- dependency-name: binggan
  dependency-type: direct:production
...

Signed-off-by: dependabot[bot] <support@github.com>
Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com>
 2024-05-21 04:36:32 +02:00
dependabot[bot]
aa26ff5029 Update binggan requirement from 0.6.2 to 0.7.0 (#2401) 
---
updated-dependencies:
- dependency-name: binggan
  dependency-type: direct:production
...

Signed-off-by: dependabot[bot] <support@github.com>
Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com>
 2024-05-17 02:53:25 +02:00
dependabot[bot]
e197b59258 Update itertools requirement from 0.12.0 to 0.13.0 (#2400) 
Updates the requirements on [itertools](https://github.com/rust-itertools/itertools) to permit the latest version.
- [Changelog](https://github.com/rust-itertools/itertools/blob/master/CHANGELOG.md)
- [Commits](https://github.com/rust-itertools/itertools/compare/v0.12.0...v0.13.0)

---
updated-dependencies:
- dependency-name: itertools
  dependency-type: direct:production
...

Signed-off-by: dependabot[bot] <support@github.com>
Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com>
 2024-05-17 02:53:02 +02:00
PSeitz
5b7cca13e5 lower contention on AggregationLimits (#2394) 
PR https://github.com/quickwit-oss/quickwit/pull/4962 fixes an issue
where the AggregationLimits are not passed correctly. Since the
AggregationLimits are shared properly we run into contention issues.

This PR includes some straightforward improvement to reduce contention,
by only calling if the memory changed and avoiding the second read.

We probably need some sharding with multiple counters or local caching before updating the
global after some threshold.
 2024-05-15 12:25:40 +02:00
dependabot[bot]
a79590477e Update binggan requirement from 0.5.2 to 0.6.2 (#2399) 
---
updated-dependencies:
- dependency-name: binggan
  dependency-type: direct:production
...

Signed-off-by: dependabot[bot] <support@github.com>
Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com>
 2024-05-15 05:40:37 +02:00
Paul Masurel
6181c1eb5e Small changes in the Executor API. (#2391) 
Warning, this change is mildly not backward compatible
so I bumped tantivy's version.
 2024-05-10 17:19:12 +09:00
Adam Reichold
1ee5f90761 Give allocation control to the caller instead of force a clone (#2389) 
Achieved by moving the boxes out of the temporary reference wrappers which are
cloneable themselves, i.e. if required the caller can clone them already or
consume them to reuse existing allocations.
 2024-05-09 16:01:13 +09:00
PSeitz
71f3b4e4e3 fix ReferenceValue API flaw (#2372) 
* fix ReferenceValue API flaw

Remove `Facet` and `TokenizedString` values from the `ReferenceValue` API,
as this requires the trait value to have them stored somewhere.

Since `TokenizedString` is quite niche, I just copy it into a Box,
instead of designing a reference API around it.

* fix comment link
 2024-05-09 06:14:42 +02:00
trinity-1686a
8cd7ddc535 run block decompression from executor (#2386) 
* run block decompression from executor

* add a wrapper with is_closed to oneshot channel

* add cancelation test to Executor::spawn_blocking
 2024-05-08 12:22:44 +02:00
Paul Masurel
2b76335a95 Removed usage of num_cpus (#2387) 
* Removed usage of num_cpus
* handling error
 2024-05-08 13:32:52 +09:00
PSeitz
c6b213d8f0 use bingang for agg benchmark (#2378) 
* use bingang for agg benchmark

use bingang for agg benchmark, which includes memory consumption

Output:
```
full
histogram                     Memory: 15.8 KB              Avg: 10.9322ms  (+5.44%)    Median: 10.8790ms  (+9.28%)     Min: 10.7470ms    Max: 11.3263ms
histogram_hard_bounds         Memory: 15.5 KB              Avg: 5.1939ms  (+6.61%)     Median: 5.1722ms  (+10.98%)     Min: 5.0432ms     Max: 5.3910ms
histogram_with_avg_sub_agg    Memory: 48.7 KB              Avg: 23.8165ms  (+4.57%)    Median: 23.7264ms  (+10.06%)    Min: 23.4995ms    Max: 24.8107ms
dense
histogram                     Memory: 17.3 KB              Avg: 15.6810ms  (-8.54%)    Median: 15.6174ms  (-8.89%)    Min: 15.4953ms    Max: 16.0702ms
histogram_hard_bounds         Memory: 15.4 KB              Avg: 10.0720ms  (-7.33%)    Median: 10.0572ms  (-7.06%)    Min: 9.8500ms     Max: 10.4819ms
histogram_with_avg_sub_agg    Memory: 50.1 KB              Avg: 33.0993ms  (-7.04%)    Median: 32.9499ms  (-6.86%)    Min: 32.8284ms    Max: 34.0529ms
sparse
histogram                     Memory: 16.3 KB              Avg: 19.2325ms  (-0.44%)    Median: 19.1211ms  (-1.26%)    Min: 19.0348ms    Max: 19.7902ms
histogram_hard_bounds         Memory: 16.1 KB              Avg: 18.5179ms  (-0.61%)    Median: 18.4552ms  (-0.90%)    Min: 18.3799ms    Max: 19.0535ms
histogram_with_avg_sub_agg    Memory: 34.7 KB              Avg: 21.2589ms  (-0.69%)    Median: 21.1867ms  (-1.05%)    Min: 21.0342ms    Max: 21.9900ms
```

* add more bench with term as sub agg
 2024-05-07 11:29:49 +02:00
PSeitz
eea70030bf cleanup top level exports (#2382) 
remove some top level exports
 2024-05-07 09:59:41 +02:00
PSeitz
92b5526310 allow more JSON values, fix i64 special case (#2383) 
This changes three things:
- Reuse positions_per_path hashmap instead of allocating one per
  indexed JSON value
- Try to cast u64 values to i64 to streamline with search behaviour
- Allow top level json values to be of any type, instead of limiting it
  to JSON objects. Remove special JSON object handling method.

TODO: We probably should also try to check f64 to i64 and u64 when
indexing, as values may get converted to f64 by the JSON parser
 2024-05-01 12:08:12 +02:00
PSeitz
99a59ad37e remove zero byte check (#2379) 
remove zero byte checks in columnar. zero bytes are converted during serialization now.
unify code paths
extend test for expected column names
 2024-04-26 06:03:28 +02:00
trinity-1686a
6a66a71cbb modify fastfield range query heuristic (#2375) 2024-04-25 10:06:11 +02:00
PSeitz
ff40764204 make convert_to_fast_value_and_append_to_json_term pub (#2370) 
* make convert_to_fast_value_and_append_to_json_term pub

* clippy
 2024-04-23 04:05:41 +02:00
PSeitz
047da20b5b add json path constructor to term (#2367) 2024-04-22 12:23:35 +02:00
PSeitz
1417eaf3a7 fix coverage (#2368) 2024-04-22 12:23:15 +02:00
PSeitz
4f8493d2de improve document docs (#2359) 2024-04-22 12:05:16 +02:00
Paul Masurel
8861366137 Owned value relying on Vec instead of BTreeMap (#2364) 
* Owned value relying on Vec instead of BTreeMap

* fmt

* fix build

* fix serialization

---------

Co-authored-by: Pascal Seitz <pascal.seitz@gmail.com>
 2024-04-22 09:38:05 +02:00
PSeitz
0e9fced336 remove JsonTermWriter (#2238) 
* remove JsonTermWriter

remove JsonTermWriter
remove path truncation logic, add assertion

* fix json_path_writer add sep logic
 2024-04-18 16:28:05 +02:00
PSeitz
b257b960b3 validate sort by field type (#2336) 
* validate sort by field type

* Update src/index/index.rs

Co-authored-by: Adam Reichold <adamreichold@users.noreply.github.com>

---------

Co-authored-by: Adam Reichold <adamreichold@users.noreply.github.com>
 2024-04-16 04:42:24 +02:00
Adam Reichold
4708171a32 Fix some of the things current Clippy complains about (#2363) 2024-04-16 04:27:06 +02:00
Adam Reichold
b493743f8d Fix trait bound of StoreReader::iter (#2360) 
* Fix trait bound of StoreReader::iter

Similar to `StoreReader::get`, `StoreReader::iter` should only require
`DocumentDeserialize` and not `Document`.

* Mark the iterator returned by SegmentReader::doc_ids_alive as Send so it can be used in impls of Stream/AsyncIterator.
 2024-04-15 15:50:02 +02:00
trinity-1686a
d2955a3fd2 extend field grouping (#2333) 
* extend field grouping
 2024-04-15 10:36:32 +02:00
290 changed files with 12721 additions and 7438 deletions
Expand all files Collapse all files

4
.github/workflows/coverage.yml vendored
View File

@@ -15,11 +15,11 @@ jobs:
steps:
- uses: actions/checkout@v4
- name: Install Rust
run: rustup toolchain install nightly-2023-09-10 --profile minimal --component llvm-tools-preview
run: rustup toolchain install nightly-2024-07-01 --profile minimal --component llvm-tools-preview
- uses: Swatinem/rust-cache@v2
- uses: taiki-e/install-action@cargo-llvm-cov
- name: Generate code coverage
run: cargo +nightly-2023-09-10 llvm-cov --all-features --workspace --doctests --lcov --output-path lcov.info
run: cargo +nightly-2024-07-01 llvm-cov --all-features --workspace --doctests --lcov --output-path lcov.info
- name: Upload coverage to Codecov
uses: codecov/codecov-action@v3
continue-on-error: true

4
ARCHITECTURE.md
View File

@@ -46,7 +46,7 @@ The file of a segment has the format
```segment-id . ext```
The extension signals which data structure (or [`SegmentComponent`](src/core/segment_component.rs)) is stored in the file.
The extension signals which data structure (or [`SegmentComponent`](src/index/segment_component.rs)) is stored in the file.
A small `meta.json` file is in charge of keeping track of the list of segments, as well as the schema.
@@ -102,7 +102,7 @@ but users can extend tantivy with their own implementation.
Tantivy's document follows a very strict schema, decided before building any index.
The schema defines all of the fields that the indexes [`Document`](src/schema/document.rs) may and should contain, their types (`text`, `i64`, `u64`, `Date`, ...) as well as how it should be indexed / represented in tantivy.
The schema defines all of the fields that the indexes [`Document`](src/schema/document/mod.rs) may and should contain, their types (`text`, `i64`, `u64`, `Date`, ...) as well as how it should be indexed / represented in tantivy.
Depending on the type of the field, you can decide to

96
CHANGELOG.md
View File

@@ -1,3 +1,79 @@
Tantivy 0.23 - Unreleased
================================
Tantivy 0.23 will be backwards compatible with indices created with v0.22 and v0.21. The new minimum rust version will be 1.75.
#### Bugfixes
- fix potential endless loop in merge [#2457](https://github.com/quickwit-oss/tantivy/pull/2457)(@PSeitz)
- fix bug that causes out-of-order sstable key. [#2445](https://github.com/quickwit-oss/tantivy/pull/2445)(@fulmicoton)
- fix ReferenceValue API flaw [#2372](https://github.com/quickwit-oss/tantivy/pull/2372)(@PSeitz)
- fix `OwnedBytes` debug panic [#2512](https://github.com/quickwit-oss/tantivy/pull/2512)(@b41sh)
#### Breaking API Changes
- remove index sorting [#2434](https://github.com/quickwit-oss/tantivy/pull/2434)(@PSeitz)
#### Features/Improvements
- **Aggregation**
- Support for cardinality aggregation [#2337](https://github.com/quickwit-oss/tantivy/pull/2337) [#2446](https://github.com/quickwit-oss/tantivy/pull/2446) (@raphaelcoeffic @PSeitz)
- Support for extended stats aggregation [#2247](https://github.com/quickwit-oss/tantivy/pull/2247)(@giovannicuccu)
- Add Key::I64 and Key::U64 variants in aggregation to avoid f64 precision issues [#2468](https://github.com/quickwit-oss/tantivy/pull/2468)(@PSeitz)
- Faster term aggregation fetch terms [#2447](https://github.com/quickwit-oss/tantivy/pull/2447)(@PSeitz)
- Improve custom order deserialization [#2451](https://github.com/quickwit-oss/tantivy/pull/2451)(@PSeitz)
- Change AggregationLimits behavior [#2495](https://github.com/quickwit-oss/tantivy/pull/2495)(@PSeitz)
- lower contention on AggregationLimits [#2394](https://github.com/quickwit-oss/tantivy/pull/2394)(@PSeitz)
- fix postcard compatibility for top_hits, add postcard test [#2346](https://github.com/quickwit-oss/tantivy/pull/2346)(@PSeitz)
- reduce top hits memory consumption [#2426](https://github.com/quickwit-oss/tantivy/pull/2426)(@PSeitz)
- check unsupported parameters top_hits [#2351](https://github.com/quickwit-oss/tantivy/pull/2351)(@PSeitz)
- Change AggregationLimits to AggregationLimitsGuard [#2495](https://github.com/quickwit-oss/tantivy/pull/2495)(@PSeitz)
- **Range Queries**
- Support fast field range queries on json fields [#2456](https://github.com/quickwit-oss/tantivy/pull/2456)(@PSeitz)
- Add support for str fast field range query [#2460](https://github.com/quickwit-oss/tantivy/pull/2460) [#2452](https://github.com/quickwit-oss/tantivy/pull/2452) [#2453](https://github.com/quickwit-oss/tantivy/pull/2453)(@PSeitz)
- modify fastfield range query heuristic [#2375](https://github.com/quickwit-oss/tantivy/pull/2375)(@trinity-1686a)
- add FastFieldRangeQuery for explicit range queries on fast field (for `RangeQuery` it is autodetected) [#2477](https://github.com/quickwit-oss/tantivy/pull/2477)(@PSeitz)
- add format backwards-compatibility tests [#2485](https://github.com/quickwit-oss/tantivy/pull/2485)(@PSeitz)
- add columnar format compatibility tests [#2433](https://github.com/quickwit-oss/tantivy/pull/2433)(@PSeitz)
- Improved snippet ranges algorithm [#2474](https://github.com/quickwit-oss/tantivy/pull/2474)(@gezihuzi)
- make find_field_with_default return json fields without path [#2476](https://github.com/quickwit-oss/tantivy/pull/2476)(@trinity-1686a)
- feat(query): Make `BooleanQuery` support `minimum_number_should_match` [#2405](https://github.com/quickwit-oss/tantivy/pull/2405)(@LebranceBW)
- **RegexPhraseQuery**
`RegexPhraseQuery` supports phrase queries with regex. E.g. query "b.* b.* wolf" matches "big bad wolf". Slop is supported as well: "b.* wolf"~2 matches "big bad wolf" [#2516](https://github.com/quickwit-oss/tantivy/pull/2516)(@PSeitz)
- **Optional Index in Multivalue Columnar Index**
For mostly empty multivalued indices there was a large overhead during creation when iterating all docids (merge case).
This is alleviated by placing an optional index in the multivalued index to mark documents that have values.
This will slightly increase space and access time. [#2439](https://github.com/quickwit-oss/tantivy/pull/2439)(@PSeitz)
- **Store DateTime as nanoseconds in doc store** DateTime in the doc store was truncated to microseconds previously. This removes this truncation, while still keeping backwards compatibility. [#2486](https://github.com/quickwit-oss/tantivy/pull/2486)(@PSeitz)
- **Performace/Memory**
- lift clauses in LogicalAst for optimized ast during execution [#2449](https://github.com/quickwit-oss/tantivy/pull/2449)(@PSeitz)
- Use Vec instead of BTreeMap to back OwnedValue object [#2364](https://github.com/quickwit-oss/tantivy/pull/2364)(@fulmicoton)
- Replace TantivyDocument with CompactDoc. CompactDoc is much smaller and provides similar performance. [#2402](https://github.com/quickwit-oss/tantivy/pull/2402)(@PSeitz)
- Recycling buffer in PrefixPhraseScorer [#2443](https://github.com/quickwit-oss/tantivy/pull/2443)(@fulmicoton)
- **Json Type**
- JSON supports now all values on the root level. Previously an object was required. This enables support for flat mixed types. allow more JSON values, fix i64 special case [#2383](https://github.com/quickwit-oss/tantivy/pull/2383)(@PSeitz)
- add json path constructor to term [#2367](https://github.com/quickwit-oss/tantivy/pull/2367)(@PSeitz)
- **QueryParser**
- fix de-escaping too much in query parser [#2427](https://github.com/quickwit-oss/tantivy/pull/2427)(@trinity-1686a)
- improve query parser [#2416](https://github.com/quickwit-oss/tantivy/pull/2416)(@trinity-1686a)
- Support field grouping `title:(return AND "pink panther")` [#2333](https://github.com/quickwit-oss/tantivy/pull/2333)(@trinity-1686a)
- add access benchmark for columnar [#2432](https://github.com/quickwit-oss/tantivy/pull/2432)(@PSeitz)
- extend indexwriter proptests [#2342](https://github.com/quickwit-oss/tantivy/pull/2342)(@PSeitz)
- add bench & test for columnar merging [#2428](https://github.com/quickwit-oss/tantivy/pull/2428)(@PSeitz)
- Change in Executor API [#2391](https://github.com/quickwit-oss/tantivy/pull/2391)(@fulmicoton)
- Removed usage of num_cpus [#2387](https://github.com/quickwit-oss/tantivy/pull/2387)(@fulmicoton)
- use bingang for agg and stacker benchmark [#2378](https://github.com/quickwit-oss/tantivy/pull/2378)[#2492](https://github.com/quickwit-oss/tantivy/pull/2492)(@PSeitz)
- cleanup top level exports [#2382](https://github.com/quickwit-oss/tantivy/pull/2382)(@PSeitz)
- make convert_to_fast_value_and_append_to_json_term pub [#2370](https://github.com/quickwit-oss/tantivy/pull/2370)(@PSeitz)
- remove JsonTermWriter [#2238](https://github.com/quickwit-oss/tantivy/pull/2238)(@PSeitz)
- validate sort by field type [#2336](https://github.com/quickwit-oss/tantivy/pull/2336)(@PSeitz)
- Fix trait bound of StoreReader::iter [#2360](https://github.com/quickwit-oss/tantivy/pull/2360)(@adamreichold)
- remove read_postings_no_deletes [#2526](https://github.com/quickwit-oss/tantivy/pull/2526)(@PSeitz)
Tantivy 0.22
================================
@@ -8,7 +84,7 @@ Tantivy 0.22 will be able to read indices created with Tantivy 0.21.
- Fix bug that can cause `get_docids_for_value_range` to panic. [#2295](https://github.com/quickwit-oss/tantivy/pull/2295)(@fulmicoton)
- Avoid 1 document indices by increase min memory to 15MB for indexing [#2176](https://github.com/quickwit-oss/tantivy/pull/2176)(@PSeitz)
- Fix merge panic for JSON fields [#2284](https://github.com/quickwit-oss/tantivy/pull/2284)(@PSeitz)
- Fix bug occuring when merging JSON object indexed with positions. [#2253](https://github.com/quickwit-oss/tantivy/pull/2253)(@fulmicoton)
- Fix bug occurring when merging JSON object indexed with positions. [#2253](https://github.com/quickwit-oss/tantivy/pull/2253)(@fulmicoton)
- Fix empty DateHistogram gap bug [#2183](https://github.com/quickwit-oss/tantivy/pull/2183)(@PSeitz)
- Fix range query end check (fields with less than 1 value per doc are affected) [#2226](https://github.com/quickwit-oss/tantivy/pull/2226)(@PSeitz)
- Handle exclusive out of bounds ranges on fastfield range queries [#2174](https://github.com/quickwit-oss/tantivy/pull/2174)(@PSeitz)
@@ -26,7 +102,7 @@ Tantivy 0.22 will be able to read indices created with Tantivy 0.21.
- Support to deserialize f64 from string [#2311](https://github.com/quickwit-oss/tantivy/pull/2311)(@PSeitz)
- Add a top_hits aggregator [#2198](https://github.com/quickwit-oss/tantivy/pull/2198)(@ditsuke)
- Support bool type in term aggregation [#2318](https://github.com/quickwit-oss/tantivy/pull/2318)(@PSeitz)
- Support ip adresses in term aggregation [#2319](https://github.com/quickwit-oss/tantivy/pull/2319)(@PSeitz)
- Support ip addresses in term aggregation [#2319](https://github.com/quickwit-oss/tantivy/pull/2319)(@PSeitz)
- Support date type in term aggregation [#2172](https://github.com/quickwit-oss/tantivy/pull/2172)(@PSeitz)
- Support escaped dot when addressing field [#2250](https://github.com/quickwit-oss/tantivy/pull/2250)(@PSeitz)
@@ -116,7 +192,7 @@ Tantivy 0.20
- Add PhrasePrefixQuery [#1842](https://github.com/quickwit-oss/tantivy/issues/1842) (@trinity-1686a)
- Add `coerce` option for text and numbers types (convert the value instead of returning an error during indexing) [#1904](https://github.com/quickwit-oss/tantivy/issues/1904) (@PSeitz)
- Add regex tokenizer [#1759](https://github.com/quickwit-oss/tantivy/issues/1759)(@mkleen)
- Move tokenizer API to seperate crate. Having a seperate crate with a stable API will allow us to use tokenizers with different tantivy versions. [#1767](https://github.com/quickwit-oss/tantivy/issues/1767) (@PSeitz)
- Move tokenizer API to separate crate. Having a separate crate with a stable API will allow us to use tokenizers with different tantivy versions. [#1767](https://github.com/quickwit-oss/tantivy/issues/1767) (@PSeitz)
- **Columnar crate**: New fast field handling (@fulmicoton @PSeitz) [#1806](https://github.com/quickwit-oss/tantivy/issues/1806)[#1809](https://github.com/quickwit-oss/tantivy/issues/1809)
- Support for fast fields with optional values. Previously tantivy supported only single-valued and multi-value fast fields. The encoding of optional fast fields is now very compact.
- Fast field Support for JSON (schemaless fast fields). Support multiple types on the same column. [#1876](https://github.com/quickwit-oss/tantivy/issues/1876) (@fulmicoton)
@@ -163,13 +239,13 @@ Tantivy 0.20
- Auto downgrade index record option, instead of vint error [#1857](https://github.com/quickwit-oss/tantivy/issues/1857) (@PSeitz)
- Enable range query on fast field for u64 compatible types [#1762](https://github.com/quickwit-oss/tantivy/issues/1762) (@PSeitz) [#1876]
- sstable
- Isolating sstable and stacker in independant crates. [#1718](https://github.com/quickwit-oss/tantivy/issues/1718) (@fulmicoton)
- Isolating sstable and stacker in independent crates. [#1718](https://github.com/quickwit-oss/tantivy/issues/1718) (@fulmicoton)
- New sstable format [#1943](https://github.com/quickwit-oss/tantivy/issues/1943)[#1953](https://github.com/quickwit-oss/tantivy/issues/1953) (@trinity-1686a)
- Use DeltaReader directly to implement Dictionnary::ord_to_term [#1928](https://github.com/quickwit-oss/tantivy/issues/1928) (@trinity-1686a)
- Use DeltaReader directly to implement Dictionnary::term_ord [#1925](https://github.com/quickwit-oss/tantivy/issues/1925) (@trinity-1686a)
- Add seperate tokenizer manager for fast fields [#2019](https://github.com/quickwit-oss/tantivy/issues/2019) (@PSeitz)
- Use DeltaReader directly to implement Dictionary::ord_to_term [#1928](https://github.com/quickwit-oss/tantivy/issues/1928) (@trinity-1686a)
- Use DeltaReader directly to implement Dictionary::term_ord [#1925](https://github.com/quickwit-oss/tantivy/issues/1925) (@trinity-1686a)
- Add separate tokenizer manager for fast fields [#2019](https://github.com/quickwit-oss/tantivy/issues/2019) (@PSeitz)
- Make construction of LevenshteinAutomatonBuilder for FuzzyTermQuery instances lazy. [#1756](https://github.com/quickwit-oss/tantivy/issues/1756) (@adamreichold)
- Added support for madvise when opening an mmaped Index [#2036](https://github.com/quickwit-oss/tantivy/issues/2036) (@fulmicoton)
- Added support for madvise when opening an mmapped Index [#2036](https://github.com/quickwit-oss/tantivy/issues/2036) (@fulmicoton)
- Rename `DatePrecision` to `DateTimePrecision` [#2051](https://github.com/quickwit-oss/tantivy/issues/2051) (@guilload)
- Query Parser
- Quotation mark can now be used for phrase queries. [#2050](https://github.com/quickwit-oss/tantivy/issues/2050) (@fulmicoton)
@@ -208,7 +284,7 @@ Tantivy 0.19
- Add support for phrase slop in query language [#1393](https://github.com/quickwit-oss/tantivy/pull/1393) (@saroh)
- Aggregation
- Add aggregation support for date type [#1693](https://github.com/quickwit-oss/tantivy/pull/1693)(@PSeitz)
- Add support for keyed parameter in range and histgram aggregations [#1424](https://github.com/quickwit-oss/tantivy/pull/1424) (@k-yomo)
- Add support for keyed parameter in range and histogram aggregations [#1424](https://github.com/quickwit-oss/tantivy/pull/1424) (@k-yomo)
- Add aggregation bucket limit [#1363](https://github.com/quickwit-oss/tantivy/pull/1363) (@PSeitz)
- Faster indexing
- [#1610](https://github.com/quickwit-oss/tantivy/pull/1610) (@PSeitz)
@@ -651,7 +727,7 @@ Tantivy 0.4.0
- Raise the limit of number of fields (previously 256 fields) (@fulmicoton)
- Removed u32 fields. They are replaced by u64 and i64 fields (#65) (@fulmicoton)
- Optimized skip in SegmentPostings (#130) (@lnicola)
- Replacing rustc_serialize by serde. Kudos to @KodrAus and @lnicola
- Replacing rustc_serialize by serde. Kudos to benchmark@KodrAus and @lnicola
- Using error-chain (@KodrAus)
- QueryParser: (@fulmicoton)
- Explicit error returned when searched for a term that is not indexed

10
CITATION.cff Normal file
View File

@@ -0,0 +1,10 @@
cff-version: 1.2.0
message: "If you use this software, please cite it as below."
authors:
- alias: Quickwit Inc.
website: "https://quickwit.io"
title: "tantivy"
version: 0.22.0
doi: 10.5281/zenodo.13942948
date-released: 2024-10-17
url: "https://github.com/quickwit-oss/tantivy"

65
Cargo.toml
View File

@@ -1,6 +1,6 @@
[package]
name = "tantivy"
version = "0.22.0"
version = "0.23.0"
authors = ["Paul Masurel <paul.masurel@gmail.com>"]
license = "MIT"
categories = ["database-implementations", "data-structures"]
@@ -11,55 +11,60 @@ repository = "https://github.com/quickwit-oss/tantivy"
readme = "README.md"
keywords = ["search", "information", "retrieval"]
edition = "2021"
rust-version = "1.63"
rust-version = "1.75"
exclude = ["benches/*.json", "benches/*.txt"]
[dependencies]
oneshot = "0.1.5"
oneshot = "0.1.7"
base64 = "0.22.0"
byteorder = "1.4.3"
crc32fast = "1.3.2"
once_cell = "1.10.0"
regex = { version = "1.5.5", default-features = false, features = ["std", "unicode"] }
regex = { version = "1.5.5", default-features = false, features = [
"std",
"unicode",
] }
aho-corasick = "1.0"
tantivy-fst = "0.5"
memmap2 = { version = "0.9.0", optional = true }
lz4_flex = { version = "0.11", default-features = false, optional = true }
zstd = { version = "0.13", optional = true, default-features = false }
tempfile = { version = "3.3.0", optional = true }
tempfile = { version = "3.12.0", optional = true }
log = "0.4.16"
serde = { version = "1.0.136", features = ["derive"] }
serde_json = "1.0.79"
num_cpus = "1.13.1"
fs4 = { version = "0.8.0", optional = true }
levenshtein_automata = "0.2.1"
uuid = { version = "1.0.0", features = ["v4", "serde"] }
crossbeam-channel = "0.5.4"
rust-stemmers = "1.2.0"
downcast-rs = "1.2.0"
bitpacking = { version = "0.9.2", default-features = false, features = ["bitpacker4x"] }
downcast-rs = "1.2.1"
bitpacking = { version = "0.9.2", default-features = false, features = [
"bitpacker4x",
] }
census = "0.4.2"
rustc-hash = "1.1.0"
thiserror = "1.0.30"
rustc-hash = "2.0.0"
thiserror = "2.0.1"
htmlescape = "0.3.1"
fail = { version = "0.5.0", optional = true }
time = { version = "0.3.10", features = ["serde-well-known"] }
time = { version = "0.3.35", features = ["serde-well-known"] }
smallvec = "1.8.0"
rayon = "1.5.2"
lru = "0.12.0"
fastdivide = "0.4.0"
itertools = "0.12.0"
itertools = "0.13.0"
measure_time = "0.8.2"
arc-swap = "1.5.0"
columnar = { version= "0.3", path="./columnar", package ="tantivy-columnar" }
sstable = { version= "0.3", path="./sstable", package ="tantivy-sstable", optional = true }
stacker = { version= "0.3", path="./stacker", package ="tantivy-stacker" }
query-grammar = { version= "0.22.0", path="./query-grammar", package = "tantivy-query-grammar" }
tantivy-bitpacker = { version= "0.6", path="./bitpacker" }
common = { version= "0.7", path = "./common/", package = "tantivy-common" }
tokenizer-api = { version= "0.3", path="./tokenizer-api", package="tantivy-tokenizer-api" }
sketches-ddsketch = { version = "0.2.1", features = ["use_serde"] }
columnar = { version = "0.3", path = "./columnar", package = "tantivy-columnar" }
sstable = { version = "0.3", path = "./sstable", package = "tantivy-sstable", optional = true }
stacker = { version = "0.3", path = "./stacker", package = "tantivy-stacker" }
query-grammar = { version = "0.22.0", path = "./query-grammar", package = "tantivy-query-grammar" }
tantivy-bitpacker = { version = "0.6", path = "./bitpacker" }
common = { version = "0.7", path = "./common/", package = "tantivy-common" }
tokenizer-api = { version = "0.3", path = "./tokenizer-api", package = "tantivy-tokenizer-api" }
sketches-ddsketch = { version = "0.3.0", features = ["use_serde"] }
hyperloglogplus = { version = "0.4.1", features = ["const-loop"] }
futures-util = { version = "0.3.28", optional = true }
fnv = "1.0.7"
@@ -67,6 +72,7 @@ fnv = "1.0.7"
winapi = "0.3.9"
[dev-dependencies]
binggan = "0.14.0"
rand = "0.8.5"
maplit = "1.0.2"
matches = "0.1.9"
@@ -112,17 +118,26 @@ lz4-compression = ["lz4_flex"]
zstd-compression = ["zstd"]
failpoints = ["fail", "fail/failpoints"]
unstable = [] # useful for benches.
unstable = [] # useful for benches.
quickwit = ["sstable", "futures-util"]
# Compares only the hash of a string when indexing data.
# Compares only the hash of a string when indexing data.
# Increases indexing speed, but may lead to extremely rare missing terms, when there's a hash collision.
# Uses 64bit ahash.
compare_hash_only = ["stacker/compare_hash_only"]
[workspace]
members = ["query-grammar", "bitpacker", "common", "ownedbytes", "stacker", "sstable", "tokenizer-api", "columnar"]
members = [
"query-grammar",
"bitpacker",
"common",
"ownedbytes",
"stacker",
"sstable",
"tokenizer-api",
"columnar",
]
# Following the "fail" crate best practises, we isolate
# tests that define specific behavior in fail check points
@@ -143,3 +158,7 @@ harness = false
[[bench]]
name = "index-bench"
harness = false
[[bench]]
name = "agg_bench"
harness = false

5
README.md
View File

@@ -18,7 +18,7 @@ Tantivy is, in fact, strongly inspired by Lucene's design.
## Benchmark
The following [benchmark](https://tantivy-search.github.io/bench/) breakdowns
The following [benchmark](https://tantivy-search.github.io/bench/) breaks down the
performance for different types of queries/collections.
Your mileage WILL vary depending on the nature of queries and their load.
@@ -101,7 +101,8 @@ cargo test
## Companies Using Tantivy
<p align="left">
<img align="center" src="doc/assets/images/etsy.png" alt="Etsy" height="25" width="auto" />&nbsp;
<img align="center" src="doc/assets/images/etsy.png" alt="Etsy" height="25" width="auto" /> &nbsp;
<img align="center" src="doc/assets/images/paradedb.png" alt="ParadeDB" height="25" width="auto" /> &nbsp;
<img align="center" src="doc/assets/images/Nuclia.png#gh-light-mode-only" alt="Nuclia" height="25" width="auto" /> &nbsp;
<img align="center" src="doc/assets/images/humanfirst.png#gh-light-mode-only" alt="Humanfirst.ai" height="30" width="auto" />
<img align="center" src="doc/assets/images/element.io.svg#gh-light-mode-only" alt="Element.io" height="25" width="auto" />

2
TODO.txt
View File

@@ -1,7 +1,7 @@
Make schema_builder API fluent.
fix doc serialization and prevent compression problems
u64 , etc. shoudl return Resutl<Option> now that we support optional missing a column is really not an error
u64 , etc. should return Result<Option> now that we support optional missing a column is really not an error
remove fastfield codecs
ditch the first_or_default trick. if it is still useful, improve its implementation.
rename FastFieldReaders::open to load

462
benches/agg_bench.rs Normal file
View File

@@ -0,0 +1,462 @@
use binggan::plugins::PeakMemAllocPlugin;
use binggan::{black_box, InputGroup, PeakMemAlloc, INSTRUMENTED_SYSTEM};
use rand::prelude::SliceRandom;
use rand::rngs::StdRng;
use rand::{Rng, SeedableRng};
use rand_distr::Distribution;
use serde_json::json;
use tantivy::aggregation::agg_req::Aggregations;
use tantivy::aggregation::AggregationCollector;
use tantivy::query::{AllQuery, TermQuery};
use tantivy::schema::{IndexRecordOption, Schema, TextFieldIndexing, FAST, STRING};
use tantivy::{doc, Index, Term};
#[global_allocator]
pub static GLOBAL: &PeakMemAlloc<std::alloc::System> = &INSTRUMENTED_SYSTEM;
/// Mini macro to register a function via its name
/// runner.register("average_u64", move |index| average_u64(index));
macro_rules! register {
($runner:expr, $func:ident) => {
$runner.register(stringify!($func), move |index| {
$func(index);
})
};
}
fn main() {
let inputs = vec![
("full", get_test_index_bench(Cardinality::Full).unwrap()),
(
"dense",
get_test_index_bench(Cardinality::OptionalDense).unwrap(),
),
(
"sparse",
get_test_index_bench(Cardinality::OptionalSparse).unwrap(),
),
(
"multivalue",
get_test_index_bench(Cardinality::Multivalued).unwrap(),
),
];
bench_agg(InputGroup::new_with_inputs(inputs));
}
fn bench_agg(mut group: InputGroup<Index>) {
group.add_plugin(PeakMemAllocPlugin::new(GLOBAL));
register!(group, average_u64);
register!(group, average_f64);
register!(group, average_f64_u64);
register!(group, stats_f64);
register!(group, extendedstats_f64);
register!(group, percentiles_f64);
register!(group, terms_few);
register!(group, terms_many);
register!(group, terms_many_top_1000);
register!(group, terms_many_order_by_term);
register!(group, terms_many_with_top_hits);
register!(group, terms_many_with_avg_sub_agg);
register!(group, terms_many_json_mixed_type_with_avg_sub_agg);
register!(group, cardinality_agg);
register!(group, terms_few_with_cardinality_agg);
register!(group, range_agg);
register!(group, range_agg_with_avg_sub_agg);
register!(group, range_agg_with_term_agg_few);
register!(group, range_agg_with_term_agg_many);
register!(group, histogram);
register!(group, histogram_hard_bounds);
register!(group, histogram_with_avg_sub_agg);
register!(group, avg_and_range_with_avg_sub_agg);
group.run();
}
fn exec_term_with_agg(index: &Index, agg_req: serde_json::Value) {
let agg_req: Aggregations = serde_json::from_value(agg_req).unwrap();
let reader = index.reader().unwrap();
let text_field = reader.searcher().schema().get_field("text").unwrap();
let term_query = TermQuery::new(
Term::from_field_text(text_field, "cool"),
IndexRecordOption::Basic,
);
let collector = get_collector(agg_req);
let searcher = reader.searcher();
black_box(searcher.search(&term_query, &collector).unwrap());
}
fn average_u64(index: &Index) {
let agg_req = json!({
"average": { "avg": { "field": "score", } }
});
exec_term_with_agg(index, agg_req)
}
fn average_f64(index: &Index) {
let agg_req = json!({
"average": { "avg": { "field": "score_f64", } }
});
exec_term_with_agg(index, agg_req)
}
fn average_f64_u64(index: &Index) {
let agg_req = json!({
"average_f64": { "avg": { "field": "score_f64" } },
"average": { "avg": { "field": "score" } },
});
exec_term_with_agg(index, agg_req)
}
fn stats_f64(index: &Index) {
let agg_req = json!({
"average_f64": { "stats": { "field": "score_f64", } }
});
exec_term_with_agg(index, agg_req)
}
fn extendedstats_f64(index: &Index) {
let agg_req = json!({
"extendedstats_f64": { "extended_stats": { "field": "score_f64", } }
});
exec_term_with_agg(index, agg_req)
}
fn percentiles_f64(index: &Index) {
let agg_req = json!({
"mypercentiles": {
"percentiles": {
"field": "score_f64",
"percents": [ 95, 99, 99.9 ]
}
}
});
execute_agg(index, agg_req);
}
fn cardinality_agg(index: &Index) {
let agg_req = json!({
"cardinality": {
"cardinality": {
"field": "text_many_terms"
},
}
});
execute_agg(index, agg_req);
}
fn terms_few_with_cardinality_agg(index: &Index) {
let agg_req = json!({
"my_texts": {
"terms": { "field": "text_few_terms" },
"aggs": {
"cardinality": {
"cardinality": {
"field": "text_many_terms"
},
}
}
},
});
execute_agg(index, agg_req);
}
fn terms_few(index: &Index) {
let agg_req = json!({
"my_texts": { "terms": { "field": "text_few_terms" } },
});
execute_agg(index, agg_req);
}
fn terms_many(index: &Index) {
let agg_req = json!({
"my_texts": { "terms": { "field": "text_many_terms" } },
});
execute_agg(index, agg_req);
}
fn terms_many_top_1000(index: &Index) {
let agg_req = json!({
"my_texts": { "terms": { "field": "text_many_terms", "size": 1000 } },
});
execute_agg(index, agg_req);
}
fn terms_many_order_by_term(index: &Index) {
let agg_req = json!({
"my_texts": { "terms": { "field": "text_many_terms", "order": { "_key": "desc" } } },
});
execute_agg(index, agg_req);
}
fn terms_many_with_top_hits(index: &Index) {
let agg_req = json!({
"my_texts": {
"terms": { "field": "text_many_terms" },
"aggs": {
"top_hits": { "top_hits":
{
"sort": [
{ "score": "desc" }
],
"size": 2,
"doc_value_fields": ["score_f64"]
}
}
}
},
});
execute_agg(index, agg_req);
}
fn terms_many_with_avg_sub_agg(index: &Index) {
let agg_req = json!({
"my_texts": {
"terms": { "field": "text_many_terms" },
"aggs": {
"average_f64": { "avg": { "field": "score_f64" } }
}
},
});
execute_agg(index, agg_req);
}
fn terms_many_json_mixed_type_with_avg_sub_agg(index: &Index) {
let agg_req = json!({
"my_texts": {
"terms": { "field": "json.mixed_type" },
"aggs": {
"average_f64": { "avg": { "field": "score_f64" } }
}
},
});
execute_agg(index, agg_req);
}
fn execute_agg(index: &Index, agg_req: serde_json::Value) {
let agg_req: Aggregations = serde_json::from_value(agg_req).unwrap();
let collector = get_collector(agg_req);
let reader = index.reader().unwrap();
let searcher = reader.searcher();
black_box(searcher.search(&AllQuery, &collector).unwrap());
}
fn range_agg(index: &Index) {
let agg_req = json!({
"range_f64": { "range": { "field": "score_f64", "ranges": [
{ "from": 3, "to": 7000 },
{ "from": 7000, "to": 20000 },
{ "from": 20000, "to": 30000 },
{ "from": 30000, "to": 40000 },
{ "from": 40000, "to": 50000 },
{ "from": 50000, "to": 60000 }
] } },
});
execute_agg(index, agg_req);
}
fn range_agg_with_avg_sub_agg(index: &Index) {
let agg_req = json!({
"rangef64": {
"range": {
"field": "score_f64",
"ranges": [
{ "from": 3, "to": 7000 },
{ "from": 7000, "to": 20000 },
{ "from": 20000, "to": 30000 },
{ "from": 30000, "to": 40000 },
{ "from": 40000, "to": 50000 },
{ "from": 50000, "to": 60000 }
]
},
"aggs": {
"average_f64": { "avg": { "field": "score_f64" } }
}
},
});
execute_agg(index, agg_req);
}
fn range_agg_with_term_agg_few(index: &Index) {
let agg_req = json!({
"rangef64": {
"range": {
"field": "score_f64",
"ranges": [
{ "from": 3, "to": 7000 },
{ "from": 7000, "to": 20000 },
{ "from": 20000, "to": 30000 },
{ "from": 30000, "to": 40000 },
{ "from": 40000, "to": 50000 },
{ "from": 50000, "to": 60000 }
]
},
"aggs": {
"my_texts": { "terms": { "field": "text_few_terms" } },
}
},
});
execute_agg(index, agg_req);
}
fn range_agg_with_term_agg_many(index: &Index) {
let agg_req = json!({
"rangef64": {
"range": {
"field": "score_f64",
"ranges": [
{ "from": 3, "to": 7000 },
{ "from": 7000, "to": 20000 },
{ "from": 20000, "to": 30000 },
{ "from": 30000, "to": 40000 },
{ "from": 40000, "to": 50000 },
{ "from": 50000, "to": 60000 }
]
},
"aggs": {
"my_texts": { "terms": { "field": "text_many_terms" } },
}
},
});
execute_agg(index, agg_req);
}
fn histogram(index: &Index) {
let agg_req = json!({
"rangef64": {
"histogram": {
"field": "score_f64",
"interval": 100 // 1000 buckets
},
}
});
execute_agg(index, agg_req);
}
fn histogram_hard_bounds(index: &Index) {
let agg_req = json!({
"rangef64": { "histogram": { "field": "score_f64", "interval": 100, "hard_bounds": { "min": 1000, "max": 300000 } } },
});
execute_agg(index, agg_req);
}
fn histogram_with_avg_sub_agg(index: &Index) {
let agg_req = json!({
"rangef64": {
"histogram": { "field": "score_f64", "interval": 100 },
"aggs": {
"average_f64": { "avg": { "field": "score_f64" } }
}
}
});
execute_agg(index, agg_req);
}
fn avg_and_range_with_avg_sub_agg(index: &Index) {
let agg_req = json!({
"rangef64": {
"range": {
"field": "score_f64",
"ranges": [
{ "from": 3, "to": 7000 },
{ "from": 7000, "to": 20000 },
{ "from": 20000, "to": 60000 }
]
},
"aggs": {
"average_in_range": { "avg": { "field": "score" } }
}
},
"average": { "avg": { "field": "score" } }
});
execute_agg(index, agg_req);
}
#[derive(Clone, Copy, Hash, Default, Debug, PartialEq, Eq, PartialOrd, Ord)]
enum Cardinality {
/// All documents contain exactly one value.
/// `Full` is the default for auto-detecting the Cardinality, since it is the most strict.
#[default]
Full = 0,
/// All documents contain at most one value.
OptionalDense = 1,
/// All documents may contain any number of values.
Multivalued = 2,
/// 1 / 20 documents has a value
OptionalSparse = 3,
}
fn get_collector(agg_req: Aggregations) -> AggregationCollector {
AggregationCollector::from_aggs(agg_req, Default::default())
}
fn get_test_index_bench(cardinality: Cardinality) -> tantivy::Result<Index> {
let mut schema_builder = Schema::builder();
let text_fieldtype = tantivy::schema::TextOptions::default()
.set_indexing_options(
TextFieldIndexing::default().set_index_option(IndexRecordOption::WithFreqs),
)
.set_stored();
let text_field = schema_builder.add_text_field("text", text_fieldtype);
let json_field = schema_builder.add_json_field("json", FAST);
let text_field_many_terms = schema_builder.add_text_field("text_many_terms", STRING | FAST);
let text_field_few_terms = schema_builder.add_text_field("text_few_terms", STRING | FAST);
let score_fieldtype = tantivy::schema::NumericOptions::default().set_fast();
let score_field = schema_builder.add_u64_field("score", score_fieldtype.clone());
let score_field_f64 = schema_builder.add_f64_field("score_f64", score_fieldtype.clone());
let score_field_i64 = schema_builder.add_i64_field("score_i64", score_fieldtype);
let index = Index::create_from_tempdir(schema_builder.build())?;
let few_terms_data = ["INFO", "ERROR", "WARN", "DEBUG"];
let lg_norm = rand_distr::LogNormal::new(2.996f64, 0.979f64).unwrap();
let many_terms_data = (0..150_000)
.map(|num| format!("author{num}"))
.collect::<Vec<_>>();
{
let mut rng = StdRng::from_seed([1u8; 32]);
let mut index_writer = index.writer_with_num_threads(1, 200_000_000)?;
// To make the different test cases comparable we just change one doc to force the
// cardinality
if cardinality == Cardinality::OptionalDense {
index_writer.add_document(doc!())?;
}
if cardinality == Cardinality::Multivalued {
index_writer.add_document(doc!(
json_field => json!({"mixed_type": 10.0}),
json_field => json!({"mixed_type": 10.0}),
text_field => "cool",
text_field => "cool",
text_field_many_terms => "cool",
text_field_many_terms => "cool",
text_field_few_terms => "cool",
text_field_few_terms => "cool",
score_field => 1u64,
score_field => 1u64,
score_field_f64 => lg_norm.sample(&mut rng),
score_field_f64 => lg_norm.sample(&mut rng),
score_field_i64 => 1i64,
score_field_i64 => 1i64,
))?;
}
let mut doc_with_value = 1_000_000;
if cardinality == Cardinality::OptionalSparse {
doc_with_value /= 20;
}
let _val_max = 1_000_000.0;
for _ in 0..doc_with_value {
let val: f64 = rng.gen_range(0.0..1_000_000.0);
let json = if rng.gen_bool(0.1) {
// 10% are numeric values
json!({ "mixed_type": val })
} else {
json!({"mixed_type": many_terms_data.choose(&mut rng).unwrap().to_string()})
};
index_writer.add_document(doc!(
text_field => "cool",
json_field => json,
text_field_many_terms => many_terms_data.choose(&mut rng).unwrap().to_string(),
text_field_few_terms => few_terms_data.choose(&mut rng).unwrap().to_string(),
score_field => val as u64,
score_field_f64 => lg_norm.sample(&mut rng),
score_field_i64 => val as i64,
))?;
if cardinality == Cardinality::OptionalSparse {
for _ in 0..20 {
index_writer.add_document(doc!(text_field => "cool"))?;
}
}
}
// writing the segment
index_writer.commit()?;
}
Ok(index)
}

14
benches/index-bench.rs
View File

@@ -18,7 +18,7 @@ fn benchmark(
benchmark_dynamic_json(b, input, schema, commit, parse_json)
} else {
_benchmark(b, input, schema, commit, parse_json, |schema, doc_json| {
TantivyDocument::parse_json(&schema, doc_json).unwrap()
TantivyDocument::parse_json(schema, doc_json).unwrap()
})
}
}
@@ -90,8 +90,7 @@ fn benchmark_dynamic_json(
) {
let json_field = schema.get_field("json").unwrap();
_benchmark(b, input, schema, commit, parse_json, |_schema, doc_json| {
let json_val: serde_json::Map<String, serde_json::Value> =
serde_json::from_str(doc_json).unwrap();
let json_val: serde_json::Value = serde_json::from_str(doc_json).unwrap();
tantivy::doc!(json_field=>json_val)
})
}
@@ -138,15 +137,16 @@ pub fn hdfs_index_benchmark(c: &mut Criterion) {
for (prefix, schema, is_dynamic) in benches {
for commit in [false, true] {
let suffix = if commit { "with-commit" } else { "no-commit" };
for parse_json in [false] {
{
let parse_json = false;
// for parse_json in [false, true] {
let suffix = if parse_json {
format!("{}-with-json-parsing", suffix)
format!("{suffix}-with-json-parsing")
} else {
format!("{}", suffix)
suffix.to_string()
};
let bench_name = format!("{}{}", prefix, suffix);
let bench_name = format!("{prefix}{suffix}");
group.bench_function(bench_name, |b| {
benchmark(b, HDFS_LOGS, schema.clone(), commit, parse_json, is_dynamic)
});

4
bitpacker/src/bitpacker.rs
View File

@@ -368,9 +368,9 @@ mod test {
for start_idx in 0u32..32u32 {
output.resize(len, 0);
bitunpacker.get_batch_u32s(start_idx, &buffer, &mut output);
for i in 0..len {
for (i, output_byte) in output.iter().enumerate() {
let expected = (start_idx + i as u32) & mask;
assert_eq!(output[i], expected);
assert_eq!(*output_byte, expected);
}
}
}

2
bitpacker/src/filter_vec/mod.rs
View File

@@ -35,8 +35,8 @@ const IMPLS: [FilterImplPerInstructionSet; 2] = [
const IMPLS: [FilterImplPerInstructionSet; 1] = [FilterImplPerInstructionSet::Scalar];
impl FilterImplPerInstructionSet {
#[allow(unused_variables)]
#[inline]
#[allow(unused_variables)] // on non-x86_64, code is unused.
fn from(code: u8) -> FilterImplPerInstructionSet {
#[cfg(target_arch = "x86_64")]
if code == FilterImplPerInstructionSet::AVX2 as u8 {

12
columnar/Cargo.toml
View File

@@ -9,7 +9,7 @@ description = "column oriented storage for tantivy"
categories = ["database-implementations", "data-structures", "compression"]
[dependencies]
itertools = "0.12.0"
itertools = "0.13.0"
fastdivide = "0.4.0"
stacker = { version= "0.3", path = "../stacker", package="tantivy-stacker"}
@@ -23,6 +23,16 @@ downcast-rs = "1.2.0"
proptest = "1"
more-asserts = "0.3.1"
rand = "0.8"
binggan = "0.14.0"
[[bench]]
name = "bench_merge"
harness = false
[[bench]]
name = "bench_access"
harness = false
[features]
unstable = []

2
columnar/README.md
View File

@@ -31,7 +31,7 @@ restriction on 50% of the values (e.g. a 64-bit hash). On the other hand, a lot
# Columnar format
This columnar format may have more than one column (with different types) associated to the same `column_name` (see [Coercion rules](#coercion-rules) above).
The `(column_name, columne_type)` couple however uniquely identifies a column.
The `(column_name, column_type)` couple however uniquely identifies a column.
That couple is serialized as a column `column_key`. The format of that key is:
`[column_name][ZERO_BYTE][column_type_header: u8]`

67
columnar/benches/bench_access.rs Normal file
View File

@@ -0,0 +1,67 @@
use binggan::{black_box, InputGroup};
use common::*;
use tantivy_columnar::Column;
pub mod common;
const NUM_DOCS: u32 = 2_000_000;
pub fn generate_columnar_and_open(card: Card, num_docs: u32) -> Column {
let reader = generate_columnar_with_name(card, num_docs, "price");
reader.read_columns("price").unwrap()[0]
.open_u64_lenient()
.unwrap()
.unwrap()
}
fn main() {
let mut inputs = Vec::new();
let mut add_card = |card1: Card| {
inputs.push((
format!("{card1}"),
generate_columnar_and_open(card1, NUM_DOCS),
));
};
add_card(Card::MultiSparse);
add_card(Card::Multi);
add_card(Card::Sparse);
add_card(Card::Dense);
add_card(Card::Full);
bench_group(InputGroup::new_with_inputs(inputs));
}
fn bench_group(mut runner: InputGroup<Column>) {
runner.register("access_values_for_doc", |column| {
let mut sum = 0;
for i in 0..NUM_DOCS {
for value in column.values_for_doc(i) {
sum += value;
}
}
black_box(sum);
});
runner.register("access_first_vals", |column| {
let mut sum = 0;
const BLOCK_SIZE: usize = 32;
let mut docs = vec![0; BLOCK_SIZE];
let mut buffer = vec![None; BLOCK_SIZE];
for i in (0..NUM_DOCS).step_by(BLOCK_SIZE) {
// fill docs
for idx in 0..BLOCK_SIZE {
docs[idx] = idx as u32 + i;
}
column.first_vals(&docs, &mut buffer);
for val in buffer.iter() {
let Some(val) = val else { continue };
sum += *val;
}
}
black_box(sum);
});
runner.run();
}

2
columnar/benches/bench_first_vals.rs
View File

@@ -31,7 +31,7 @@ fn get_test_columns() -> Columns {
}
let mut buffer: Vec<u8> = Vec::new();
dataframe_writer
.serialize(data.len() as u32, None, &mut buffer)
.serialize(data.len() as u32, &mut buffer)
.unwrap();
let columnar = ColumnarReader::open(buffer).unwrap();

49
columnar/benches/bench_merge.rs Normal file
View File

@@ -0,0 +1,49 @@
pub mod common;
use binggan::BenchRunner;
use common::{generate_columnar_with_name, Card};
use tantivy_columnar::*;
const NUM_DOCS: u32 = 100_000;
fn main() {
let mut inputs = Vec::new();
let mut add_combo = |card1: Card, card2: Card| {
inputs.push((
format!("merge_{card1}_and_{card2}"),
vec![
generate_columnar_with_name(card1, NUM_DOCS, "price"),
generate_columnar_with_name(card2, NUM_DOCS, "price"),
],
));
};
add_combo(Card::Multi, Card::Multi);
add_combo(Card::MultiSparse, Card::MultiSparse);
add_combo(Card::Dense, Card::Dense);
add_combo(Card::Sparse, Card::Sparse);
add_combo(Card::Sparse, Card::Dense);
add_combo(Card::MultiSparse, Card::Dense);
add_combo(Card::MultiSparse, Card::Sparse);
add_combo(Card::Multi, Card::Dense);
add_combo(Card::Multi, Card::Sparse);
let mut runner: BenchRunner = BenchRunner::new();
let mut group = runner.new_group();
for (input_name, columnar_readers) in inputs.iter() {
group.register_with_input(
input_name,
columnar_readers,
move |columnar_readers: &Vec<ColumnarReader>| {
let mut out = Vec::new();
let columnar_readers = columnar_readers.iter().collect::<Vec<_>>();
let merge_row_order = StackMergeOrder::stack(&columnar_readers[..]);
merge_columnar(&columnar_readers, &[], merge_row_order.into(), &mut out).unwrap();
Some(out.len() as u64)
},
);
}
group.run();
}

59
columnar/benches/common.rs Normal file
View File

@@ -0,0 +1,59 @@
extern crate tantivy_columnar;
use core::fmt;
use std::fmt::{Display, Formatter};
use tantivy_columnar::{ColumnarReader, ColumnarWriter};
pub enum Card {
MultiSparse,
Multi,
Sparse,
Dense,
Full,
}
impl Display for Card {
fn fmt(&self, f: &mut Formatter) -> fmt::Result {
match self {
Card::MultiSparse => write!(f, "multi sparse 1/13"),
Card::Multi => write!(f, "multi 2x"),
Card::Sparse => write!(f, "sparse 1/13"),
Card::Dense => write!(f, "dense 1/12"),
Card::Full => write!(f, "full"),
}
}
}
pub fn generate_columnar_with_name(card: Card, num_docs: u32, column_name: &str) -> ColumnarReader {
let mut columnar_writer = ColumnarWriter::default();
if let Card::MultiSparse = card {
columnar_writer.record_numerical(0, column_name, 10u64);
columnar_writer.record_numerical(0, column_name, 10u64);
}
for i in 0..num_docs {
match card {
Card::MultiSparse | Card::Sparse => {
if i % 13 == 0 {
columnar_writer.record_numerical(i, column_name, i as u64);
}
}
Card::Dense => {
if i % 12 == 0 {
columnar_writer.record_numerical(i, column_name, i as u64);
}
}
Card::Full => {
columnar_writer.record_numerical(i, column_name, i as u64);
}
Card::Multi => {
columnar_writer.record_numerical(i, column_name, i as u64);
columnar_writer.record_numerical(i, column_name, i as u64);
}
}
}
let mut wrt: Vec<u8> = Vec::new();
columnar_writer.serialize(num_docs, &mut wrt).unwrap();
ColumnarReader::open(wrt).unwrap()
}

BIN
columnar/compat_tests_data/v1.columnar Normal file
View File

Binary file not shown.

BIN
columnar/compat_tests_data/v2.columnar Normal file
View File

Binary file not shown.

6
columnar/src/TODO.md
View File

@@ -10,7 +10,7 @@
# Perf and Size
* remove alloc in `ord_to_term`
+ multivaued range queries restrat frm the beginning all of the time.
+ multivaued range queries restart from the beginning all of the time.
* re-add ZSTD compression for dictionaries
no systematic monotonic mapping
consider removing multilinear
@@ -30,7 +30,7 @@ investigate if should have better errors? io::Error is overused at the moment.
rename rank/select in unit tests
Review the public API via cargo doc
go through TODOs
remove all doc_id occurences -> row_id
remove all doc_id occurrences -> row_id
use the rank & select naming in unit tests branch.
multi-linear -> blockwise
linear codec -> simply a multiplication for the index column
@@ -43,5 +43,5 @@ isolate u128_based and uniform naming
# Other
fix enhance column-cli
# Santa claus
# Santa Claus
autodetect datetime ipaddr, plug customizable tokenizer.

2
columnar/src/block_accessor.rs
View File

@@ -66,7 +66,7 @@ impl<T: PartialOrd + Copy + std::fmt::Debug + Send + Sync + 'static + Default>
&'a self,
docs: &'a [u32],
accessor: &Column<T>,
) -> impl Iterator<Item = (DocId, T)> + '_ {
) -> impl Iterator<Item = (DocId, T)> + 'a {
if accessor.index.get_cardinality().is_full() {
docs.iter().cloned().zip(self.val_cache.iter().cloned())
} else {

2
columnar/src/column/mod.rs
View File

@@ -136,7 +136,7 @@ impl<T: PartialOrd + Copy + Debug + Send + Sync + 'static> Column<T> {
.map(|value_row_id: RowId| self.values.get_val(value_row_id))
}
/// Get the docids of values which are in the provided value range.
/// Get the docids of values which are in the provided value and docid range.
#[inline]
pub fn get_docids_for_value_range(
&self,

59
columnar/src/column/serialize.rs
View File

@@ -12,7 +12,7 @@ use crate::column_values::{
CodecType, MonotonicallyMappableToU128, MonotonicallyMappableToU64,
};
use crate::iterable::Iterable;
use crate::StrColumn;
use crate::{StrColumn, Version};
pub fn serialize_column_mappable_to_u128<T: MonotonicallyMappableToU128>(
column_index: SerializableColumnIndex<'_>,
@@ -40,25 +40,9 @@ pub fn serialize_column_mappable_to_u64<T: MonotonicallyMappableToU64>(
Ok(())
}
pub fn open_column_u64<T: MonotonicallyMappableToU64>(bytes: OwnedBytes) -> io::Result<Column<T>> {
let (body, column_index_num_bytes_payload) = bytes.rsplit(4);
let column_index_num_bytes = u32::from_le_bytes(
column_index_num_bytes_payload
.as_slice()
.try_into()
.unwrap(),
);
let (column_index_data, column_values_data) = body.split(column_index_num_bytes as usize);
let column_index = crate::column_index::open_column_index(column_index_data)?;
let column_values = load_u64_based_column_values(column_values_data)?;
Ok(Column {
index: column_index,
values: column_values,
})
}
pub fn open_column_u128<T: MonotonicallyMappableToU128>(
pub fn open_column_u64<T: MonotonicallyMappableToU64>(
bytes: OwnedBytes,
format_version: Version,
) -> io::Result<Column<T>> {
let (body, column_index_num_bytes_payload) = bytes.rsplit(4);
let column_index_num_bytes = u32::from_le_bytes(
@@ -68,7 +52,27 @@ pub fn open_column_u128<T: MonotonicallyMappableToU128>(
.unwrap(),
);
let (column_index_data, column_values_data) = body.split(column_index_num_bytes as usize);
let column_index = crate::column_index::open_column_index(column_index_data)?;
let column_index = crate::column_index::open_column_index(column_index_data, format_version)?;
let column_values = load_u64_based_column_values(column_values_data)?;
Ok(Column {
index: column_index,
values: column_values,
})
}
pub fn open_column_u128<T: MonotonicallyMappableToU128>(
bytes: OwnedBytes,
format_version: Version,
) -> io::Result<Column<T>> {
let (body, column_index_num_bytes_payload) = bytes.rsplit(4);
let column_index_num_bytes = u32::from_le_bytes(
column_index_num_bytes_payload
.as_slice()
.try_into()
.unwrap(),
);
let (column_index_data, column_values_data) = body.split(column_index_num_bytes as usize);
let column_index = crate::column_index::open_column_index(column_index_data, format_version)?;
let column_values = crate::column_values::open_u128_mapped(column_values_data)?;
Ok(Column {
index: column_index,
@@ -79,7 +83,10 @@ pub fn open_column_u128<T: MonotonicallyMappableToU128>(
/// Open the column as u64.
///
/// See [`open_u128_as_compact_u64`] for more details.
pub fn open_column_u128_as_compact_u64(bytes: OwnedBytes) -> io::Result<Column<u64>> {
pub fn open_column_u128_as_compact_u64(
bytes: OwnedBytes,
format_version: Version,
) -> io::Result<Column<u64>> {
let (body, column_index_num_bytes_payload) = bytes.rsplit(4);
let column_index_num_bytes = u32::from_le_bytes(
column_index_num_bytes_payload
@@ -88,7 +95,7 @@ pub fn open_column_u128_as_compact_u64(bytes: OwnedBytes) -> io::Result<Column<u
.unwrap(),
);
let (column_index_data, column_values_data) = body.split(column_index_num_bytes as usize);
let column_index = crate::column_index::open_column_index(column_index_data)?;
let column_index = crate::column_index::open_column_index(column_index_data, format_version)?;
let column_values = crate::column_values::open_u128_as_compact_u64(column_values_data)?;
Ok(Column {
index: column_index,
@@ -96,19 +103,19 @@ pub fn open_column_u128_as_compact_u64(bytes: OwnedBytes) -> io::Result<Column<u
})
}
pub fn open_column_bytes(data: OwnedBytes) -> io::Result<BytesColumn> {
pub fn open_column_bytes(data: OwnedBytes, format_version: Version) -> io::Result<BytesColumn> {
let (body, dictionary_len_bytes) = data.rsplit(4);
let dictionary_len = u32::from_le_bytes(dictionary_len_bytes.as_slice().try_into().unwrap());
let (dictionary_bytes, column_bytes) = body.split(dictionary_len as usize);
let dictionary = Arc::new(Dictionary::from_bytes(dictionary_bytes)?);
let term_ord_column = crate::column::open_column_u64::<u64>(column_bytes)?;
let term_ord_column = crate::column::open_column_u64::<u64>(column_bytes, format_version)?;
Ok(BytesColumn {
dictionary,
term_ord_column,
})
}
pub fn open_column_str(data: OwnedBytes) -> io::Result<StrColumn> {
let bytes_column = open_column_bytes(data)?;
pub fn open_column_str(data: OwnedBytes, format_version: Version) -> io::Result<StrColumn> {
let bytes_column = open_column_bytes(data, format_version)?;
Ok(StrColumn::wrap(bytes_column))
}

23
columnar/src/column_index/merge/mod.rs
View File

@@ -95,8 +95,12 @@ pub fn merge_column_index<'a>(
#[cfg(test)]
mod tests {
use common::OwnedBytes;
use crate::column_index::merge::detect_cardinality;
use crate::column_index::multivalued_index::MultiValueIndex;
use crate::column_index::multivalued_index::{
open_multivalued_index, serialize_multivalued_index, MultiValueIndex,
};
use crate::column_index::{merge_column_index, OptionalIndex, SerializableColumnIndex};
use crate::{
Cardinality, ColumnIndex, MergeRowOrder, RowAddr, RowId, ShuffleMergeOrder, StackMergeOrder,
@@ -169,9 +173,13 @@ mod tests {
.into();
let merged_column_index = merge_column_index(&column_indexes[..], &merge_row_order);
let SerializableColumnIndex::Multivalued(start_index_iterable) = merged_column_index else {
panic!("Excpected a multivalued index")
panic!("Expected a multivalued index")
};
let start_indexes: Vec<RowId> = start_index_iterable.boxed_iter().collect();
let mut output = Vec::new();
serialize_multivalued_index(&start_index_iterable, &mut output).unwrap();
let multivalue =
open_multivalued_index(OwnedBytes::new(output), crate::Version::V2).unwrap();
let start_indexes: Vec<RowId> = multivalue.get_start_index_column().iter().collect();
assert_eq!(&start_indexes, &[0, 3, 5]);
}
@@ -200,11 +208,16 @@ mod tests {
],
)
.into();
let merged_column_index = merge_column_index(&column_indexes[..], &merge_row_order);
let SerializableColumnIndex::Multivalued(start_index_iterable) = merged_column_index else {
panic!("Excpected a multivalued index")
panic!("Expected a multivalued index")
};
let start_indexes: Vec<RowId> = start_index_iterable.boxed_iter().collect();
let mut output = Vec::new();
serialize_multivalued_index(&start_index_iterable, &mut output).unwrap();
let multivalue =
open_multivalued_index(OwnedBytes::new(output), crate::Version::V2).unwrap();
let start_indexes: Vec<RowId> = multivalue.get_start_index_column().iter().collect();
assert_eq!(&start_indexes, &[0, 3, 5, 6]);
}
}

48
columnar/src/column_index/merge/shuffled.rs
View File

@@ -1,6 +1,8 @@
use std::iter;
use crate::column_index::{SerializableColumnIndex, Set};
use crate::column_index::{
SerializableColumnIndex, SerializableMultivalueIndex, SerializableOptionalIndex, Set,
};
use crate::iterable::Iterable;
use crate::{Cardinality, ColumnIndex, RowId, ShuffleMergeOrder};
@@ -14,15 +16,24 @@ pub fn merge_column_index_shuffled<'a>(
Cardinality::Optional => {
let non_null_row_ids =
merge_column_index_shuffled_optional(column_indexes, shuffle_merge_order);
SerializableColumnIndex::Optional {
SerializableColumnIndex::Optional(SerializableOptionalIndex {
non_null_row_ids,
num_rows: shuffle_merge_order.num_rows(),
}
})
}
Cardinality::Multivalued => {
let multivalue_start_index =
merge_column_index_shuffled_multivalued(column_indexes, shuffle_merge_order);
SerializableColumnIndex::Multivalued(multivalue_start_index)
let non_null_row_ids =
merge_column_index_shuffled_optional(column_indexes, shuffle_merge_order);
SerializableColumnIndex::Multivalued(SerializableMultivalueIndex {
doc_ids_with_values: SerializableOptionalIndex {
non_null_row_ids,
num_rows: shuffle_merge_order.num_rows(),
},
start_offsets: merge_column_index_shuffled_multivalued(
column_indexes,
shuffle_merge_order,
),
})
}
}
}
@@ -47,7 +58,7 @@ struct ShuffledIndex<'a> {
merge_order: &'a ShuffleMergeOrder,
}
impl<'a> Iterable<u32> for ShuffledIndex<'a> {
impl Iterable<u32> for ShuffledIndex<'_> {
fn boxed_iter(&self) -> Box<dyn Iterator<Item = u32> + '_> {
Box::new(
self.merge_order
@@ -102,14 +113,21 @@ fn iter_num_values<'a>(
/// Transforms an iterator containing the number of vals per row (with `num_rows` elements)
/// into a `start_offset` iterator starting at 0 and (with `num_rows + 1` element)
///
/// This will filter values with 0 values as these are covered by the optional index in the
/// multivalue index.
fn integrate_num_vals(num_vals: impl Iterator<Item = u32>) -> impl Iterator<Item = RowId> {
iter::once(0u32).chain(num_vals.scan(0, |state, num_vals| {
*state += num_vals;
Some(*state)
}))
iter::once(0u32).chain(
num_vals
.filter(|num_vals| *num_vals != 0)
.scan(0, |state, num_vals| {
*state += num_vals;
Some(*state)
}),
)
}
impl<'a> Iterable<u32> for ShuffledMultivaluedIndex<'a> {
impl Iterable<u32> for ShuffledMultivaluedIndex<'_> {
fn boxed_iter(&self) -> Box<dyn Iterator<Item = u32> + '_> {
let num_vals_per_row = iter_num_values(self.column_indexes, self.merge_order);
Box::new(integrate_num_vals(num_vals_per_row))
@@ -134,7 +152,7 @@ mod tests {
#[test]
fn test_integrate_num_vals_several() {
assert!(integrate_num_vals([3, 0, 10, 20].into_iter()).eq([0, 3, 3, 13, 33].into_iter()));
assert!(integrate_num_vals([3, 0, 10, 20].into_iter()).eq([0, 3, 13, 33].into_iter()));
}
#[test]
@@ -157,10 +175,10 @@ mod tests {
Cardinality::Optional,
&shuffle_merge_order,
);
let SerializableColumnIndex::Optional {
let SerializableColumnIndex::Optional(SerializableOptionalIndex {
non_null_row_ids,
num_rows,
} = serializable_index
}) = serializable_index
else {
panic!()
};

230
columnar/src/column_index/merge/stacked.rs
View File

@@ -1,6 +1,8 @@
use std::iter;
use std::ops::Range;
use crate::column_index::{SerializableColumnIndex, Set};
use crate::column_index::multivalued_index::{MultiValueIndex, SerializableMultivalueIndex};
use crate::column_index::serialize::SerializableOptionalIndex;
use crate::column_index::SerializableColumnIndex;
use crate::iterable::Iterable;
use crate::{Cardinality, ColumnIndex, RowId, StackMergeOrder};
@@ -15,23 +17,149 @@ pub fn merge_column_index_stacked<'a>(
) -> SerializableColumnIndex<'a> {
match cardinality_after_merge {
Cardinality::Full => SerializableColumnIndex::Full,
Cardinality::Optional => SerializableColumnIndex::Optional {
Cardinality::Optional => SerializableColumnIndex::Optional(SerializableOptionalIndex {
non_null_row_ids: Box::new(StackedOptionalIndex {
columns,
stack_merge_order,
}),
num_rows: stack_merge_order.num_rows(),
},
}),
Cardinality::Multivalued => {
let stacked_multivalued_index = StackedMultivaluedIndex {
columns,
stack_merge_order,
};
SerializableColumnIndex::Multivalued(Box::new(stacked_multivalued_index))
let serializable_multivalue_index =
make_serializable_multivalued_index(columns, stack_merge_order);
SerializableColumnIndex::Multivalued(serializable_multivalue_index)
}
}
}
struct StackedDocIdsWithValues<'a> {
column_indexes: &'a [ColumnIndex],
stack_merge_order: &'a StackMergeOrder,
}
impl Iterable<u32> for StackedDocIdsWithValues<'_> {
fn boxed_iter(&self) -> Box<dyn Iterator<Item = u32> + '_> {
Box::new((0..self.column_indexes.len()).flat_map(|i| {
let column_index = &self.column_indexes[i];
let doc_range = self.stack_merge_order.columnar_range(i);
get_doc_ids_with_values(column_index, doc_range)
}))
}
}
fn get_doc_ids_with_values<'a>(
column_index: &'a ColumnIndex,
doc_range: Range<u32>,
) -> Box<dyn Iterator<Item = u32> + 'a> {
match column_index {
ColumnIndex::Empty { .. } => Box::new(0..0),
ColumnIndex::Full => Box::new(doc_range),
ColumnIndex::Optional(optional_index) => Box::new(
optional_index
.iter_rows()
.map(move |row| row + doc_range.start),
),
ColumnIndex::Multivalued(multivalued_index) => match multivalued_index {
MultiValueIndex::MultiValueIndexV1(multivalued_index) => {
Box::new((0..multivalued_index.num_docs()).filter_map(move |docid| {
let range = multivalued_index.range(docid);
if range.is_empty() {
None
} else {
Some(docid + doc_range.start)
}
}))
}
MultiValueIndex::MultiValueIndexV2(multivalued_index) => Box::new(
multivalued_index
.optional_index
.iter_rows()
.map(move |row| row + doc_range.start),
),
},
}
}
fn stack_doc_ids_with_values<'a>(
column_indexes: &'a [ColumnIndex],
stack_merge_order: &'a StackMergeOrder,
) -> SerializableOptionalIndex<'a> {
let num_rows = stack_merge_order.num_rows();
SerializableOptionalIndex {
non_null_row_ids: Box::new(StackedDocIdsWithValues {
column_indexes,
stack_merge_order,
}),
num_rows,
}
}
struct StackedStartOffsets<'a> {
column_indexes: &'a [ColumnIndex],
stack_merge_order: &'a StackMergeOrder,
}
fn get_num_values_iterator<'a>(
column_index: &'a ColumnIndex,
num_docs: u32,
) -> Box<dyn Iterator<Item = u32> + 'a> {
match column_index {
ColumnIndex::Empty { .. } => Box::new(std::iter::empty()),
ColumnIndex::Full => Box::new(std::iter::repeat(1u32).take(num_docs as usize)),
ColumnIndex::Optional(optional_index) => {
Box::new(std::iter::repeat(1u32).take(optional_index.num_non_nulls() as usize))
}
ColumnIndex::Multivalued(multivalued_index) => Box::new(
multivalued_index
.get_start_index_column()
.iter()
.scan(0u32, |previous_start_offset, current_start_offset| {
let num_vals = current_start_offset - *previous_start_offset;
*previous_start_offset = current_start_offset;
Some(num_vals)
})
.skip(1),
),
}
}
impl Iterable<u32> for StackedStartOffsets<'_> {
fn boxed_iter(&self) -> Box<dyn Iterator<Item = u32> + '_> {
let num_values_it = (0..self.column_indexes.len()).flat_map(|columnar_id| {
let num_docs = self.stack_merge_order.columnar_range(columnar_id).len() as u32;
let column_index = &self.column_indexes[columnar_id];
get_num_values_iterator(column_index, num_docs)
});
Box::new(std::iter::once(0u32).chain(num_values_it.into_iter().scan(
0u32,
|cumulated, el| {
*cumulated += el;
Some(*cumulated)
},
)))
}
}
fn stack_start_offsets<'a>(
column_indexes: &'a [ColumnIndex],
stack_merge_order: &'a StackMergeOrder,
) -> Box<dyn Iterable<u32> + 'a> {
Box::new(StackedStartOffsets {
column_indexes,
stack_merge_order,
})
}
fn make_serializable_multivalued_index<'a>(
columns: &'a [ColumnIndex],
stack_merge_order: &'a StackMergeOrder,
) -> SerializableMultivalueIndex<'a> {
SerializableMultivalueIndex {
doc_ids_with_values: stack_doc_ids_with_values(columns, stack_merge_order),
start_offsets: stack_start_offsets(columns, stack_merge_order),
}
}
struct StackedOptionalIndex<'a> {
columns: &'a [ColumnIndex],
stack_merge_order: &'a StackMergeOrder,
@@ -62,87 +190,3 @@ impl<'a> Iterable<RowId> for StackedOptionalIndex<'a> {
)
}
}
#[derive(Clone, Copy)]
struct StackedMultivaluedIndex<'a> {
columns: &'a [ColumnIndex],
stack_merge_order: &'a StackMergeOrder,
}
fn convert_column_opt_to_multivalued_index<'a>(
column_index_opt: &'a ColumnIndex,
num_rows: RowId,
) -> Box<dyn Iterator<Item = RowId> + 'a> {
match column_index_opt {
ColumnIndex::Empty { .. } => Box::new(iter::repeat(0u32).take(num_rows as usize + 1)),
ColumnIndex::Full => Box::new(0..num_rows + 1),
ColumnIndex::Optional(optional_index) => {
Box::new(
(0..num_rows)
// TODO optimize
.map(|row_id| optional_index.rank(row_id))
.chain(std::iter::once(optional_index.num_non_nulls())),
)
}
ColumnIndex::Multivalued(multivalued_index) => multivalued_index.start_index_column.iter(),
}
}
impl<'a> Iterable<RowId> for StackedMultivaluedIndex<'a> {
fn boxed_iter(&self) -> Box<dyn Iterator<Item = RowId> + '_> {
let multivalued_indexes =
self.columns
.iter()
.enumerate()
.map(|(columnar_id, column_opt)| {
let num_rows =
self.stack_merge_order.columnar_range(columnar_id).len() as RowId;
convert_column_opt_to_multivalued_index(column_opt, num_rows)
});
stack_multivalued_indexes(multivalued_indexes)
}
}
// Refactor me
fn stack_multivalued_indexes<'a>(
mut multivalued_indexes: impl Iterator<Item = Box<dyn Iterator<Item = RowId> + 'a>> + 'a,
) -> Box<dyn Iterator<Item = RowId> + 'a> {
let mut offset = 0;
let mut last_row_id = 0;
let mut current_it = multivalued_indexes.next();
Box::new(std::iter::from_fn(move || loop {
if let Some(row_id) = current_it.as_mut()?.next() {
last_row_id = offset + row_id;
return Some(last_row_id);
}
offset = last_row_id;
loop {
current_it = multivalued_indexes.next();
if current_it.as_mut()?.next().is_some() {
break;
}
}
}))
}
#[cfg(test)]
mod tests {
use crate::RowId;
fn it<'a>(row_ids: &'a [RowId]) -> Box<dyn Iterator<Item = RowId> + 'a> {
Box::new(row_ids.iter().copied())
}
#[test]
fn test_stack() {
let columns = [
it(&[0u32, 0u32]),
it(&[0u32, 1u32, 1u32, 4u32]),
it(&[0u32, 3u32, 5u32]),
it(&[0u32, 4u32]),
]
.into_iter();
let start_offsets: Vec<RowId> = super::stack_multivalued_indexes(columns).collect();
assert_eq!(start_offsets, &[0, 0, 1, 1, 4, 7, 9, 13]);
}
}

49
columnar/src/column_index/mod.rs
View File

@@ -11,8 +11,11 @@ mod serialize;
use std::ops::Range;
pub use merge::merge_column_index;
pub(crate) use multivalued_index::SerializableMultivalueIndex;
pub use optional_index::{OptionalIndex, Set};
pub use serialize::{open_column_index, serialize_column_index, SerializableColumnIndex};
pub use serialize::{
open_column_index, serialize_column_index, SerializableColumnIndex, SerializableOptionalIndex,
};
use crate::column_index::multivalued_index::MultiValueIndex;
use crate::{Cardinality, DocId, RowId};
@@ -25,7 +28,7 @@ pub enum ColumnIndex {
Full,
Optional(OptionalIndex),
/// In addition, at index num_rows, an extra value is added
/// containing the overal number of values.
/// containing the overall number of values.
Multivalued(MultiValueIndex),
}
@@ -131,15 +134,41 @@ impl ColumnIndex {
let row_end = optional_index.rank(doc_id_range.end);
row_start..row_end
}
ColumnIndex::Multivalued(multivalued_index) => {
let end_docid = doc_id_range.end.min(multivalued_index.num_docs() - 1) + 1;
let start_docid = doc_id_range.start.min(end_docid);
ColumnIndex::Multivalued(multivalued_index) => match multivalued_index {
MultiValueIndex::MultiValueIndexV1(index) => {
let row_start = index.start_index_column.get_val(doc_id_range.start);
let row_end = index.start_index_column.get_val(doc_id_range.end);
row_start..row_end
}
MultiValueIndex::MultiValueIndexV2(index) => {
// In this case we will use the optional_index select the next values
// that are valid. There are different cases to consider:
// Not exists below means does not exist in the optional
// index, because it has no values.
// * doc_id_range may cover a range of docids which are non existent
// => rank
// will give us the next document outside the range with a value. They both
// get the same rank and therefore return a zero range
//
// * doc_id_range.start and doc_id_range.end may not exist, but docids in
// between may have values
// => rank will give us the next document outside the range with a value.
//
// * doc_id_range.start may be not existent but doc_id_range.end may exist
// * doc_id_range.start may exist but doc_id_range.end may not exist
// * doc_id_range.start and doc_id_range.end may exist
// => rank on doc_id_range.end will give use the next value, which matches
// how the `start_index_column` works, so we get the value start of the next
// docid which we use to create the exclusive range.
//
let rank_start = index.optional_index.rank(doc_id_range.start);
let row_start = index.start_index_column.get_val(rank_start);
let rank_end = index.optional_index.rank(doc_id_range.end);
let row_end = index.start_index_column.get_val(rank_end);
let row_start = multivalued_index.start_index_column.get_val(start_docid);
let row_end = multivalued_index.start_index_column.get_val(end_docid);
row_start..row_end
}
row_start..row_end
}
},
}
}

318
columnar/src/column_index/multivalued_index.rs
View File

@@ -3,64 +3,98 @@ use std::io::Write;
use std::ops::Range;
use std::sync::Arc;
use common::OwnedBytes;
use common::{CountingWriter, OwnedBytes};
use super::optional_index::{open_optional_index, serialize_optional_index};
use super::{OptionalIndex, SerializableOptionalIndex, Set};
use crate::column_values::{
load_u64_based_column_values, serialize_u64_based_column_values, CodecType, ColumnValues,
};
use crate::iterable::Iterable;
use crate::{DocId, RowId};
use crate::{DocId, RowId, Version};
pub struct SerializableMultivalueIndex<'a> {
pub doc_ids_with_values: SerializableOptionalIndex<'a>,
pub start_offsets: Box<dyn Iterable<u32> + 'a>,
}
pub fn serialize_multivalued_index(
multivalued_index: &dyn Iterable<RowId>,
multivalued_index: &SerializableMultivalueIndex,
output: &mut impl Write,
) -> io::Result<()> {
let SerializableMultivalueIndex {
doc_ids_with_values,
start_offsets,
} = multivalued_index;
let mut count_writer = CountingWriter::wrap(output);
let SerializableOptionalIndex {
non_null_row_ids,
num_rows,
} = doc_ids_with_values;
serialize_optional_index(&**non_null_row_ids, *num_rows, &mut count_writer)?;
let optional_len = count_writer.written_bytes() as u32;
let output = count_writer.finish();
serialize_u64_based_column_values(
multivalued_index,
&**start_offsets,
&[CodecType::Bitpacked, CodecType::Linear],
output,
)?;
output.write_all(&optional_len.to_le_bytes())?;
Ok(())
}
pub fn open_multivalued_index(bytes: OwnedBytes) -> io::Result<MultiValueIndex> {
let start_index_column: Arc<dyn ColumnValues<RowId>> = load_u64_based_column_values(bytes)?;
Ok(MultiValueIndex { start_index_column })
pub fn open_multivalued_index(
bytes: OwnedBytes,
format_version: Version,
) -> io::Result<MultiValueIndex> {
match format_version {
Version::V1 => {
let start_index_column: Arc<dyn ColumnValues<RowId>> =
load_u64_based_column_values(bytes)?;
Ok(MultiValueIndex::MultiValueIndexV1(MultiValueIndexV1 {
start_index_column,
}))
}
Version::V2 => {
let (body_bytes, optional_index_len) = bytes.rsplit(4);
let optional_index_len =
u32::from_le_bytes(optional_index_len.as_slice().try_into().unwrap());
let (optional_index_bytes, start_index_bytes) =
body_bytes.split(optional_index_len as usize);
let optional_index = open_optional_index(optional_index_bytes)?;
let start_index_column: Arc<dyn ColumnValues<RowId>> =
load_u64_based_column_values(start_index_bytes)?;
Ok(MultiValueIndex::MultiValueIndexV2(MultiValueIndexV2 {
optional_index,
start_index_column,
}))
}
}
}
#[derive(Clone)]
/// Index to resolve value range for given doc_id.
/// Starts at 0.
pub struct MultiValueIndex {
pub enum MultiValueIndex {
MultiValueIndexV1(MultiValueIndexV1),
MultiValueIndexV2(MultiValueIndexV2),
}
#[derive(Clone)]
/// Index to resolve value range for given doc_id.
/// Starts at 0.
pub struct MultiValueIndexV1 {
pub start_index_column: Arc<dyn crate::ColumnValues<RowId>>,
}
impl std::fmt::Debug for MultiValueIndex {
fn fmt(&self, f: &mut std::fmt::Formatter) -> std::fmt::Result {
f.debug_struct("MultiValuedIndex")
.field("num_rows", &self.start_index_column.num_vals())
.finish_non_exhaustive()
}
}
impl From<Arc<dyn ColumnValues<RowId>>> for MultiValueIndex {
fn from(start_index_column: Arc<dyn ColumnValues<RowId>>) -> Self {
MultiValueIndex { start_index_column }
}
}
impl MultiValueIndex {
pub fn for_test(start_offsets: &[RowId]) -> MultiValueIndex {
let mut buffer = Vec::new();
serialize_multivalued_index(&start_offsets, &mut buffer).unwrap();
let bytes = OwnedBytes::new(buffer);
open_multivalued_index(bytes).unwrap()
}
impl MultiValueIndexV1 {
/// Returns `[start, end)`, such that the values associated with
/// the given document are `start..end`.
#[inline]
pub(crate) fn range(&self, doc_id: DocId) -> Range<RowId> {
if doc_id >= self.num_docs() {
return 0..0;
}
let start = self.start_index_column.get_val(doc_id);
let end = self.start_index_column.get_val(doc_id + 1);
start..end
@@ -83,7 +117,6 @@ impl MultiValueIndex {
///
/// TODO: Instead of a linear scan we can employ a exponential search into binary search to
/// match a docid to its value position.
#[allow(clippy::bool_to_int_with_if)]
pub(crate) fn select_batch_in_place(&self, docid_start: DocId, ranks: &mut Vec<u32>) {
if ranks.is_empty() {
return;
@@ -111,11 +144,170 @@ impl MultiValueIndex {
}
}
#[derive(Clone)]
/// Index to resolve value range for given doc_id.
/// Starts at 0.
pub struct MultiValueIndexV2 {
pub optional_index: OptionalIndex,
pub start_index_column: Arc<dyn crate::ColumnValues<RowId>>,
}
impl std::fmt::Debug for MultiValueIndex {
fn fmt(&self, f: &mut std::fmt::Formatter) -> std::fmt::Result {
let index = match self {
MultiValueIndex::MultiValueIndexV1(idx) => &idx.start_index_column,
MultiValueIndex::MultiValueIndexV2(idx) => &idx.start_index_column,
};
f.debug_struct("MultiValuedIndex")
.field("num_rows", &index.num_vals())
.finish_non_exhaustive()
}
}
impl MultiValueIndex {
pub fn for_test(start_offsets: &[RowId]) -> MultiValueIndex {
assert!(!start_offsets.is_empty());
assert_eq!(start_offsets[0], 0);
let mut doc_with_values = Vec::new();
let mut compact_start_offsets: Vec<u32> = vec![0];
for doc in 0..start_offsets.len() - 1 {
if start_offsets[doc] < start_offsets[doc + 1] {
doc_with_values.push(doc as RowId);
compact_start_offsets.push(start_offsets[doc + 1]);
}
}
let serializable_multivalued_index = SerializableMultivalueIndex {
doc_ids_with_values: SerializableOptionalIndex {
non_null_row_ids: Box::new(&doc_with_values[..]),
num_rows: start_offsets.len() as u32 - 1,
},
start_offsets: Box::new(&compact_start_offsets[..]),
};
let mut buffer = Vec::new();
serialize_multivalued_index(&serializable_multivalued_index, &mut buffer).unwrap();
let bytes = OwnedBytes::new(buffer);
open_multivalued_index(bytes, Version::V2).unwrap()
}
pub fn get_start_index_column(&self) -> &Arc<dyn crate::ColumnValues<RowId>> {
match self {
MultiValueIndex::MultiValueIndexV1(idx) => &idx.start_index_column,
MultiValueIndex::MultiValueIndexV2(idx) => &idx.start_index_column,
}
}
/// Returns `[start, end)` values range, such that the values associated with
/// the given document are `start..end`.
#[inline]
pub(crate) fn range(&self, doc_id: DocId) -> Range<RowId> {
match self {
MultiValueIndex::MultiValueIndexV1(idx) => idx.range(doc_id),
MultiValueIndex::MultiValueIndexV2(idx) => idx.range(doc_id),
}
}
/// Returns the number of documents in the index.
#[inline]
pub fn num_docs(&self) -> u32 {
match self {
MultiValueIndex::MultiValueIndexV1(idx) => idx.start_index_column.num_vals() - 1,
MultiValueIndex::MultiValueIndexV2(idx) => idx.optional_index.num_docs(),
}
}
/// Converts a list of ranks (row ids of values) in a 1:n index to the corresponding list of
/// docids. Positions are converted inplace to docids.
///
/// Since there is no index for value pos -> docid, but docid -> value pos range, we scan the
/// index.
///
/// Correctness: positions needs to be sorted. idx_reader needs to contain monotonically
/// increasing positions.
///
/// TODO: Instead of a linear scan we can employ a exponential search into binary search to
/// match a docid to its value position.
pub(crate) fn select_batch_in_place(&self, docid_start: DocId, ranks: &mut Vec<u32>) {
match self {
MultiValueIndex::MultiValueIndexV1(idx) => {
idx.select_batch_in_place(docid_start, ranks)
}
MultiValueIndex::MultiValueIndexV2(idx) => {
idx.select_batch_in_place(docid_start, ranks)
}
}
}
}
impl MultiValueIndexV2 {
/// Returns `[start, end)`, such that the values associated with
/// the given document are `start..end`.
#[inline]
pub(crate) fn range(&self, doc_id: DocId) -> Range<RowId> {
let Some(rank) = self.optional_index.rank_if_exists(doc_id) else {
return 0..0;
};
let start = self.start_index_column.get_val(rank);
let end = self.start_index_column.get_val(rank + 1);
start..end
}
/// Returns the number of documents in the index.
#[inline]
pub fn num_docs(&self) -> u32 {
self.optional_index.num_docs()
}
/// Converts a list of ranks (row ids of values) in a 1:n index to the corresponding list of
/// docids. Positions are converted inplace to docids.
///
/// Since there is no index for value pos -> docid, but docid -> value pos range, we scan the
/// index.
///
/// Correctness: positions needs to be sorted. idx_reader needs to contain monotonically
/// increasing positions.
///
/// TODO: Instead of a linear scan we can employ a exponential search into binary search to
/// match a docid to its value position.
pub(crate) fn select_batch_in_place(&self, docid_start: DocId, ranks: &mut Vec<u32>) {
if ranks.is_empty() {
return;
}
let mut cur_pos_in_idx = self.optional_index.rank(docid_start);
let mut last_doc = None;
assert!(cur_pos_in_idx <= ranks[0]);
let mut write_doc_pos = 0;
for i in 0..ranks.len() {
let pos = ranks[i];
loop {
let end = self.start_index_column.get_val(cur_pos_in_idx + 1);
if end > pos {
ranks[write_doc_pos] = cur_pos_in_idx;
write_doc_pos += if last_doc == Some(cur_pos_in_idx) {
0
} else {
1
};
last_doc = Some(cur_pos_in_idx);
break;
}
cur_pos_in_idx += 1;
}
}
ranks.truncate(write_doc_pos);
for rank in ranks.iter_mut() {
*rank = self.optional_index.select(*rank);
}
}
}
#[cfg(test)]
mod tests {
use std::ops::Range;
use super::MultiValueIndex;
use crate::{ColumnarReader, DynamicColumn};
fn index_to_pos_helper(
index: &MultiValueIndex,
@@ -134,6 +326,7 @@ mod tests {
let positions = &[10u32, 11, 15, 20, 21, 22];
assert_eq!(index_to_pos_helper(&index, 0..5, positions), vec![1, 3, 4]);
assert_eq!(index_to_pos_helper(&index, 1..5, positions), vec![1, 3, 4]);
assert_eq!(index_to_pos_helper(&index, 0..5, &[9]), vec![0]);
assert_eq!(index_to_pos_helper(&index, 1..5, &[10]), vec![1]);
assert_eq!(index_to_pos_helper(&index, 1..5, &[11]), vec![1]);
@@ -141,4 +334,67 @@ mod tests {
assert_eq!(index_to_pos_helper(&index, 2..5, &[12, 14]), vec![2]);
assert_eq!(index_to_pos_helper(&index, 2..5, &[12, 14, 15]), vec![2, 3]);
}
#[test]
fn test_range_to_rowids() {
use crate::ColumnarWriter;
let mut columnar_writer = ColumnarWriter::default();
// This column gets coerced to u64
columnar_writer.record_numerical(1, "full", u64::MAX);
columnar_writer.record_numerical(1, "full", u64::MAX);
columnar_writer.record_numerical(5, "full", u64::MAX);
columnar_writer.record_numerical(5, "full", u64::MAX);
let mut wrt: Vec<u8> = Vec::new();
columnar_writer.serialize(7, &mut wrt).unwrap();
let reader = ColumnarReader::open(wrt).unwrap();
// Open the column as u64
let column = reader.read_columns("full").unwrap()[0]
.open()
.unwrap()
.coerce_numerical(crate::NumericalType::U64)
.unwrap();
let DynamicColumn::U64(column) = column else {
panic!();
};
let row_id_range = column.index.docid_range_to_rowids(1..2);
assert_eq!(row_id_range, 0..2);
let row_id_range = column.index.docid_range_to_rowids(0..2);
assert_eq!(row_id_range, 0..2);
let row_id_range = column.index.docid_range_to_rowids(0..4);
assert_eq!(row_id_range, 0..2);
let row_id_range = column.index.docid_range_to_rowids(3..4);
assert_eq!(row_id_range, 2..2);
let row_id_range = column.index.docid_range_to_rowids(1..6);
assert_eq!(row_id_range, 0..4);
let row_id_range = column.index.docid_range_to_rowids(3..6);
assert_eq!(row_id_range, 2..4);
let row_id_range = column.index.docid_range_to_rowids(0..6);
assert_eq!(row_id_range, 0..4);
let row_id_range = column.index.docid_range_to_rowids(0..6);
assert_eq!(row_id_range, 0..4);
let check = |range, expected| {
let full_range = 0..=u64::MAX;
let mut docids = Vec::new();
column.get_docids_for_value_range(full_range, range, &mut docids);
assert_eq!(docids, expected);
};
// check(0..1, vec![]);
// check(0..2, vec![1]);
check(1..2, vec![1]);
}
}

21
columnar/src/column_index/optional_index/mod.rs
View File

@@ -86,8 +86,14 @@ pub struct OptionalIndex {
block_metas: Arc<[BlockMeta]>,
}
impl Iterable<u32> for &OptionalIndex {
fn boxed_iter(&self) -> Box<dyn Iterator<Item = u32> + '_> {
Box::new(self.iter_rows())
}
}
impl std::fmt::Debug for OptionalIndex {
fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result {
fn fmt(&self, f: &mut std::fmt::Formatter) -> std::fmt::Result {
f.debug_struct("OptionalIndex")
.field("num_rows", &self.num_rows)
.field("num_non_null_rows", &self.num_non_null_rows)
@@ -117,7 +123,7 @@ enum BlockSelectCursor<'a> {
Sparse(<SparseBlock<'a> as Set<u16>>::SelectCursor<'a>),
}
impl<'a> BlockSelectCursor<'a> {
impl BlockSelectCursor<'_> {
fn select(&mut self, rank: u16) -> u16 {
match self {
BlockSelectCursor::Dense(dense_select_cursor) => dense_select_cursor.select(rank),
@@ -135,7 +141,7 @@ pub struct OptionalIndexSelectCursor<'a> {
num_null_rows_before_block: RowId,
}
impl<'a> OptionalIndexSelectCursor<'a> {
impl OptionalIndexSelectCursor<'_> {
fn search_and_load_block(&mut self, rank: RowId) {
if rank < self.current_block_end_rank {
// we are already in the right block
@@ -159,7 +165,7 @@ impl<'a> OptionalIndexSelectCursor<'a> {
}
}
impl<'a> SelectCursor<RowId> for OptionalIndexSelectCursor<'a> {
impl SelectCursor<RowId> for OptionalIndexSelectCursor<'_> {
fn select(&mut self, rank: RowId) -> RowId {
self.search_and_load_block(rank);
let index_in_block = (rank - self.num_null_rows_before_block) as u16;
@@ -168,7 +174,9 @@ impl<'a> SelectCursor<RowId> for OptionalIndexSelectCursor<'a> {
}
impl Set<RowId> for OptionalIndex {
type SelectCursor<'b> = OptionalIndexSelectCursor<'b> where Self: 'b;
type SelectCursor<'b>
= OptionalIndexSelectCursor<'b>
where Self: 'b;
// Check if value at position is not null.
#[inline]
fn contains(&self, row_id: RowId) -> bool {
@@ -196,6 +204,7 @@ impl Set<RowId> for OptionalIndex {
} = row_addr_from_row_id(doc_id);
let block_meta = self.block_metas[block_id as usize];
let block = self.block(block_meta);
let block_offset_row_id = match block {
Block::Dense(dense_block) => dense_block.rank(in_block_row_id),
Block::Sparse(sparse_block) => sparse_block.rank(in_block_row_id),
@@ -496,7 +505,7 @@ fn deserialize_optional_index_block_metadatas(
non_null_rows_before_block += num_non_null_rows;
}
block_metas.resize(
((num_rows + ELEMENTS_PER_BLOCK - 1) / ELEMENTS_PER_BLOCK) as usize,
num_rows.div_ceil(ELEMENTS_PER_BLOCK) as usize,
BlockMeta {
non_null_rows_before_block,
start_byte_offset,

5
columnar/src/column_index/optional_index/set.rs
View File

@@ -28,10 +28,11 @@ pub trait Set<T> {
/// Returns true if the elements is contained in the Set
fn contains(&self, el: T) -> bool;
/// Returns the number of rows in the set that are < `el`
/// Returns the element's rank (its position in the set).
/// If the set does not contain the element, it will return the next existing elements rank.
fn rank(&self, el: T) -> T;
/// If the set contains `el` returns the element rank.
/// If the set contains `el`, returns the element's rank (its position in the set).
/// If the set does not contain the element, it returns `None`.
fn rank_if_exists(&self, el: T) -> Option<T>;

9
columnar/src/column_index/optional_index/set_block/dense.rs
View File

@@ -23,7 +23,6 @@ fn set_bit_at(input: &mut u64, n: u16) {
///
/// When translating a dense index to the original index, we can use the offset to find the correct
/// block. Direct computation is not possible, but we can employ a linear or binary search.
const ELEMENTS_PER_MINI_BLOCK: u16 = 64;
const MINI_BLOCK_BITVEC_NUM_BYTES: usize = 8;
const MINI_BLOCK_OFFSET_NUM_BYTES: usize = 2;
@@ -109,7 +108,7 @@ pub struct DenseBlockSelectCursor<'a> {
dense_block: DenseBlock<'a>,
}
impl<'a> SelectCursor<u16> for DenseBlockSelectCursor<'a> {
impl SelectCursor<u16> for DenseBlockSelectCursor<'_> {
#[inline]
fn select(&mut self, rank: u16) -> u16 {
self.block_id = self
@@ -123,7 +122,9 @@ impl<'a> SelectCursor<u16> for DenseBlockSelectCursor<'a> {
}
impl<'a> Set<u16> for DenseBlock<'a> {
type SelectCursor<'b> = DenseBlockSelectCursor<'a> where Self: 'b;
type SelectCursor<'b>
= DenseBlockSelectCursor<'a>
where Self: 'b;
#[inline(always)]
fn contains(&self, el: u16) -> bool {
@@ -173,7 +174,7 @@ impl<'a> Set<u16> for DenseBlock<'a> {
}
}
impl<'a> DenseBlock<'a> {
impl DenseBlock<'_> {
#[inline]
fn mini_block(&self, mini_block_id: u16) -> DenseMiniBlock {
let data_start_pos = mini_block_id as usize * MINI_BLOCK_NUM_BYTES;

10
columnar/src/column_index/optional_index/set_block/sparse.rs
View File

@@ -31,8 +31,10 @@ impl<'a> SelectCursor<u16> for SparseBlock<'a> {
}
}
impl<'a> Set<u16> for SparseBlock<'a> {
type SelectCursor<'b> = Self where Self: 'b;
impl Set<u16> for SparseBlock<'_> {
type SelectCursor<'b>
= Self
where Self: 'b;
#[inline(always)]
fn contains(&self, el: u16) -> bool {
@@ -67,7 +69,7 @@ fn get_u16(data: &[u8], byte_position: usize) -> u16 {
u16::from_le_bytes(bytes)
}
impl<'a> SparseBlock<'a> {
impl SparseBlock<'_> {
#[inline(always)]
fn value_at_idx(&self, data: &[u8], idx: u16) -> u16 {
let start_offset: usize = idx as usize * 2;
@@ -80,7 +82,7 @@ impl<'a> SparseBlock<'a> {
}
#[inline]
#[allow(clippy::comparison_chain)]
#[expect(clippy::comparison_chain)]
// Looks for the element in the block. Returns the positions if found.
fn binary_search(&self, target: u16) -> Result<u16, u16> {
let data = &self.0;

42
columnar/src/column_index/optional_index/set_block/tests.rs
View File

@@ -22,8 +22,8 @@ fn test_set_helper<C: SetCodec<Item = u16>>(vals: &[u16]) -> usize {
vals.iter().cloned().take_while(|v| *v < val).count() as u16
);
}
for rank in 0..vals.len() {
assert_eq!(tested_set.select(rank as u16), vals[rank]);
for (rank, val) in vals.iter().enumerate() {
assert_eq!(tested_set.select(rank as u16), *val);
}
buffer.len()
}
@@ -107,3 +107,41 @@ fn test_simple_translate_codec_idx_to_original_idx_dense() {
assert_eq!(i, select_cursor.select(i));
}
}
#[test]
fn test_simple_translate_idx_to_value_idx_dense() {
let mut buffer = Vec::new();
DenseBlockCodec::serialize([1, 10].iter().copied(), &mut buffer).unwrap();
let tested_set = DenseBlockCodec::open(buffer.as_slice());
assert!(tested_set.contains(1));
assert!(!tested_set.contains(2));
assert_eq!(tested_set.rank(0), 0);
assert_eq!(tested_set.rank(1), 0);
for rank in 2..10 {
// ranks that don't exist select the next highest one
assert_eq!(tested_set.rank_if_exists(rank), None);
assert_eq!(tested_set.rank(rank), 1);
}
assert_eq!(tested_set.rank(10), 1);
}
#[test]
fn test_simple_translate_idx_to_value_idx_sparse() {
let mut buffer = Vec::new();
SparseBlockCodec::serialize([1, 10].iter().copied(), &mut buffer).unwrap();
let tested_set = SparseBlockCodec::open(buffer.as_slice());
assert!(tested_set.contains(1));
assert!(!tested_set.contains(2));
assert_eq!(tested_set.rank(0), 0);
assert_eq!(tested_set.select(tested_set.rank(0)), 1);
assert_eq!(tested_set.rank(1), 0);
assert_eq!(tested_set.select(tested_set.rank(1)), 1);
for rank in 2..10 {
// ranks that don't exist select the next highest one
assert_eq!(tested_set.rank_if_exists(rank), None);
assert_eq!(tested_set.rank(rank), 1);
assert_eq!(tested_set.select(tested_set.rank(rank)), 10);
}
assert_eq!(tested_set.rank(10), 1);
assert_eq!(tested_set.select(tested_set.rank(10)), 10);
}

8
columnar/src/column_index/optional_index/tests.rs
View File

@@ -15,9 +15,7 @@ fn test_optional_index_with_num_docs(num_docs: u32) {
let mut dataframe_writer = ColumnarWriter::default();
dataframe_writer.record_numerical(100, "score", 80i64);
let mut buffer: Vec<u8> = Vec::new();
dataframe_writer
.serialize(num_docs, None, &mut buffer)
.unwrap();
dataframe_writer.serialize(num_docs, &mut buffer).unwrap();
let columnar = ColumnarReader::open(buffer).unwrap();
assert_eq!(columnar.num_columns(), 1);
let cols: Vec<DynamicColumnHandle> = columnar.read_columns("score").unwrap();
@@ -112,8 +110,8 @@ fn test_null_index(data: &[bool]) {
.map(|(pos, _val)| pos as u32)
.collect();
let mut select_iter = null_index.select_cursor();
for i in 0..orig_idx_with_value.len() {
assert_eq!(select_iter.select(i as u32), orig_idx_with_value[i]);
for (i, expected) in orig_idx_with_value.iter().enumerate() {
assert_eq!(select_iter.select(i as u32), *expected);
}
let step_size = (orig_idx_with_value.len() / 100).max(1);

45
columnar/src/column_index/serialize.rs
View File

@@ -3,28 +3,39 @@ use std::io::Write;
use common::{CountingWriter, OwnedBytes};
use super::multivalued_index::SerializableMultivalueIndex;
use super::OptionalIndex;
use crate::column_index::multivalued_index::serialize_multivalued_index;
use crate::column_index::optional_index::serialize_optional_index;
use crate::column_index::ColumnIndex;
use crate::iterable::Iterable;
use crate::{Cardinality, RowId};
use crate::{Cardinality, RowId, Version};
pub struct SerializableOptionalIndex<'a> {
pub non_null_row_ids: Box<dyn Iterable<RowId> + 'a>,
pub num_rows: RowId,
}
impl<'a> From<&'a OptionalIndex> for SerializableOptionalIndex<'a> {
fn from(optional_index: &'a OptionalIndex) -> Self {
SerializableOptionalIndex {
non_null_row_ids: Box::new(optional_index),
num_rows: optional_index.num_docs(),
}
}
}
pub enum SerializableColumnIndex<'a> {
Full,
Optional {
non_null_row_ids: Box<dyn Iterable<RowId> + 'a>,
num_rows: RowId,
},
// TODO remove the Arc<dyn> apart from serialization this is not
// dynamic at all.
Multivalued(Box<dyn Iterable<RowId> + 'a>),
Optional(SerializableOptionalIndex<'a>),
Multivalued(SerializableMultivalueIndex<'a>),
}
impl<'a> SerializableColumnIndex<'a> {
impl SerializableColumnIndex<'_> {
pub fn get_cardinality(&self) -> Cardinality {
match self {
SerializableColumnIndex::Full => Cardinality::Full,
SerializableColumnIndex::Optional { .. } => Cardinality::Optional,
SerializableColumnIndex::Optional(_) => Cardinality::Optional,
SerializableColumnIndex::Multivalued(_) => Cardinality::Multivalued,
}
}
@@ -40,12 +51,12 @@ pub fn serialize_column_index(
output.write_all(&[cardinality])?;
match column_index {
SerializableColumnIndex::Full => {}
SerializableColumnIndex::Optional {
SerializableColumnIndex::Optional(SerializableOptionalIndex {
non_null_row_ids,
num_rows,
} => serialize_optional_index(non_null_row_ids.as_ref(), num_rows, &mut output)?,
}) => serialize_optional_index(non_null_row_ids.as_ref(), num_rows, &mut output)?,
SerializableColumnIndex::Multivalued(multivalued_index) => {
serialize_multivalued_index(&*multivalued_index, &mut output)?
serialize_multivalued_index(&multivalued_index, &mut output)?
}
}
let column_index_num_bytes = output.written_bytes() as u32;
@@ -53,7 +64,10 @@ pub fn serialize_column_index(
}
/// Open a serialized column index.
pub fn open_column_index(mut bytes: OwnedBytes) -> io::Result<ColumnIndex> {
pub fn open_column_index(
mut bytes: OwnedBytes,
format_version: Version,
) -> io::Result<ColumnIndex> {
if bytes.is_empty() {
return Err(io::Error::new(
io::ErrorKind::UnexpectedEof,
@@ -70,7 +84,8 @@ pub fn open_column_index(mut bytes: OwnedBytes) -> io::Result<ColumnIndex> {
Ok(ColumnIndex::Optional(optional_index))
}
Cardinality::Multivalued => {
let multivalue_index = super::multivalued_index::open_multivalued_index(bytes)?;
let multivalue_index =
super::multivalued_index::open_multivalued_index(bytes, format_version)?;
Ok(ColumnIndex::Multivalued(multivalue_index))
}
}

6
columnar/src/column_values/bench.rs
View File

@@ -34,6 +34,7 @@ fn compute_stats(vals: impl Iterator<Item = u64>) -> ColumnStats {
fn value_iter() -> impl Iterator<Item = u64> {
0..20_000
}
fn get_reader_for_bench<Codec: ColumnCodec>(data: &[u64]) -> Codec::ColumnValues {
let mut bytes = Vec::new();
let stats = compute_stats(data.iter().cloned());
@@ -41,10 +42,13 @@ fn get_reader_for_bench<Codec: ColumnCodec>(data: &[u64]) -> Codec::ColumnValues
for val in data {
codec_serializer.collect(*val);
}
codec_serializer.serialize(&stats, Box::new(data.iter().copied()).as_mut(), &mut bytes);
codec_serializer
.serialize(&stats, Box::new(data.iter().copied()).as_mut(), &mut bytes)
.unwrap();
Codec::load(OwnedBytes::new(bytes)).unwrap()
}
fn bench_get<Codec: ColumnCodec>(b: &mut Bencher, data: &[u64]) {
let col = get_reader_for_bench::<Codec>(data);
b.iter(|| {

2
columnar/src/column_values/merge.rs
View File

@@ -10,7 +10,7 @@ pub(crate) struct MergedColumnValues<'a, T> {
pub(crate) merge_row_order: &'a MergeRowOrder,
}
impl<'a, T: Copy + PartialOrd + Debug + 'static> Iterable<T> for MergedColumnValues<'a, T> {
impl<T: Copy + PartialOrd + Debug + 'static> Iterable<T> for MergedColumnValues<'_, T> {
fn boxed_iter(&self) -> Box<dyn Iterator<Item = T> + '_> {
match self.merge_row_order {
MergeRowOrder::Stack(_) => Box::new(

2
columnar/src/column_values/u128_based/compact_space/build_compact_space.rs
View File

@@ -184,7 +184,7 @@ impl CompactSpaceBuilder {
let mut covered_space = Vec::with_capacity(self.blanks.len());
// begining of the blanks
// beginning of the blanks
if let Some(first_blank_start) = self.blanks.first().map(RangeInclusive::start) {
if *first_blank_start != 0 {
covered_space.push(0..=first_blank_start - 1);

2
columnar/src/column_values/u128_based/mod.rs
View File

@@ -128,7 +128,7 @@ pub fn open_u128_as_compact_u64(mut bytes: OwnedBytes) -> io::Result<Arc<dyn Col
}
#[cfg(test)]
pub mod tests {
pub(crate) mod tests {
use super::*;
use crate::column_values::u64_based::{
serialize_and_load_u64_based_column_values, serialize_u64_based_column_values,

2
columnar/src/column_values/u64_based/blockwise_linear.rs
View File

@@ -39,7 +39,7 @@ impl BinarySerializable for Block {
}
fn compute_num_blocks(num_vals: u32) -> u32 {
(num_vals + BLOCK_SIZE - 1) / BLOCK_SIZE
num_vals.div_ceil(BLOCK_SIZE)
}
pub struct BlockwiseLinearEstimator {

7
columnar/src/column_values/u64_based/line.rs
View File

@@ -122,12 +122,11 @@ impl Line {
line
}
/// Returns a line that attemps to approximate a function
/// Returns a line that attempts to approximate a function
/// f: i in 0..[ys.num_vals()) -> ys[i].
///
/// - The approximation is always lower than the actual value.
/// Or more rigorously, formally `f(i).wrapping_sub(ys[i])` is small
/// for any i in [0..ys.len()).
/// - The approximation is always lower than the actual value. Or more rigorously, formally
/// `f(i).wrapping_sub(ys[i])` is small for any i in [0..ys.len()).
/// - It computes without panicking for any value of it.
///
/// This function is only invariable by translation if all of the

25
columnar/src/columnar/format_version.rs
View File

@@ -1,3 +1,6 @@
use core::fmt;
use std::fmt::{Display, Formatter};
use crate::InvalidData;
pub const VERSION_FOOTER_NUM_BYTES: usize = MAGIC_BYTES.len() + std::mem::size_of::<u32>();
@@ -8,7 +11,7 @@ const MAGIC_BYTES: [u8; 4] = [2, 113, 119, 66];
pub fn footer() -> [u8; VERSION_FOOTER_NUM_BYTES] {
let mut footer_bytes = [0u8; VERSION_FOOTER_NUM_BYTES];
footer_bytes[0..4].copy_from_slice(&Version::V1.to_bytes());
footer_bytes[0..4].copy_from_slice(&CURRENT_VERSION.to_bytes());
footer_bytes[4..8].copy_from_slice(&MAGIC_BYTES[..]);
footer_bytes
}
@@ -20,10 +23,22 @@ pub fn parse_footer(footer_bytes: [u8; VERSION_FOOTER_NUM_BYTES]) -> Result<Vers
Version::try_from_bytes(footer_bytes[0..4].try_into().unwrap())
}
pub const CURRENT_VERSION: Version = Version::V2;
#[derive(Debug, Copy, Clone, Eq, PartialEq)]
#[repr(u32)]
pub enum Version {
V1 = 1u32,
V2 = 2u32,
}
impl Display for Version {
fn fmt(&self, f: &mut Formatter) -> fmt::Result {
match self {
Version::V1 => write!(f, "v1"),
Version::V2 => write!(f, "v2"),
}
}
}
impl Version {
@@ -35,6 +50,7 @@ impl Version {
let code = u32::from_le_bytes(bytes);
match code {
1u32 => Ok(Version::V1),
2u32 => Ok(Version::V2),
_ => Err(InvalidData),
}
}
@@ -47,9 +63,9 @@ mod tests {
use super::*;
#[test]
fn test_footer_dserialization() {
fn test_footer_deserialization() {
let parsed_version: Version = parse_footer(footer()).unwrap();
assert_eq!(Version::V1, parsed_version);
assert_eq!(Version::V2, parsed_version);
}
#[test]
@@ -63,11 +79,10 @@ mod tests {
for &i in &version_to_tests {
let version_res = Version::try_from_bytes(i.to_le_bytes());
if let Ok(version) = version_res {
assert_eq!(version, Version::V1);
assert_eq!(version.to_bytes(), i.to_le_bytes());
valid_versions.insert(i);
}
}
assert_eq!(valid_versions.len(), 1);
assert_eq!(valid_versions.len(), 2);
}
}

4
columnar/src/columnar/merge/merge_dict_column.rs
View File

@@ -39,7 +39,7 @@ struct RemappedTermOrdinalsValues<'a> {
merge_row_order: &'a MergeRowOrder,
}
impl<'a> Iterable for RemappedTermOrdinalsValues<'a> {
impl Iterable for RemappedTermOrdinalsValues<'_> {
fn boxed_iter(&self) -> Box<dyn Iterator<Item = u64> + '_> {
match self.merge_row_order {
MergeRowOrder::Stack(_) => self.boxed_iter_stacked(),
@@ -50,7 +50,7 @@ impl<'a> Iterable for RemappedTermOrdinalsValues<'a> {
}
}
impl<'a> RemappedTermOrdinalsValues<'a> {
impl RemappedTermOrdinalsValues<'_> {
fn boxed_iter_stacked(&self) -> Box<dyn Iterator<Item = u64> + '_> {
let iter = self
.bytes_columns

28
columnar/src/columnar/merge/mod.rs
View File

@@ -7,7 +7,6 @@ use std::io;
use std::net::Ipv6Addr;
use std::sync::Arc;
use itertools::Itertools;
pub use merge_mapping::{MergeRowOrder, ShuffleMergeOrder, StackMergeOrder};
use super::writer::ColumnarSerializer;
@@ -26,7 +25,7 @@ use crate::{
/// After merge, all columns belonging to the same category are coerced to
/// the same column type.
///
/// In practise, today, only Numerical colummns are coerced into one type today.
/// In practise, today, only Numerical columns are coerced into one type today.
///
/// See also [README.md].
///
@@ -64,11 +63,10 @@ impl From<ColumnType> for ColumnTypeCategory {
/// `require_columns` makes it possible to ensure that some columns will be present in the
/// resulting columnar. When a required column is a numerical column type, one of two things can
/// happen:
/// - If the required column type is compatible with all of the input columnar, the resulsting
/// merged
/// columnar will simply coerce the input column and use the required column type.
/// - If the required column type is incompatible with one of the input columnar, the merged
/// will fail with an InvalidData error.
/// - If the required column type is compatible with all of the input columnar, the resulting merged
/// columnar will simply coerce the input column and use the required column type.
/// - If the required column type is incompatible with one of the input columnar, the merged will
/// fail with an InvalidData error.
///
/// `merge_row_order` makes it possible to remove or reorder row in the resulting
/// `Columnar` table.
@@ -371,20 +369,8 @@ fn is_empty_after_merge(
true
}
ColumnIndex::Multivalued(multivalued_index) => {
for (doc_id, (start_index, end_index)) in multivalued_index
.start_index_column
.iter()
.tuple_windows()
.enumerate()
{
let doc_id = doc_id as u32;
if start_index == end_index {
// There are no values in this document
continue;
}
// The document contains values and is present in the alive bitset.
// The column is therefore not empty.
if alive_bitset.contains(doc_id) {
for alive_docid in alive_bitset.iter() {
if !multivalued_index.range(alive_docid).is_empty() {
return false;
}
}

7
columnar/src/columnar/merge/term_merger.rs
View File

@@ -10,13 +10,13 @@ pub struct HeapItem<'a> {
pub segment_ord: usize,
}
impl<'a> PartialEq for HeapItem<'a> {
impl PartialEq for HeapItem<'_> {
fn eq(&self, other: &Self) -> bool {
self.segment_ord == other.segment_ord
}
}
impl<'a> Eq for HeapItem<'a> {}
impl Eq for HeapItem<'_> {}
impl<'a> PartialOrd for HeapItem<'a> {
fn partial_cmp(&self, other: &HeapItem<'a>) -> Option<Ordering> {
@@ -35,8 +35,7 @@ impl<'a> Ord for HeapItem<'a> {
///
/// The item yield is actually a pair with
/// - the term
/// - a slice with the ordinal of the segments containing
/// the terms.
/// - a slice with the ordinal of the segments containing the terms.
pub struct TermMerger<'a> {
heap: BinaryHeap<HeapItem<'a>>,
current_streamers: Vec<HeapItem<'a>>,

16
columnar/src/columnar/merge/tests.rs
View File

@@ -1,3 +1,5 @@
use itertools::Itertools;
use super::*;
use crate::{Cardinality, ColumnarWriter, HasAssociatedColumnType, RowId};
@@ -12,7 +14,7 @@ fn make_columnar<T: Into<NumericalValue> + HasAssociatedColumnType + Copy>(
}
let mut buffer: Vec<u8> = Vec::new();
dataframe_writer
.serialize(vals.len() as RowId, None, &mut buffer)
.serialize(vals.len() as RowId, &mut buffer)
.unwrap();
ColumnarReader::open(buffer).unwrap()
}
@@ -157,9 +159,7 @@ fn make_numerical_columnar_multiple_columns(
.max()
.unwrap_or(0u32);
let mut buffer: Vec<u8> = Vec::new();
dataframe_writer
.serialize(num_rows, None, &mut buffer)
.unwrap();
dataframe_writer.serialize(num_rows, &mut buffer).unwrap();
ColumnarReader::open(buffer).unwrap()
}
@@ -182,9 +182,7 @@ fn make_byte_columnar_multiple_columns(
}
}
let mut buffer: Vec<u8> = Vec::new();
dataframe_writer
.serialize(num_rows, None, &mut buffer)
.unwrap();
dataframe_writer.serialize(num_rows, &mut buffer).unwrap();
ColumnarReader::open(buffer).unwrap()
}
@@ -203,9 +201,7 @@ fn make_text_columnar_multiple_columns(columns: &[(&str, &[&[&str]])]) -> Column
.max()
.unwrap_or(0u32);
let mut buffer: Vec<u8> = Vec::new();
dataframe_writer
.serialize(num_rows, None, &mut buffer)
.unwrap();
dataframe_writer.serialize(num_rows, &mut buffer).unwrap();
ColumnarReader::open(buffer).unwrap()
}

1
columnar/src/columnar/mod.rs
View File

@@ -5,6 +5,7 @@ mod reader;
mod writer;
pub use column_type::{ColumnType, HasAssociatedColumnType};
pub use format_version::{Version, CURRENT_VERSION};
#[cfg(test)]
pub(crate) use merge::ColumnTypeCategory;
pub use merge::{merge_columnar, MergeRowOrder, ShuffleMergeOrder, StackMergeOrder};

17
columnar/src/columnar/reader/mod.rs
View File

@@ -6,7 +6,7 @@ use sstable::{Dictionary, RangeSSTable};
use crate::columnar::{format_version, ColumnType};
use crate::dynamic_column::DynamicColumnHandle;
use crate::RowId;
use crate::{RowId, Version};
fn io_invalid_data(msg: String) -> io::Error {
io::Error::new(io::ErrorKind::InvalidData, msg)
@@ -19,6 +19,7 @@ pub struct ColumnarReader {
column_dictionary: Dictionary<RangeSSTable>,
column_data: FileSlice,
num_rows: RowId,
format_version: Version,
}
impl fmt::Debug for ColumnarReader {
@@ -53,6 +54,7 @@ impl fmt::Debug for ColumnarReader {
fn read_all_columns_in_stream(
mut stream: sstable::Streamer<'_, RangeSSTable>,
column_data: &FileSlice,
format_version: Version,
) -> io::Result<Vec<DynamicColumnHandle>> {
let mut results = Vec::new();
while stream.advance() {
@@ -67,6 +69,7 @@ fn read_all_columns_in_stream(
let dynamic_column_handle = DynamicColumnHandle {
file_slice,
column_type,
format_version,
};
results.push(dynamic_column_handle);
}
@@ -88,7 +91,7 @@ impl ColumnarReader {
let num_rows = u32::deserialize(&mut &footer_bytes[8..12])?;
let version_footer_bytes: [u8; format_version::VERSION_FOOTER_NUM_BYTES] =
footer_bytes[12..].try_into().unwrap();
let _version = format_version::parse_footer(version_footer_bytes)?;
let format_version = format_version::parse_footer(version_footer_bytes)?;
let (column_data, sstable) =
file_slice_without_sstable_len.split_from_end(sstable_len as usize);
let column_dictionary = Dictionary::open(sstable)?;
@@ -96,6 +99,7 @@ impl ColumnarReader {
column_dictionary,
column_data,
num_rows,
format_version,
})
}
@@ -126,6 +130,7 @@ impl ColumnarReader {
let column_handle = DynamicColumnHandle {
file_slice,
column_type,
format_version: self.format_version,
};
Some((column_name, column_handle))
} else {
@@ -167,7 +172,7 @@ impl ColumnarReader {
.stream_for_column_range(column_name)
.into_stream_async()
.await?;
read_all_columns_in_stream(stream, &self.column_data)
read_all_columns_in_stream(stream, &self.column_data, self.format_version)
}
/// Get all columns for the given column name.
@@ -176,7 +181,7 @@ impl ColumnarReader {
/// different types.
pub fn read_columns(&self, column_name: &str) -> io::Result<Vec<DynamicColumnHandle>> {
let stream = self.stream_for_column_range(column_name).into_stream()?;
read_all_columns_in_stream(stream, &self.column_data)
read_all_columns_in_stream(stream, &self.column_data, self.format_version)
}
/// Return the number of columns in the columnar.
@@ -195,7 +200,7 @@ mod tests {
columnar_writer.record_column_type("col1", ColumnType::Str, false);
columnar_writer.record_column_type("col2", ColumnType::U64, false);
let mut buffer = Vec::new();
columnar_writer.serialize(1, None, &mut buffer).unwrap();
columnar_writer.serialize(1, &mut buffer).unwrap();
let columnar = ColumnarReader::open(buffer).unwrap();
let columns = columnar.list_columns().unwrap();
assert_eq!(columns.len(), 2);
@@ -211,7 +216,7 @@ mod tests {
columnar_writer.record_column_type("count", ColumnType::U64, false);
columnar_writer.record_numerical(1, "count", 1u64);
let mut buffer = Vec::new();
columnar_writer.serialize(2, None, &mut buffer).unwrap();
columnar_writer.serialize(2, &mut buffer).unwrap();
let columnar = ColumnarReader::open(buffer).unwrap();
let columns = columnar.list_columns().unwrap();
assert_eq!(columns.len(), 1);

3
columnar/src/columnar/writer/column_operation.rs
View File

@@ -87,7 +87,7 @@ impl<V: SymbolValue> ColumnOperation<V> {
minibuf
}
/// Deserialize a colummn operation.
/// Deserialize a column operation.
/// Returns None if the buffer is empty.
///
/// Panics if the payload is invalid:
@@ -122,7 +122,6 @@ impl<T> From<T> for ColumnOperation<T> {
// In order to limit memory usage, and in order
// to benefit from the stacker, we do this by serialization our data
// as "Symbols".
#[allow(clippy::from_over_into)]
pub(super) trait SymbolValue: Clone + Copy {
// Serializes the symbol into the given buffer.
// Returns the number of bytes written into the buffer.

29
columnar/src/columnar/writer/column_writers.rs
View File

@@ -41,31 +41,10 @@ impl ColumnWriter {
pub(super) fn operation_iterator<'a, V: SymbolValue>(
&self,
arena: &MemoryArena,
old_to_new_ids_opt: Option<&[RowId]>,
buffer: &'a mut Vec<u8>,
) -> impl Iterator<Item = ColumnOperation<V>> + 'a {
buffer.clear();
self.values.read_to_end(arena, buffer);
if let Some(old_to_new_ids) = old_to_new_ids_opt {
// TODO avoid the extra deserialization / serialization.
let mut sorted_ops: Vec<(RowId, ColumnOperation<V>)> = Vec::new();
let mut new_doc = 0u32;
let mut cursor = &buffer[..];
for op in std::iter::from_fn(|| ColumnOperation::<V>::deserialize(&mut cursor)) {
if let ColumnOperation::NewDoc(doc) = &op {
new_doc = old_to_new_ids[*doc as usize];
sorted_ops.push((new_doc, ColumnOperation::NewDoc(new_doc)));
} else {
sorted_ops.push((new_doc, op));
}
}
// stable sort is crucial here.
sorted_ops.sort_by_key(|(new_doc_id, _)| *new_doc_id);
buffer.clear();
for (_, op) in sorted_ops {
buffer.extend_from_slice(op.serialize().as_ref());
}
}
let mut cursor: &[u8] = &buffer[..];
std::iter::from_fn(move || ColumnOperation::deserialize(&mut cursor))
}
@@ -231,11 +210,9 @@ impl NumericalColumnWriter {
pub(super) fn operation_iterator<'a>(
self,
arena: &MemoryArena,
old_to_new_ids: Option<&[RowId]>,
buffer: &'a mut Vec<u8>,
) -> impl Iterator<Item = ColumnOperation<NumericalValue>> + 'a {
self.column_writer
.operation_iterator(arena, old_to_new_ids, buffer)
self.column_writer.operation_iterator(arena, buffer)
}
}
@@ -277,11 +254,9 @@ impl StrOrBytesColumnWriter {
pub(super) fn operation_iterator<'a>(
&self,
arena: &MemoryArena,
old_to_new_ids: Option<&[RowId]>,
byte_buffer: &'a mut Vec<u8>,
) -> impl Iterator<Item = ColumnOperation<UnorderedId>> + 'a {
self.column_writer
.operation_iterator(arena, old_to_new_ids, byte_buffer)
self.column_writer.operation_iterator(arena, byte_buffer)
}
}

249
columnar/src/columnar/writer/mod.rs
View File

@@ -8,11 +8,12 @@ use std::net::Ipv6Addr;
use column_operation::ColumnOperation;
pub(crate) use column_writers::CompatibleNumericalTypes;
use common::json_path_writer::JSON_END_OF_PATH;
use common::CountingWriter;
pub(crate) use serializer::ColumnarSerializer;
use stacker::{Addr, ArenaHashMap, MemoryArena};
use crate::column_index::SerializableColumnIndex;
use crate::column_index::{SerializableColumnIndex, SerializableOptionalIndex};
use crate::column_values::{MonotonicallyMappableToU128, MonotonicallyMappableToU64};
use crate::columnar::column_type::ColumnType;
use crate::columnar::writer::column_writers::{
@@ -43,7 +44,7 @@ struct SpareBuffers {
/// columnar_writer.record_str(1u32 /* doc id */, "product_name", "Apple");
/// columnar_writer.record_numerical(0u32 /* doc id */, "price", 10.5f64); //< uh oh we ended up mixing integer and floats.
/// let mut wrt: Vec<u8> = Vec::new();
/// columnar_writer.serialize(2u32, None, &mut wrt).unwrap();
/// columnar_writer.serialize(2u32, &mut wrt).unwrap();
/// ```
#[derive(Default)]
pub struct ColumnarWriter {
@@ -59,22 +60,6 @@ pub struct ColumnarWriter {
buffers: SpareBuffers,
}
#[inline]
fn mutate_or_create_column<V, TMutator>(
arena_hash_map: &mut ArenaHashMap,
column_name: &str,
updater: TMutator,
) where
V: Copy + 'static,
TMutator: FnMut(Option<V>) -> V,
{
assert!(
!column_name.as_bytes().contains(&0u8),
"key may not contain the 0 byte"
);
arena_hash_map.mutate_or_create(column_name.as_bytes(), updater);
}
impl ColumnarWriter {
pub fn mem_usage(&self) -> usize {
self.arena.mem_usage()
@@ -91,63 +76,6 @@ impl ColumnarWriter {
.sum::<usize>()
}
/// Returns the list of doc ids from 0..num_docs sorted by the `sort_field`
/// column.
///
/// If the column is multivalued, use the first value for scoring.
/// If no value is associated to a specific row, the document is assigned
/// the lowest possible score.
///
/// The sort applied is stable.
pub fn sort_order(&self, sort_field: &str, num_docs: RowId, reversed: bool) -> Vec<u32> {
let Some(numerical_col_writer) = self
.numerical_field_hash_map
.get::<NumericalColumnWriter>(sort_field.as_bytes())
.or_else(|| {
self.datetime_field_hash_map
.get::<NumericalColumnWriter>(sort_field.as_bytes())
})
else {
return Vec::new();
};
let mut symbols_buffer = Vec::new();
let mut values = Vec::new();
let mut start_doc_check_fill = 0;
let mut current_doc_opt: Option<RowId> = None;
// Assumption: NewDoc will never call the same doc twice and is strictly increasing between
// calls
for op in numerical_col_writer.operation_iterator(&self.arena, None, &mut symbols_buffer) {
match op {
ColumnOperation::NewDoc(doc) => {
current_doc_opt = Some(doc);
}
ColumnOperation::Value(numerical_value) => {
if let Some(current_doc) = current_doc_opt {
// Fill up with 0.0 since last doc
values.extend((start_doc_check_fill..current_doc).map(|doc| (0.0, doc)));
start_doc_check_fill = current_doc + 1;
// handle multi values
current_doc_opt = None;
let score: f32 = f64::coerce(numerical_value) as f32;
values.push((score, current_doc));
}
}
}
}
for doc in values.len() as u32..num_docs {
values.push((0.0f32, doc));
}
values.sort_by(|(left_score, _), (right_score, _)| {
if reversed {
right_score.total_cmp(left_score)
} else {
left_score.total_cmp(right_score)
}
});
values.into_iter().map(|(_score, doc)| doc).collect()
}
/// Records a column type. This is useful to bypass the coercion process,
/// makes sure the empty is present in the resulting columnar, or set
/// the `sort_values_within_row`.
@@ -175,9 +103,8 @@ impl ColumnarWriter {
},
&mut self.dictionaries,
);
mutate_or_create_column(
hash_map,
column_name,
hash_map.mutate_or_create(
column_name.as_bytes(),
|column_opt: Option<StrOrBytesColumnWriter>| {
let mut column_writer = if let Some(column_writer) = column_opt {
column_writer
@@ -192,24 +119,21 @@ impl ColumnarWriter {
);
}
ColumnType::Bool => {
mutate_or_create_column(
&mut self.bool_field_hash_map,
column_name,
self.bool_field_hash_map.mutate_or_create(
column_name.as_bytes(),
|column_opt: Option<ColumnWriter>| column_opt.unwrap_or_default(),
);
}
ColumnType::DateTime => {
mutate_or_create_column(
&mut self.datetime_field_hash_map,
column_name,
self.datetime_field_hash_map.mutate_or_create(
column_name.as_bytes(),
|column_opt: Option<ColumnWriter>| column_opt.unwrap_or_default(),
);
}
ColumnType::I64 | ColumnType::F64 | ColumnType::U64 => {
let numerical_type = column_type.numerical_type().unwrap();
mutate_or_create_column(
&mut self.numerical_field_hash_map,
column_name,
self.numerical_field_hash_map.mutate_or_create(
column_name.as_bytes(),
|column_opt: Option<NumericalColumnWriter>| {
let mut column: NumericalColumnWriter = column_opt.unwrap_or_default();
column.force_numerical_type(numerical_type);
@@ -217,9 +141,8 @@ impl ColumnarWriter {
},
);
}
ColumnType::IpAddr => mutate_or_create_column(
&mut self.ip_addr_field_hash_map,
column_name,
ColumnType::IpAddr => self.ip_addr_field_hash_map.mutate_or_create(
column_name.as_bytes(),
|column_opt: Option<ColumnWriter>| column_opt.unwrap_or_default(),
),
}
@@ -232,9 +155,8 @@ impl ColumnarWriter {
numerical_value: T,
) {
let (hash_map, arena) = (&mut self.numerical_field_hash_map, &mut self.arena);
mutate_or_create_column(
hash_map,
column_name,
hash_map.mutate_or_create(
column_name.as_bytes(),
|column_opt: Option<NumericalColumnWriter>| {
let mut column: NumericalColumnWriter = column_opt.unwrap_or_default();
column.record_numerical_value(doc, numerical_value.into(), arena);
@@ -244,10 +166,6 @@ impl ColumnarWriter {
}
pub fn record_ip_addr(&mut self, doc: RowId, column_name: &str, ip_addr: Ipv6Addr) {
assert!(
!column_name.as_bytes().contains(&0u8),
"key may not contain the 0 byte"
);
let (hash_map, arena) = (&mut self.ip_addr_field_hash_map, &mut self.arena);
hash_map.mutate_or_create(
column_name.as_bytes(),
@@ -261,24 +179,30 @@ impl ColumnarWriter {
pub fn record_bool(&mut self, doc: RowId, column_name: &str, val: bool) {
let (hash_map, arena) = (&mut self.bool_field_hash_map, &mut self.arena);
mutate_or_create_column(hash_map, column_name, |column_opt: Option<ColumnWriter>| {
let mut column: ColumnWriter = column_opt.unwrap_or_default();
column.record(doc, val, arena);
column
});
hash_map.mutate_or_create(
column_name.as_bytes(),
|column_opt: Option<ColumnWriter>| {
let mut column: ColumnWriter = column_opt.unwrap_or_default();
column.record(doc, val, arena);
column
},
);
}
pub fn record_datetime(&mut self, doc: RowId, column_name: &str, datetime: common::DateTime) {
let (hash_map, arena) = (&mut self.datetime_field_hash_map, &mut self.arena);
mutate_or_create_column(hash_map, column_name, |column_opt: Option<ColumnWriter>| {
let mut column: ColumnWriter = column_opt.unwrap_or_default();
column.record(
doc,
NumericalValue::I64(datetime.into_timestamp_nanos()),
arena,
);
column
});
hash_map.mutate_or_create(
column_name.as_bytes(),
|column_opt: Option<ColumnWriter>| {
let mut column: ColumnWriter = column_opt.unwrap_or_default();
column.record(
doc,
NumericalValue::I64(datetime.into_timestamp_nanos()),
arena,
);
column
},
);
}
pub fn record_str(&mut self, doc: RowId, column_name: &str, value: &str) {
@@ -303,10 +227,6 @@ impl ColumnarWriter {
}
pub fn record_bytes(&mut self, doc: RowId, column_name: &str, value: &[u8]) {
assert!(
!column_name.as_bytes().contains(&0u8),
"key may not contain the 0 byte"
);
let (hash_map, arena, dictionaries) = (
&mut self.bytes_field_hash_map,
&mut self.arena,
@@ -326,13 +246,9 @@ impl ColumnarWriter {
},
);
}
pub fn serialize(
&mut self,
num_docs: RowId,
old_to_new_row_ids: Option<&[RowId]>,
wrt: &mut dyn io::Write,
) -> io::Result<()> {
pub fn serialize(&mut self, num_docs: RowId, wrt: &mut dyn io::Write) -> io::Result<()> {
let mut serializer = ColumnarSerializer::new(wrt);
let mut columns: Vec<(&[u8], ColumnType, Addr)> = self
.numerical_field_hash_map
.iter()
@@ -346,7 +262,7 @@ impl ColumnarWriter {
columns.extend(
self.bytes_field_hash_map
.iter()
.map(|(term, addr)| (term, ColumnType::Bytes, addr)),
.map(|(column_name, addr)| (column_name, ColumnType::Bytes, addr)),
);
columns.extend(
self.str_field_hash_map
@@ -373,6 +289,12 @@ impl ColumnarWriter {
let (arena, buffers, dictionaries) = (&self.arena, &mut self.buffers, &self.dictionaries);
let mut symbol_byte_buffer: Vec<u8> = Vec::new();
for (column_name, column_type, addr) in columns {
if column_name.contains(&JSON_END_OF_PATH) {
// Tantivy uses b'0' as a separator for nested fields in JSON.
// Column names with a b'0' are not simply ignored by the columnar (and the inverted
// index).
continue;
}
match column_type {
ColumnType::Bool => {
let column_writer: ColumnWriter = self.bool_field_hash_map.read(addr);
@@ -382,11 +304,7 @@ impl ColumnarWriter {
serialize_bool_column(
cardinality,
num_docs,
column_writer.operation_iterator(
arena,
old_to_new_row_ids,
&mut symbol_byte_buffer,
),
column_writer.operation_iterator(arena, &mut symbol_byte_buffer),
buffers,
&mut column_serializer,
)?;
@@ -400,11 +318,7 @@ impl ColumnarWriter {
serialize_ip_addr_column(
cardinality,
num_docs,
column_writer.operation_iterator(
arena,
old_to_new_row_ids,
&mut symbol_byte_buffer,
),
column_writer.operation_iterator(arena, &mut symbol_byte_buffer),
buffers,
&mut column_serializer,
)?;
@@ -429,11 +343,8 @@ impl ColumnarWriter {
num_docs,
str_or_bytes_column_writer.sort_values_within_row,
dictionary_builder,
str_or_bytes_column_writer.operation_iterator(
arena,
old_to_new_row_ids,
&mut symbol_byte_buffer,
),
str_or_bytes_column_writer
.operation_iterator(arena, &mut symbol_byte_buffer),
buffers,
&self.arena,
&mut column_serializer,
@@ -451,11 +362,7 @@ impl ColumnarWriter {
cardinality,
num_docs,
numerical_type,
numerical_column_writer.operation_iterator(
arena,
old_to_new_row_ids,
&mut symbol_byte_buffer,
),
numerical_column_writer.operation_iterator(arena, &mut symbol_byte_buffer),
buffers,
&mut column_serializer,
)?;
@@ -470,11 +377,7 @@ impl ColumnarWriter {
cardinality,
num_docs,
NumericalType::I64,
column_writer.operation_iterator(
arena,
old_to_new_row_ids,
&mut symbol_byte_buffer,
),
column_writer.operation_iterator(arena, &mut symbol_byte_buffer),
buffers,
&mut column_serializer,
)?;
@@ -489,7 +392,7 @@ impl ColumnarWriter {
// Serialize [Dictionary, Column, dictionary num bytes U32::LE]
// Column: [Column Index, Column Values, column index num bytes U32::LE]
#[allow(clippy::too_many_arguments)]
#[expect(clippy::too_many_arguments)]
fn serialize_bytes_or_str_column(
cardinality: Cardinality,
num_docs: RowId,
@@ -659,16 +562,16 @@ fn send_to_serialize_column_mappable_to_u128<
let optional_index_builder = value_index_builders.borrow_optional_index_builder();
consume_operation_iterator(op_iterator, optional_index_builder, values);
let optional_index = optional_index_builder.finish(num_rows);
SerializableColumnIndex::Optional {
SerializableColumnIndex::Optional(SerializableOptionalIndex {
num_rows,
non_null_row_ids: Box::new(optional_index),
}
})
}
Cardinality::Multivalued => {
let multivalued_index_builder = value_index_builders.borrow_multivalued_index_builder();
consume_operation_iterator(op_iterator, multivalued_index_builder, values);
let multivalued_index = multivalued_index_builder.finish(num_rows);
SerializableColumnIndex::Multivalued(Box::new(multivalued_index))
let serializable_multivalued_index = multivalued_index_builder.finish(num_rows);
SerializableColumnIndex::Multivalued(serializable_multivalued_index)
}
};
crate::column::serialize_column_mappable_to_u128(
@@ -679,15 +582,6 @@ fn send_to_serialize_column_mappable_to_u128<
Ok(())
}
fn sort_values_within_row_in_place(multivalued_index: &[RowId], values: &mut [u64]) {
let mut start_index: usize = 0;
for end_index in multivalued_index.iter().copied() {
let end_index = end_index as usize;
values[start_index..end_index].sort_unstable();
start_index = end_index;
}
}
fn send_to_serialize_column_mappable_to_u64(
op_iterator: impl Iterator<Item = ColumnOperation<u64>>,
cardinality: Cardinality,
@@ -711,19 +605,22 @@ fn send_to_serialize_column_mappable_to_u64(
let optional_index_builder = value_index_builders.borrow_optional_index_builder();
consume_operation_iterator(op_iterator, optional_index_builder, values);
let optional_index = optional_index_builder.finish(num_rows);
SerializableColumnIndex::Optional {
SerializableColumnIndex::Optional(SerializableOptionalIndex {
non_null_row_ids: Box::new(optional_index),
num_rows,
}
})
}
Cardinality::Multivalued => {
let multivalued_index_builder = value_index_builders.borrow_multivalued_index_builder();
consume_operation_iterator(op_iterator, multivalued_index_builder, values);
let multivalued_index = multivalued_index_builder.finish(num_rows);
let serializable_multivalued_index = multivalued_index_builder.finish(num_rows);
if sort_values_within_row {
sort_values_within_row_in_place(multivalued_index, values);
sort_values_within_row_in_place(
serializable_multivalued_index.start_offsets.boxed_iter(),
values,
);
}
SerializableColumnIndex::Multivalued(Box::new(multivalued_index))
SerializableColumnIndex::Multivalued(serializable_multivalued_index)
}
};
crate::column::serialize_column_mappable_to_u64(
@@ -734,6 +631,18 @@ fn send_to_serialize_column_mappable_to_u64(
Ok(())
}
fn sort_values_within_row_in_place(
multivalued_index: impl Iterator<Item = RowId>,
values: &mut [u64],
) {
let mut start_index: usize = 0;
for end_index in multivalued_index {
let end_index = end_index as usize;
values[start_index..end_index].sort_unstable();
start_index = end_index;
}
}
fn coerce_numerical_symbol<T>(
operation_iterator: impl Iterator<Item = ColumnOperation<NumericalValue>>,
) -> impl Iterator<Item = ColumnOperation<u64>>
@@ -781,7 +690,7 @@ mod tests {
assert_eq!(column_writer.get_cardinality(3), Cardinality::Full);
let mut buffer = Vec::new();
let symbols: Vec<ColumnOperation<NumericalValue>> = column_writer
.operation_iterator(&arena, None, &mut buffer)
.operation_iterator(&arena, &mut buffer)
.collect();
assert_eq!(symbols.len(), 6);
assert!(matches!(symbols[0], ColumnOperation::NewDoc(0u32)));
@@ -810,7 +719,7 @@ mod tests {
assert_eq!(column_writer.get_cardinality(3), Cardinality::Optional);
let mut buffer = Vec::new();
let symbols: Vec<ColumnOperation<NumericalValue>> = column_writer
.operation_iterator(&arena, None, &mut buffer)
.operation_iterator(&arena, &mut buffer)
.collect();
assert_eq!(symbols.len(), 4);
assert!(matches!(symbols[0], ColumnOperation::NewDoc(1u32)));
@@ -833,7 +742,7 @@ mod tests {
assert_eq!(column_writer.get_cardinality(2), Cardinality::Optional);
let mut buffer = Vec::new();
let symbols: Vec<ColumnOperation<NumericalValue>> = column_writer
.operation_iterator(&arena, None, &mut buffer)
.operation_iterator(&arena, &mut buffer)
.collect();
assert_eq!(symbols.len(), 2);
assert!(matches!(symbols[0], ColumnOperation::NewDoc(0u32)));
@@ -852,7 +761,7 @@ mod tests {
assert_eq!(column_writer.get_cardinality(1), Cardinality::Multivalued);
let mut buffer = Vec::new();
let symbols: Vec<ColumnOperation<NumericalValue>> = column_writer
.operation_iterator(&arena, None, &mut buffer)
.operation_iterator(&arena, &mut buffer)
.collect();
assert_eq!(symbols.len(), 3);
assert!(matches!(symbols[0], ColumnOperation::NewDoc(0u32)));

29
columnar/src/columnar/writer/serializer.rs
View File

@@ -1,6 +1,7 @@
use std::io;
use std::io::Write;
use common::json_path_writer::JSON_END_OF_PATH;
use common::{BinarySerializable, CountingWriter};
use sstable::value::RangeValueWriter;
use sstable::RangeSSTable;
@@ -18,13 +19,8 @@ pub struct ColumnarSerializer<W: io::Write> {
/// code.
fn prepare_key(key: &[u8], column_type: ColumnType, buffer: &mut Vec<u8>) {
buffer.clear();
// Convert 0 bytes to '0' string, as 0 bytes are reserved for the end of the path.
if key.contains(&0u8) {
buffer.extend(key.iter().map(|&b| if b == 0 { b'0' } else { b }));
} else {
buffer.extend_from_slice(key);
}
buffer.push(0u8);
buffer.extend_from_slice(key);
buffer.push(JSON_END_OF_PATH);
buffer.push(column_type.to_code());
}
@@ -71,7 +67,7 @@ pub struct ColumnSerializer<'a, W: io::Write> {
start_offset: u64,
}
impl<'a, W: io::Write> ColumnSerializer<'a, W> {
impl<W: io::Write> ColumnSerializer<'_, W> {
pub fn finalize(self) -> io::Result<()> {
let end_offset: u64 = self.columnar_serializer.wrt.written_bytes();
let byte_range = self.start_offset..end_offset;
@@ -84,7 +80,7 @@ impl<'a, W: io::Write> ColumnSerializer<'a, W> {
}
}
impl<'a, W: io::Write> io::Write for ColumnSerializer<'a, W> {
impl<W: io::Write> io::Write for ColumnSerializer<'_, W> {
fn write(&mut self, buf: &[u8]) -> io::Result<usize> {
self.columnar_serializer.wrt.write(buf)
}
@@ -97,18 +93,3 @@ impl<'a, W: io::Write> io::Write for ColumnSerializer<'a, W> {
self.columnar_serializer.wrt.write_all(buf)
}
}
#[cfg(test)]
mod tests {
use super::*;
#[test]
fn test_prepare_key_bytes() {
let mut buffer: Vec<u8> = b"somegarbage".to_vec();
prepare_key(b"root\0child", ColumnType::Str, &mut buffer);
assert_eq!(buffer.len(), 12);
assert_eq!(&buffer[..10], b"root0child");
assert_eq!(buffer[10], 0u8);
assert_eq!(buffer[11], ColumnType::Str.to_code());
}
}

81
columnar/src/columnar/writer/value_index.rs
View File

@@ -1,3 +1,4 @@
use crate::column_index::{SerializableMultivalueIndex, SerializableOptionalIndex};
use crate::iterable::Iterable;
use crate::RowId;
@@ -59,31 +60,47 @@ impl IndexBuilder for OptionalIndexBuilder {
#[derive(Default)]
pub struct MultivaluedIndexBuilder {
start_offsets: Vec<RowId>,
doc_with_values: Vec<RowId>,
start_offsets: Vec<u32>,
total_num_vals_seen: u32,
current_row: RowId,
current_row_has_value: bool,
}
impl MultivaluedIndexBuilder {
pub fn finish(&mut self, num_docs: RowId) -> &[u32] {
self.start_offsets
.resize(num_docs as usize + 1, self.total_num_vals_seen);
&self.start_offsets[..]
pub fn finish(&mut self, num_docs: RowId) -> SerializableMultivalueIndex<'_> {
self.start_offsets.push(self.total_num_vals_seen);
let non_null_row_ids: Box<dyn Iterable<RowId>> = Box::new(&self.doc_with_values[..]);
SerializableMultivalueIndex {
doc_ids_with_values: SerializableOptionalIndex {
non_null_row_ids,
num_rows: num_docs,
},
start_offsets: Box::new(&self.start_offsets[..]),
}
}
fn reset(&mut self) {
self.doc_with_values.clear();
self.start_offsets.clear();
self.start_offsets.push(0u32);
self.total_num_vals_seen = 0;
self.current_row = 0;
self.current_row_has_value = false;
}
}
impl IndexBuilder for MultivaluedIndexBuilder {
fn record_row(&mut self, row_id: RowId) {
self.start_offsets
.resize(row_id as usize + 1, self.total_num_vals_seen);
self.current_row = row_id;
self.current_row_has_value = false;
}
fn record_value(&mut self) {
if !self.current_row_has_value {
self.current_row_has_value = true;
self.doc_with_values.push(self.current_row);
self.start_offsets.push(self.total_num_vals_seen);
}
self.total_num_vals_seen += 1;
}
}
@@ -141,6 +158,32 @@ mod tests {
);
}
#[test]
fn test_multivalued_value_index_builder_simple() {
let mut multivalued_value_index_builder = MultivaluedIndexBuilder::default();
{
multivalued_value_index_builder.record_row(0u32);
multivalued_value_index_builder.record_value();
multivalued_value_index_builder.record_value();
let serialized_multivalue_index = multivalued_value_index_builder.finish(1u32);
let start_offsets: Vec<u32> = serialized_multivalue_index
.start_offsets
.boxed_iter()
.collect();
assert_eq!(&start_offsets, &[0, 2]);
}
multivalued_value_index_builder.reset();
multivalued_value_index_builder.record_row(0u32);
multivalued_value_index_builder.record_value();
multivalued_value_index_builder.record_value();
let serialized_multivalue_index = multivalued_value_index_builder.finish(1u32);
let start_offsets: Vec<u32> = serialized_multivalue_index
.start_offsets
.boxed_iter()
.collect();
assert_eq!(&start_offsets, &[0, 2]);
}
#[test]
fn test_multivalued_value_index_builder() {
let mut multivalued_value_index_builder = MultivaluedIndexBuilder::default();
@@ -149,17 +192,15 @@ mod tests {
multivalued_value_index_builder.record_value();
multivalued_value_index_builder.record_row(2u32);
multivalued_value_index_builder.record_value();
assert_eq!(
multivalued_value_index_builder.finish(4u32).to_vec(),
vec![0, 0, 2, 3, 3]
);
multivalued_value_index_builder.reset();
multivalued_value_index_builder.record_row(2u32);
multivalued_value_index_builder.record_value();
multivalued_value_index_builder.record_value();
assert_eq!(
multivalued_value_index_builder.finish(4u32).to_vec(),
vec![0, 0, 0, 2, 2]
);
let SerializableMultivalueIndex {
doc_ids_with_values,
start_offsets,
} = multivalued_value_index_builder.finish(4u32);
assert_eq!(doc_ids_with_values.num_rows, 4u32);
let doc_ids_with_values: Vec<u32> =
doc_ids_with_values.non_null_row_ids.boxed_iter().collect();
assert_eq!(&doc_ids_with_values, &[1u32, 2u32]);
let start_offsets: Vec<u32> = start_offsets.boxed_iter().collect();
assert_eq!(&start_offsets[..], &[0, 2, 3]);
}
}

183
columnar/src/compat_tests.rs Normal file
View File

@@ -0,0 +1,183 @@
use std::path::PathBuf;
use itertools::Itertools;
use crate::{
merge_columnar, Cardinality, Column, ColumnarReader, DynamicColumn, StackMergeOrder,
CURRENT_VERSION,
};
const NUM_DOCS: u32 = u16::MAX as u32;
fn generate_columnar(num_docs: u32, value_offset: u64) -> Vec<u8> {
use crate::ColumnarWriter;
let mut columnar_writer = ColumnarWriter::default();
for i in 0..num_docs {
if i % 100 == 0 {
columnar_writer.record_numerical(i, "sparse", value_offset + i as u64);
}
if i % 5 == 0 {
columnar_writer.record_numerical(i, "dense", value_offset + i as u64);
}
columnar_writer.record_numerical(i, "full", value_offset + i as u64);
columnar_writer.record_numerical(i, "multi", value_offset + i as u64);
columnar_writer.record_numerical(i, "multi", value_offset + i as u64);
}
let mut wrt: Vec<u8> = Vec::new();
columnar_writer.serialize(num_docs, &mut wrt).unwrap();
wrt
}
#[test]
/// Writes a columnar for the CURRENT_VERSION to disk.
fn create_format() {
let version = CURRENT_VERSION.to_string();
let file_path = path_for_version(&version);
if PathBuf::from(file_path.clone()).exists() {
return;
}
let columnar = generate_columnar(NUM_DOCS, 0);
std::fs::write(file_path, columnar).unwrap();
}
fn path_for_version(version: &str) -> String {
format!("./compat_tests_data/{}.columnar", version)
}
#[test]
fn test_format_v1() {
let path = path_for_version("v1");
test_format(&path);
}
#[test]
fn test_format_v2() {
let path = path_for_version("v2");
test_format(&path);
}
fn test_format(path: &str) {
let file_content = std::fs::read(path).unwrap();
let reader = ColumnarReader::open(file_content).unwrap();
check_columns(&reader);
// Test merge
let reader2 = ColumnarReader::open(generate_columnar(NUM_DOCS, NUM_DOCS as u64)).unwrap();
let columnar_readers = vec![&reader, &reader2];
let merge_row_order = StackMergeOrder::stack(&columnar_readers[..]);
let mut out = Vec::new();
merge_columnar(&columnar_readers, &[], merge_row_order.into(), &mut out).unwrap();
let reader = ColumnarReader::open(out).unwrap();
check_columns(&reader);
}
fn check_columns(reader: &ColumnarReader) {
let column = open_column(reader, "full");
check_column(&column, |doc_id| vec![(doc_id, doc_id as u64).into()]);
assert_eq!(column.get_cardinality(), Cardinality::Full);
let column = open_column(reader, "multi");
check_column(&column, |doc_id| {
vec![
(doc_id * 2, doc_id as u64).into(),
(doc_id * 2 + 1, doc_id as u64).into(),
]
});
assert_eq!(column.get_cardinality(), Cardinality::Multivalued);
let column = open_column(reader, "sparse");
check_column(&column, |doc_id| {
if doc_id % 100 == 0 {
vec![(doc_id / 100, doc_id as u64).into()]
} else {
vec![]
}
});
assert_eq!(column.get_cardinality(), Cardinality::Optional);
let column = open_column(reader, "dense");
check_column(&column, |doc_id| {
if doc_id % 5 == 0 {
vec![(doc_id / 5, doc_id as u64).into()]
} else {
vec![]
}
});
assert_eq!(column.get_cardinality(), Cardinality::Optional);
}
struct RowIdAndValue {
row_id: u32,
value: u64,
}
impl From<(u32, u64)> for RowIdAndValue {
fn from((row_id, value): (u32, u64)) -> Self {
Self { row_id, value }
}
}
fn check_column<F: Fn(u32) -> Vec<RowIdAndValue>>(column: &Column<u64>, expected: F) {
let num_docs = column.num_docs();
let test_doc = |doc: u32| {
if expected(doc).is_empty() {
assert_eq!(column.first(doc), None);
} else {
assert_eq!(column.first(doc), Some(expected(doc)[0].value));
}
let values = column.values_for_doc(doc).collect_vec();
assert_eq!(values, expected(doc).iter().map(|x| x.value).collect_vec());
let mut row_ids = Vec::new();
column.row_ids_for_docs(&[doc], &mut vec![], &mut row_ids);
assert_eq!(
row_ids,
expected(doc).iter().map(|x| x.row_id).collect_vec()
);
let values = column.values_for_doc(doc).collect_vec();
assert_eq!(values, expected(doc).iter().map(|x| x.value).collect_vec());
// Docid rowid conversion
let mut row_ids = Vec::new();
let safe_next_doc = |doc: u32| (doc + 1).min(num_docs - 1);
column
.index
.docids_to_rowids(&[doc, safe_next_doc(doc)], &mut vec![], &mut row_ids);
let expected_rowids = expected(doc)
.iter()
.map(|x| x.row_id)
.chain(expected(safe_next_doc(doc)).iter().map(|x| x.row_id))
.collect_vec();
assert_eq!(row_ids, expected_rowids);
let rowid_range = column
.index
.docid_range_to_rowids(doc..safe_next_doc(doc) + 1);
if expected_rowids.is_empty() {
assert!(rowid_range.is_empty());
} else {
assert_eq!(
rowid_range,
expected_rowids[0]..expected_rowids.last().unwrap() + 1
);
}
};
test_doc(0);
test_doc(num_docs - 1);
test_doc(num_docs - 2);
test_doc(65000);
}
fn open_column(reader: &ColumnarReader, name: &str) -> Column<u64> {
let column = reader.read_columns(name).unwrap()[0]
.open()
.unwrap()
.coerce_numerical(crate::NumericalType::U64)
.unwrap();
let DynamicColumn::U64(column) = column else {
panic!();
};
column
}

46
columnar/src/dynamic_column.rs
View File

@@ -8,7 +8,7 @@ use common::{ByteCount, DateTime, HasLen, OwnedBytes};
use crate::column::{BytesColumn, Column, StrColumn};
use crate::column_values::{monotonic_map_column, StrictlyMonotonicFn};
use crate::columnar::ColumnType;
use crate::{Cardinality, ColumnIndex, ColumnValues, NumericalType};
use crate::{Cardinality, ColumnIndex, ColumnValues, NumericalType, Version};
#[derive(Clone)]
pub enum DynamicColumn {
@@ -232,6 +232,7 @@ static_dynamic_conversions!(Column<Ipv6Addr>, IpAddr);
pub struct DynamicColumnHandle {
pub(crate) file_slice: FileSlice,
pub(crate) column_type: ColumnType,
pub(crate) format_version: Version,
}
impl DynamicColumnHandle {
@@ -260,11 +261,15 @@ impl DynamicColumnHandle {
let column_bytes = self.file_slice.read_bytes()?;
match self.column_type {
ColumnType::Str | ColumnType::Bytes => {
let column: BytesColumn = crate::column::open_column_bytes(column_bytes)?;
let column: BytesColumn =
crate::column::open_column_bytes(column_bytes, self.format_version)?;
Ok(Some(column.term_ord_column))
}
ColumnType::IpAddr => {
let column = crate::column::open_column_u128_as_compact_u64(column_bytes)?;
let column = crate::column::open_column_u128_as_compact_u64(
column_bytes,
self.format_version,
)?;
Ok(Some(column))
}
ColumnType::Bool
@@ -272,7 +277,8 @@ impl DynamicColumnHandle {
| ColumnType::U64
| ColumnType::F64
| ColumnType::DateTime => {
let column = crate::column::open_column_u64::<u64>(column_bytes)?;
let column =
crate::column::open_column_u64::<u64>(column_bytes, self.format_version)?;
Ok(Some(column))
}
}
@@ -280,15 +286,31 @@ impl DynamicColumnHandle {
fn open_internal(&self, column_bytes: OwnedBytes) -> io::Result<DynamicColumn> {
let dynamic_column: DynamicColumn = match self.column_type {
ColumnType::Bytes => crate::column::open_column_bytes(column_bytes)?.into(),
ColumnType::Str => crate::column::open_column_str(column_bytes)?.into(),
ColumnType::I64 => crate::column::open_column_u64::<i64>(column_bytes)?.into(),
ColumnType::U64 => crate::column::open_column_u64::<u64>(column_bytes)?.into(),
ColumnType::F64 => crate::column::open_column_u64::<f64>(column_bytes)?.into(),
ColumnType::Bool => crate::column::open_column_u64::<bool>(column_bytes)?.into(),
ColumnType::IpAddr => crate::column::open_column_u128::<Ipv6Addr>(column_bytes)?.into(),
ColumnType::Bytes => {
crate::column::open_column_bytes(column_bytes, self.format_version)?.into()
}
ColumnType::Str => {
crate::column::open_column_str(column_bytes, self.format_version)?.into()
}
ColumnType::I64 => {
crate::column::open_column_u64::<i64>(column_bytes, self.format_version)?.into()
}
ColumnType::U64 => {
crate::column::open_column_u64::<u64>(column_bytes, self.format_version)?.into()
}
ColumnType::F64 => {
crate::column::open_column_u64::<f64>(column_bytes, self.format_version)?.into()
}
ColumnType::Bool => {
crate::column::open_column_u64::<bool>(column_bytes, self.format_version)?.into()
}
ColumnType::IpAddr => {
crate::column::open_column_u128::<Ipv6Addr>(column_bytes, self.format_version)?
.into()
}
ColumnType::DateTime => {
crate::column::open_column_u64::<DateTime>(column_bytes)?.into()
crate::column::open_column_u64::<DateTime>(column_bytes, self.format_version)?
.into()
}
};
Ok(dynamic_column)

11
columnar/src/iterable.rs
View File

@@ -1,10 +1,13 @@
use std::ops::Range;
use std::sync::Arc;
use crate::{ColumnValues, RowId};
pub trait Iterable<T = u64> {
fn boxed_iter(&self) -> Box<dyn Iterator<Item = T> + '_>;
}
impl<'a, T: Copy> Iterable<T> for &'a [T] {
impl<T: Copy> Iterable<T> for &[T] {
fn boxed_iter(&self) -> Box<dyn Iterator<Item = T> + '_> {
Box::new(self.iter().copied())
}
@@ -17,3 +20,9 @@ where Range<T>: Iterator<Item = T>
Box::new(self.clone())
}
}
impl Iterable for Arc<dyn crate::ColumnValues<RowId>> {
fn boxed_iter(&self) -> Box<dyn Iterator<Item = u64> + '_> {
Box::new(self.iter().map(|row_id| row_id as u64))
}
}

5
columnar/src/lib.rs
View File

@@ -48,7 +48,7 @@ pub use column_values::{
};
pub use columnar::{
merge_columnar, ColumnType, ColumnarReader, ColumnarWriter, HasAssociatedColumnType,
MergeRowOrder, ShuffleMergeOrder, StackMergeOrder,
MergeRowOrder, ShuffleMergeOrder, StackMergeOrder, Version, CURRENT_VERSION,
};
use sstable::VoidSSTable;
pub use value::{NumericalType, NumericalValue};
@@ -131,3 +131,6 @@ impl Cardinality {
#[cfg(test)]
mod tests;
#[cfg(test)]
mod compat_tests;

205
columnar/src/tests.rs
View File

@@ -21,7 +21,7 @@ fn test_dataframe_writer_str() {
dataframe_writer.record_str(1u32, "my_string", "hello");
dataframe_writer.record_str(3u32, "my_string", "helloeee");
let mut buffer: Vec<u8> = Vec::new();
dataframe_writer.serialize(5, None, &mut buffer).unwrap();
dataframe_writer.serialize(5, &mut buffer).unwrap();
let columnar = ColumnarReader::open(buffer).unwrap();
assert_eq!(columnar.num_columns(), 1);
let cols: Vec<DynamicColumnHandle> = columnar.read_columns("my_string").unwrap();
@@ -35,7 +35,7 @@ fn test_dataframe_writer_bytes() {
dataframe_writer.record_bytes(1u32, "my_string", b"hello");
dataframe_writer.record_bytes(3u32, "my_string", b"helloeee");
let mut buffer: Vec<u8> = Vec::new();
dataframe_writer.serialize(5, None, &mut buffer).unwrap();
dataframe_writer.serialize(5, &mut buffer).unwrap();
let columnar = ColumnarReader::open(buffer).unwrap();
assert_eq!(columnar.num_columns(), 1);
let cols: Vec<DynamicColumnHandle> = columnar.read_columns("my_string").unwrap();
@@ -49,7 +49,7 @@ fn test_dataframe_writer_bool() {
dataframe_writer.record_bool(1u32, "bool.value", false);
dataframe_writer.record_bool(3u32, "bool.value", true);
let mut buffer: Vec<u8> = Vec::new();
dataframe_writer.serialize(5, None, &mut buffer).unwrap();
dataframe_writer.serialize(5, &mut buffer).unwrap();
let columnar = ColumnarReader::open(buffer).unwrap();
assert_eq!(columnar.num_columns(), 1);
let cols: Vec<DynamicColumnHandle> = columnar.read_columns("bool.value").unwrap();
@@ -74,12 +74,12 @@ fn test_dataframe_writer_u64_multivalued() {
dataframe_writer.record_numerical(6u32, "divisor", 2u64);
dataframe_writer.record_numerical(6u32, "divisor", 3u64);
let mut buffer: Vec<u8> = Vec::new();
dataframe_writer.serialize(7, None, &mut buffer).unwrap();
dataframe_writer.serialize(7, &mut buffer).unwrap();
let columnar = ColumnarReader::open(buffer).unwrap();
assert_eq!(columnar.num_columns(), 1);
let cols: Vec<DynamicColumnHandle> = columnar.read_columns("divisor").unwrap();
assert_eq!(cols.len(), 1);
assert_eq!(cols[0].num_bytes(), 29);
assert_eq!(cols[0].num_bytes(), 50);
let dyn_i64_col = cols[0].open().unwrap();
let DynamicColumn::I64(divisor_col) = dyn_i64_col else {
panic!();
@@ -97,7 +97,7 @@ fn test_dataframe_writer_ip_addr() {
dataframe_writer.record_ip_addr(1, "ip_addr", Ipv6Addr::from_u128(1001));
dataframe_writer.record_ip_addr(3, "ip_addr", Ipv6Addr::from_u128(1050));
let mut buffer: Vec<u8> = Vec::new();
dataframe_writer.serialize(5, None, &mut buffer).unwrap();
dataframe_writer.serialize(5, &mut buffer).unwrap();
let columnar = ColumnarReader::open(buffer).unwrap();
assert_eq!(columnar.num_columns(), 1);
let cols: Vec<DynamicColumnHandle> = columnar.read_columns("ip_addr").unwrap();
@@ -128,7 +128,7 @@ fn test_dataframe_writer_numerical() {
dataframe_writer.record_numerical(2u32, "srical.value", NumericalValue::U64(13u64));
dataframe_writer.record_numerical(4u32, "srical.value", NumericalValue::U64(15u64));
let mut buffer: Vec<u8> = Vec::new();
dataframe_writer.serialize(6, None, &mut buffer).unwrap();
dataframe_writer.serialize(6, &mut buffer).unwrap();
let columnar = ColumnarReader::open(buffer).unwrap();
assert_eq!(columnar.num_columns(), 1);
let cols: Vec<DynamicColumnHandle> = columnar.read_columns("srical.value").unwrap();
@@ -153,46 +153,6 @@ fn test_dataframe_writer_numerical() {
assert_eq!(column_i64.first(6), None); //< we can change the spec for that one.
}
#[test]
fn test_dataframe_sort_by_full() {
let mut dataframe_writer = ColumnarWriter::default();
dataframe_writer.record_numerical(0u32, "value", NumericalValue::U64(1));
dataframe_writer.record_numerical(1u32, "value", NumericalValue::U64(2));
let data = dataframe_writer.sort_order("value", 2, false);
assert_eq!(data, vec![0, 1]);
}
#[test]
fn test_dataframe_sort_by_opt() {
let mut dataframe_writer = ColumnarWriter::default();
dataframe_writer.record_numerical(1u32, "value", NumericalValue::U64(3));
dataframe_writer.record_numerical(3u32, "value", NumericalValue::U64(2));
let data = dataframe_writer.sort_order("value", 5, false);
// 0, 2, 4 is 0.0
assert_eq!(data, vec![0, 2, 4, 3, 1]);
let data = dataframe_writer.sort_order("value", 5, true);
assert_eq!(
data,
vec![4, 2, 0, 3, 1].into_iter().rev().collect::<Vec<_>>()
);
}
#[test]
fn test_dataframe_sort_by_multi() {
let mut dataframe_writer = ColumnarWriter::default();
// valid for sort
dataframe_writer.record_numerical(1u32, "value", NumericalValue::U64(2));
// those are ignored for sort
dataframe_writer.record_numerical(1u32, "value", NumericalValue::U64(4));
dataframe_writer.record_numerical(1u32, "value", NumericalValue::U64(4));
// valid for sort
dataframe_writer.record_numerical(3u32, "value", NumericalValue::U64(3));
// ignored, would change sort order
dataframe_writer.record_numerical(3u32, "value", NumericalValue::U64(1));
let data = dataframe_writer.sort_order("value", 4, false);
assert_eq!(data, vec![0, 2, 1, 3]);
}
#[test]
fn test_dictionary_encoded_str() {
let mut buffer = Vec::new();
@@ -201,7 +161,7 @@ fn test_dictionary_encoded_str() {
columnar_writer.record_str(3, "my.column", "c");
columnar_writer.record_str(3, "my.column2", "different_column!");
columnar_writer.record_str(4, "my.column", "b");
columnar_writer.serialize(5, None, &mut buffer).unwrap();
columnar_writer.serialize(5, &mut buffer).unwrap();
let columnar_reader = ColumnarReader::open(buffer).unwrap();
assert_eq!(columnar_reader.num_columns(), 2);
let col_handles = columnar_reader.read_columns("my.column").unwrap();
@@ -235,7 +195,7 @@ fn test_dictionary_encoded_bytes() {
columnar_writer.record_bytes(3, "my.column", b"c");
columnar_writer.record_bytes(3, "my.column2", b"different_column!");
columnar_writer.record_bytes(4, "my.column", b"b");
columnar_writer.serialize(5, None, &mut buffer).unwrap();
columnar_writer.serialize(5, &mut buffer).unwrap();
let columnar_reader = ColumnarReader::open(buffer).unwrap();
assert_eq!(columnar_reader.num_columns(), 2);
let col_handles = columnar_reader.read_columns("my.column").unwrap();
@@ -344,7 +304,7 @@ fn column_value_strategy() -> impl Strategy<Value = ColumnValue> {
ip_addr_byte
))),
1 => any::<bool>().prop_map(ColumnValue::Bool),
1 => (0_679_723_993i64..1_679_723_995i64)
1 => (679_723_993i64..1_679_723_995i64)
.prop_map(|val| { ColumnValue::DateTime(DateTime::from_timestamp_secs(val)) })
]
}
@@ -369,26 +329,12 @@ fn columnar_docs_strategy() -> impl Strategy<Value = Vec<Vec<(&'static str, Colu
.prop_flat_map(|num_docs| proptest::collection::vec(doc_strategy(), num_docs))
}
fn columnar_docs_and_mapping_strategy(
) -> impl Strategy<Value = (Vec<Vec<(&'static str, ColumnValue)>>, Vec<RowId>)> {
columnar_docs_strategy().prop_flat_map(|docs| {
permutation_strategy(docs.len()).prop_map(move |permutation| (docs.clone(), permutation))
})
}
fn permutation_strategy(n: usize) -> impl Strategy<Value = Vec<RowId>> {
Just((0u32..n as RowId).collect()).prop_shuffle()
}
fn permutation_and_subset_strategy(n: usize) -> impl Strategy<Value = Vec<usize>> {
let vals: Vec<usize> = (0..n).collect();
subsequence(vals, 0..=n).prop_shuffle()
}
fn build_columnar_with_mapping(
docs: &[Vec<(&'static str, ColumnValue)>],
old_to_new_row_ids_opt: Option<&[RowId]>,
) -> ColumnarReader {
fn build_columnar_with_mapping(docs: &[Vec<(&'static str, ColumnValue)>]) -> ColumnarReader {
let num_docs = docs.len() as u32;
let mut buffer = Vec::new();
let mut columnar_writer = ColumnarWriter::default();
@@ -416,15 +362,13 @@ fn build_columnar_with_mapping(
}
}
}
columnar_writer
.serialize(num_docs, old_to_new_row_ids_opt, &mut buffer)
.unwrap();
columnar_writer.serialize(num_docs, &mut buffer).unwrap();
ColumnarReader::open(buffer).unwrap()
}
fn build_columnar(docs: &[Vec<(&'static str, ColumnValue)>]) -> ColumnarReader {
build_columnar_with_mapping(docs, None)
build_columnar_with_mapping(docs)
}
fn assert_columnar_eq_strict(left: &ColumnarReader, right: &ColumnarReader) {
@@ -448,6 +392,7 @@ fn assert_columnar_eq(
}
}
#[track_caller]
fn assert_column_eq<T: Copy + PartialOrd + Debug + Send + Sync + 'static>(
left: &Column<T>,
right: &Column<T>,
@@ -683,54 +628,6 @@ proptest! {
}
}
// Same as `test_single_columnar_builder_proptest` but with a shuffling mapping.
proptest! {
#![proptest_config(ProptestConfig::with_cases(500))]
#[test]
fn test_single_columnar_builder_with_shuffle_proptest((docs, mapping) in columnar_docs_and_mapping_strategy()) {
let columnar = build_columnar_with_mapping(&docs[..], Some(&mapping));
assert_eq!(columnar.num_rows() as usize, docs.len());
let mut expected_columns: HashMap<(&str, ColumnTypeCategory), HashMap<u32, Vec<&ColumnValue>> > = Default::default();
for (doc_id, doc_vals) in docs.iter().enumerate() {
for (col_name, col_val) in doc_vals {
expected_columns
.entry((col_name, col_val.column_type_category()))
.or_default()
.entry(mapping[doc_id])
.or_default()
.push(col_val);
}
}
let column_list = columnar.list_columns().unwrap();
assert_eq!(expected_columns.len(), column_list.len());
for (column_name, column) in column_list {
let dynamic_column = column.open().unwrap();
let col_category: ColumnTypeCategory = dynamic_column.column_type().into();
let expected_col_values: &HashMap<u32, Vec<&ColumnValue>> = expected_columns.get(&(column_name.as_str(), col_category)).unwrap();
for _doc_id in 0..columnar.num_rows() {
match &dynamic_column {
DynamicColumn::Bool(col) =>
assert_column_values(col, expected_col_values),
DynamicColumn::I64(col) =>
assert_column_values(col, expected_col_values),
DynamicColumn::U64(col) =>
assert_column_values(col, expected_col_values),
DynamicColumn::F64(col) =>
assert_column_values(col, expected_col_values),
DynamicColumn::IpAddr(col) =>
assert_column_values(col, expected_col_values),
DynamicColumn::DateTime(col) =>
assert_column_values(col, expected_col_values),
DynamicColumn::Bytes(col) =>
assert_bytes_column_values(col, expected_col_values, false),
DynamicColumn::Str(col) =>
assert_bytes_column_values(col, expected_col_values, true),
}
}
}
}
}
// This tests create 2 or 3 random small columnar and attempts to merge them.
// It compares the resulting merged dataframe with what would have been obtained by building the
// dataframe from the concatenated rows to begin with.
@@ -844,24 +741,68 @@ fn columnar_docs_and_remap(
proptest! {
#![proptest_config(ProptestConfig::with_cases(1000))]
#[test]
fn test_columnar_merge_and_remap_proptest((columnar_docs, shuffle_merge_order) in columnar_docs_and_remap()) {
let shuffled_rows: Vec<Vec<(&'static str, ColumnValue)>> = shuffle_merge_order.iter()
.map(|row_addr| columnar_docs[row_addr.segment_ord as usize][row_addr.row_id as usize].clone())
.collect();
let expected_merged_columnar = build_columnar(&shuffled_rows[..]);
let columnar_readers: Vec<ColumnarReader> = columnar_docs.iter()
.map(|docs| build_columnar(&docs[..]))
.collect::<Vec<_>>();
let columnar_readers_arr: Vec<&ColumnarReader> = columnar_readers.iter().collect();
let mut output: Vec<u8> = Vec::new();
let segment_num_rows: Vec<RowId> = columnar_docs.iter().map(|docs| docs.len() as RowId).collect();
let shuffle_merge_order = ShuffleMergeOrder::for_test(&segment_num_rows, shuffle_merge_order);
crate::merge_columnar(&columnar_readers_arr[..], &[], shuffle_merge_order.into(), &mut output).unwrap();
let merged_columnar = ColumnarReader::open(output).unwrap();
assert_columnar_eq(&merged_columnar, &expected_merged_columnar, true);
fn test_columnar_merge_and_remap_proptest((columnar_docs, shuffle_merge_order) in
columnar_docs_and_remap()) {
test_columnar_merge_and_remap(columnar_docs, shuffle_merge_order);
}
}
fn test_columnar_merge_and_remap(
columnar_docs: Vec<Vec<Vec<(&'static str, ColumnValue)>>>,
shuffle_merge_order: Vec<RowAddr>,
) {
let shuffled_rows: Vec<Vec<(&'static str, ColumnValue)>> = shuffle_merge_order
.iter()
.map(|row_addr| {
columnar_docs[row_addr.segment_ord as usize][row_addr.row_id as usize].clone()
})
.collect();
let expected_merged_columnar = build_columnar(&shuffled_rows[..]);
let columnar_readers: Vec<ColumnarReader> = columnar_docs
.iter()
.map(|docs| build_columnar(&docs[..]))
.collect::<Vec<_>>();
let columnar_readers_ref: Vec<&ColumnarReader> = columnar_readers.iter().collect();
let mut output: Vec<u8> = Vec::new();
let segment_num_rows: Vec<RowId> = columnar_docs
.iter()
.map(|docs| docs.len() as RowId)
.collect();
let shuffle_merge_order = ShuffleMergeOrder::for_test(&segment_num_rows, shuffle_merge_order);
crate::merge_columnar(
&columnar_readers_ref[..],
&[],
shuffle_merge_order.into(),
&mut output,
)
.unwrap();
let merged_columnar = ColumnarReader::open(output).unwrap();
assert_columnar_eq(&merged_columnar, &expected_merged_columnar, true);
}
#[test]
fn test_columnar_merge_and_remap_bug_1() {
let columnar_docs = vec![vec![
vec![
("c1", ColumnValue::Numerical(NumericalValue::U64(0))),
("c1", ColumnValue::Numerical(NumericalValue::U64(0))),
],
vec![],
]];
let shuffle_merge_order: Vec<RowAddr> = vec![
RowAddr {
segment_ord: 0,
row_id: 1,
},
RowAddr {
segment_ord: 0,
row_id: 0,
},
];
test_columnar_merge_and_remap(columnar_docs, shuffle_merge_order);
}
#[test]
fn test_columnar_merge_empty() {
let columnar_reader_1 = build_columnar(&[]);

25
columnar/src/value.rs
View File

@@ -17,6 +17,31 @@ impl NumericalValue {
NumericalValue::F64(_) => NumericalType::F64,
}
}
/// Tries to normalize the numerical value in the following priorities:
/// i64, i64, f64
pub fn normalize(self) -> Self {
match self {
NumericalValue::U64(val) => {
if val <= i64::MAX as u64 {
NumericalValue::I64(val as i64)
} else {
NumericalValue::F64(val as f64)
}
}
NumericalValue::I64(val) => NumericalValue::I64(val),
NumericalValue::F64(val) => {
let fract = val.fract();
if fract == 0.0 && val >= i64::MIN as f64 && val <= i64::MAX as f64 {
NumericalValue::I64(val as i64)
} else if fract == 0.0 && val >= u64::MIN as f64 && val <= u64::MAX as f64 {
NumericalValue::U64(val as u64)
} else {
NumericalValue::F64(val)
}
}
}
}
}
impl From<u64> for NumericalValue {

3
common/Cargo.toml
View File

@@ -9,7 +9,6 @@ documentation = "https://docs.rs/tantivy_common/"
homepage = "https://github.com/quickwit-oss/tantivy"
repository = "https://github.com/quickwit-oss/tantivy"
# See more keys and their definitions at https://doc.rust-lang.org/cargo/reference/manifest.html
[dependencies]
@@ -20,5 +19,7 @@ time = { version = "0.3.10", features = ["serde-well-known"] }
serde = { version = "1.0.136", features = ["derive"] }
[dev-dependencies]
binggan = "0.14.0"
proptest = "1.0.0"
rand = "0.8.4"

95
common/benches/bench.rs
View File

@@ -1,39 +1,64 @@
#![feature(test)]
use binggan::{black_box, BenchRunner};
use rand::seq::IteratorRandom;
use rand::thread_rng;
use tantivy_common::{serialize_vint_u32, BitSet, TinySet};
extern crate test;
fn bench_vint() {
let mut runner = BenchRunner::new();
#[cfg(test)]
mod tests {
use rand::seq::IteratorRandom;
use rand::thread_rng;
use tantivy_common::serialize_vint_u32;
use test::Bencher;
let vals: Vec<u32> = (0..20_000).collect();
runner.bench_function("bench_vint", move |_| {
let mut out = 0u64;
for val in vals.iter().cloned() {
let mut buf = [0u8; 8];
serialize_vint_u32(val, &mut buf);
out += u64::from(buf[0]);
}
black_box(out);
});
#[bench]
fn bench_vint(b: &mut Bencher) {
let vals: Vec<u32> = (0..20_000).collect();
b.iter(|| {
let mut out = 0u64;
for val in vals.iter().cloned() {
let mut buf = [0u8; 8];
serialize_vint_u32(val, &mut buf);
out += u64::from(buf[0]);
}
out
});
}
#[bench]
fn bench_vint_rand(b: &mut Bencher) {
let vals: Vec<u32> = (0..20_000).choose_multiple(&mut thread_rng(), 100_000);
b.iter(|| {
let mut out = 0u64;
for val in vals.iter().cloned() {
let mut buf = [0u8; 8];
serialize_vint_u32(val, &mut buf);
out += u64::from(buf[0]);
}
out
});
}
let vals: Vec<u32> = (0..20_000).choose_multiple(&mut thread_rng(), 100_000);
runner.bench_function("bench_vint_rand", move |_| {
let mut out = 0u64;
for val in vals.iter().cloned() {
let mut buf = [0u8; 8];
serialize_vint_u32(val, &mut buf);
out += u64::from(buf[0]);
}
black_box(out);
});
}
fn bench_bitset() {
let mut runner = BenchRunner::new();
runner.bench_function("bench_tinyset_pop", move |_| {
let mut tinyset = TinySet::singleton(black_box(31u32));
tinyset.pop_lowest();
tinyset.pop_lowest();
tinyset.pop_lowest();
tinyset.pop_lowest();
tinyset.pop_lowest();
tinyset.pop_lowest();
black_box(tinyset);
});
let tiny_set = TinySet::empty().insert(10u32).insert(14u32).insert(21u32);
runner.bench_function("bench_tinyset_sum", move |_| {
assert_eq!(black_box(tiny_set).into_iter().sum::<u32>(), 45u32);
});
let v = [10u32, 14u32, 21u32];
runner.bench_function("bench_tinyarr_sum", move |_| {
black_box(v.iter().cloned().sum::<u32>());
});
runner.bench_function("bench_bitset_initialize", move |_| {
black_box(BitSet::with_max_value(1_000_000));
});
}
fn main() {
bench_vint();
bench_bitset();
}

42
common/src/bitset.rs
View File

@@ -1,5 +1,5 @@
use std::io::Write;
use std::{fmt, io, u64};
use std::{fmt, io};
use ownedbytes::OwnedBytes;
@@ -696,43 +696,3 @@ mod tests {
}
}
}
#[cfg(all(test, feature = "unstable"))]
mod bench {
use test;
use super::{BitSet, TinySet};
#[bench]
fn bench_tinyset_pop(b: &mut test::Bencher) {
b.iter(|| {
let mut tinyset = TinySet::singleton(test::black_box(31u32));
tinyset.pop_lowest();
tinyset.pop_lowest();
tinyset.pop_lowest();
tinyset.pop_lowest();
tinyset.pop_lowest();
tinyset.pop_lowest();
});
}
#[bench]
fn bench_tinyset_sum(b: &mut test::Bencher) {
let tiny_set = TinySet::empty().insert(10u32).insert(14u32).insert(21u32);
b.iter(|| {
assert_eq!(test::black_box(tiny_set).into_iter().sum::<u32>(), 45u32);
});
}
#[bench]
fn bench_tinyarr_sum(b: &mut test::Bencher) {
let v = [10u32, 14u32, 21u32];
b.iter(|| test::black_box(v).iter().cloned().sum::<u32>());
}
#[bench]
fn bench_bitset_initialize(b: &mut test::Bencher) {
b.iter(|| BitSet::with_max_value(1_000_000));
}
}

130
common/src/bounds.rs Normal file
View File

@@ -0,0 +1,130 @@
use std::io;
use std::ops::Bound;
#[derive(Clone, Debug)]
pub struct BoundsRange<T> {
pub lower_bound: Bound<T>,
pub upper_bound: Bound<T>,
}
impl<T> BoundsRange<T> {
pub fn new(lower_bound: Bound<T>, upper_bound: Bound<T>) -> Self {
BoundsRange {
lower_bound,
upper_bound,
}
}
pub fn is_unbounded(&self) -> bool {
matches!(self.lower_bound, Bound::Unbounded) && matches!(self.upper_bound, Bound::Unbounded)
}
pub fn map_bound<TTo>(&self, transform: impl Fn(&T) -> TTo) -> BoundsRange<TTo> {
BoundsRange {
lower_bound: map_bound(&self.lower_bound, &transform),
upper_bound: map_bound(&self.upper_bound, &transform),
}
}
pub fn map_bound_res<TTo, Err>(
&self,
transform: impl Fn(&T) -> Result<TTo, Err>,
) -> Result<BoundsRange<TTo>, Err> {
Ok(BoundsRange {
lower_bound: map_bound_res(&self.lower_bound, &transform)?,
upper_bound: map_bound_res(&self.upper_bound, &transform)?,
})
}
pub fn transform_inner<TTo>(
&self,
transform_lower: impl Fn(&T) -> TransformBound<TTo>,
transform_upper: impl Fn(&T) -> TransformBound<TTo>,
) -> BoundsRange<TTo> {
BoundsRange {
lower_bound: transform_bound_inner(&self.lower_bound, &transform_lower),
upper_bound: transform_bound_inner(&self.upper_bound, &transform_upper),
}
}
/// Returns the first set inner value
pub fn get_inner(&self) -> Option<&T> {
inner_bound(&self.lower_bound).or(inner_bound(&self.upper_bound))
}
}
pub enum TransformBound<T> {
/// Overwrite the bounds
NewBound(Bound<T>),
/// Use Existing bounds with new value
Existing(T),
}
/// Takes a bound and transforms the inner value into a new bound via a closure.
/// The bound variant may change by the value returned value from the closure.
pub fn transform_bound_inner_res<TFrom, TTo>(
bound: &Bound<TFrom>,
transform: impl Fn(&TFrom) -> io::Result<TransformBound<TTo>>,
) -> io::Result<Bound<TTo>> {
use self::Bound::*;
Ok(match bound {
Excluded(ref from_val) => match transform(from_val)? {
TransformBound::NewBound(new_val) => new_val,
TransformBound::Existing(new_val) => Excluded(new_val),
},
Included(ref from_val) => match transform(from_val)? {
TransformBound::NewBound(new_val) => new_val,
TransformBound::Existing(new_val) => Included(new_val),
},
Unbounded => Unbounded,
})
}
/// Takes a bound and transforms the inner value into a new bound via a closure.
/// The bound variant may change by the value returned value from the closure.
pub fn transform_bound_inner<TFrom, TTo>(
bound: &Bound<TFrom>,
transform: impl Fn(&TFrom) -> TransformBound<TTo>,
) -> Bound<TTo> {
use self::Bound::*;
match bound {
Excluded(ref from_val) => match transform(from_val) {
TransformBound::NewBound(new_val) => new_val,
TransformBound::Existing(new_val) => Excluded(new_val),
},
Included(ref from_val) => match transform(from_val) {
TransformBound::NewBound(new_val) => new_val,
TransformBound::Existing(new_val) => Included(new_val),
},
Unbounded => Unbounded,
}
}
/// Returns the inner value of a `Bound`
pub fn inner_bound<T>(val: &Bound<T>) -> Option<&T> {
match val {
Bound::Included(term) | Bound::Excluded(term) => Some(term),
Bound::Unbounded => None,
}
}
pub fn map_bound<TFrom, TTo>(
bound: &Bound<TFrom>,
transform: impl Fn(&TFrom) -> TTo,
) -> Bound<TTo> {
use self::Bound::*;
match bound {
Excluded(ref from_val) => Bound::Excluded(transform(from_val)),
Included(ref from_val) => Bound::Included(transform(from_val)),
Unbounded => Unbounded,
}
}
pub fn map_bound_res<TFrom, TTo, Err>(
bound: &Bound<TFrom>,
transform: impl Fn(&TFrom) -> Result<TTo, Err>,
) -> Result<Bound<TTo>, Err> {
use self::Bound::*;
Ok(match bound {
Excluded(ref from_val) => Excluded(transform(from_val)?),
Included(ref from_val) => Included(transform(from_val)?),
Unbounded => Unbounded,
})
}

36
common/src/json_path_writer.rs
View File

@@ -5,6 +5,12 @@ pub const JSON_PATH_SEGMENT_SEP: u8 = 1u8;
pub const JSON_PATH_SEGMENT_SEP_STR: &str =
unsafe { std::str::from_utf8_unchecked(&[JSON_PATH_SEGMENT_SEP]) };
/// Separates the json path and the value in
/// a JSON term binary representation.
pub const JSON_END_OF_PATH: u8 = 0u8;
pub const JSON_END_OF_PATH_STR: &str =
unsafe { std::str::from_utf8_unchecked(&[JSON_END_OF_PATH]) };
/// Create a new JsonPathWriter, that creates flattened json paths for tantivy.
#[derive(Clone, Debug, Default)]
pub struct JsonPathWriter {
@@ -14,6 +20,14 @@ pub struct JsonPathWriter {
}
impl JsonPathWriter {
pub fn with_expand_dots(expand_dots: bool) -> Self {
JsonPathWriter {
path: String::new(),
indices: Vec::new(),
expand_dots,
}
}
pub fn new() -> Self {
JsonPathWriter {
path: String::new(),
@@ -39,8 +53,8 @@ impl JsonPathWriter {
pub fn push(&mut self, segment: &str) {
let len_path = self.path.len();
self.indices.push(len_path);
if !self.path.is_empty() {
self.path.push_str(JSON_PATH_SEGMENT_SEP_STR);
if self.indices.len() > 1 {
self.path.push(JSON_PATH_SEGMENT_SEP as char);
}
self.path.push_str(segment);
if self.expand_dots {
@@ -55,6 +69,12 @@ impl JsonPathWriter {
}
}
/// Set the end of JSON path marker.
#[inline]
pub fn set_end(&mut self) {
self.path.push_str(JSON_END_OF_PATH_STR);
}
/// Remove the last segment. Does nothing if the path is empty.
#[inline]
pub fn pop(&mut self) {
@@ -91,6 +111,7 @@ mod tests {
#[test]
fn json_path_writer_test() {
let mut writer = JsonPathWriter::new();
writer.set_expand_dots(false);
writer.push("root");
assert_eq!(writer.as_str(), "root");
@@ -109,4 +130,15 @@ mod tests {
writer.push("k8s.node.id");
assert_eq!(writer.as_str(), "root\u{1}k8s\u{1}node\u{1}id");
}
#[test]
fn test_json_path_expand_dots_enabled_pop_segment() {
let mut json_writer = JsonPathWriter::with_expand_dots(true);
json_writer.push("hello");
assert_eq!(json_writer.as_str(), "hello");
json_writer.push("color.hue");
assert_eq!(json_writer.as_str(), "hello\x01color\x01hue");
json_writer.pop();
assert_eq!(json_writer.as_str(), "hello");
}
}

13
common/src/lib.rs
View File

@@ -5,11 +5,12 @@ use std::ops::Deref;
pub use byteorder::LittleEndian as Endianness;
mod bitset;
pub mod bounds;
mod byte_count;
mod datetime;
pub mod file_slice;
mod group_by;
mod json_path_writer;
pub mod json_path_writer;
mod serialize;
mod vint;
mod writer;
@@ -129,11 +130,11 @@ pub fn replace_in_place(needle: u8, replacement: u8, bytes: &mut [u8]) {
}
#[cfg(test)]
pub mod test {
pub(crate) mod test {
use proptest::prelude::*;
use super::{f64_to_u64, i64_to_u64, u64_to_f64, u64_to_i64, BinarySerializable, FixedSize};
use super::{f64_to_u64, i64_to_u64, u64_to_f64, u64_to_i64};
fn test_i64_converter_helper(val: i64) {
assert_eq!(u64_to_i64(i64_to_u64(val)), val);
@@ -143,12 +144,6 @@ pub mod test {
assert_eq!(u64_to_f64(f64_to_u64(val)), val);
}
pub fn fixed_size_test<O: BinarySerializable + FixedSize + Default>() {
let mut buffer = Vec::new();
O::default().serialize(&mut buffer).unwrap();
assert_eq!(buffer.len(), O::SIZE_IN_BYTES);
}
proptest! {
#[test]
fn test_f64_converter_monotonicity_proptest((left, right) in (proptest::num::f64::NORMAL, proptest::num::f64::NORMAL)) {

20
common/src/serialize.rs
View File

@@ -74,14 +74,14 @@ impl FixedSize for () {
impl<T: BinarySerializable> BinarySerializable for Vec<T> {
fn serialize<W: Write + ?Sized>(&self, writer: &mut W) -> io::Result<()> {
VInt(self.len() as u64).serialize(writer)?;
BinarySerializable::serialize(&VInt(self.len() as u64), writer)?;
for it in self {
it.serialize(writer)?;
}
Ok(())
}
fn deserialize<R: Read>(reader: &mut R) -> io::Result<Vec<T>> {
let num_items = VInt::deserialize(reader)?.val();
let num_items = <VInt as BinarySerializable>::deserialize(reader)?.val();
let mut items: Vec<T> = Vec::with_capacity(num_items as usize);
for _ in 0..num_items {
let item = T::deserialize(reader)?;
@@ -236,12 +236,12 @@ impl FixedSize for bool {
impl BinarySerializable for String {
fn serialize<W: Write + ?Sized>(&self, writer: &mut W) -> io::Result<()> {
let data: &[u8] = self.as_bytes();
VInt(data.len() as u64).serialize(writer)?;
BinarySerializable::serialize(&VInt(data.len() as u64), writer)?;
writer.write_all(data)
}
fn deserialize<R: Read>(reader: &mut R) -> io::Result<String> {
let string_length = VInt::deserialize(reader)?.val() as usize;
let string_length = <VInt as BinarySerializable>::deserialize(reader)?.val() as usize;
let mut result = String::with_capacity(string_length);
reader
.take(string_length as u64)
@@ -253,12 +253,12 @@ impl BinarySerializable for String {
impl<'a> BinarySerializable for Cow<'a, str> {
fn serialize<W: Write + ?Sized>(&self, writer: &mut W) -> io::Result<()> {
let data: &[u8] = self.as_bytes();
VInt(data.len() as u64).serialize(writer)?;
BinarySerializable::serialize(&VInt(data.len() as u64), writer)?;
writer.write_all(data)
}
fn deserialize<R: Read>(reader: &mut R) -> io::Result<Cow<'a, str>> {
let string_length = VInt::deserialize(reader)?.val() as usize;
let string_length = <VInt as BinarySerializable>::deserialize(reader)?.val() as usize;
let mut result = String::with_capacity(string_length);
reader
.take(string_length as u64)
@@ -269,18 +269,18 @@ impl<'a> BinarySerializable for Cow<'a, str> {
impl<'a> BinarySerializable for Cow<'a, [u8]> {
fn serialize<W: Write + ?Sized>(&self, writer: &mut W) -> io::Result<()> {
VInt(self.len() as u64).serialize(writer)?;
BinarySerializable::serialize(&VInt(self.len() as u64), writer)?;
for it in self.iter() {
it.serialize(writer)?;
BinarySerializable::serialize(it, writer)?;
}
Ok(())
}
fn deserialize<R: Read>(reader: &mut R) -> io::Result<Cow<'a, [u8]>> {
let num_items = VInt::deserialize(reader)?.val();
let num_items = <VInt as BinarySerializable>::deserialize(reader)?.val();
let mut items: Vec<u8> = Vec::with_capacity(num_items as usize);
for _ in 0..num_items {
let item = u8::deserialize(reader)?;
let item = <u8 as BinarySerializable>::deserialize(reader)?;
items.push(item);
}
Ok(Cow::Owned(items))

2
common/src/vint.rs
View File

@@ -151,7 +151,7 @@ pub fn read_u32_vint_no_advance(data: &[u8]) -> (u32, usize) {
(result, vlen)
}
/// Write a `u32` as a vint payload.
pub fn write_u32_vint<W: io::Write>(val: u32, writer: &mut W) -> io::Result<()> {
pub fn write_u32_vint<W: io::Write + ?Sized>(val: u32, writer: &mut W) -> io::Result<()> {
let mut buf = [0u8; 8];
let data = serialize_vint_u32(val, &mut buf);
writer.write_all(data)

2
common/src/writer.rs
View File

@@ -87,7 +87,7 @@ impl<W: TerminatingWrite> TerminatingWrite for BufWriter<W> {
}
}
impl<'a> TerminatingWrite for &'a mut Vec<u8> {
impl TerminatingWrite for &mut Vec<u8> {
fn terminate_ref(&mut self, _a: AntiCallToken) -> io::Result<()> {
self.flush()
}

BIN
doc/assets/images/paradedb.png Normal file
View File

Binary file not shown.
Side by Side

After

Width:  |  Height:  |  Size: 30 KiB

6
doc/src/avant-propos.md
View File

@@ -2,7 +2,7 @@
> Tantivy is a **search** engine **library** for Rust.
If you are familiar with Lucene, it's an excellent approximation to consider tantivy as Lucene for rust. tantivy is heavily inspired by Lucene's design and
If you are familiar with Lucene, it's an excellent approximation to consider tantivy as Lucene for Rust. Tantivy is heavily inspired by Lucene's design and
they both have the same scope and targeted use cases.
If you are not familiar with Lucene, let's break down our little tagline.
@@ -17,7 +17,7 @@ relevancy, collapsing, highlighting, spatial search.
experience. But keep in mind this is just a toolbox.
Which bring us to the second keyword...
- **Library** means that you will have to write code. tantivy is not an *all-in-one* server solution like elastic search for instance.
- **Library** means that you will have to write code. Tantivy is not an *all-in-one* server solution like Elasticsearch for instance.
Sometimes a functionality will not be available in tantivy because it is too
specific to your use case. By design, tantivy should make it possible to extend
@@ -31,4 +31,4 @@ relevancy, collapsing, highlighting, spatial search.
index from a different format.
Tantivy exposes a lot of low level API to do all of these things.

2
doc/src/basis.md
View File

@@ -11,7 +11,7 @@ directory shipped with tantivy is the `MmapDirectory`.
While this design has some downsides, this greatly simplifies the source code of
tantivy. Caching is also entirely delegated to the OS.
`tantivy` works entirely (or almost) by directly reading the datastructures as they are laid on disk. As a result, the act of opening an indexing does not involve loading different datastructures from the disk into random access memory : starting a process, opening an index, and performing your first query can typically be done in a matter of milliseconds.
Tantivy works entirely (or almost) by directly reading the datastructures as they are laid on disk. As a result, the act of opening an indexing does not involve loading different datastructures from the disk into random access memory : starting a process, opening an index, and performing your first query can typically be done in a matter of milliseconds.
This is an interesting property for a command line search engine, or for some multi-tenant log search engine : spawning a new process for each new query can be a perfectly sensible solution in some use case.

11
doc/src/index_sorting.md
View File

@@ -7,6 +7,11 @@
- [Other](#other)
- [Usage](#usage)
# Index Sorting has been removed!
More infos here:
https://github.com/quickwit-oss/tantivy/issues/2352
# Index Sorting
Tantivy allows you to sort the index according to a property.
@@ -26,13 +31,13 @@ Compression ratio is mainly affected on the fast field of the sorted property, e
When data is presorted by a field and search queries request sorting by the same field, we can leverage the natural order of the documents.
E.g. if the data is sorted by timestamp and want the top n newest docs containing a term, we can simply leveraging the order of the docids.
Note: Tantivy 0.16 does not do this optimization yet.
Note: tantivy 0.16 does not do this optimization yet.
### Pruning
Let's say we want all documents and want to apply the filter `>= 2010-08-11`. When the data is sorted, we could make a lookup in the fast field to find the docid range and use this as the filter.
Note: Tantivy 0.16 does not do this optimization yet.
Note: tantivy 0.16 does not do this optimization yet.
### Other?
@@ -40,7 +45,7 @@ In principle there are many algorithms possible that exploit the monotonically i
## Usage
The index sorting can be configured setting [`sort_by_field`](https://github.com/quickwit-oss/tantivy/blob/000d76b11a139a84b16b9b95060a1c93e8b9851c/src/core/index_meta.rs#L238) on `IndexSettings` and passing it to a `IndexBuilder`. As of Tantivy 0.16 only fast fields are allowed to be used.
The index sorting can be configured setting [`sort_by_field`](https://github.com/quickwit-oss/tantivy/blob/000d76b11a139a84b16b9b95060a1c93e8b9851c/src/core/index_meta.rs#L238) on `IndexSettings` and passing it to a `IndexBuilder`. As of tantivy 0.16 only fast fields are allowed to be used.
```rust
let settings = IndexSettings {

2
doc/src/json.md
View File

@@ -39,7 +39,7 @@ Its representation is done by separating segments by a unicode char `\x01`, and
- `value`: The value representation is just the regular Value representation.
This representation is designed to align the natural sort of Terms with the lexicographical sort
of their binary representation (Tantivy's dictionary (whether fst or sstable) is sorted and does prefix encoding).
of their binary representation (tantivy's dictionary (whether fst or sstable) is sorted and does prefix encoding).
In the example above, the terms will be sorted as

3
examples/custom_collector.rs
View File

@@ -11,9 +11,10 @@ use columnar::Column;
// ---
// Importing tantivy...
use tantivy::collector::{Collector, SegmentCollector};
use tantivy::index::SegmentReader;
use tantivy::query::QueryParser;
use tantivy::schema::{Schema, FAST, INDEXED, TEXT};
use tantivy::{doc, Index, IndexWriter, Score, SegmentReader};
use tantivy::{doc, Index, IndexWriter, Score};
#[derive(Default)]
struct Stats {

14
examples/date_time_field.rs
View File

@@ -4,7 +4,7 @@
use tantivy::collector::TopDocs;
use tantivy::query::QueryParser;
use tantivy::schema::{DateOptions, Document, OwnedValue, Schema, INDEXED, STORED, STRING};
use tantivy::schema::{DateOptions, Document, Schema, Value, INDEXED, STORED, STRING};
use tantivy::{Index, IndexWriter, TantivyDocument};
fn main() -> tantivy::Result<()> {
@@ -13,7 +13,7 @@ fn main() -> tantivy::Result<()> {
let opts = DateOptions::from(INDEXED)
.set_stored()
.set_fast()
.set_precision(tantivy::DateTimePrecision::Seconds);
.set_precision(tantivy::schema::DateTimePrecision::Seconds);
// Add `occurred_at` date field type
let occurred_at = schema_builder.add_date_field("occurred_at", opts);
let event_type = schema_builder.add_text_field("event", STRING | STORED);
@@ -61,10 +61,12 @@ fn main() -> tantivy::Result<()> {
assert_eq!(count_docs.len(), 1);
for (_score, doc_address) in count_docs {
let retrieved_doc = searcher.doc::<TantivyDocument>(doc_address)?;
assert!(matches!(
retrieved_doc.get_first(occurred_at),
Some(OwnedValue::Date(_))
));
assert!(retrieved_doc
.get_first(occurred_at)
.unwrap()
.as_value()
.as_datetime()
.is_some(),);
assert_eq!(
retrieved_doc.to_json(&schema),
r#"{"event":["comment"],"occurred_at":["2022-06-22T13:00:00.22Z"]}"#

5
examples/faceted_search_with_tweaked_score.rs
View File

@@ -51,7 +51,7 @@ fn main() -> tantivy::Result<()> {
let reader = index.reader()?;
let searcher = reader.searcher();
{
let facets = vec![
let facets = [
Facet::from("/ingredient/egg"),
Facet::from("/ingredient/oil"),
Facet::from("/ingredient/garlic"),
@@ -94,9 +94,8 @@ fn main() -> tantivy::Result<()> {
.doc::<TantivyDocument>(*doc_id)
.unwrap()
.get_first(title)
.and_then(|v| v.as_str())
.and_then(|v| v.as_str().map(|el| el.to_string()))
.unwrap()
.to_owned()
})
.collect();
assert_eq!(titles, vec!["Fried egg", "Egg rolls"]);

4
examples/index_from_multiple_threads.rs
View File

@@ -61,7 +61,7 @@ fn main() -> tantivy::Result<()> {
debris of the winter’s flooding; and sycamores with mottled, white, recumbent \
limbs and branches that arch over the pool"
))?;
println!("add doc {} from thread 1 - opstamp {}", i, opstamp);
println!("add doc {i} from thread 1 - opstamp {opstamp}");
thread::sleep(Duration::from_millis(20));
}
Result::<(), TantivyError>::Ok(())
@@ -82,7 +82,7 @@ fn main() -> tantivy::Result<()> {
body => "Some great book description..."
))?
};
println!("add doc {} from thread 2 - opstamp {}", i, opstamp);
println!("add doc {i} from thread 2 - opstamp {opstamp}");
thread::sleep(Duration::from_millis(10));
}
Result::<(), TantivyError>::Ok(())

9
examples/integer_range_search.rs
View File

@@ -1,3 +1,5 @@
use std::ops::Bound;
// # Searching a range on an indexed int field.
//
// Below is an example of creating an indexed integer field in your schema
@@ -5,7 +7,7 @@
use tantivy::collector::Count;
use tantivy::query::RangeQuery;
use tantivy::schema::{Schema, INDEXED};
use tantivy::{doc, Index, IndexWriter, Result};
use tantivy::{doc, Index, IndexWriter, Result, Term};
fn main() -> Result<()> {
// For the sake of simplicity, this schema will only have 1 field
@@ -27,7 +29,10 @@ fn main() -> Result<()> {
reader.reload()?;
let searcher = reader.searcher();
// The end is excluded i.e. here we are searching up to 1969
let docs_in_the_sixties = RangeQuery::new_u64("year".to_string(), 1960..1970);
let docs_in_the_sixties = RangeQuery::new(
Bound::Included(Term::from_field_u64(year_field, 1960)),
Bound::Excluded(Term::from_field_u64(year_field, 1970)),
);
// Uses a Count collector to sum the total number of docs in the range
let num_60s_books = searcher.search(&docs_in_the_sixties, &Count)?;
assert_eq!(num_60s_books, 10);

5
examples/iterating_docs_and_positions.rs
View File

@@ -7,10 +7,11 @@
// the list of documents containing a term, getting
// its term frequency, and accessing its positions.
use tantivy::postings::Postings;
// ---
// Importing tantivy...
use tantivy::schema::*;
use tantivy::{doc, DocSet, Index, IndexWriter, Postings, TERMINATED};
use tantivy::{doc, DocSet, Index, IndexWriter, TERMINATED};
fn main() -> tantivy::Result<()> {
// We first create a schema for the sake of the
@@ -27,7 +28,7 @@ fn main() -> tantivy::Result<()> {
let mut index_writer: IndexWriter = index.writer_with_num_threads(1, 50_000_000)?;
index_writer.add_document(doc!(title => "The Old Man and the Sea"))?;
index_writer.add_document(doc!(title => "Of Mice and Men"))?;
index_writer.add_document(doc!(title => "The modern Promotheus"))?;
index_writer.add_document(doc!(title => "The modern Prometheus"))?;
index_writer.commit()?;
let reader = index.reader()?;

3
examples/warmer.rs
View File

@@ -3,10 +3,11 @@ use std::collections::{HashMap, HashSet};
use std::sync::{Arc, RwLock, Weak};
use tantivy::collector::TopDocs;
use tantivy::index::SegmentId;
use tantivy::query::QueryParser;
use tantivy::schema::{Schema, FAST, TEXT};
use tantivy::{
doc, DocAddress, DocId, Index, IndexWriter, Opstamp, Searcher, SearcherGeneration, SegmentId,
doc, DocAddress, DocId, Index, IndexWriter, Opstamp, Searcher, SearcherGeneration,
SegmentReader, Warmer,
};

7
ownedbytes/src/lib.rs
View File

@@ -151,7 +151,7 @@ impl fmt::Debug for OwnedBytes {
fn fmt(&self, f: &mut fmt::Formatter) -> fmt::Result {
// We truncate the bytes in order to make sure the debug string
// is not too long.
let bytes_truncated: &[u8] = if self.len() > 8 {
let bytes_truncated: &[u8] = if self.len() > 10 {
&self.as_slice()[..10]
} else {
self.as_slice()
@@ -252,6 +252,11 @@ mod tests {
format!("{short_bytes:?}"),
"OwnedBytes([97, 98, 99, 100], len=4)"
);
let medium_bytes = OwnedBytes::new(b"abcdefghi".as_ref());
assert_eq!(
format!("{medium_bytes:?}"),
"OwnedBytes([97, 98, 99, 100, 101, 102, 103, 104, 105], len=9)"
);
let long_bytes = OwnedBytes::new(b"abcdefghijklmnopq".as_ref());
assert_eq!(
format!("{long_bytes:?}"),

2
query-grammar/src/infallible.rs
View File

@@ -109,6 +109,8 @@ where F: nom::Parser<I, (O, ErrorList), Infallible> {
move |input: I| match f.parse(input) {
Ok((input, (output, _err))) => Ok((input, output)),
Err(Err::Incomplete(needed)) => Err(Err::Incomplete(needed)),
// old versions don't understand this is uninhabited and need the empty match to help,
// newer versions warn because this arm is unreachable (which it is indeed).
Err(Err::Error(val)) | Err(Err::Failure(val)) => match val {},
}
}

6
query-grammar/src/occur.rs
View File

@@ -6,12 +6,12 @@ use std::fmt::Write;
#[derive(Debug, Clone, Hash, Copy, Eq, PartialEq)]
pub enum Occur {
/// For a given document to be considered for scoring,
/// at least one of the terms with the Should or the Must
/// at least one of the queries with the Should or the Must
/// Occur constraint must be within the document.
Should,
/// Document without the term are excluded from the search.
/// Document without the queries are excluded from the search.
Must,
/// Document that contain the term are excluded from the
/// Document that contain the query are excluded from the
/// search.
MustNot,
}

265
query-grammar/src/query_grammar.rs
View File

@@ -1,3 +1,4 @@
use std::borrow::Cow;
use std::iter::once;
use nom::branch::alt;
@@ -5,7 +6,7 @@ use nom::bytes::complete::tag;
use nom::character::complete::{
anychar, char, digit1, multispace0, multispace1, none_of, one_of, satisfy, u32,
};
use nom::combinator::{eof, map, map_res, opt, peek, recognize, value, verify};
use nom::combinator::{eof, map, map_res, not, opt, peek, recognize, value, verify};
use nom::error::{Error, ErrorKind};
use nom::multi::{many0, many1, separated_list0};
use nom::sequence::{delimited, preceded, separated_pair, terminated, tuple};
@@ -19,7 +20,7 @@ use crate::Occur;
// Note: '-' char is only forbidden at the beginning of a field name, would be clearer to add it to
// special characters.
const SPECIAL_CHARS: &[char] = &[
'+', '^', '`', ':', '{', '}', '"', '[', ']', '(', ')', '!', '\\', '*', ' ',
'+', '^', '`', ':', '{', '}', '"', '\'', '[', ']', '(', ')', '!', '\\', ' ',
];
/// consume a field name followed by colon. Return the field name with escape sequence
@@ -41,36 +42,92 @@ fn field_name(inp: &str) -> IResult<&str, String> {
)(inp)
}
const ESCAPE_IN_WORD: &[char] = &['^', '`', ':', '{', '}', '"', '\'', '[', ']', '(', ')', '\\'];
fn interpret_escape(source: &str) -> String {
let mut res = String::with_capacity(source.len());
let mut in_escape = false;
let require_escape = |c: char| c.is_whitespace() || ESCAPE_IN_WORD.contains(&c) || c == '-';
for c in source.chars() {
if in_escape {
if !require_escape(c) {
// we re-add the escape sequence
res.push('\\');
}
res.push(c);
in_escape = false;
} else if c == '\\' {
in_escape = true;
} else {
res.push(c);
}
}
res
}
/// Consume a word outside of any context.
// TODO should support escape sequences
fn word(inp: &str) -> IResult<&str, &str> {
fn word(inp: &str) -> IResult<&str, Cow<str>> {
map_res(
recognize(tuple((
satisfy(|c| {
!c.is_whitespace()
&& !['-', '^', '`', ':', '{', '}', '"', '[', ']', '(', ')'].contains(&c)
}),
many0(satisfy(|c: char| {
!c.is_whitespace() && ![':', '^', '{', '}', '"', '[', ']', '(', ')'].contains(&c)
})),
alt((
preceded(char('\\'), anychar),
satisfy(|c| !c.is_whitespace() && !ESCAPE_IN_WORD.contains(&c) && c != '-'),
)),
many0(alt((
preceded(char('\\'), anychar),
satisfy(|c: char| !c.is_whitespace() && !ESCAPE_IN_WORD.contains(&c)),
))),
))),
|s| match s {
"OR" | "AND" | "NOT" | "IN" => Err(Error::new(inp, ErrorKind::Tag)),
_ => Ok(s),
s if s.contains('\\') => Ok(Cow::Owned(interpret_escape(s))),
s => Ok(Cow::Borrowed(s)),
},
)(inp)
}
fn word_infallible(delimiter: &str) -> impl Fn(&str) -> JResult<&str, Option<&str>> + '_ {
|inp| {
opt_i_err(
preceded(
multispace0,
recognize(many1(satisfy(|c| {
!c.is_whitespace() && !delimiter.contains(c)
}))),
fn word_infallible(
delimiter: &str,
emit_error: bool,
) -> impl Fn(&str) -> JResult<&str, Option<Cow<str>>> + '_ {
// emit error is set when receiving an unescaped `:` should emit an error
move |inp| {
map(
opt_i_err(
preceded(
multispace0,
recognize(many1(alt((
preceded(char::<&str, _>('\\'), anychar),
satisfy(|c| !c.is_whitespace() && !delimiter.contains(c)),
)))),
),
"expected word",
),
"expected word",
|(opt_s, mut errors)| match opt_s {
Some(s) => {
if emit_error
&& (s
.as_bytes()
.windows(2)
.any(|window| window[0] != b'\\' && window[1] == b':')
|| s.starts_with(':'))
{
errors.push(LenientErrorInternal {
pos: inp.len(),
message: "parsed possible invalid field as term".to_string(),
});
}
if s.contains('\\') {
(Some(Cow::Owned(interpret_escape(s))), errors)
} else {
(Some(Cow::Borrowed(s)), errors)
}
}
None => (None, errors),
},
)(inp)
}
}
@@ -159,7 +216,7 @@ fn simple_term_infallible(
(value((), char('\'')), simple_quotes),
),
// numbers are parsed with words in this case, as we allow string starting with a -
map(word_infallible(delimiter), |(text, errors)| {
map(word_infallible(delimiter, true), |(text, errors)| {
(text.map(|text| (Delimiter::None, text.to_string())), errors)
}),
)(inp)
@@ -218,27 +275,14 @@ fn term_or_phrase_infallible(inp: &str) -> JResult<&str, Option<UserInputLeaf>>
}
fn term_group(inp: &str) -> IResult<&str, UserInputAst> {
let occur_symbol = alt((
value(Occur::MustNot, char('-')),
value(Occur::Must, char('+')),
));
map(
tuple((
terminated(field_name, multispace0),
delimited(
tuple((char('('), multispace0)),
separated_list0(multispace1, tuple((opt(occur_symbol), term_or_phrase))),
char(')'),
),
delimited(tuple((char('('), multispace0)), ast, char(')')),
)),
|(field_name, terms)| {
UserInputAst::Clause(
terms
.into_iter()
.map(|(occur, leaf)| (occur, leaf.set_field(Some(field_name.clone())).into()))
.collect(),
)
|(field_name, mut ast)| {
ast.set_default_field(field_name);
ast
},
)(inp)
}
@@ -258,46 +302,18 @@ fn term_group_precond(inp: &str) -> IResult<&str, (), ()> {
}
fn term_group_infallible(inp: &str) -> JResult<&str, UserInputAst> {
let (mut inp, (field_name, _, _, _)) =
let (inp, (field_name, _, _, _)) =
tuple((field_name, multispace0, char('('), multispace0))(inp).expect("precondition failed");
let mut terms = Vec::new();
let mut errs = Vec::new();
let mut first_round = true;
loop {
let mut space_error = if first_round {
first_round = false;
Vec::new()
} else {
let (rest, (_, err)) = space1_infallible(inp)?;
inp = rest;
err
};
if inp.is_empty() {
errs.push(LenientErrorInternal {
pos: inp.len(),
message: "missing )".to_string(),
});
break Ok((inp, (UserInputAst::Clause(terms), errs)));
}
if let Some(inp) = inp.strip_prefix(')') {
break Ok((inp, (UserInputAst::Clause(terms), errs)));
}
// only append missing space error if we did not reach the end of group
errs.append(&mut space_error);
// here we do the assumption term_or_phrase_infallible always consume something if the
// first byte is not `)` or ' '. If it did not, we would end up looping.
let (rest, ((occur, leaf), mut err)) =
tuple_infallible((occur_symbol, term_or_phrase_infallible))(inp)?;
errs.append(&mut err);
if let Some(leaf) = leaf {
terms.push((occur, leaf.set_field(Some(field_name.clone())).into()));
}
inp = rest;
}
let res = delimited_infallible(
nothing,
map(ast_infallible, |(mut ast, errors)| {
ast.set_default_field(field_name.to_string());
(ast, errors)
}),
opt_i_err(char(')'), "expected ')'"),
)(inp);
res
}
fn exists(inp: &str) -> IResult<&str, UserInputLeaf> {
@@ -363,15 +379,6 @@ fn literal_no_group_infallible(inp: &str) -> JResult<&str, Option<UserInputAst>>
|((field_name, _, leaf), mut errors)| {
(
leaf.map(|leaf| {
if matches!(&leaf, UserInputLeaf::Literal(literal)
if literal.phrase.contains(':') && literal.delimiter == Delimiter::None)
&& field_name.is_none()
{
errors.push(LenientErrorInternal {
pos: inp.len(),
message: "parsed possible invalid field as term".to_string(),
});
}
if matches!(&leaf, UserInputLeaf::Literal(literal)
if literal.phrase == "NOT" && literal.delimiter == Delimiter::None)
&& field_name.is_none()
@@ -490,20 +497,20 @@ fn range_infallible(inp: &str) -> JResult<&str, UserInputLeaf> {
tuple_infallible((
opt_i(anychar),
space0_infallible,
word_infallible("]}"),
word_infallible("]}", false),
space1_infallible,
opt_i_err(
terminated(tag("TO"), alt((value((), multispace1), value((), eof)))),
"missing keyword TO",
),
word_infallible("]}"),
word_infallible("]}", false),
opt_i_err(one_of("]}"), "missing range delimiter"),
)),
|(
(lower_bound_kind, _multispace0, lower, _multispace1, to, upper, upper_bound_kind),
errs,
)| {
let lower_bound = match (lower_bound_kind, lower) {
let lower_bound = match (lower_bound_kind, lower.as_deref()) {
(_, Some("*")) => UserInputBound::Unbounded,
(_, None) => UserInputBound::Unbounded,
// if it is some, TO was actually the bound (i.e. [TO TO something])
@@ -512,7 +519,7 @@ fn range_infallible(inp: &str) -> JResult<&str, UserInputLeaf> {
(Some('{'), Some(bound)) => UserInputBound::Exclusive(bound.to_string()),
_ => unreachable!("precondition failed, range did not start with [ or {{"),
};
let upper_bound = match (upper_bound_kind, upper) {
let upper_bound = match (upper_bound_kind, upper.as_deref()) {
(_, Some("*")) => UserInputBound::Unbounded,
(_, None) => UserInputBound::Unbounded,
(Some(']'), Some(bound)) => UserInputBound::Inclusive(bound.to_string()),
@@ -529,7 +536,7 @@ fn range_infallible(inp: &str) -> JResult<&str, UserInputLeaf> {
(
(
value((), tag(">=")),
map(word_infallible(""), |(bound, err)| {
map(word_infallible("", false), |(bound, err)| {
(
(
bound
@@ -543,7 +550,7 @@ fn range_infallible(inp: &str) -> JResult<&str, UserInputLeaf> {
),
(
value((), tag("<=")),
map(word_infallible(""), |(bound, err)| {
map(word_infallible("", false), |(bound, err)| {
(
(
UserInputBound::Unbounded,
@@ -557,7 +564,7 @@ fn range_infallible(inp: &str) -> JResult<&str, UserInputLeaf> {
),
(
value((), tag(">")),
map(word_infallible(""), |(bound, err)| {
map(word_infallible("", false), |(bound, err)| {
(
(
bound
@@ -571,7 +578,7 @@ fn range_infallible(inp: &str) -> JResult<&str, UserInputLeaf> {
),
(
value((), tag("<")),
map(word_infallible(""), |(bound, err)| {
map(word_infallible("", false), |(bound, err)| {
(
(
UserInputBound::Unbounded,
@@ -672,7 +679,10 @@ fn negate(expr: UserInputAst) -> UserInputAst {
fn leaf(inp: &str) -> IResult<&str, UserInputAst> {
alt((
delimited(char('('), ast, char(')')),
map(char('*'), |_| UserInputAst::from(UserInputLeaf::All)),
preceded(
peek(not(tag("*:"))),
map(char('*'), |_| UserInputAst::from(UserInputLeaf::All)),
),
map(preceded(tuple((tag("NOT"), multispace1)), leaf), negate),
literal,
))(inp)
@@ -693,7 +703,13 @@ fn leaf_infallible(inp: &str) -> JResult<&str, Option<UserInputAst>> {
),
),
(
value((), char('*')),
value(
(),
preceded(
peek(not(tag("*:"))), // Fail if `*:` is detected
char('*'), // Match standalone `*`
),
),
map(nothing, |_| {
(Some(UserInputAst::from(UserInputLeaf::All)), Vec::new())
}),
@@ -760,7 +776,7 @@ fn occur_leaf(inp: &str) -> IResult<&str, (Option<Occur>, UserInputAst)> {
tuple((fallible(occur_symbol), boosted_leaf))(inp)
}
#[allow(clippy::type_complexity)]
#[expect(clippy::type_complexity)]
fn operand_occur_leaf_infallible(
inp: &str,
) -> JResult<&str, (Option<BinaryOperand>, Option<Occur>, Option<UserInputAst>)> {
@@ -826,7 +842,7 @@ fn aggregate_infallible_expressions(
if early_operand {
err.push(LenientErrorInternal {
pos: 0,
message: "Found unexpeted boolean operator before term".to_string(),
message: "Found unexpected boolean operator before term".to_string(),
});
}
@@ -849,7 +865,7 @@ fn aggregate_infallible_expressions(
_ => Some(Occur::Should),
};
if occur == &Some(Occur::MustNot) && default_op == Some(Occur::Should) {
// if occur is MustNot *and* operation is OR, we synthetize a ShouldNot
// if occur is MustNot *and* operation is OR, we synthesize a ShouldNot
clauses.push(vec![(
Some(Occur::Should),
ast.clone().unary(Occur::MustNot),
@@ -865,7 +881,7 @@ fn aggregate_infallible_expressions(
None => None,
};
if occur == &Some(Occur::MustNot) && default_op == Some(Occur::Should) {
// if occur is MustNot *and* operation is OR, we synthetize a ShouldNot
// if occur is MustNot *and* operation is OR, we synthesize a ShouldNot
clauses.push(vec![(
Some(Occur::Should),
ast.clone().unary(Occur::MustNot),
@@ -890,7 +906,7 @@ fn aggregate_infallible_expressions(
}
Some(BinaryOperand::Or) => {
if last_occur == Some(Occur::MustNot) {
// if occur is MustNot *and* operation is OR, we synthetize a ShouldNot
// if occur is MustNot *and* operation is OR, we synthesize a ShouldNot
clauses.push(vec![(Some(Occur::Should), last_ast.unary(Occur::MustNot))]);
} else {
clauses.push(vec![(last_occur.or(Some(Occur::Should)), last_ast)]);
@@ -1050,7 +1066,7 @@ mod test {
valid_parse("1", 1.0, "");
valid_parse("0.234234 aaa", 0.234234f64, " aaa");
error_parse(".3332");
// TODO trinity-1686a: I disagree that it should fail, I think it should succeeed,
// TODO trinity-1686a: I disagree that it should fail, I think it should succeed,
// consuming only "1", and leave "." for the next thing (which will likely fail then)
// error_parse("1.");
error_parse("-1.");
@@ -1198,6 +1214,12 @@ mod test {
test_parse_query_to_ast_helper("weight: <= 70", "\"weight\":{\"*\" TO \"70\"]");
test_parse_query_to_ast_helper("weight: <= 70.5", "\"weight\":{\"*\" TO \"70.5\"]");
test_parse_query_to_ast_helper(">a", "{\"a\" TO \"*\"}");
test_parse_query_to_ast_helper(">=a", "[\"a\" TO \"*\"}");
test_parse_query_to_ast_helper("<a", "{\"*\" TO \"a\"}");
test_parse_query_to_ast_helper("<=a", "{\"*\" TO \"a\"]");
test_parse_query_to_ast_helper("<=bsd", "{\"*\" TO \"bsd\"]");
}
#[test]
@@ -1209,6 +1231,7 @@ mod test {
#[test]
fn test_field_name() {
assert_eq!(super::field_name("*:a"), Ok(("a", "*".to_string())));
assert_eq!(
super::field_name(".my.field.name:a"),
Ok(("a", ".my.field.name".to_string()))
@@ -1454,7 +1477,7 @@ mod test {
}
#[test]
fn test_parse_query_to_triming_spaces() {
fn test_parse_query_to_trimming_spaces() {
test_parse_query_to_ast_helper(" abc", "abc");
test_parse_query_to_ast_helper("abc ", "abc");
test_parse_query_to_ast_helper("( a OR abc)", "(?a ?abc)");
@@ -1468,12 +1491,27 @@ mod test {
#[test]
fn test_parse_query_term_group() {
test_parse_query_to_ast_helper(r#"field:(abc)"#, r#"(*"field":abc)"#);
test_parse_query_to_ast_helper(r#"field:(abc)"#, r#""field":abc"#);
test_parse_query_to_ast_helper(r#"field:(+a -"b c")"#, r#"(+"field":a -"field":"b c")"#);
test_parse_query_to_ast_helper(r#"field:(a AND "b c")"#, r#"(+"field":a +"field":"b c")"#);
test_parse_query_to_ast_helper(r#"field:(a OR "b c")"#, r#"(?"field":a ?"field":"b c")"#);
test_parse_query_to_ast_helper(
r#"field:(a OR (b AND c))"#,
r#"(?"field":a ?(+"field":b +"field":c))"#,
);
test_parse_query_to_ast_helper(
r#"field:(a [b TO c])"#,
r#"(*"field":a *"field":["b" TO "c"])"#,
);
test_is_parse_err(r#"field:(+a -"b c""#, r#"(+"field":a -"field":"b c")"#);
}
#[test]
fn field_re_specification() {
test_parse_query_to_ast_helper(r#"field:(abc AND b:cde)"#, r#"(+"field":abc +"b":cde)"#);
}
#[test]
fn test_parse_query_single_term() {
test_parse_query_to_ast_helper("abc", "abc");
@@ -1499,6 +1537,11 @@ mod test {
test_parse_query_to_ast_helper("abc:toto", "\"abc\":toto");
}
#[test]
fn all_field_star() {
test_parse_query_to_ast_helper("*:toto", "\"*\":toto");
}
#[test]
fn test_phrase_with_field() {
test_parse_query_to_ast_helper("abc:\"happy tax payer\"", "\"abc\":\"happy tax payer\"");
@@ -1621,5 +1664,21 @@ mod test {
r#"myfield:'hello\"happy\'tax'"#,
r#""myfield":'hello"happy'tax'"#,
);
// we don't process escape sequence for chars which don't require it
test_parse_query_to_ast_helper(r#"abc\*"#, r#"abc\*"#);
}
#[test]
fn test_queries_with_colons() {
test_parse_query_to_ast_helper(r#""abc:def""#, r#""abc:def""#);
test_parse_query_to_ast_helper(r#"'abc:def'"#, r#"'abc:def'"#);
test_parse_query_to_ast_helper(r#"abc\:def"#, r#"abc:def"#);
test_parse_query_to_ast_helper(r#""abc\:def""#, r#""abc:def""#);
test_parse_query_to_ast_helper(r#"'abc\:def'"#, r#"'abc:def'"#);
}
#[test]
fn test_invalid_field() {
test_is_parse_err(r#"!bc:def"#, "!bc:def");
}
}

32
query-grammar/src/user_input_ast.rs
View File

@@ -44,6 +44,26 @@ impl UserInputLeaf {
},
}
}
pub(crate) fn set_default_field(&mut self, default_field: String) {
match self {
UserInputLeaf::Literal(ref mut literal) if literal.field_name.is_none() => {
literal.field_name = Some(default_field)
}
UserInputLeaf::All => {
*self = UserInputLeaf::Exists {
field: default_field,
}
}
UserInputLeaf::Range { ref mut field, .. } if field.is_none() => {
*field = Some(default_field)
}
UserInputLeaf::Set { ref mut field, .. } if field.is_none() => {
*field = Some(default_field)
}
_ => (), // field was already set, do nothing
}
}
}
impl Debug for UserInputLeaf {
@@ -205,6 +225,16 @@ impl UserInputAst {
pub fn or(asts: Vec<UserInputAst>) -> UserInputAst {
UserInputAst::compose(Occur::Should, asts)
}
pub(crate) fn set_default_field(&mut self, field: String) {
match self {
UserInputAst::Clause(clauses) => clauses
.iter_mut()
.for_each(|(_, ast)| ast.set_default_field(field.clone())),
UserInputAst::Leaf(leaf) => leaf.set_default_field(field),
UserInputAst::Boost(ref mut ast, _) => ast.set_default_field(field),
}
}
}
impl From<UserInputLiteral> for UserInputLeaf {
@@ -237,7 +267,7 @@ impl fmt::Debug for UserInputAst {
match *self {
UserInputAst::Clause(ref subqueries) => {
if subqueries.is_empty() {
// TODO this will break ast reserialization, is writing "( )" enought?
// TODO this will break ast reserialization, is writing "( )" enough?
write!(formatter, "<emptyclause>")?;
} else {
write!(formatter, "(")?;

585
src/aggregation/agg_bench.rs
View File

@@ -1,585 +0,0 @@
#[cfg(all(test, feature = "unstable"))]
mod bench {
use rand::prelude::SliceRandom;
use rand::rngs::StdRng;
use rand::{Rng, SeedableRng};
use rand_distr::Distribution;
use serde_json::json;
use test::{self, Bencher};
use crate::aggregation::agg_req::Aggregations;
use crate::aggregation::AggregationCollector;
use crate::query::{AllQuery, TermQuery};
use crate::schema::{IndexRecordOption, Schema, TextFieldIndexing, FAST, STRING};
use crate::{Index, Term};
#[derive(Clone, Copy, Hash, Default, Debug, PartialEq, Eq, PartialOrd, Ord)]
enum Cardinality {
/// All documents contain exactly one value.
/// `Full` is the default for auto-detecting the Cardinality, since it is the most strict.
#[default]
Full = 0,
/// All documents contain at most one value.
Optional = 1,
/// All documents may contain any number of values.
Multivalued = 2,
/// 1 / 20 documents has a value
Sparse = 3,
}
fn get_collector(agg_req: Aggregations) -> AggregationCollector {
AggregationCollector::from_aggs(agg_req, Default::default())
}
fn get_test_index_bench(cardinality: Cardinality) -> crate::Result<Index> {
let mut schema_builder = Schema::builder();
let text_fieldtype = crate::schema::TextOptions::default()
.set_indexing_options(
TextFieldIndexing::default().set_index_option(IndexRecordOption::WithFreqs),
)
.set_stored();
let text_field = schema_builder.add_text_field("text", text_fieldtype);
let json_field = schema_builder.add_json_field("json", FAST);
let text_field_many_terms = schema_builder.add_text_field("text_many_terms", STRING | FAST);
let text_field_few_terms = schema_builder.add_text_field("text_few_terms", STRING | FAST);
let score_fieldtype = crate::schema::NumericOptions::default().set_fast();
let score_field = schema_builder.add_u64_field("score", score_fieldtype.clone());
let score_field_f64 = schema_builder.add_f64_field("score_f64", score_fieldtype.clone());
let score_field_i64 = schema_builder.add_i64_field("score_i64", score_fieldtype);
let index = Index::create_from_tempdir(schema_builder.build())?;
let few_terms_data = ["INFO", "ERROR", "WARN", "DEBUG"];
let lg_norm = rand_distr::LogNormal::new(2.996f64, 0.979f64).unwrap();
let many_terms_data = (0..150_000)
.map(|num| format!("author{}", num))
.collect::<Vec<_>>();
{
let mut rng = StdRng::from_seed([1u8; 32]);
let mut index_writer = index.writer_with_num_threads(1, 200_000_000)?;
// To make the different test cases comparable we just change one doc to force the
// cardinality
if cardinality == Cardinality::Optional {
index_writer.add_document(doc!())?;
}
if cardinality == Cardinality::Multivalued {
index_writer.add_document(doc!(
json_field => json!({"mixed_type": 10.0}),
json_field => json!({"mixed_type": 10.0}),
text_field => "cool",
text_field => "cool",
text_field_many_terms => "cool",
text_field_many_terms => "cool",
text_field_few_terms => "cool",
text_field_few_terms => "cool",
score_field => 1u64,
score_field => 1u64,
score_field_f64 => lg_norm.sample(&mut rng),
score_field_f64 => lg_norm.sample(&mut rng),
score_field_i64 => 1i64,
score_field_i64 => 1i64,
))?;
}
let mut doc_with_value = 1_000_000;
if cardinality == Cardinality::Sparse {
doc_with_value /= 20;
}
let _val_max = 1_000_000.0;
for _ in 0..doc_with_value {
let val: f64 = rng.gen_range(0.0..1_000_000.0);
let json = if rng.gen_bool(0.1) {
// 10% are numeric values
json!({ "mixed_type": val })
} else {
json!({"mixed_type": many_terms_data.choose(&mut rng).unwrap().to_string()})
};
index_writer.add_document(doc!(
text_field => "cool",
json_field => json,
text_field_many_terms => many_terms_data.choose(&mut rng).unwrap().to_string(),
text_field_few_terms => few_terms_data.choose(&mut rng).unwrap().to_string(),
score_field => val as u64,
score_field_f64 => lg_norm.sample(&mut rng),
score_field_i64 => val as i64,
))?;
if cardinality == Cardinality::Sparse {
for _ in 0..20 {
index_writer.add_document(doc!(text_field => "cool"))?;
}
}
}
// writing the segment
index_writer.commit()?;
}
Ok(index)
}
use paste::paste;
#[macro_export]
macro_rules! bench_all_cardinalities {
( $x:ident ) => {
paste! {
#[bench]
fn $x(b: &mut Bencher) {
[<$x _card>](b, Cardinality::Full)
}
#[bench]
fn [<$x _opt>](b: &mut Bencher) {
[<$x _card>](b, Cardinality::Optional)
}
#[bench]
fn [<$x _multi>](b: &mut Bencher) {
[<$x _card>](b, Cardinality::Multivalued)
}
#[bench]
fn [<$x _sparse>](b: &mut Bencher) {
[<$x _card>](b, Cardinality::Sparse)
}
}
};
}
bench_all_cardinalities!(bench_aggregation_average_u64);
fn bench_aggregation_average_u64_card(b: &mut Bencher, cardinality: Cardinality) {
let index = get_test_index_bench(cardinality).unwrap();
let reader = index.reader().unwrap();
let text_field = reader.searcher().schema().get_field("text").unwrap();
b.iter(|| {
let term_query = TermQuery::new(
Term::from_field_text(text_field, "cool"),
IndexRecordOption::Basic,
);
let agg_req_1: Aggregations = serde_json::from_value(json!({
"average": { "avg": { "field": "score", } }
}))
.unwrap();
let collector = get_collector(agg_req_1);
let searcher = reader.searcher();
searcher.search(&term_query, &collector).unwrap()
});
}
bench_all_cardinalities!(bench_aggregation_stats_f64);
fn bench_aggregation_stats_f64_card(b: &mut Bencher, cardinality: Cardinality) {
let index = get_test_index_bench(cardinality).unwrap();
let reader = index.reader().unwrap();
let text_field = reader.searcher().schema().get_field("text").unwrap();
b.iter(|| {
let term_query = TermQuery::new(
Term::from_field_text(text_field, "cool"),
IndexRecordOption::Basic,
);
let agg_req_1: Aggregations = serde_json::from_value(json!({
"average_f64": { "stats": { "field": "score_f64", } }
}))
.unwrap();
let collector = get_collector(agg_req_1);
let searcher = reader.searcher();
searcher.search(&term_query, &collector).unwrap()
});
}
bench_all_cardinalities!(bench_aggregation_average_f64);
fn bench_aggregation_average_f64_card(b: &mut Bencher, cardinality: Cardinality) {
let index = get_test_index_bench(cardinality).unwrap();
let reader = index.reader().unwrap();
let text_field = reader.searcher().schema().get_field("text").unwrap();
b.iter(|| {
let term_query = TermQuery::new(
Term::from_field_text(text_field, "cool"),
IndexRecordOption::Basic,
);
let agg_req_1: Aggregations = serde_json::from_value(json!({
"average_f64": { "avg": { "field": "score_f64", } }
}))
.unwrap();
let collector = get_collector(agg_req_1);
let searcher = reader.searcher();
searcher.search(&term_query, &collector).unwrap()
});
}
bench_all_cardinalities!(bench_aggregation_percentiles_f64);
fn bench_aggregation_percentiles_f64_card(b: &mut Bencher, cardinality: Cardinality) {
let index = get_test_index_bench(cardinality).unwrap();
let reader = index.reader().unwrap();
b.iter(|| {
let agg_req_str = r#"
{
"mypercentiles": {
"percentiles": {
"field": "score_f64",
"percents": [ 95, 99, 99.9 ]
}
}
} "#;
let agg_req_1: Aggregations = serde_json::from_str(agg_req_str).unwrap();
let collector = get_collector(agg_req_1);
let searcher = reader.searcher();
searcher.search(&AllQuery, &collector).unwrap()
});
}
bench_all_cardinalities!(bench_aggregation_average_u64_and_f64);
fn bench_aggregation_average_u64_and_f64_card(b: &mut Bencher, cardinality: Cardinality) {
let index = get_test_index_bench(cardinality).unwrap();
let reader = index.reader().unwrap();
let text_field = reader.searcher().schema().get_field("text").unwrap();
b.iter(|| {
let term_query = TermQuery::new(
Term::from_field_text(text_field, "cool"),
IndexRecordOption::Basic,
);
let agg_req_1: Aggregations = serde_json::from_value(json!({
"average_f64": { "avg": { "field": "score_f64" } },
"average": { "avg": { "field": "score" } },
}))
.unwrap();
let collector = get_collector(agg_req_1);
let searcher = reader.searcher();
searcher.search(&term_query, &collector).unwrap()
});
}
bench_all_cardinalities!(bench_aggregation_terms_few);
fn bench_aggregation_terms_few_card(b: &mut Bencher, cardinality: Cardinality) {
let index = get_test_index_bench(cardinality).unwrap();
let reader = index.reader().unwrap();
b.iter(|| {
let agg_req: Aggregations = serde_json::from_value(json!({
"my_texts": { "terms": { "field": "text_few_terms" } },
}))
.unwrap();
let collector = get_collector(agg_req);
let searcher = reader.searcher();
searcher.search(&AllQuery, &collector).unwrap()
});
}
bench_all_cardinalities!(bench_aggregation_terms_many_with_top_hits_agg);
fn bench_aggregation_terms_many_with_top_hits_agg_card(
b: &mut Bencher,
cardinality: Cardinality,
) {
let index = get_test_index_bench(cardinality).unwrap();
let reader = index.reader().unwrap();
b.iter(|| {
let agg_req: Aggregations = serde_json::from_value(json!({
"my_texts": {
"terms": { "field": "text_many_terms" },
"aggs": {
"top_hits": { "top_hits":
{
"sort": [
{ "score": "desc" }
],
"size": 2,
"doc_value_fields": ["score_f64"]
}
}
}
},
}))
.unwrap();
let collector = get_collector(agg_req);
let searcher = reader.searcher();
searcher.search(&AllQuery, &collector).unwrap()
});
}
bench_all_cardinalities!(bench_aggregation_terms_many_with_sub_agg);
fn bench_aggregation_terms_many_with_sub_agg_card(b: &mut Bencher, cardinality: Cardinality) {
let index = get_test_index_bench(cardinality).unwrap();
let reader = index.reader().unwrap();
b.iter(|| {
let agg_req: Aggregations = serde_json::from_value(json!({
"my_texts": {
"terms": { "field": "text_many_terms" },
"aggs": {
"average_f64": { "avg": { "field": "score_f64" } }
}
},
}))
.unwrap();
let collector = get_collector(agg_req);
let searcher = reader.searcher();
searcher.search(&AllQuery, &collector).unwrap()
});
}
bench_all_cardinalities!(bench_aggregation_terms_many_json_mixed_type_with_sub_agg);
fn bench_aggregation_terms_many_json_mixed_type_with_sub_agg_card(
b: &mut Bencher,
cardinality: Cardinality,
) {
let index = get_test_index_bench(cardinality).unwrap();
let reader = index.reader().unwrap();
b.iter(|| {
let agg_req: Aggregations = serde_json::from_value(json!({
"my_texts": {
"terms": { "field": "json.mixed_type" },
"aggs": {
"average_f64": { "avg": { "field": "score_f64" } }
}
},
}))
.unwrap();
let collector = get_collector(agg_req);
let searcher = reader.searcher();
searcher.search(&AllQuery, &collector).unwrap()
});
}
bench_all_cardinalities!(bench_aggregation_terms_many2);
fn bench_aggregation_terms_many2_card(b: &mut Bencher, cardinality: Cardinality) {
let index = get_test_index_bench(cardinality).unwrap();
let reader = index.reader().unwrap();
b.iter(|| {
let agg_req: Aggregations = serde_json::from_value(json!({
"my_texts": { "terms": { "field": "text_many_terms" } },
}))
.unwrap();
let collector = get_collector(agg_req);
let searcher = reader.searcher();
searcher.search(&AllQuery, &collector).unwrap()
});
}
bench_all_cardinalities!(bench_aggregation_terms_many_order_by_term);
fn bench_aggregation_terms_many_order_by_term_card(b: &mut Bencher, cardinality: Cardinality) {
let index = get_test_index_bench(cardinality).unwrap();
let reader = index.reader().unwrap();
b.iter(|| {
let agg_req: Aggregations = serde_json::from_value(json!({
"my_texts": { "terms": { "field": "text_many_terms", "order": { "_key": "desc" } } },
}))
.unwrap();
let collector = get_collector(agg_req);
let searcher = reader.searcher();
searcher.search(&AllQuery, &collector).unwrap()
});
}
bench_all_cardinalities!(bench_aggregation_range_only);
fn bench_aggregation_range_only_card(b: &mut Bencher, cardinality: Cardinality) {
let index = get_test_index_bench(cardinality).unwrap();
let reader = index.reader().unwrap();
b.iter(|| {
let agg_req_1: Aggregations = serde_json::from_value(json!({
"range_f64": { "range": { "field": "score_f64", "ranges": [
{ "from": 3, "to": 7000 },
{ "from": 7000, "to": 20000 },
{ "from": 20000, "to": 30000 },
{ "from": 30000, "to": 40000 },
{ "from": 40000, "to": 50000 },
{ "from": 50000, "to": 60000 }
] } },
}))
.unwrap();
let collector = get_collector(agg_req_1);
let searcher = reader.searcher();
searcher.search(&AllQuery, &collector).unwrap()
});
}
bench_all_cardinalities!(bench_aggregation_range_with_avg);
fn bench_aggregation_range_with_avg_card(b: &mut Bencher, cardinality: Cardinality) {
let index = get_test_index_bench(cardinality).unwrap();
let reader = index.reader().unwrap();
b.iter(|| {
let agg_req_1: Aggregations = serde_json::from_value(json!({
"rangef64": {
"range": {
"field": "score_f64",
"ranges": [
{ "from": 3, "to": 7000 },
{ "from": 7000, "to": 20000 },
{ "from": 20000, "to": 30000 },
{ "from": 30000, "to": 40000 },
{ "from": 40000, "to": 50000 },
{ "from": 50000, "to": 60000 }
]
},
"aggs": {
"average_f64": { "avg": { "field": "score_f64" } }
}
},
}))
.unwrap();
let collector = get_collector(agg_req_1);
let searcher = reader.searcher();
searcher.search(&AllQuery, &collector).unwrap()
});
}
// hard bounds has a different algorithm, because it actually limits collection range
//
bench_all_cardinalities!(bench_aggregation_histogram_only_hard_bounds);
fn bench_aggregation_histogram_only_hard_bounds_card(
b: &mut Bencher,
cardinality: Cardinality,
) {
let index = get_test_index_bench(cardinality).unwrap();
let reader = index.reader().unwrap();
b.iter(|| {
let agg_req_1: Aggregations = serde_json::from_value(json!({
"rangef64": { "histogram": { "field": "score_f64", "interval": 100, "hard_bounds": { "min": 1000, "max": 300000 } } },
}))
.unwrap();
let collector = get_collector(agg_req_1);
let searcher = reader.searcher();
searcher.search(&AllQuery, &collector).unwrap()
});
}
bench_all_cardinalities!(bench_aggregation_histogram_with_avg);
fn bench_aggregation_histogram_with_avg_card(b: &mut Bencher, cardinality: Cardinality) {
let index = get_test_index_bench(cardinality).unwrap();
let reader = index.reader().unwrap();
b.iter(|| {
let agg_req_1: Aggregations = serde_json::from_value(json!({
"rangef64": {
"histogram": { "field": "score_f64", "interval": 100 },
"aggs": {
"average_f64": { "avg": { "field": "score_f64" } }
}
}
}))
.unwrap();
let collector = get_collector(agg_req_1);
let searcher = reader.searcher();
searcher.search(&AllQuery, &collector).unwrap()
});
}
bench_all_cardinalities!(bench_aggregation_histogram_only);
fn bench_aggregation_histogram_only_card(b: &mut Bencher, cardinality: Cardinality) {
let index = get_test_index_bench(cardinality).unwrap();
let reader = index.reader().unwrap();
b.iter(|| {
let agg_req_1: Aggregations = serde_json::from_value(json!({
"rangef64": {
"histogram": {
"field": "score_f64",
"interval": 100 // 1000 buckets
},
}
}))
.unwrap();
let collector = get_collector(agg_req_1);
let searcher = reader.searcher();
searcher.search(&AllQuery, &collector).unwrap()
});
}
bench_all_cardinalities!(bench_aggregation_avg_and_range_with_avg);
fn bench_aggregation_avg_and_range_with_avg_card(b: &mut Bencher, cardinality: Cardinality) {
let index = get_test_index_bench(cardinality).unwrap();
let reader = index.reader().unwrap();
let text_field = reader.searcher().schema().get_field("text").unwrap();
b.iter(|| {
let term_query = TermQuery::new(
Term::from_field_text(text_field, "cool"),
IndexRecordOption::Basic,
);
let agg_req_1: Aggregations = serde_json::from_value(json!({
"rangef64": {
"range": {
"field": "score_f64",
"ranges": [
{ "from": 3, "to": 7000 },
{ "from": 7000, "to": 20000 },
{ "from": 20000, "to": 60000 }
]
},
"aggs": {
"average_in_range": { "avg": { "field": "score" } }
}
},
"average": { "avg": { "field": "score" } }
}))
.unwrap();
let collector = get_collector(agg_req_1);
let searcher = reader.searcher();
searcher.search(&term_query, &collector).unwrap()
});
}
}

75
src/aggregation/agg_limits.rs
View File

@@ -21,7 +21,10 @@ impl<K, V, S> MemoryConsumption for HashMap<K, V, S> {
/// Aggregation memory limit after which the request fails. Defaults to DEFAULT_MEMORY_LIMIT
/// (500MB). The limit is shared by all SegmentCollectors
pub struct AggregationLimits {
///
/// The memory limit is also a guard, which tracks how much it allocated and releases it's memory
/// on the shared counter. Cloning will create a new guard.
pub struct AggregationLimitsGuard {
/// The counter which is shared between the aggregations for one request.
memory_consumption: Arc<AtomicU64>,
/// The memory_limit in bytes
@@ -29,28 +32,41 @@ pub struct AggregationLimits {
/// The maximum number of buckets _returned_
/// This is not counting intermediate buckets.
bucket_limit: u32,
/// Allocated memory with this guard.
allocated_with_the_guard: u64,
}
impl Clone for AggregationLimits {
impl Clone for AggregationLimitsGuard {
fn clone(&self) -> Self {
Self {
memory_consumption: Arc::clone(&self.memory_consumption),
memory_limit: self.memory_limit,
bucket_limit: self.bucket_limit,
allocated_with_the_guard: 0,
}
}
}
impl Default for AggregationLimits {
impl Drop for AggregationLimitsGuard {
/// Removes the memory consumed tracked by this _instance_ of AggregationLimits.
/// This is used to clear the segment specific memory consumption all at once.
fn drop(&mut self) {
self.memory_consumption
.fetch_sub(self.allocated_with_the_guard, Ordering::Relaxed);
}
}
impl Default for AggregationLimitsGuard {
fn default() -> Self {
Self {
memory_consumption: Default::default(),
memory_limit: DEFAULT_MEMORY_LIMIT.into(),
bucket_limit: DEFAULT_BUCKET_LIMIT,
allocated_with_the_guard: 0,
}
}
}
impl AggregationLimits {
impl AggregationLimitsGuard {
/// *memory_limit*
/// memory_limit is defined in bytes.
/// Aggregation fails when the estimated memory consumption of the aggregation is higher than
@@ -67,24 +83,16 @@ impl AggregationLimits {
memory_consumption: Default::default(),
memory_limit: memory_limit.unwrap_or(DEFAULT_MEMORY_LIMIT).into(),
bucket_limit: bucket_limit.unwrap_or(DEFAULT_BUCKET_LIMIT),
}
}
/// Create a new ResourceLimitGuard, that will release the memory when dropped.
pub fn new_guard(&self) -> ResourceLimitGuard {
ResourceLimitGuard {
// The counter which is shared between the aggregations for one request.
memory_consumption: Arc::clone(&self.memory_consumption),
// The memory_limit in bytes
memory_limit: self.memory_limit,
allocated_with_the_guard: 0,
}
}
pub(crate) fn add_memory_consumed(&self, num_bytes: u64) -> crate::Result<()> {
self.memory_consumption
.fetch_add(num_bytes, Ordering::Relaxed);
validate_memory_consumption(&self.memory_consumption, self.memory_limit)?;
pub(crate) fn add_memory_consumed(&mut self, add_num_bytes: u64) -> crate::Result<()> {
let prev_value = self
.memory_consumption
.fetch_add(add_num_bytes, Ordering::Relaxed);
self.allocated_with_the_guard += add_num_bytes;
validate_memory_consumption(prev_value + add_num_bytes, self.memory_limit)?;
Ok(())
}
@@ -94,11 +102,11 @@ impl AggregationLimits {
}
fn validate_memory_consumption(
memory_consumption: &AtomicU64,
memory_consumption: u64,
memory_limit: ByteCount,
) -> Result<(), AggregationError> {
// Load the estimated memory consumed by the aggregations
let memory_consumed: ByteCount = memory_consumption.load(Ordering::Relaxed).into();
let memory_consumed: ByteCount = memory_consumption.into();
if memory_consumed > memory_limit {
return Err(AggregationError::MemoryExceeded {
limit: memory_limit,
@@ -108,33 +116,6 @@ fn validate_memory_consumption(
Ok(())
}
pub struct ResourceLimitGuard {
/// The counter which is shared between the aggregations for one request.
memory_consumption: Arc<AtomicU64>,
/// The memory_limit in bytes
memory_limit: ByteCount,
/// Allocated memory with this guard.
allocated_with_the_guard: u64,
}
impl ResourceLimitGuard {
pub(crate) fn add_memory_consumed(&self, num_bytes: u64) -> crate::Result<()> {
self.memory_consumption
.fetch_add(num_bytes, Ordering::Relaxed);
validate_memory_consumption(&self.memory_consumption, self.memory_limit)?;
Ok(())
}
}
impl Drop for ResourceLimitGuard {
/// Removes the memory consumed tracked by this _instance_ of AggregationLimits.
/// This is used to clear the segment specific memory consumption all at once.
fn drop(&mut self) {
self.memory_consumption
.fetch_sub(self.allocated_with_the_guard, Ordering::Relaxed);
}
}
#[cfg(test)]
mod tests {
use crate::aggregation::tests::exec_request_with_query;

23
src/aggregation/agg_req.rs
View File

@@ -34,8 +34,9 @@ use super::bucket::{
DateHistogramAggregationReq, HistogramAggregation, RangeAggregation, TermsAggregation,
};
use super::metric::{
AverageAggregation, CountAggregation, MaxAggregation, MinAggregation,
PercentilesAggregationReq, StatsAggregation, SumAggregation, TopHitsAggregation,
AverageAggregation, CardinalityAggregationReq, CountAggregation, ExtendedStatsAggregation,
MaxAggregation, MinAggregation, PercentilesAggregationReq, StatsAggregation, SumAggregation,
TopHitsAggregationReq,
};
/// The top-level aggregation request structure, which contains [`Aggregation`] and their user
@@ -146,6 +147,11 @@ pub enum AggregationVariants {
/// extracted values.
#[serde(rename = "stats")]
Stats(StatsAggregation),
/// Computes a collection of estended statistics (`min`, `max`, `sum`, `count`, `avg`,
/// `sum_of_squares`, `variance`, `variance_sampling`, `std_deviation`,
/// `std_deviation_sampling`) over the extracted values.
#[serde(rename = "extended_stats")]
ExtendedStats(ExtendedStatsAggregation),
/// Computes the sum of the extracted values.
#[serde(rename = "sum")]
Sum(SumAggregation),
@@ -154,7 +160,10 @@ pub enum AggregationVariants {
Percentiles(PercentilesAggregationReq),
/// Finds the top k values matching some order
#[serde(rename = "top_hits")]
TopHits(TopHitsAggregation),
TopHits(TopHitsAggregationReq),
/// Computes an estimate of the number of unique values
#[serde(rename = "cardinality")]
Cardinality(CardinalityAggregationReq),
}
impl AggregationVariants {
@@ -170,9 +179,11 @@ impl AggregationVariants {
AggregationVariants::Max(max) => vec![max.field_name()],
AggregationVariants::Min(min) => vec![min.field_name()],
AggregationVariants::Stats(stats) => vec![stats.field_name()],
AggregationVariants::ExtendedStats(extended_stats) => vec![extended_stats.field_name()],
AggregationVariants::Sum(sum) => vec![sum.field_name()],
AggregationVariants::Percentiles(per) => vec![per.field_name()],
AggregationVariants::TopHits(top_hits) => top_hits.field_names(),
AggregationVariants::Cardinality(per) => vec![per.field_name()],
}
}
@@ -197,6 +208,12 @@ impl AggregationVariants {
_ => None,
}
}
pub(crate) fn as_top_hits(&self) -> Option<&TopHitsAggregationReq> {
match &self {
AggregationVariants::TopHits(top_hits) => Some(top_hits),
_ => None,
}
}
pub(crate) fn as_percentile(&self) -> Option<&PercentilesAggregationReq> {
match &self {

76
src/aggregation/agg_req_with_accessor.rs
View File

@@ -5,19 +5,19 @@ use std::io;
use columnar::{Column, ColumnBlockAccessor, ColumnType, DynamicColumn, StrColumn};
use super::agg_limits::ResourceLimitGuard;
use super::agg_req::{Aggregation, AggregationVariants, Aggregations};
use super::bucket::{
DateHistogramAggregationReq, HistogramAggregation, RangeAggregation, TermsAggregation,
};
use super::metric::{
AverageAggregation, CountAggregation, MaxAggregation, MinAggregation, StatsAggregation,
SumAggregation,
AverageAggregation, CardinalityAggregationReq, CountAggregation, ExtendedStatsAggregation,
MaxAggregation, MinAggregation, StatsAggregation, SumAggregation,
};
use super::segment_agg_result::AggregationLimits;
use super::segment_agg_result::AggregationLimitsGuard;
use super::VecWithNames;
use crate::aggregation::{f64_to_fastfield_u64, Key};
use crate::{SegmentOrdinal, SegmentReader};
use crate::index::SegmentReader;
use crate::SegmentOrdinal;
#[derive(Default)]
pub(crate) struct AggregationsWithAccessor {
@@ -45,7 +45,7 @@ pub struct AggregationWithAccessor {
pub(crate) str_dict_column: Option<StrColumn>,
pub(crate) field_type: ColumnType,
pub(crate) sub_aggregation: AggregationsWithAccessor,
pub(crate) limits: ResourceLimitGuard,
pub(crate) limits: AggregationLimitsGuard,
pub(crate) column_block_accessor: ColumnBlockAccessor<u64>,
/// Used for missing term aggregation, which checks all columns for existence.
/// And also for `top_hits` aggregation, which may sort on multiple fields.
@@ -68,7 +68,7 @@ impl AggregationWithAccessor {
sub_aggregation: &Aggregations,
reader: &SegmentReader,
segment_ordinal: SegmentOrdinal,
limits: AggregationLimits,
limits: AggregationLimitsGuard,
) -> crate::Result<Vec<AggregationWithAccessor>> {
let mut agg = agg.clone();
@@ -90,7 +90,7 @@ impl AggregationWithAccessor {
&limits,
)?,
agg: agg.clone(),
limits: limits.new_guard(),
limits: limits.clone(),
missing_value_for_accessor: None,
str_dict_column: None,
column_block_accessor: Default::default(),
@@ -105,6 +105,7 @@ impl AggregationWithAccessor {
value_accessors: HashMap<String, Vec<DynamicColumn>>|
-> crate::Result<()> {
let (accessor, field_type) = accessors.first().expect("at least one accessor");
let limits = limits.clone();
let res = AggregationWithAccessor {
segment_ordinal,
// TODO: We should do away with the `accessor` field altogether
@@ -119,7 +120,7 @@ impl AggregationWithAccessor {
&limits,
)?,
agg: agg.clone(),
limits: limits.new_guard(),
limits,
missing_value_for_accessor: None,
str_dict_column: None,
column_block_accessor: Default::default(),
@@ -161,6 +162,11 @@ impl AggregationWithAccessor {
field: ref field_name,
ref missing,
..
})
| Cardinality(CardinalityAggregationReq {
field: ref field_name,
ref missing,
..
}) => {
let str_dict_column = reader.fast_fields().str(field_name)?;
let allowed_column_types = [
@@ -180,6 +186,8 @@ impl AggregationWithAccessor {
.map(|missing| match missing {
Key::Str(_) => ColumnType::Str,
Key::F64(_) => ColumnType::F64,
Key::I64(_) => ColumnType::I64,
Key::U64(_) => ColumnType::U64,
})
.unwrap_or(ColumnType::U64);
let column_and_types = get_all_ff_reader_or_empty(
@@ -226,14 +234,18 @@ impl AggregationWithAccessor {
missing.clone()
};
let missing_value_for_accessor = if let Some(missing) =
missing_value_term_agg.as_ref()
{
get_missing_val(column_type, missing, agg.agg.get_fast_field_names()[0])?
} else {
None
};
let missing_value_for_accessor =
if let Some(missing) = missing_value_term_agg.as_ref() {
get_missing_val_as_u64_lenient(
column_type,
missing,
agg.agg.get_fast_field_names()[0],
)?
} else {
None
};
let limits = limits.clone();
let agg = AggregationWithAccessor {
segment_ordinal,
missing_value_for_accessor,
@@ -249,7 +261,7 @@ impl AggregationWithAccessor {
)?,
agg: agg.clone(),
str_dict_column: str_dict_column.clone(),
limits: limits.new_guard(),
limits,
column_block_accessor: Default::default(),
};
res.push(agg);
@@ -275,6 +287,10 @@ impl AggregationWithAccessor {
field: ref field_name,
..
})
| ExtendedStats(ExtendedStatsAggregation {
field: ref field_name,
..
})
| Sum(SumAggregation {
field: ref field_name,
..
@@ -320,7 +336,14 @@ impl AggregationWithAccessor {
}
}
fn get_missing_val(
/// Get the missing value as internal u64 representation
///
/// For terms we use u64::MAX as sentinel value
/// For numerical data we convert the value into the representation
/// we would get from the fast field, when we open it as u64_lenient_for_type.
///
/// That way we can use it the same way as if it would come from the fastfield.
fn get_missing_val_as_u64_lenient(
column_type: ColumnType,
missing: &Key,
field_name: &str,
@@ -329,13 +352,22 @@ fn get_missing_val(
Key::Str(_) if column_type == ColumnType::Str => Some(u64::MAX),
// Allow fallback to number on text fields
Key::F64(_) if column_type == ColumnType::Str => Some(u64::MAX),
Key::U64(_) if column_type == ColumnType::Str => Some(u64::MAX),
Key::I64(_) if column_type == ColumnType::Str => Some(u64::MAX),
Key::F64(val) if column_type.numerical_type().is_some() => {
f64_to_fastfield_u64(*val, &column_type)
}
// NOTE: We may loose precision of the passed missing value by casting i64 and u64 to f64.
Key::I64(val) if column_type.numerical_type().is_some() => {
f64_to_fastfield_u64(*val as f64, &column_type)
}
Key::U64(val) if column_type.numerical_type().is_some() => {
f64_to_fastfield_u64(*val as f64, &column_type)
}
_ => {
return Err(crate::TantivyError::InvalidArgument(format!(
"Missing value {:?} for field {} is not supported for column type {:?}",
missing, field_name, column_type
"Missing value {missing:?} for field {field_name} is not supported for column \
type {column_type:?}"
)));
}
};
@@ -355,7 +387,7 @@ pub(crate) fn get_aggs_with_segment_accessor_and_validate(
aggs: &Aggregations,
reader: &SegmentReader,
segment_ordinal: SegmentOrdinal,
limits: &AggregationLimits,
limits: &AggregationLimitsGuard,
) -> crate::Result<AggregationsWithAccessor> {
let mut aggss = Vec::new();
for (key, agg) in aggs.iter() {
@@ -402,7 +434,7 @@ fn get_dynamic_columns(
.iter()
.map(|h| h.open())
.collect::<io::Result<_>>()?;
assert!(!ff_fields.is_empty(), "field {} not found", field_name);
assert!(!ff_fields.is_empty(), "field {field_name} not found");
Ok(cols)
}

13
src/aggregation/agg_result.rs
View File

@@ -1,4 +1,5 @@
//! Contains the final aggregation tree.
//!
//! This tree can be converted via the `into()` method from `IntermediateAggregationResults`.
//! This conversion computes the final result. For example: The intermediate result contains
//! intermediate average results, which is the sum and the number of values. The actual average is
@@ -8,7 +9,9 @@ use rustc_hash::FxHashMap;
use serde::{Deserialize, Serialize};
use super::bucket::GetDocCount;
use super::metric::{PercentilesMetricResult, SingleMetricResult, Stats, TopHitsMetricResult};
use super::metric::{
ExtendedStats, PercentilesMetricResult, SingleMetricResult, Stats, TopHitsMetricResult,
};
use super::{AggregationError, Key};
use crate::TantivyError;
@@ -88,12 +91,16 @@ pub enum MetricResult {
Min(SingleMetricResult),
/// Stats metric result.
Stats(Stats),
/// ExtendedStats metric result.
ExtendedStats(Box<ExtendedStats>),
/// Sum metric result.
Sum(SingleMetricResult),
/// Percentiles metric result.
Percentiles(PercentilesMetricResult),
/// Top hits metric result
TopHits(TopHitsMetricResult),
/// Cardinality metric result
Cardinality(SingleMetricResult),
}
impl MetricResult {
@@ -104,6 +111,7 @@ impl MetricResult {
MetricResult::Max(max) => Ok(max.value),
MetricResult::Min(min) => Ok(min.value),
MetricResult::Stats(stats) => stats.get_value(agg_property),
MetricResult::ExtendedStats(extended_stats) => extended_stats.get_value(agg_property),
MetricResult::Sum(sum) => Ok(sum.value),
MetricResult::Percentiles(_) => Err(TantivyError::AggregationError(
AggregationError::InvalidRequest("percentiles can't be used to order".to_string()),
@@ -111,6 +119,7 @@ impl MetricResult {
MetricResult::TopHits(_) => Err(TantivyError::AggregationError(
AggregationError::InvalidRequest("top_hits can't be used to order".to_string()),
)),
MetricResult::Cardinality(card) => Ok(card.value),
}
}
}
@@ -179,7 +188,7 @@ pub enum BucketEntries<T> {
}
impl<T> BucketEntries<T> {
fn iter<'a>(&'a self) -> Box<dyn Iterator<Item = &T> + 'a> {
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()),

84
src/aggregation/agg_tests.rs
View File

@@ -5,7 +5,7 @@ use crate::aggregation::agg_result::AggregationResults;
use crate::aggregation::buf_collector::DOC_BLOCK_SIZE;
use crate::aggregation::collector::AggregationCollector;
use crate::aggregation::intermediate_agg_result::IntermediateAggregationResults;
use crate::aggregation::segment_agg_result::AggregationLimits;
use crate::aggregation::segment_agg_result::AggregationLimitsGuard;
use crate::aggregation::tests::{get_test_index_2_segments, get_test_index_from_values_and_terms};
use crate::aggregation::DistributedAggregationCollector;
use crate::query::{AllQuery, TermQuery};
@@ -110,6 +110,16 @@ fn test_aggregation_flushing(
}
}
}
},
"cardinality_string_id":{
"cardinality": {
"field": "string_id"
}
},
"cardinality_score":{
"cardinality": {
"field": "score"
}
}
});
@@ -120,7 +130,7 @@ fn test_aggregation_flushing(
let agg_res: AggregationResults = if use_distributed_collector {
let collector = DistributedAggregationCollector::from_aggs(
agg_req.clone(),
AggregationLimits::default(),
AggregationLimitsGuard::default(),
);
let searcher = reader.searcher();
@@ -136,7 +146,7 @@ fn test_aggregation_flushing(
.expect("Post deserialization failed");
intermediate_agg_result
.into_final_result(agg_req, &Default::default())
.into_final_result(agg_req, Default::default())
.unwrap()
} else {
let collector = get_collector(agg_req);
@@ -212,6 +222,9 @@ fn test_aggregation_flushing(
)
);
assert_eq!(res["cardinality_string_id"]["value"], 2.0);
assert_eq!(res["cardinality_score"]["value"], 80.0);
Ok(())
}
@@ -447,7 +460,7 @@ fn test_aggregation_level2(
let searcher = reader.searcher();
let res = searcher.search(&term_query, &collector).unwrap();
res.into_final_result(agg_req.clone(), &Default::default())
res.into_final_result(agg_req.clone(), Default::default())
.unwrap()
} else {
let collector = get_collector(agg_req.clone());
@@ -857,7 +870,7 @@ fn test_aggregation_on_json_object_mixed_types() {
.add_document(doc!(json => json!({"mixed_type": "blue", "mixed_price": 5.0})))
.unwrap();
index_writer.commit().unwrap();
// => Segment with all boolen
// => Segment with all boolean
index_writer
.add_document(doc!(json => json!({"mixed_type": true, "mixed_price": "no_price"})))
.unwrap();
@@ -926,11 +939,11 @@ fn test_aggregation_on_json_object_mixed_types() {
},
"termagg": {
"buckets": [
{ "doc_count": 1, "key": 10.0, "min_price": { "value": 10.0 } },
{ "doc_count": 1, "key": 10, "min_price": { "value": 10.0 } },
{ "doc_count": 3, "key": "blue", "min_price": { "value": 5.0 } },
{ "doc_count": 2, "key": "red", "min_price": { "value": 1.0 } },
{ "doc_count": 1, "key": -20.5, "min_price": { "value": -20.5 } },
{ "doc_count": 2, "key": 1.0, "key_as_string": "true", "min_price": { "value": null } },
{ "doc_count": 2, "key": 1, "key_as_string": "true", "min_price": { "value": null } },
],
"sum_other_doc_count": 0
}
@@ -938,3 +951,60 @@ fn test_aggregation_on_json_object_mixed_types() {
)
);
}
#[test]
fn test_aggregation_on_json_object_mixed_numerical_segments() {
let mut schema_builder = Schema::builder();
let json = schema_builder.add_json_field("json", FAST);
let schema = schema_builder.build();
let index = Index::create_in_ram(schema);
let mut index_writer: IndexWriter = index.writer_for_tests().unwrap();
// => Segment with all values f64 numeric
index_writer
.add_document(doc!(json => json!({"mixed_price": 10.5})))
.unwrap();
// Gets converted to f64!
index_writer
.add_document(doc!(json => json!({"mixed_price": 10})))
.unwrap();
index_writer.commit().unwrap();
// => Segment with all values i64 numeric
index_writer
.add_document(doc!(json => json!({"mixed_price": 10})))
.unwrap();
index_writer.commit().unwrap();
index_writer.commit().unwrap();
// All bucket types
let agg_req_str = r#"
{
"termagg": {
"terms": {
"field": "json.mixed_price"
}
}
} "#;
let agg: Aggregations = serde_json::from_str(agg_req_str).unwrap();
let aggregation_collector = get_collector(agg);
let reader = index.reader().unwrap();
let searcher = reader.searcher();
let aggregation_results = searcher.search(&AllQuery, &aggregation_collector).unwrap();
let aggregation_res_json = serde_json::to_value(aggregation_results).unwrap();
use pretty_assertions::assert_eq;
assert_eq!(
&aggregation_res_json,
&serde_json::json!({
"termagg": {
"buckets": [
{ "doc_count": 2, "key": 10},
{ "doc_count": 1, "key": 10.5},
],
"doc_count_error_upper_bound": 0,
"sum_other_doc_count": 0
}
}
)
);
}

2
src/aggregation/bucket/histogram/date_histogram.rs
View File

@@ -244,7 +244,7 @@ fn parse_into_milliseconds(input: &str) -> Result<i64, AggregationError> {
}
#[cfg(test)]
pub mod tests {
pub(crate) mod tests {
use pretty_assertions::assert_eq;
use super::*;

14
src/aggregation/bucket/histogram/histogram.rs
View File

@@ -331,9 +331,11 @@ impl SegmentAggregationCollector for SegmentHistogramCollector {
}
let mem_delta = self.get_memory_consumption() - mem_pre;
bucket_agg_accessor
.limits
.add_memory_consumed(mem_delta as u64)?;
if mem_delta > 0 {
bucket_agg_accessor
.limits
.add_memory_consumed(mem_delta as u64)?;
}
Ok(())
}
@@ -436,7 +438,7 @@ fn intermediate_buckets_to_final_buckets_fill_gaps(
buckets: Vec<IntermediateHistogramBucketEntry>,
histogram_req: &HistogramAggregation,
sub_aggregation: &Aggregations,
limits: &AggregationLimits,
limits: &mut AggregationLimitsGuard,
) -> crate::Result<Vec<BucketEntry>> {
// Generate the full list of buckets without gaps.
//
@@ -494,7 +496,7 @@ pub(crate) fn intermediate_histogram_buckets_to_final_buckets(
is_date_agg: bool,
histogram_req: &HistogramAggregation,
sub_aggregation: &Aggregations,
limits: &AggregationLimits,
limits: &mut AggregationLimitsGuard,
) -> crate::Result<Vec<BucketEntry>> {
// Normalization is column type dependent.
// The request used in the the call to final is not yet be normalized.
@@ -748,7 +750,7 @@ mod tests {
agg_req,
&index,
None,
AggregationLimits::new(Some(5_000), None),
AggregationLimitsGuard::new(Some(5_000), None),
)
.unwrap_err();
assert!(res.to_string().starts_with(

87
src/aggregation/bucket/mod.rs
View File

@@ -28,6 +28,7 @@ mod term_agg;
mod term_missing_agg;
use std::collections::HashMap;
use std::fmt;
pub use histogram::*;
pub use range::*;
@@ -72,12 +73,12 @@ impl From<&str> for OrderTarget {
}
}
impl ToString for OrderTarget {
fn to_string(&self) -> String {
impl fmt::Display for OrderTarget {
fn fmt(&self, f: &mut fmt::Formatter) -> fmt::Result {
match self {
OrderTarget::Key => "_key".to_string(),
OrderTarget::Count => "_count".to_string(),
OrderTarget::SubAggregation(agg) => agg.to_string(),
OrderTarget::Key => f.write_str("_key"),
OrderTarget::Count => f.write_str("_count"),
OrderTarget::SubAggregation(agg) => agg.fmt(f),
}
}
}
@@ -111,18 +112,64 @@ impl Serialize for CustomOrder {
impl<'de> Deserialize<'de> for CustomOrder {
fn deserialize<D>(deserializer: D) -> Result<CustomOrder, D::Error>
where D: Deserializer<'de> {
HashMap::<String, Order>::deserialize(deserializer).and_then(|map| {
if let Some((key, value)) = map.into_iter().next() {
let value = serde_json::Value::deserialize(deserializer)?;
let return_err = |message, val: serde_json::Value| {
de::Error::custom(format!(
"{}, but got {}",
message,
serde_json::to_string(&val).unwrap()
))
};
match value {
serde_json::Value::Object(map) => {
if map.len() != 1 {
return Err(return_err(
"expected exactly one key-value pair in the order map",
map.into(),
));
}
let (key, value) = map.into_iter().next().unwrap();
let order = serde_json::from_value(value).map_err(de::Error::custom)?;
Ok(CustomOrder {
target: key.as_str().into(),
order: value,
order,
})
} else {
Err(de::Error::custom(
"unexpected empty map in order".to_string(),
))
}
})
serde_json::Value::Array(arr) => {
if arr.is_empty() {
return Err(return_err("unexpected empty array in order", arr.into()));
}
if arr.len() != 1 {
return Err(return_err(
"only one sort order supported currently",
arr.into(),
));
}
let entry = arr.into_iter().next().unwrap();
let map = entry
.as_object()
.ok_or_else(|| return_err("expected object as sort order", entry.clone()))?;
let (key, value) = map.into_iter().next().ok_or_else(|| {
return_err(
"expected exactly one key-value pair in the order map",
entry.clone(),
)
})?;
let order = serde_json::from_value(value.clone()).map_err(de::Error::custom)?;
Ok(CustomOrder {
target: key.as_str().into(),
order,
})
}
_ => Err(return_err(
"unexpected type, expected an object or array",
value,
)),
}
}
}
@@ -137,11 +184,23 @@ fn custom_order_serde_test() {
assert_eq!(order_str, "{\"_key\":\"desc\"}");
let order_deser = serde_json::from_str(&order_str).unwrap();
assert_eq!(order, order_deser);
let order_deser: CustomOrder = serde_json::from_str("[{\"_key\":\"desc\"}]").unwrap();
assert_eq!(order, order_deser);
let order_deser: serde_json::Result<CustomOrder> = serde_json::from_str("{}");
assert!(order_deser.is_err());
let order_deser: serde_json::Result<CustomOrder> = serde_json::from_str("[]");
assert!(order_deser.is_err());
assert!(order_deser
.unwrap_err()
.to_string()
.contains("unexpected empty array in order"));
let order_deser: serde_json::Result<CustomOrder> =
serde_json::from_str(r#"[{"_key":"desc"},{"_key":"desc"}]"#);
assert_eq!(
order_deser.unwrap_err().to_string(),
r#"only one sort order supported currently, but got [{"_key":"desc"},{"_key":"desc"}]"#
);
}

6
src/aggregation/bucket/range.rs
View File

@@ -4,7 +4,6 @@ use std::ops::Range;
use rustc_hash::FxHashMap;
use serde::{Deserialize, Serialize};
use crate::aggregation::agg_limits::ResourceLimitGuard;
use crate::aggregation::agg_req_with_accessor::AggregationsWithAccessor;
use crate::aggregation::intermediate_agg_result::{
IntermediateAggregationResult, IntermediateAggregationResults, IntermediateBucketResult,
@@ -17,6 +16,7 @@ use crate::aggregation::*;
use crate::TantivyError;
/// Provide user-defined buckets to aggregate on.
///
/// Two special buckets will automatically be created to cover the whole range of values.
/// The provided buckets have to be continuous.
/// During the aggregation, the values extracted from the fast_field `field` will be checked
@@ -270,7 +270,7 @@ impl SegmentRangeCollector {
pub(crate) fn from_req_and_validate(
req: &RangeAggregation,
sub_aggregation: &mut AggregationsWithAccessor,
limits: &ResourceLimitGuard,
limits: &mut AggregationLimitsGuard,
field_type: ColumnType,
accessor_idx: usize,
) -> crate::Result<Self> {
@@ -471,7 +471,7 @@ mod tests {
SegmentRangeCollector::from_req_and_validate(
&req,
&mut Default::default(),
&AggregationLimits::default().new_guard(),
&mut AggregationLimitsGuard::default(),
field_type,
0,
)

197
src/aggregation/bucket/term_agg.rs
View File

@@ -1,9 +1,10 @@
use std::fmt::Debug;
use std::io;
use std::net::Ipv6Addr;
use columnar::column_values::CompactSpaceU64Accessor;
use columnar::{
BytesColumn, ColumnType, MonotonicallyMappableToU128, MonotonicallyMappableToU64, StrColumn,
ColumnType, Dictionary, MonotonicallyMappableToU128, MonotonicallyMappableToU64, NumericalValue,
};
use rustc_hash::FxHashMap;
use serde::{Deserialize, Serialize};
@@ -20,11 +21,11 @@ use crate::aggregation::intermediate_agg_result::{
use crate::aggregation::segment_agg_result::{
build_segment_agg_collector, SegmentAggregationCollector,
};
use crate::aggregation::{f64_from_fastfield_u64, format_date, Key};
use crate::aggregation::{format_date, Key};
use crate::error::DataCorruption;
use crate::TantivyError;
/// Creates a bucket for every unique term and counts the number of occurences.
/// Creates a bucket for every unique term and counts the number of occurrences.
/// Note that doc_count in the response buckets equals term count here.
///
/// If the text is untokenized and single value, that means one term per document and therefore it
@@ -157,7 +158,7 @@ pub struct TermsAggregation {
/// when loading the text.
/// Special Case 1:
/// If we have multiple columns on one field, we need to have a union on the indices on both
/// columns, to find docids without a value. That requires a special missing aggreggation.
/// columns, to find docids without a value. That requires a special missing aggregation.
/// Special Case 2: if the key is of type text and the column is numerical, we also need to use
/// the special missing aggregation, since there is no mechanism in the numerical column to
/// add text.
@@ -324,9 +325,11 @@ impl SegmentAggregationCollector for SegmentTermCollector {
}
let mem_delta = self.get_memory_consumption() - mem_pre;
bucket_agg_accessor
.limits
.add_memory_consumed(mem_delta as u64)?;
if mem_delta > 0 {
bucket_agg_accessor
.limits
.add_memory_consumed(mem_delta as u64)?;
}
Ok(())
}
@@ -355,14 +358,13 @@ impl SegmentTermCollector {
) -> crate::Result<Self> {
if field_type == ColumnType::Bytes {
return Err(TantivyError::InvalidArgument(format!(
"terms aggregation is not supported for column type {:?}",
field_type
"terms aggregation is not supported for column type {field_type:?}"
)));
}
let term_buckets = TermBuckets::default();
if let Some(custom_order) = req.order.as_ref() {
// Validate sub aggregtion exists
// Validate sub aggregation exists
if let OrderTarget::SubAggregation(sub_agg_name) = &custom_order.target {
let (agg_name, _agg_property) = get_agg_name_and_property(sub_agg_name);
@@ -465,49 +467,72 @@ impl SegmentTermCollector {
};
if self.column_type == ColumnType::Str {
let fallback_dict = Dictionary::empty();
let term_dict = agg_with_accessor
.str_dict_column
.as_ref()
.cloned()
.unwrap_or_else(|| {
StrColumn::wrap(BytesColumn::empty(agg_with_accessor.accessor.num_docs()))
});
let mut buffer = String::new();
for (term_id, doc_count) in entries {
let intermediate_entry = into_intermediate_bucket_entry(term_id, doc_count)?;
// Special case for missing key
if term_id == u64::MAX {
let missing_key = self
.req
.missing
.as_ref()
.expect("Found placeholder term_id but `missing` is None");
match missing_key {
Key::Str(missing) => {
buffer.clear();
buffer.push_str(missing);
dict.insert(
IntermediateKey::Str(buffer.to_string()),
intermediate_entry,
);
}
Key::F64(val) => {
buffer.push_str(&val.to_string());
dict.insert(IntermediateKey::F64(*val), intermediate_entry);
}
.map(|el| el.dictionary())
.unwrap_or_else(|| &fallback_dict);
let mut buffer = Vec::new();
// special case for missing key
if let Some(index) = entries.iter().position(|value| value.0 == u64::MAX) {
let entry = entries[index];
let intermediate_entry = into_intermediate_bucket_entry(entry.0, entry.1)?;
let missing_key = self
.req
.missing
.as_ref()
.expect("Found placeholder term_id but `missing` is None");
match missing_key {
Key::Str(missing) => {
buffer.clear();
buffer.extend_from_slice(missing.as_bytes());
dict.insert(
IntermediateKey::Str(
String::from_utf8(buffer.to_vec())
.expect("could not convert to String"),
),
intermediate_entry,
);
}
} else {
if !term_dict.ord_to_str(term_id, &mut buffer)? {
return Err(TantivyError::InternalError(format!(
"Couldn't find term_id {term_id} in dict"
)));
Key::F64(val) => {
dict.insert(IntermediateKey::F64(*val), intermediate_entry);
}
Key::U64(val) => {
dict.insert(IntermediateKey::U64(*val), intermediate_entry);
}
Key::I64(val) => {
dict.insert(IntermediateKey::I64(*val), intermediate_entry);
}
dict.insert(IntermediateKey::Str(buffer.to_string()), intermediate_entry);
}
entries.swap_remove(index);
}
// Sort by term ord
entries.sort_unstable_by_key(|bucket| bucket.0);
let mut idx = 0;
term_dict.sorted_ords_to_term_cb(
entries.iter().map(|(term_id, _)| *term_id),
|term| {
let entry = entries[idx];
let intermediate_entry = into_intermediate_bucket_entry(entry.0, entry.1)
.map_err(|err| io::Error::new(io::ErrorKind::Other, err))?;
dict.insert(
IntermediateKey::Str(
String::from_utf8(term.to_vec()).expect("could not convert to String"),
),
intermediate_entry,
);
idx += 1;
Ok(())
},
)?;
if self.req.min_doc_count == 0 {
// TODO: Handle rev streaming for descending sorting by keys
let mut stream = term_dict.dictionary().stream()?;
let mut stream = term_dict.stream()?;
let empty_sub_aggregation = IntermediateAggregationResults::empty_from_req(
agg_with_accessor.agg.sub_aggregation(),
);
@@ -566,8 +591,26 @@ impl SegmentTermCollector {
} else {
for (val, doc_count) in entries {
let intermediate_entry = into_intermediate_bucket_entry(val, doc_count)?;
let val = f64_from_fastfield_u64(val, &self.column_type);
dict.insert(IntermediateKey::F64(val), intermediate_entry);
if self.column_type == ColumnType::U64 {
dict.insert(IntermediateKey::U64(val), intermediate_entry);
} else if self.column_type == ColumnType::I64 {
dict.insert(IntermediateKey::I64(i64::from_u64(val)), intermediate_entry);
} else {
let val = f64::from_u64(val);
let val: NumericalValue = val.into();
match val.normalize() {
NumericalValue::U64(val) => {
dict.insert(IntermediateKey::U64(val), intermediate_entry);
}
NumericalValue::I64(val) => {
dict.insert(IntermediateKey::I64(val), intermediate_entry);
}
NumericalValue::F64(val) => {
dict.insert(IntermediateKey::F64(val), intermediate_entry);
}
}
};
}
};
@@ -626,7 +669,7 @@ mod tests {
exec_request, exec_request_with_query, exec_request_with_query_and_memory_limit,
get_test_index_from_terms, get_test_index_from_values_and_terms,
};
use crate::aggregation::AggregationLimits;
use crate::aggregation::AggregationLimitsGuard;
use crate::indexer::NoMergePolicy;
use crate::schema::{IntoIpv6Addr, Schema, FAST, STRING};
use crate::{Index, IndexWriter};
@@ -1189,8 +1232,8 @@ mod tests {
#[test]
fn terms_aggregation_min_doc_count_special_case() -> crate::Result<()> {
let terms_per_segment = vec![
vec!["terma", "terma", "termb", "termb", "termb", "termc"],
vec!["terma", "terma", "termb", "termc", "termc"],
vec!["terma", "terma", "termb", "termb", "termb"],
vec!["terma", "terma", "termb"],
];
let index = get_test_index_from_terms(false, &terms_per_segment)?;
@@ -1212,8 +1255,6 @@ mod tests {
assert_eq!(res["my_texts"]["buckets"][0]["doc_count"], 4);
assert_eq!(res["my_texts"]["buckets"][1]["key"], "termb");
assert_eq!(res["my_texts"]["buckets"][1]["doc_count"], 0);
assert_eq!(res["my_texts"]["buckets"][2]["key"], "termc");
assert_eq!(res["my_texts"]["buckets"][2]["doc_count"], 0);
assert_eq!(res["my_texts"]["sum_other_doc_count"], 0);
assert_eq!(res["my_texts"]["doc_count_error_upper_bound"], 0);
@@ -1381,7 +1422,7 @@ mod tests {
agg_req,
&index,
None,
AggregationLimits::new(Some(50_000), None),
AggregationLimitsGuard::new(Some(50_000), None),
)
.unwrap_err();
assert!(res
@@ -1642,7 +1683,7 @@ mod tests {
res["my_texts"]["buckets"][2]["key"],
serde_json::Value::Null
);
// text field with numner as missing fallback
// text field with number as missing fallback
assert_eq!(res["my_texts2"]["buckets"][0]["key"], "Hello Hello");
assert_eq!(res["my_texts2"]["buckets"][0]["doc_count"], 5);
assert_eq!(res["my_texts2"]["buckets"][1]["key"], 1337.0);
@@ -1702,6 +1743,54 @@ mod tests {
Ok(())
}
#[test]
fn terms_aggregation_u64_value() -> crate::Result<()> {
// Make sure that large u64 are not truncated
let mut schema_builder = Schema::builder();
let id_field = schema_builder.add_u64_field("id", FAST);
let index = Index::create_in_ram(schema_builder.build());
{
let mut index_writer = index.writer_with_num_threads(1, 20_000_000)?;
index_writer.set_merge_policy(Box::new(NoMergePolicy));
index_writer.add_document(doc!(
id_field => 9_223_372_036_854_775_807u64,
))?;
index_writer.add_document(doc!(
id_field => 1_769_070_189_829_214_202u64,
))?;
index_writer.add_document(doc!(
id_field => 1_769_070_189_829_214_202u64,
))?;
index_writer.commit()?;
}
let agg_req: Aggregations = serde_json::from_value(json!({
"my_ids": {
"terms": {
"field": "id"
},
}
}))
.unwrap();
let res = exec_request_with_query(agg_req, &index, None)?;
// id field
assert_eq!(
res["my_ids"]["buckets"][0]["key"],
1_769_070_189_829_214_202u64
);
assert_eq!(res["my_ids"]["buckets"][0]["doc_count"], 2);
assert_eq!(
res["my_ids"]["buckets"][1]["key"],
9_223_372_036_854_775_807u64
);
assert_eq!(res["my_ids"]["buckets"][1]["doc_count"], 1);
assert_eq!(res["my_ids"]["buckets"][2]["key"], serde_json::Value::Null);
Ok(())
}
#[test]
fn terms_aggregation_missing1() -> crate::Result<()> {
let mut schema_builder = Schema::builder();
@@ -1768,7 +1857,7 @@ mod tests {
res["my_texts"]["buckets"][2]["key"],
serde_json::Value::Null
);
// text field with numner as missing fallback
// text field with number as missing fallback
assert_eq!(res["my_texts2"]["buckets"][0]["key"], "Hello Hello");
assert_eq!(res["my_texts2"]["buckets"][0]["doc_count"], 4);
assert_eq!(res["my_texts2"]["buckets"][1]["key"], 1337.0);

1
src/aggregation/bucket/term_missing_agg.rs
View File

@@ -70,7 +70,6 @@ impl SegmentAggregationCollector for TermMissingAgg {
)?;
missing_entry.sub_aggregation = res;
}
entries.insert(missing.into(), missing_entry);
let bucket = IntermediateBucketResult::Terms {

17
src/aggregation/collector.rs
View File

@@ -4,11 +4,12 @@ use super::agg_result::AggregationResults;
use super::buf_collector::BufAggregationCollector;
use super::intermediate_agg_result::IntermediateAggregationResults;
use super::segment_agg_result::{
build_segment_agg_collector, AggregationLimits, SegmentAggregationCollector,
build_segment_agg_collector, AggregationLimitsGuard, SegmentAggregationCollector,
};
use crate::aggregation::agg_req_with_accessor::get_aggs_with_segment_accessor_and_validate;
use crate::collector::{Collector, SegmentCollector};
use crate::{DocId, SegmentOrdinal, SegmentReader, TantivyError};
use crate::index::SegmentReader;
use crate::{DocId, SegmentOrdinal, TantivyError};
/// The default max bucket count, before the aggregation fails.
pub const DEFAULT_BUCKET_LIMIT: u32 = 65000;
@@ -21,7 +22,7 @@ pub const DEFAULT_MEMORY_LIMIT: u64 = 500_000_000;
/// The collector collects all aggregations by the underlying aggregation request.
pub struct AggregationCollector {
agg: Aggregations,
limits: AggregationLimits,
limits: AggregationLimitsGuard,
}
impl AggregationCollector {
@@ -29,7 +30,7 @@ impl AggregationCollector {
///
/// Aggregation fails when the limits in `AggregationLimits` is exceeded. (memory limit and
/// bucket limit)
pub fn from_aggs(agg: Aggregations, limits: AggregationLimits) -> Self {
pub fn from_aggs(agg: Aggregations, limits: AggregationLimitsGuard) -> Self {
Self { agg, limits }
}
}
@@ -44,7 +45,7 @@ impl AggregationCollector {
/// into the final `AggregationResults` via the `into_final_result()` method.
pub struct DistributedAggregationCollector {
agg: Aggregations,
limits: AggregationLimits,
limits: AggregationLimitsGuard,
}
impl DistributedAggregationCollector {
@@ -52,7 +53,7 @@ impl DistributedAggregationCollector {
///
/// Aggregation fails when the limits in `AggregationLimits` is exceeded. (memory limit and
/// bucket limit)
pub fn from_aggs(agg: Aggregations, limits: AggregationLimits) -> Self {
pub fn from_aggs(agg: Aggregations, limits: AggregationLimitsGuard) -> Self {
Self { agg, limits }
}
}
@@ -114,7 +115,7 @@ impl Collector for AggregationCollector {
segment_fruits: Vec<<Self::Child as SegmentCollector>::Fruit>,
) -> crate::Result<Self::Fruit> {
let res = merge_fruits(segment_fruits)?;
res.into_final_result(self.agg.clone(), &self.limits)
res.into_final_result(self.agg.clone(), self.limits.clone())
}
}
@@ -146,7 +147,7 @@ impl AggregationSegmentCollector {
agg: &Aggregations,
reader: &SegmentReader,
segment_ordinal: SegmentOrdinal,
limits: &AggregationLimits,
limits: &AggregationLimitsGuard,
) -> crate::Result<Self> {
let mut aggs_with_accessor =
get_aggs_with_segment_accessor_and_validate(agg, reader, segment_ordinal, limits)?;

81
src/aggregation/intermediate_agg_result.rs
View File

@@ -19,13 +19,14 @@ use super::bucket::{
GetDocCount, Order, OrderTarget, RangeAggregation, TermsAggregation,
};
use super::metric::{
IntermediateAverage, IntermediateCount, IntermediateMax, IntermediateMin, IntermediateStats,
IntermediateSum, PercentilesCollector, TopHitsTopNComputer,
IntermediateAverage, IntermediateCount, IntermediateExtendedStats, IntermediateMax,
IntermediateMin, IntermediateStats, IntermediateSum, PercentilesCollector, TopHitsTopNComputer,
};
use super::segment_agg_result::AggregationLimits;
use super::segment_agg_result::AggregationLimitsGuard;
use super::{format_date, AggregationError, Key, SerializedKey};
use crate::aggregation::agg_result::{AggregationResults, BucketEntries, BucketEntry};
use crate::aggregation::bucket::TermsAggregationInternal;
use crate::aggregation::metric::CardinalityCollector;
use crate::TantivyError;
/// Contains the intermediate aggregation result, which is optimized to be merged with other
@@ -50,12 +51,18 @@ pub enum IntermediateKey {
Str(String),
/// `f64` key
F64(f64),
/// `i64` key
I64(i64),
/// `u64` key
U64(u64),
}
impl From<Key> for IntermediateKey {
fn from(value: Key) -> Self {
match value {
Key::Str(s) => Self::Str(s),
Key::F64(f) => Self::F64(f),
Key::U64(f) => Self::U64(f),
Key::I64(f) => Self::I64(f),
}
}
}
@@ -72,7 +79,9 @@ impl From<IntermediateKey> for Key {
}
}
IntermediateKey::F64(f) => Self::F64(f),
IntermediateKey::Bool(f) => Self::F64(f as u64 as f64),
IntermediateKey::Bool(f) => Self::U64(f as u64),
IntermediateKey::U64(f) => Self::U64(f),
IntermediateKey::I64(f) => Self::I64(f),
}
}
}
@@ -85,6 +94,8 @@ impl std::hash::Hash for IntermediateKey {
match self {
IntermediateKey::Str(text) => text.hash(state),
IntermediateKey::F64(val) => val.to_bits().hash(state),
IntermediateKey::U64(val) => val.hash(state),
IntermediateKey::I64(val) => val.hash(state),
IntermediateKey::Bool(val) => val.hash(state),
IntermediateKey::IpAddr(val) => val.hash(state),
}
@@ -111,9 +122,9 @@ impl IntermediateAggregationResults {
pub fn into_final_result(
self,
req: Aggregations,
limits: &AggregationLimits,
mut limits: AggregationLimitsGuard,
) -> crate::Result<AggregationResults> {
let res = self.into_final_result_internal(&req, limits)?;
let res = self.into_final_result_internal(&req, &mut limits)?;
let bucket_count = res.get_bucket_count() as u32;
if bucket_count > limits.get_bucket_limit() {
return Err(TantivyError::AggregationError(
@@ -130,7 +141,7 @@ impl IntermediateAggregationResults {
pub(crate) fn into_final_result_internal(
self,
req: &Aggregations,
limits: &AggregationLimits,
limits: &mut AggregationLimitsGuard,
) -> crate::Result<AggregationResults> {
let mut results: FxHashMap<String, AggregationResult> = FxHashMap::default();
for (key, agg_res) in self.aggs_res.into_iter() {
@@ -215,6 +226,9 @@ pub(crate) fn empty_from_req(req: &Aggregation) -> IntermediateAggregationResult
Stats(_) => IntermediateAggregationResult::Metric(IntermediateMetricResult::Stats(
IntermediateStats::default(),
)),
ExtendedStats(_) => IntermediateAggregationResult::Metric(
IntermediateMetricResult::ExtendedStats(IntermediateExtendedStats::default()),
),
Sum(_) => IntermediateAggregationResult::Metric(IntermediateMetricResult::Sum(
IntermediateSum::default(),
)),
@@ -222,7 +236,10 @@ pub(crate) fn empty_from_req(req: &Aggregation) -> IntermediateAggregationResult
IntermediateMetricResult::Percentiles(PercentilesCollector::default()),
),
TopHits(ref req) => IntermediateAggregationResult::Metric(
IntermediateMetricResult::TopHits(TopHitsTopNComputer::new(req.clone())),
IntermediateMetricResult::TopHits(TopHitsTopNComputer::new(req)),
),
Cardinality(_) => IntermediateAggregationResult::Metric(
IntermediateMetricResult::Cardinality(CardinalityCollector::default()),
),
}
}
@@ -240,7 +257,7 @@ impl IntermediateAggregationResult {
pub(crate) fn into_final_result(
self,
req: &Aggregation,
limits: &AggregationLimits,
limits: &mut AggregationLimitsGuard,
) -> crate::Result<AggregationResult> {
let res = match self {
IntermediateAggregationResult::Bucket(bucket) => {
@@ -282,10 +299,14 @@ pub enum IntermediateMetricResult {
Min(IntermediateMin),
/// Intermediate stats result.
Stats(IntermediateStats),
/// Intermediate stats result.
ExtendedStats(IntermediateExtendedStats),
/// Intermediate sum result.
Sum(IntermediateSum),
/// Intermediate top_hits result
TopHits(TopHitsTopNComputer),
/// Intermediate cardinality result
Cardinality(CardinalityCollector),
}
impl IntermediateMetricResult {
@@ -306,6 +327,9 @@ impl IntermediateMetricResult {
IntermediateMetricResult::Stats(intermediate_stats) => {
MetricResult::Stats(intermediate_stats.finalize())
}
IntermediateMetricResult::ExtendedStats(intermediate_stats) => {
MetricResult::ExtendedStats(intermediate_stats.finalize())
}
IntermediateMetricResult::Sum(intermediate_sum) => {
MetricResult::Sum(intermediate_sum.finalize().into())
}
@@ -316,6 +340,9 @@ impl IntermediateMetricResult {
IntermediateMetricResult::TopHits(top_hits) => {
MetricResult::TopHits(top_hits.into_final_result())
}
IntermediateMetricResult::Cardinality(cardinality) => {
MetricResult::Cardinality(cardinality.finalize().into())
}
}
}
@@ -346,6 +373,12 @@ impl IntermediateMetricResult {
) => {
stats_left.merge_fruits(stats_right);
}
(
IntermediateMetricResult::ExtendedStats(extended_stats_left),
IntermediateMetricResult::ExtendedStats(extended_stats_right),
) => {
extended_stats_left.merge_fruits(extended_stats_right);
}
(IntermediateMetricResult::Sum(sum_left), IntermediateMetricResult::Sum(sum_right)) => {
sum_left.merge_fruits(sum_right);
}
@@ -358,6 +391,12 @@ impl IntermediateMetricResult {
(IntermediateMetricResult::TopHits(left), IntermediateMetricResult::TopHits(right)) => {
left.merge_fruits(right)?;
}
(
IntermediateMetricResult::Cardinality(left),
IntermediateMetricResult::Cardinality(right),
) => {
left.merge_fruits(right)?;
}
_ => {
panic!("incompatible fruit types in tree or missing merge_fruits handler");
}
@@ -393,7 +432,7 @@ impl IntermediateBucketResult {
pub(crate) fn into_final_bucket_result(
self,
req: &Aggregation,
limits: &AggregationLimits,
limits: &mut AggregationLimitsGuard,
) -> crate::Result<BucketResult> {
match self {
IntermediateBucketResult::Range(range_res) => {
@@ -557,7 +596,7 @@ impl IntermediateTermBucketResult {
self,
req: &TermsAggregation,
sub_aggregation_req: &Aggregations,
limits: &AggregationLimits,
limits: &mut AggregationLimitsGuard,
) -> crate::Result<BucketResult> {
let req = TermsAggregationInternal::from_req(req);
let mut buckets: Vec<BucketEntry> = self
@@ -684,7 +723,7 @@ impl IntermediateHistogramBucketEntry {
pub(crate) fn into_final_bucket_entry(
self,
req: &Aggregations,
limits: &AggregationLimits,
limits: &mut AggregationLimitsGuard,
) -> crate::Result<BucketEntry> {
Ok(BucketEntry {
key_as_string: None,
@@ -719,7 +758,7 @@ impl IntermediateRangeBucketEntry {
req: &Aggregations,
_range_req: &RangeAggregation,
column_type: Option<ColumnType>,
limits: &AggregationLimits,
limits: &mut AggregationLimitsGuard,
) -> crate::Result<RangeBucketEntry> {
let mut range_bucket_entry = RangeBucketEntry {
key: self.key.into(),
@@ -821,7 +860,7 @@ mod tests {
}
}
fn get_intermediat_tree_with_ranges(
fn get_intermediate_tree_with_ranges(
data: &[(String, u64, String, u64)],
) -> IntermediateAggregationResults {
let mut map = HashMap::new();
@@ -857,18 +896,18 @@ mod tests {
#[test]
fn test_merge_fruits_tree_1() {
let mut tree_left = get_intermediat_tree_with_ranges(&[
let mut tree_left = get_intermediate_tree_with_ranges(&[
("red".to_string(), 50, "1900".to_string(), 25),
("blue".to_string(), 30, "1900".to_string(), 30),
]);
let tree_right = get_intermediat_tree_with_ranges(&[
let tree_right = get_intermediate_tree_with_ranges(&[
("red".to_string(), 60, "1900".to_string(), 30),
("blue".to_string(), 25, "1900".to_string(), 50),
]);
tree_left.merge_fruits(tree_right).unwrap();
let tree_expected = get_intermediat_tree_with_ranges(&[
let tree_expected = get_intermediate_tree_with_ranges(&[
("red".to_string(), 110, "1900".to_string(), 55),
("blue".to_string(), 55, "1900".to_string(), 80),
]);
@@ -878,18 +917,18 @@ mod tests {
#[test]
fn test_merge_fruits_tree_2() {
let mut tree_left = get_intermediat_tree_with_ranges(&[
let mut tree_left = get_intermediate_tree_with_ranges(&[
("red".to_string(), 50, "1900".to_string(), 25),
("blue".to_string(), 30, "1900".to_string(), 30),
]);
let tree_right = get_intermediat_tree_with_ranges(&[
let tree_right = get_intermediate_tree_with_ranges(&[
("red".to_string(), 60, "1900".to_string(), 30),
("green".to_string(), 25, "1900".to_string(), 50),
]);
tree_left.merge_fruits(tree_right).unwrap();
let tree_expected = get_intermediat_tree_with_ranges(&[
let tree_expected = get_intermediate_tree_with_ranges(&[
("red".to_string(), 110, "1900".to_string(), 55),
("blue".to_string(), 30, "1900".to_string(), 30),
("green".to_string(), 25, "1900".to_string(), 50),
@@ -900,7 +939,7 @@ mod tests {
#[test]
fn test_merge_fruits_tree_empty() {
let mut tree_left = get_intermediat_tree_with_ranges(&[
let mut tree_left = get_intermediate_tree_with_ranges(&[
("red".to_string(), 50, "1900".to_string(), 25),
("blue".to_string(), 30, "1900".to_string(), 30),
]);

473
src/aggregation/metric/cardinality.rs Normal file
View File

@@ -0,0 +1,473 @@
use std::collections::hash_map::DefaultHasher;
use std::hash::{BuildHasher, Hasher};
use columnar::column_values::CompactSpaceU64Accessor;
use columnar::Dictionary;
use common::f64_to_u64;
use hyperloglogplus::{HyperLogLog, HyperLogLogPlus};
use rustc_hash::FxHashSet;
use serde::{Deserialize, Serialize};
use crate::aggregation::agg_req_with_accessor::{
AggregationWithAccessor, AggregationsWithAccessor,
};
use crate::aggregation::intermediate_agg_result::{
IntermediateAggregationResult, IntermediateAggregationResults, IntermediateMetricResult,
};
use crate::aggregation::segment_agg_result::SegmentAggregationCollector;
use crate::aggregation::*;
use crate::TantivyError;
#[derive(Clone, Debug, Serialize, Deserialize)]
struct BuildSaltedHasher {
salt: u8,
}
impl BuildHasher for BuildSaltedHasher {
type Hasher = DefaultHasher;
fn build_hasher(&self) -> Self::Hasher {
let mut hasher = DefaultHasher::new();
hasher.write_u8(self.salt);
hasher
}
}
/// # Cardinality
///
/// The cardinality aggregation allows for computing an estimate
/// of the number of different values in a data set based on the
/// HyperLogLog++ algorithm. This is particularly useful for understanding the
/// uniqueness of values in a large dataset where counting each unique value
/// individually would be computationally expensive.
///
/// For example, you might use a cardinality aggregation to estimate the number
/// of unique visitors to a website by aggregating on a field that contains
/// user IDs or session IDs.
///
/// To use the cardinality aggregation, you'll need to provide a field to
/// aggregate on. The following example demonstrates a request for the cardinality
/// of the "user_id" field:
///
/// ```JSON
/// {
/// "cardinality": {
/// "field": "user_id"
/// }
/// }
/// ```
///
/// This request will return an estimate of the number of unique values in the
/// "user_id" field.
///
/// ## Missing Values
///
/// The `missing` parameter defines how documents that are missing a value should be treated.
/// By default, documents without a value for the specified field are ignored. However, you can
/// specify a default value for these documents using the `missing` parameter. This can be useful
/// when you want to include documents with missing values in the aggregation.
///
/// For example, the following request treats documents with missing values in the "user_id"
/// field as if they had a value of "unknown":
///
/// ```JSON
/// {
/// "cardinality": {
/// "field": "user_id",
/// "missing": "unknown"
/// }
/// }
/// ```
///
/// # Estimation Accuracy
///
/// The cardinality aggregation provides an approximate count, which is usually
/// accurate within a small error range. This trade-off allows for efficient
/// computation even on very large datasets.
#[derive(Clone, Debug, PartialEq, Serialize, Deserialize)]
pub struct CardinalityAggregationReq {
/// The field name to compute the percentiles on.
pub field: String,
/// The missing parameter defines how documents that are missing a value should be treated.
/// By default they will be ignored but it is also possible to treat them as if they had a
/// value. Examples in JSON format:
/// { "field": "my_numbers", "missing": "10.0" }
#[serde(skip_serializing_if = "Option::is_none", default)]
pub missing: Option<Key>,
}
impl CardinalityAggregationReq {
/// Creates a new [`CardinalityAggregationReq`] instance from a field name.
pub fn from_field_name(field_name: String) -> Self {
Self {
field: field_name,
missing: None,
}
}
/// Returns the field name the aggregation is computed on.
pub fn field_name(&self) -> &str {
&self.field
}
}
#[derive(Clone, Debug, PartialEq)]
pub(crate) struct SegmentCardinalityCollector {
cardinality: CardinalityCollector,
entries: FxHashSet<u64>,
column_type: ColumnType,
accessor_idx: usize,
missing: Option<Key>,
}
impl SegmentCardinalityCollector {
pub fn from_req(column_type: ColumnType, accessor_idx: usize, missing: &Option<Key>) -> Self {
Self {
cardinality: CardinalityCollector::new(column_type as u8),
entries: Default::default(),
column_type,
accessor_idx,
missing: missing.clone(),
}
}
fn fetch_block_with_field(
&mut self,
docs: &[crate::DocId],
agg_accessor: &mut AggregationWithAccessor,
) {
if let Some(missing) = agg_accessor.missing_value_for_accessor {
agg_accessor.column_block_accessor.fetch_block_with_missing(
docs,
&agg_accessor.accessor,
missing,
);
} else {
agg_accessor
.column_block_accessor
.fetch_block(docs, &agg_accessor.accessor);
}
}
fn into_intermediate_metric_result(
mut self,
agg_with_accessor: &AggregationWithAccessor,
) -> crate::Result<IntermediateMetricResult> {
if self.column_type == ColumnType::Str {
let fallback_dict = Dictionary::empty();
let dict = agg_with_accessor
.str_dict_column
.as_ref()
.map(|el| el.dictionary())
.unwrap_or_else(|| &fallback_dict);
let mut has_missing = false;
// TODO: replace FxHashSet with something that allows iterating in order
// (e.g. sparse bitvec)
let mut term_ids = Vec::new();
for term_ord in self.entries.into_iter() {
if term_ord == u64::MAX {
has_missing = true;
} else {
// we can reasonably exclude values above u32::MAX
term_ids.push(term_ord as u32);
}
}
term_ids.sort_unstable();
dict.sorted_ords_to_term_cb(term_ids.iter().map(|term| *term as u64), |term| {
self.cardinality.sketch.insert_any(&term);
Ok(())
})?;
if has_missing {
// Replace missing with the actual value provided
let missing_key = self
.missing
.as_ref()
.expect("Found sentinel value u64::MAX for term_ord but `missing` is not set");
match missing_key {
Key::Str(missing) => {
self.cardinality.sketch.insert_any(&missing);
}
Key::F64(val) => {
let val = f64_to_u64(*val);
self.cardinality.sketch.insert_any(&val);
}
Key::U64(val) => {
self.cardinality.sketch.insert_any(&val);
}
Key::I64(val) => {
self.cardinality.sketch.insert_any(&val);
}
}
}
}
Ok(IntermediateMetricResult::Cardinality(self.cardinality))
}
}
impl SegmentAggregationCollector for SegmentCardinalityCollector {
fn add_intermediate_aggregation_result(
self: Box<Self>,
agg_with_accessor: &AggregationsWithAccessor,
results: &mut IntermediateAggregationResults,
) -> crate::Result<()> {
let name = agg_with_accessor.aggs.keys[self.accessor_idx].to_string();
let agg_with_accessor = &agg_with_accessor.aggs.values[self.accessor_idx];
let intermediate_result = self.into_intermediate_metric_result(agg_with_accessor)?;
results.push(
name,
IntermediateAggregationResult::Metric(intermediate_result),
)?;
Ok(())
}
fn collect(
&mut self,
doc: crate::DocId,
agg_with_accessor: &mut AggregationsWithAccessor,
) -> crate::Result<()> {
self.collect_block(&[doc], agg_with_accessor)
}
fn collect_block(
&mut self,
docs: &[crate::DocId],
agg_with_accessor: &mut AggregationsWithAccessor,
) -> crate::Result<()> {
let bucket_agg_accessor = &mut agg_with_accessor.aggs.values[self.accessor_idx];
self.fetch_block_with_field(docs, bucket_agg_accessor);
let col_block_accessor = &bucket_agg_accessor.column_block_accessor;
if self.column_type == ColumnType::Str {
for term_ord in col_block_accessor.iter_vals() {
self.entries.insert(term_ord);
}
} else if self.column_type == ColumnType::IpAddr {
let compact_space_accessor = bucket_agg_accessor
.accessor
.values
.clone()
.downcast_arc::<CompactSpaceU64Accessor>()
.map_err(|_| {
TantivyError::AggregationError(
crate::aggregation::AggregationError::InternalError(
"Type mismatch: Could not downcast to CompactSpaceU64Accessor"
.to_string(),
),
)
})?;
for val in col_block_accessor.iter_vals() {
let val: u128 = compact_space_accessor.compact_to_u128(val as u32);
self.cardinality.sketch.insert_any(&val);
}
} else {
for val in col_block_accessor.iter_vals() {
self.cardinality.sketch.insert_any(&val);
}
}
Ok(())
}
}
#[derive(Clone, Debug, Serialize, Deserialize)]
/// The percentiles collector used during segment collection and for merging results.
pub struct CardinalityCollector {
sketch: HyperLogLogPlus<u64, BuildSaltedHasher>,
}
impl Default for CardinalityCollector {
fn default() -> Self {
Self::new(0)
}
}
impl PartialEq for CardinalityCollector {
fn eq(&self, _other: &Self) -> bool {
false
}
}
impl CardinalityCollector {
/// Compute the final cardinality estimate.
pub fn finalize(self) -> Option<f64> {
Some(self.sketch.clone().count().trunc())
}
fn new(salt: u8) -> Self {
Self {
sketch: HyperLogLogPlus::new(16, BuildSaltedHasher { salt }).unwrap(),
}
}
pub(crate) fn merge_fruits(&mut self, right: CardinalityCollector) -> crate::Result<()> {
self.sketch.merge(&right.sketch).map_err(|err| {
TantivyError::AggregationError(AggregationError::InternalError(format!(
"Error while merging cardinality {err:?}"
)))
})?;
Ok(())
}
}
#[cfg(test)]
mod tests {
use std::net::IpAddr;
use std::str::FromStr;
use columnar::MonotonicallyMappableToU64;
use crate::aggregation::agg_req::Aggregations;
use crate::aggregation::tests::{exec_request, get_test_index_from_terms};
use crate::schema::{IntoIpv6Addr, Schema, FAST};
use crate::Index;
#[test]
fn cardinality_aggregation_test_empty_index() -> crate::Result<()> {
let values = vec![];
let index = get_test_index_from_terms(false, &values)?;
let agg_req: Aggregations = serde_json::from_value(json!({
"cardinality": {
"cardinality": {
"field": "string_id",
}
},
}))
.unwrap();
let res = exec_request(agg_req, &index)?;
assert_eq!(res["cardinality"]["value"], 0.0);
Ok(())
}
#[test]
fn cardinality_aggregation_test_single_segment() -> crate::Result<()> {
cardinality_aggregation_test_merge_segment(true)
}
#[test]
fn cardinality_aggregation_test() -> crate::Result<()> {
cardinality_aggregation_test_merge_segment(false)
}
fn cardinality_aggregation_test_merge_segment(merge_segments: bool) -> crate::Result<()> {
let segment_and_terms = vec![
vec!["terma"],
vec!["termb"],
vec!["termc"],
vec!["terma"],
vec!["terma"],
vec!["terma"],
vec!["termb"],
vec!["terma"],
];
let index = get_test_index_from_terms(merge_segments, &segment_and_terms)?;
let agg_req: Aggregations = serde_json::from_value(json!({
"cardinality": {
"cardinality": {
"field": "string_id",
}
},
}))
.unwrap();
let res = exec_request(agg_req, &index)?;
assert_eq!(res["cardinality"]["value"], 3.0);
Ok(())
}
#[test]
fn cardinality_aggregation_u64() -> crate::Result<()> {
let mut schema_builder = Schema::builder();
let id_field = schema_builder.add_u64_field("id", FAST);
let index = Index::create_in_ram(schema_builder.build());
{
let mut writer = index.writer_for_tests()?;
writer.add_document(doc!(id_field => 1u64))?;
writer.add_document(doc!(id_field => 2u64))?;
writer.add_document(doc!(id_field => 3u64))?;
writer.add_document(doc!())?;
writer.commit()?;
}
let agg_req: Aggregations = serde_json::from_value(json!({
"cardinality": {
"cardinality": {
"field": "id",
"missing": 0u64
},
}
}))
.unwrap();
let res = exec_request(agg_req, &index)?;
assert_eq!(res["cardinality"]["value"], 4.0);
Ok(())
}
#[test]
fn cardinality_aggregation_ip_addr() -> crate::Result<()> {
let mut schema_builder = Schema::builder();
let field = schema_builder.add_ip_addr_field("ip_field", FAST);
let index = Index::create_in_ram(schema_builder.build());
{
let mut writer = index.writer_for_tests()?;
// IpV6 loopback
writer.add_document(doc!(field=>IpAddr::from_str("::1").unwrap().into_ipv6_addr()))?;
writer.add_document(doc!(field=>IpAddr::from_str("::1").unwrap().into_ipv6_addr()))?;
// IpV4
writer.add_document(
doc!(field=>IpAddr::from_str("127.0.0.1").unwrap().into_ipv6_addr()),
)?;
writer.commit()?;
}
let agg_req: Aggregations = serde_json::from_value(json!({
"cardinality": {
"cardinality": {
"field": "ip_field"
},
}
}))
.unwrap();
let res = exec_request(agg_req, &index)?;
assert_eq!(res["cardinality"]["value"], 2.0);
Ok(())
}
#[test]
fn cardinality_aggregation_json() -> crate::Result<()> {
let mut schema_builder = Schema::builder();
let field = schema_builder.add_json_field("json", FAST);
let index = Index::create_in_ram(schema_builder.build());
{
let mut writer = index.writer_for_tests()?;
writer.add_document(doc!(field => json!({"value": false})))?;
writer.add_document(doc!(field => json!({"value": true})))?;
writer.add_document(doc!(field => json!({"value": i64::from_u64(0u64)})))?;
writer.add_document(doc!(field => json!({"value": i64::from_u64(1u64)})))?;
writer.commit()?;
}
let agg_req: Aggregations = serde_json::from_value(json!({
"cardinality": {
"cardinality": {
"field": "json.value"
},
}
}))
.unwrap();
let res = exec_request(agg_req, &index)?;
assert_eq!(res["cardinality"]["value"], 4.0);
Ok(())
}
}

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