rust/tantivy
1
0
Fork 0
mirror of https://github.com/quickwit-oss/tantivy.git synced 2025-12-31 06:22:54 +00:00
Code Issues Packages Projects Releases Wiki Activity

Compare commits

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

38 Commits
numeric_wi ... index_writ

Author SHA1 Message Date
Pascal Seitz
dc90e4f8e9 change Err from Debug to Display 
as reported by a user on Discord, the output of the Debug on error is a useless: 'Any { .. }'
switch to `Display` via to_string
 2023-09-27 11:02:53 +08:00
PSeitz
34920d31f5 Fix DateHistogram bucket gap (#2183) 
* Fix DateHistogram bucket gap

Fixes a computation issue of the number of buckets needed in the
DateHistogram.

This is due to a missing normalization from request values (ms) to fast field
values (ns), when converting an intermediate result to the final result.
This results in a wrong computation by a factor 1_000_000.
The Histogram normalizes values to nanoseconds, to make the user input like
extended_bounds (ms precision) and the values from the fast field (ns precision for date type) compatible.
This normalization happens only for date type fields, as other field types don't have precision settings.
The normalization does not happen due a missing `column_type`, which is not
correctly passed after merging an empty aggregation (which does not have a `column_type` set), with a regular aggregation.

Another related issue is an empty aggregation, which will not have
`column_type` set, will not convert the result to human readable format.

This PR fixes the issue by:
- Limit the allowed field types of DateHistogram to DateType
- Instead of passing the column_type, which is only available on the segment level, we flag the aggregation as `is_date_agg`.
- Fix the merge logic

Add a flag to to normalization only once. This is not an issue
currently, but it could become easily one.

closes https://github.com/quickwit-oss/quickwit/issues/3837

* use older nightly for time crate (breaks build)
 2023-09-21 10:41:35 +02:00
trinity-1686a
0241a05b90 add support for exists query syntax in query parser (#2170) 
* add support for exists query syntax in query parser

* rustfmt

* make Exists require a field
 2023-09-19 11:10:39 +02:00
PSeitz
e125f3b041 fix test (#2178) 2023-09-19 08:21:50 +02:00
PSeitz
c520ac46fc add support for date in term agg (#2172) 
support DateTime in TermsAggregation
Format dates with Rfc3339
 2023-09-14 09:22:18 +02:00
PSeitz
2d7390341c increase min memory to 15MB for indexing (#2176) 
With tantivy 0.20 the minimum memory consumption per SegmentWriter increased to
12MB. 7MB are for the different fast field collectors types (they could be
lazily created). Increase the minimum memory from 3MB to 15MB.

Change memory variable naming from arena to budget.

closes #2156
 2023-09-13 07:38:34 +02:00
dependabot[bot]
03fcdce016 Bump actions/checkout from 3 to 4 (#2171) 
Bumps [actions/checkout](https://github.com/actions/checkout) from 3 to 4.
- [Release notes](https://github.com/actions/checkout/releases)
- [Changelog](https://github.com/actions/checkout/blob/main/CHANGELOG.md)
- [Commits](https://github.com/actions/checkout/compare/v3...v4)

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

Signed-off-by: dependabot[bot] <support@github.com>
Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com>
 2023-09-11 10:47:33 +02:00
Ping Xia
e4e416ac42 extend FuzzyTermQuery to support json field (#2173) 
* extend fuzzy search for json field

* comments

* comments

* fmt fix

* comments
 2023-09-11 05:59:40 +02:00
Igor Motov
19325132b7 Fast-field based implementation of ExistsQuery (#2160) 
Adds an implementation of ExistsQuery that takes advantage of fast fields.

Fixes #2159
 2023-09-07 11:51:49 +09:00
Paul Masurel
389d36f760 Added comments 2023-09-04 11:06:56 +09:00
PSeitz
49448b31c6 chore: Release (#2168) 
* chore: Release

* update CHANGELOG
 2023-09-01 13:58:58 +02:00
PSeitz
ebede0bed7 update CHANGELOG (#2167) 2023-08-31 10:01:44 +02:00
PSeitz
b1d8b072db add missing aggregation part 2 (#2149) 
* add missing aggregation part 2

Add missing support for:
- Mixed types columns
- Key of type string on numerical fields

The special aggregation is slower than the integrated one in TermsAggregation and therefore not
chosen by default, although it can cover all use cases.

* simplify, add num_docs to empty
 2023-08-31 07:55:33 +02:00
ethever.eth
ee6a7c2bbb fix a small typo (#2165) 
Co-authored-by: famouscat <onismaa@gmail.com>
 2023-08-30 20:14:26 +02:00
PSeitz
c4e2708901 fix clippy, fmt (#2162) 2023-08-30 08:04:26 +02:00
PSeitz
5c8cfa50eb add missing parameter for percentiles (#2157) 2023-08-29 13:04:24 +02:00
PSeitz
73cb71762f add missing parameter for stats,min,max,count,sum,avg (#2151) 
* add missing parameter for stats,min,max,count,sum,avg

add missing parameter for stats,min,max,count,sum,avg
closes #1913
partially #1789

* Apply suggestions from code review

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

---------

Co-authored-by: Paul Masurel <paul@quickwit.io>
 2023-08-28 08:59:51 +02:00
Harrison Burt
267dfe58d7 Fix testing on windows (#2155) 
* Fix missing trait imports

* Fix building tests on windows

* Revert other PR change
 2023-08-27 09:20:44 +09:00
Harrison Burt
131c10d318 Fix missing trait imports (#2154) 2023-08-27 09:20:26 +09:00
Chris Tam
e6cacc40a9 Remove outdated fast field documentation (#2145) 2023-08-24 07:49:49 +02:00
PSeitz
48d4847b38 Improve aggregation error message (#2150) 
* Improve aggregation error message

Improve aggregation error message by wrapping the deserialization with a
custom struct. This deserialization variant is slower, since we need to
keep the deserialized data around twice with this approach.
For now the valid variants list is manually updated. This could be
replaced with a proc macro.
closes #2143

* Simpler implementation

---------

Co-authored-by: Paul Masurel <paul@quickwit.io>
 2023-08-23 20:52:15 +02:00
PSeitz
59460c767f delayed column opening during merge (#2132) 
* lazy columnar merge

This is the first part of addressing #3633
Instead of loading all Column into memory for the merge, only the current column_name
group is loaded. This can be done since the sstable streams the columns lexicographically.

* refactor

* add rustdoc

* replace iterator with BTreeMap
 2023-08-21 08:55:35 +02:00
Paul Masurel
756156beaf Fix doc 2023-08-17 17:47:45 +09:00
PSeitz
480763db0d track memory arena memory usage (#2148) 2023-08-16 18:19:42 +02:00
PSeitz
62ece86f24 track ff dictionary indexing memory consumption (#2147) 2023-08-16 14:00:08 +02:00
Caleb Hattingh
52d9e6f298 Fix doc typos in count aggregation metric (#2127) 2023-08-15 08:50:23 +02:00
Caleb Hattingh
47b315ff18 doc: escape the backslash (#2144) 2023-08-14 19:10:07 +02:00
PSeitz
ed1deee902 fix sort index by date (#2124) 
closes #2112
 2023-08-14 17:36:52 +02:00
PSeitz
2e109018b7 add missing parameter to term agg (#2103) 
* add missing parameter to term agg

* move missing handling to block accessor

* add multivalue test, fix multivalue case, add comments

* add documentation, deactivate special case

* cargo fmt

* resolve merge conflict
 2023-08-14 14:22:18 +02:00
Adam Reichold
22c35b1e00 Fix explanation of boost queries seeking beyond query result. (#2142) 
* Make current nightly Clippy happy.

* Fix explanation of boost queries seeking beyond query result.
 2023-08-14 11:59:11 +09:00
trinity-1686a
b92082b748 implement lenient parser (#2129) 
* move query parser to nom

* add suupport for term grouping

* initial work on infallible parser

* fmt

* add tests and fix minor parsing bugs

* address review comments

* add support for lenient queries in tantivy

* make lenient parser report errors

* allow mixing occur and bool in query
 2023-08-08 15:41:29 +02:00
PSeitz
c2be6603a2 alternative mixed field aggregation collection (#2135) 
* alternative mixed field aggregation collection

instead of having multiple accessor in one AggregationWithAccessor split it into
multiple independent AggregationWithAccessor

* Update src/aggregation/agg_req_with_accessor.rs

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

---------

Co-authored-by: Paul Masurel <paul@quickwit.io>
 2023-07-27 12:25:31 +02:00
Adam Reichold
c805f08ca7 Fix a few more upcoming Clippy lints (#2133) 2023-07-24 17:07:57 +09:00
Adam Reichold
ccc0335158 Minor improvements to OwnedBytes (#2134) 
This makes it obvious where the `StableDerefTrait` is invoked and avoids
`transmute` when only a lifetime needs to be extended. Furthermore, it makes use
of `slice::split_at` where that seemed appropriate.
 2023-07-24 17:06:33 +09:00
Adam Reichold
42acd334f4 Fixes the new deny-by-default incorrect_partial_ord_impl_on_ord_type Clippy lint (#2131) 2023-07-21 11:36:17 +09:00
Adam Reichold
820f126075 Remove support for Brotli and Snappy compression (#2123) 
LZ4 provides fast and simple compression whereas Zstd is exceptionally flexible
so that the additional support for Brotli and Snappy does not really add
any distinct functionality on top of those two algorithms.

Removing them reduces our maintenance burden and reduces the number of choices
users have to make when setting up their project based on Tantivy.
 2023-07-14 16:54:59 +09:00
Adam Reichold
7e6c4a1856 Include only built-in compression algorithms as enum variants (#2121) 
* Include only built-in compression algorithms as enum variants

This enables compile-time errors when a compression algorithm is requested which
is not actually enabled for the current Cargo project. The cost is that indexes
using other compression algorithms cannot even be loaded (even though they
are not fully accessible in any case).

As a drive-by, this also fixes `--no-default-features` on `cfg(unix)`.

* Provide more instructive error messages for unsupported, but not unknown compression variants.
 2023-07-14 11:02:49 +09:00
Adam Reichold
5fafe4b1ab Add missing query_terms impl for TermSetQuery. (#2120) 2023-07-13 14:54:29 +02:00
116 changed files with 5023 additions and 1429 deletions
Expand all files Collapse all files

6
.github/workflows/coverage.yml vendored
View File

@@ -15,13 +15,13 @@ jobs:
coverage:
runs-on: ubuntu-latest
steps:
- uses: actions/checkout@v3
- uses: actions/checkout@v4
- name: Install Rust
run: rustup toolchain install nightly --profile minimal --component llvm-tools-preview
run: rustup toolchain install nightly-2023-09-10 --profile minimal --component llvm-tools-preview
- uses: Swatinem/rust-cache@v2
- uses: taiki-e/install-action@cargo-llvm-cov
- name: Generate code coverage
run: cargo +nightly llvm-cov --all-features --workspace --doctests --lcov --output-path lcov.info
run: cargo +nightly-2023-09-10 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

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

@@ -19,7 +19,7 @@ jobs:
runs-on: ubuntu-latest
steps:
- uses: actions/checkout@v3
- uses: actions/checkout@v4
- name: Install stable
uses: actions-rs/toolchain@v1
with:

6
.github/workflows/test.yml vendored
View File

@@ -20,7 +20,7 @@ jobs:
runs-on: ubuntu-latest
steps:
- uses: actions/checkout@v3
- uses: actions/checkout@v4
- name: Install nightly
uses: actions-rs/toolchain@v1
@@ -53,14 +53,14 @@ jobs:
strategy:
matrix:
features: [
{ label: "all", flags: "mmap,stopwords,brotli-compression,lz4-compression,snappy-compression,zstd-compression,failpoints" },
{ label: "all", flags: "mmap,stopwords,lz4-compression,zstd-compression,failpoints" },
{ label: "quickwit", flags: "mmap,quickwit,failpoints" }
]
name: test-${{ matrix.features.label}}
steps:
- uses: actions/checkout@v3
- uses: actions/checkout@v4
- name: Install stable
uses: actions-rs/toolchain@v1

27
CHANGELOG.md
View File

@@ -1,3 +1,30 @@
Tantivy 0.21
================================
#### Bugfixes
- Fix track fast field memory consumption, which led to higher memory consumption than the budget allowed during indexing [#2148](https://github.com/quickwit-oss/tantivy/issues/2148)[#2147](https://github.com/quickwit-oss/tantivy/issues/2147)(@PSeitz)
- Fix a regression from 0.20 where sort index by date wasn't working anymore [#2124](https://github.com/quickwit-oss/tantivy/issues/2124)(@PSeitz)
- Fix getting the root facet on the `FacetCollector`. [#2086](https://github.com/quickwit-oss/tantivy/issues/2086)(@adamreichold)
- Align numerical type priority order of columnar and query. [#2088](https://github.com/quickwit-oss/tantivy/issues/2088)(@fmassot)
#### Breaking Changes
- Remove support for Brotli and Snappy compression [#2123](https://github.com/quickwit-oss/tantivy/issues/2123)(@adamreichold)
#### Features/Improvements
- Implement lenient query parser [#2129](https://github.com/quickwit-oss/tantivy/pull/2129)(@trinity-1686a)
- order_by_u64_field and order_by_fast_field allow sorting in ascending and descending order [#2111](https://github.com/quickwit-oss/tantivy/issues/2111)(@naveenann)
- Allow dynamic filters in text analyzer builder [#2110](https://github.com/quickwit-oss/tantivy/issues/2110)(@fulmicoton @fmassot)
- **Aggregation**
- Add missing parameter for term aggregation [#2149](https://github.com/quickwit-oss/tantivy/issues/2149)[#2103](https://github.com/quickwit-oss/tantivy/issues/2103)(@PSeitz)
- Add missing parameter for percentiles [#2157](https://github.com/quickwit-oss/tantivy/issues/2157)(@PSeitz)
- Add missing parameter for stats,min,max,count,sum,avg [#2151](https://github.com/quickwit-oss/tantivy/issues/2151)(@PSeitz)
- Improve aggregation deserialization error message [#2150](https://github.com/quickwit-oss/tantivy/issues/2150)(@PSeitz)
- Add validation for type Bytes to term_agg [#2077](https://github.com/quickwit-oss/tantivy/issues/2077)(@PSeitz)
- Alternative mixed field collection [#2135](https://github.com/quickwit-oss/tantivy/issues/2135)(@PSeitz)
- Add missing query_terms impl for TermSetQuery. [#2120](https://github.com/quickwit-oss/tantivy/issues/2120)(@adamreichold)
- Minor improvements to OwnedBytes [#2134](https://github.com/quickwit-oss/tantivy/issues/2134)(@adamreichold)
- Remove allocations in split compound words [#2080](https://github.com/quickwit-oss/tantivy/issues/2080)(@PSeitz)
- Ngram tokenizer now returns an error with invalid arguments [#2102](https://github.com/quickwit-oss/tantivy/issues/2102)(@fmassot)
- Make TextAnalyzerBuilder public [#2097](https://github.com/quickwit-oss/tantivy/issues/2097)(@adamreichold)
- Return an error when tokenizer is not found while indexing [#2093](https://github.com/quickwit-oss/tantivy/issues/2093)(@naveenann)
- Delayed column opening during merge [#2132](https://github.com/quickwit-oss/tantivy/issues/2132)(@PSeitz)
Tantivy 0.20.2
================================

28
Cargo.toml
View File

@@ -1,6 +1,6 @@
[package]
name = "tantivy"
version = "0.20.2"
version = "0.21.0"
authors = ["Paul Masurel <paul.masurel@gmail.com>"]
license = "MIT"
categories = ["database-implementations", "data-structures"]
@@ -25,9 +25,7 @@ aho-corasick = "1.0"
tantivy-fst = "0.4.0"
memmap2 = { version = "0.7.1", optional = true }
lz4_flex = { version = "0.11", default-features = false, optional = true }
brotli = { version = "3.3.4", optional = true }
zstd = { version = "0.12", optional = true, default-features = false }
snap = { version = "1.0.5", optional = true }
tempfile = { version = "3.3.0", optional = true }
log = "0.4.16"
serde = { version = "1.0.136", features = ["derive"] }
@@ -56,13 +54,13 @@ measure_time = "0.8.2"
async-trait = "0.1.53"
arc-swap = "1.5.0"
columnar = { version= "0.1", path="./columnar", package ="tantivy-columnar" }
sstable = { version= "0.1", path="./sstable", package ="tantivy-sstable", optional = true }
stacker = { version= "0.1", path="./stacker", package ="tantivy-stacker" }
query-grammar = { version= "0.20.0", path="./query-grammar", package = "tantivy-query-grammar" }
tantivy-bitpacker = { version= "0.4", path="./bitpacker" }
common = { version= "0.5", path = "./common/", package = "tantivy-common" }
tokenizer-api = { version= "0.1", path="./tokenizer-api", package="tantivy-tokenizer-api" }
columnar = { version= "0.2", path="./columnar", package ="tantivy-columnar" }
sstable = { version= "0.2", path="./sstable", package ="tantivy-sstable", optional = true }
stacker = { version= "0.2", path="./stacker", package ="tantivy-stacker" }
query-grammar = { version= "0.21.0", path="./query-grammar", package = "tantivy-query-grammar" }
tantivy-bitpacker = { version= "0.5", path="./bitpacker" }
common = { version= "0.6", path = "./common/", package = "tantivy-common" }
tokenizer-api = { version= "0.2", path="./tokenizer-api", package="tantivy-tokenizer-api" }
sketches-ddsketch = { version = "0.2.1", features = ["use_serde"] }
futures-util = { version = "0.3.28", optional = true }
@@ -75,15 +73,17 @@ maplit = "1.0.2"
matches = "0.1.9"
pretty_assertions = "1.2.1"
proptest = "1.0.0"
criterion = "0.5"
test-log = "0.2.10"
env_logger = "0.10.0"
pprof = { git = "https://github.com/PSeitz/pprof-rs/", rev = "53af24b", features = ["flamegraph", "criterion"] } # temp fork that works with criterion 0.5
futures = "0.3.21"
paste = "1.0.11"
more-asserts = "0.3.1"
rand_distr = "0.4.3"
[target.'cfg(not(windows))'.dev-dependencies]
criterion = "0.5"
pprof = { git = "https://github.com/PSeitz/pprof-rs/", rev = "53af24b", features = ["flamegraph", "criterion"] } # temp fork that works with criterion 0.5
[dev-dependencies.fail]
version = "0.5.0"
features = ["failpoints"]
@@ -107,9 +107,7 @@ default = ["mmap", "stopwords", "lz4-compression"]
mmap = ["fs4", "tempfile", "memmap2"]
stopwords = []
brotli-compression = ["brotli"]
lz4-compression = ["lz4_flex"]
snappy-compression = ["snap"]
zstd-compression = ["zstd"]
failpoints = ["fail", "fail/failpoints"]
@@ -130,7 +128,7 @@ members = ["query-grammar", "bitpacker", "common", "ownedbytes", "stacker", "sst
[[test]]
name = "failpoints"
path = "tests/failpoints/mod.rs"
required-features = ["fail/failpoints"]
required-features = ["failpoints"]
[[bench]]
name = "analyzer"

2
README.md
View File

@@ -44,7 +44,7 @@ Details about the benchmark can be found at this [repository](https://github.com
- Single valued and multivalued u64, i64, and f64 fast fields (equivalent of doc values in Lucene)
- `&[u8]` fast fields
- Text, i64, u64, f64, dates, ip, bool, and hierarchical facet fields
- Compressed document store (LZ4, Zstd, None, Brotli, Snap)
- Compressed document store (LZ4, Zstd, None)
- Range queries
- Faceted search
- Configurable indexing (optional term frequency and position indexing)

2
bitpacker/Cargo.toml
View File

@@ -1,6 +1,6 @@
[package]
name = "tantivy-bitpacker"
version = "0.4.0"
version = "0.5.0"
edition = "2021"
authors = ["Paul Masurel <paul.masurel@gmail.com>"]
license = "MIT"

4
bitpacker/src/blocked_bitpacker.rs
View File

@@ -64,10 +64,8 @@ fn mem_usage<T>(items: &Vec<T>) -> usize {
impl BlockedBitpacker {
pub fn new() -> Self {
let mut compressed_blocks = vec![];
compressed_blocks.resize(8, 0);
Self {
compressed_blocks,
compressed_blocks: vec![0; 8],
buffer: vec![],
offset_and_bits: vec![],
}

1
cliff.toml
View File

@@ -32,6 +32,7 @@ postprocessors = [
{ pattern = 'Michael Kleen', replace = "mkleen"}, # replace with github user
{ pattern = 'Adrien Guillo', replace = "guilload"}, # replace with github user
{ pattern = 'François Massot', replace = "fmassot"}, # replace with github user
{ pattern = 'Naveen Aiathurai', replace = "naveenann"}, # replace with github user
{ pattern = '', replace = ""}, # replace with github user
]

10
columnar/Cargo.toml
View File

@@ -1,6 +1,6 @@
[package]
name = "tantivy-columnar"
version = "0.1.0"
version = "0.2.0"
edition = "2021"
license = "MIT"
homepage = "https://github.com/quickwit-oss/tantivy"
@@ -13,10 +13,10 @@ itertools = "0.11.0"
fnv = "1.0.7"
fastdivide = "0.4.0"
stacker = { version= "0.1", path = "../stacker", package="tantivy-stacker"}
sstable = { version= "0.1", path = "../sstable", package = "tantivy-sstable" }
common = { version= "0.5", path = "../common", package = "tantivy-common" }
tantivy-bitpacker = { version= "0.4", path = "../bitpacker/" }
stacker = { version= "0.2", path = "../stacker", package="tantivy-stacker"}
sstable = { version= "0.2", path = "../sstable", package = "tantivy-sstable" }
common = { version= "0.6", path = "../common", package = "tantivy-common" }
tantivy-bitpacker = { version= "0.5", path = "../bitpacker/" }
serde = "1.0.152"
[dev-dependencies]

96
columnar/src/block_accessor.rs
View File

@@ -1,9 +1,12 @@
use std::cmp::Ordering;
use crate::{Column, DocId, RowId};
#[derive(Debug, Default, Clone)]
pub struct ColumnBlockAccessor<T> {
val_cache: Vec<T>,
docid_cache: Vec<DocId>,
missing_docids_cache: Vec<DocId>,
row_id_cache: Vec<RowId>,
}
@@ -20,6 +23,20 @@ impl<T: PartialOrd + Copy + std::fmt::Debug + Send + Sync + 'static + Default>
.values
.get_vals(&self.row_id_cache, &mut self.val_cache);
}
#[inline]
pub fn fetch_block_with_missing(&mut self, docs: &[u32], accessor: &Column<T>, missing: T) {
self.fetch_block(docs, accessor);
// We can compare docid_cache with docs to find missing docs
if docs.len() != self.docid_cache.len() || accessor.index.is_multivalue() {
self.missing_docids_cache.clear();
find_missing_docs(docs, &self.docid_cache, |doc| {
self.missing_docids_cache.push(doc);
self.val_cache.push(missing);
});
self.docid_cache
.extend_from_slice(&self.missing_docids_cache);
}
}
#[inline]
pub fn iter_vals(&self) -> impl Iterator<Item = T> + '_ {
@@ -34,3 +51,82 @@ impl<T: PartialOrd + Copy + std::fmt::Debug + Send + Sync + 'static + Default>
.zip(self.val_cache.iter().cloned())
}
}
/// Given two sorted lists of docids `docs` and `hits`, hits is a subset of `docs`.
/// Return all docs that are not in `hits`.
fn find_missing_docs<F>(docs: &[u32], hits: &[u32], mut callback: F)
where F: FnMut(u32) {
let mut docs_iter = docs.iter();
let mut hits_iter = hits.iter();
let mut doc = docs_iter.next();
let mut hit = hits_iter.next();
while let (Some(&current_doc), Some(&current_hit)) = (doc, hit) {
match current_doc.cmp(&current_hit) {
Ordering::Less => {
callback(current_doc);
doc = docs_iter.next();
}
Ordering::Equal => {
doc = docs_iter.next();
hit = hits_iter.next();
}
Ordering::Greater => {
hit = hits_iter.next();
}
}
}
while let Some(&current_doc) = doc {
callback(current_doc);
doc = docs_iter.next();
}
}
#[cfg(test)]
mod tests {
use super::*;
#[test]
fn test_find_missing_docs() {
let docs: Vec<u32> = vec![1, 2, 3, 4, 5, 6, 7, 8, 9, 10];
let hits: Vec<u32> = vec![2, 4, 6, 8, 10];
let mut missing_docs: Vec<u32> = Vec::new();
find_missing_docs(&docs, &hits, |missing_doc| {
missing_docs.push(missing_doc);
});
assert_eq!(missing_docs, vec![1, 3, 5, 7, 9]);
}
#[test]
fn test_find_missing_docs_empty() {
let docs: Vec<u32> = Vec::new();
let hits: Vec<u32> = vec![2, 4, 6, 8, 10];
let mut missing_docs: Vec<u32> = Vec::new();
find_missing_docs(&docs, &hits, |missing_doc| {
missing_docs.push(missing_doc);
});
assert_eq!(missing_docs, vec![]);
}
#[test]
fn test_find_missing_docs_all_missing() {
let docs: Vec<u32> = vec![1, 2, 3, 4, 5];
let hits: Vec<u32> = Vec::new();
let mut missing_docs: Vec<u32> = Vec::new();
find_missing_docs(&docs, &hits, |missing_doc| {
missing_docs.push(missing_doc);
});
assert_eq!(missing_docs, vec![1, 2, 3, 4, 5]);
}
}

9
columnar/src/column/dictionary_encoded.rs
View File

@@ -30,6 +30,13 @@ impl fmt::Debug for BytesColumn {
}
impl BytesColumn {
pub fn empty(num_docs: u32) -> BytesColumn {
BytesColumn {
dictionary: Arc::new(Dictionary::empty()),
term_ord_column: Column::build_empty_column(num_docs),
}
}
/// Fills the given `output` buffer with the term associated to the ordinal `ord`.
///
/// Returns `false` if the term does not exist (e.g. `term_ord` is greater or equal to the
@@ -77,7 +84,7 @@ impl From<StrColumn> for BytesColumn {
}
impl StrColumn {
pub(crate) fn wrap(bytes_column: BytesColumn) -> StrColumn {
pub fn wrap(bytes_column: BytesColumn) -> StrColumn {
StrColumn(bytes_column)
}

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

@@ -130,7 +130,7 @@ impl<T: PartialOrd + Copy + Debug + Send + Sync + 'static> Column<T> {
.select_batch_in_place(selected_docid_range.start, doc_ids);
}
/// Fils the output vector with the (possibly multiple values that are associated_with
/// Fills the output vector with the (possibly multiple values that are associated_with
/// `row_id`.
///
/// This method clears the `output` vector.

4
columnar/src/column_index/mod.rs
View File

@@ -37,6 +37,10 @@ impl From<MultiValueIndex> for ColumnIndex {
}
impl ColumnIndex {
#[inline]
pub fn is_multivalue(&self) -> bool {
matches!(self, ColumnIndex::Multivalued(_))
}
// Returns the cardinality of the column index.
//
// By convention, if the column contains no docs, we consider that it is

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

@@ -38,6 +38,6 @@ impl Ord for BlankRange {
}
impl PartialOrd for BlankRange {
fn partial_cmp(&self, other: &Self) -> Option<std::cmp::Ordering> {
Some(self.blank_size().cmp(&other.blank_size()))
Some(self.cmp(other))
}
}

186
columnar/src/columnar/merge/mod.rs
View File

@@ -2,7 +2,7 @@ mod merge_dict_column;
mod merge_mapping;
mod term_merger;
use std::collections::{BTreeMap, HashMap, HashSet};
use std::collections::{BTreeMap, HashSet};
use std::io;
use std::net::Ipv6Addr;
use std::sync::Arc;
@@ -18,7 +18,8 @@ use crate::columnar::writer::CompatibleNumericalTypes;
use crate::columnar::ColumnarReader;
use crate::dynamic_column::DynamicColumn;
use crate::{
BytesColumn, Column, ColumnIndex, ColumnType, ColumnValues, NumericalType, NumericalValue,
BytesColumn, Column, ColumnIndex, ColumnType, ColumnValues, DynamicColumnHandle, NumericalType,
NumericalValue,
};
/// Column types are grouped into different categories.
@@ -28,14 +29,16 @@ use crate::{
/// In practise, today, only Numerical colummns are coerced into one type today.
///
/// See also [README.md].
#[derive(Copy, Clone, Eq, PartialEq, Hash, Debug)]
///
/// The ordering has to match the ordering of the variants in [ColumnType].
#[derive(Copy, Clone, Eq, PartialOrd, Ord, PartialEq, Hash, Debug)]
pub(crate) enum ColumnTypeCategory {
Bool,
Str,
Numerical,
DateTime,
Bytes,
Str,
Bool,
IpAddr,
DateTime,
}
impl From<ColumnType> for ColumnTypeCategory {
@@ -83,9 +86,20 @@ pub fn merge_columnar(
.iter()
.map(|reader| reader.num_rows())
.collect::<Vec<u32>>();
let columns_to_merge =
group_columns_for_merge(columnar_readers, required_columns, &merge_row_order)?;
for ((column_name, column_type), columns) in columns_to_merge {
for res in columns_to_merge {
let ((column_name, _column_type_category), grouped_columns) = res;
let grouped_columns = grouped_columns.open(&merge_row_order)?;
if grouped_columns.is_empty() {
continue;
}
let column_type = grouped_columns.column_type_after_merge();
let mut columns = grouped_columns.columns;
coerce_columns(column_type, &mut columns)?;
let mut column_serializer =
serializer.start_serialize_column(column_name.as_bytes(), column_type);
merge_column(
@@ -97,6 +111,7 @@ pub fn merge_columnar(
)?;
column_serializer.finalize()?;
}
serializer.finalize(merge_row_order.num_rows())?;
Ok(())
}
@@ -210,40 +225,12 @@ fn merge_column(
struct GroupedColumns {
required_column_type: Option<ColumnType>,
columns: Vec<Option<DynamicColumn>>,
column_category: ColumnTypeCategory,
}
impl GroupedColumns {
fn for_category(column_category: ColumnTypeCategory, num_columnars: usize) -> Self {
GroupedColumns {
required_column_type: None,
columns: vec![None; num_columnars],
column_category,
}
}
/// Set the dynamic column for a given columnar.
fn set_column(&mut self, columnar_id: usize, column: DynamicColumn) {
self.columns[columnar_id] = Some(column);
}
/// Force the existence of a column, as well as its type.
fn require_type(&mut self, required_type: ColumnType) -> io::Result<()> {
if let Some(existing_required_type) = self.required_column_type {
if existing_required_type == required_type {
// This was just a duplicate in the `required_columns`.
// Nothing to do.
return Ok(());
} else {
return Err(io::Error::new(
io::ErrorKind::InvalidInput,
"Required column conflicts with another required column of the same type \
category.",
));
}
}
self.required_column_type = Some(required_type);
Ok(())
/// Check is column group can be skipped during serialization.
fn is_empty(&self) -> bool {
self.required_column_type.is_none() && self.columns.iter().all(Option::is_none)
}
/// Returns the column type after merge.
@@ -265,11 +252,76 @@ impl GroupedColumns {
}
// At the moment, only the numerical categorical column type has more than one possible
// column type.
assert_eq!(self.column_category, ColumnTypeCategory::Numerical);
assert!(self
.columns
.iter()
.flatten()
.all(|el| ColumnTypeCategory::from(el.column_type()) == ColumnTypeCategory::Numerical));
merged_numerical_columns_type(self.columns.iter().flatten()).into()
}
}
struct GroupedColumnsHandle {
required_column_type: Option<ColumnType>,
columns: Vec<Option<DynamicColumnHandle>>,
}
impl GroupedColumnsHandle {
fn new(num_columnars: usize) -> Self {
GroupedColumnsHandle {
required_column_type: None,
columns: vec![None; num_columnars],
}
}
fn open(self, merge_row_order: &MergeRowOrder) -> io::Result<GroupedColumns> {
let mut columns: Vec<Option<DynamicColumn>> = Vec::new();
for (columnar_id, column) in self.columns.iter().enumerate() {
if let Some(column) = column {
let column = column.open()?;
// We skip columns that end up with 0 documents.
// That way, we make sure they don't end up influencing the merge type or
// creating empty columns.
if is_empty_after_merge(merge_row_order, &column, columnar_id) {
columns.push(None);
} else {
columns.push(Some(column));
}
} else {
columns.push(None);
}
}
Ok(GroupedColumns {
required_column_type: self.required_column_type,
columns,
})
}
/// Set the dynamic column for a given columnar.
fn set_column(&mut self, columnar_id: usize, column: DynamicColumnHandle) {
self.columns[columnar_id] = Some(column);
}
/// Force the existence of a column, as well as its type.
fn require_type(&mut self, required_type: ColumnType) -> io::Result<()> {
if let Some(existing_required_type) = self.required_column_type {
if existing_required_type == required_type {
// This was just a duplicate in the `required_columns`.
// Nothing to do.
return Ok(());
} else {
return Err(io::Error::new(
io::ErrorKind::InvalidInput,
"Required column conflicts with another required column of the same type \
category.",
));
}
}
self.required_column_type = Some(required_type);
Ok(())
}
}
/// Returns the type of the merged numerical column.
///
/// This function picks the first numerical type out of i64, u64, f64 (order matters
@@ -293,7 +345,7 @@ fn merged_numerical_columns_type<'a>(
fn is_empty_after_merge(
merge_row_order: &MergeRowOrder,
column: &DynamicColumn,
columnar_id: usize,
columnar_ord: usize,
) -> bool {
if column.num_values() == 0u32 {
// It was empty before the merge.
@@ -305,7 +357,7 @@ fn is_empty_after_merge(
false
}
MergeRowOrder::Shuffled(shuffled) => {
if let Some(alive_bitset) = &shuffled.alive_bitsets[columnar_id] {
if let Some(alive_bitset) = &shuffled.alive_bitsets[columnar_ord] {
let column_index = column.column_index();
match column_index {
ColumnIndex::Empty { .. } => true,
@@ -348,56 +400,34 @@ fn is_empty_after_merge(
}
}
#[allow(clippy::type_complexity)]
fn group_columns_for_merge(
columnar_readers: &[&ColumnarReader],
required_columns: &[(String, ColumnType)],
merge_row_order: &MergeRowOrder,
) -> io::Result<BTreeMap<(String, ColumnType), Vec<Option<DynamicColumn>>>> {
// Each column name may have multiple types of column associated.
// For merging we are interested in the same column type category since they can be merged.
let mut columns_grouped: HashMap<(String, ColumnTypeCategory), GroupedColumns> = HashMap::new();
/// Iterates over the columns of the columnar readers, grouped by column name.
/// Key functionality is that `open` of the Columns is done lazy per group.
fn group_columns_for_merge<'a>(
columnar_readers: &'a [&'a ColumnarReader],
required_columns: &'a [(String, ColumnType)],
_merge_row_order: &'a MergeRowOrder,
) -> io::Result<BTreeMap<(String, ColumnTypeCategory), GroupedColumnsHandle>> {
let mut columns: BTreeMap<(String, ColumnTypeCategory), GroupedColumnsHandle> = BTreeMap::new();
for &(ref column_name, column_type) in required_columns {
columns_grouped
columns
.entry((column_name.clone(), column_type.into()))
.or_insert_with(|| {
GroupedColumns::for_category(column_type.into(), columnar_readers.len())
})
.or_insert_with(|| GroupedColumnsHandle::new(columnar_readers.len()))
.require_type(column_type)?;
}
for (columnar_id, columnar_reader) in columnar_readers.iter().enumerate() {
let column_name_and_handle = columnar_reader.list_columns()?;
// We skip columns that end up with 0 documents.
// That way, we make sure they don't end up influencing the merge type or
// creating empty columns.
let column_name_and_handle = columnar_reader.iter_columns()?;
for (column_name, handle) in column_name_and_handle {
let column_category: ColumnTypeCategory = handle.column_type().into();
let column = handle.open()?;
if is_empty_after_merge(merge_row_order, &column, columnar_id) {
continue;
}
columns_grouped
columns
.entry((column_name, column_category))
.or_insert_with(|| {
GroupedColumns::for_category(column_category, columnar_readers.len())
})
.set_column(columnar_id, column);
.or_insert_with(|| GroupedColumnsHandle::new(columnar_readers.len()))
.set_column(columnar_id, handle);
}
}
let mut merge_columns: BTreeMap<(String, ColumnType), Vec<Option<DynamicColumn>>> =
Default::default();
for ((column_name, _), mut grouped_columns) in columns_grouped {
let column_type = grouped_columns.column_type_after_merge();
coerce_columns(column_type, &mut grouped_columns.columns)?;
merge_columns.insert((column_name, column_type), grouped_columns.columns);
}
Ok(merge_columns)
Ok(columns)
}
fn coerce_columns(

92
columnar/src/columnar/merge/tests.rs
View File

@@ -1,3 +1,5 @@
use std::collections::BTreeMap;
use itertools::Itertools;
use super::*;
@@ -27,22 +29,10 @@ fn test_column_coercion_to_u64() {
let columnar2 = make_columnar("numbers", &[u64::MAX]);
let columnars = &[&columnar1, &columnar2];
let merge_order = StackMergeOrder::stack(columnars).into();
let column_map: BTreeMap<(String, ColumnType), Vec<Option<DynamicColumn>>> =
let column_map: BTreeMap<(String, ColumnTypeCategory), GroupedColumnsHandle> =
group_columns_for_merge(columnars, &[], &merge_order).unwrap();
assert_eq!(column_map.len(), 1);
assert!(column_map.contains_key(&("numbers".to_string(), ColumnType::U64)));
}
#[test]
fn test_column_no_coercion_if_all_the_same() {
let columnar1 = make_columnar("numbers", &[1u64]);
let columnar2 = make_columnar("numbers", &[2u64]);
let columnars = &[&columnar1, &columnar2];
let merge_order = StackMergeOrder::stack(columnars).into();
let column_map: BTreeMap<(String, ColumnType), Vec<Option<DynamicColumn>>> =
group_columns_for_merge(columnars, &[], &merge_order).unwrap();
assert_eq!(column_map.len(), 1);
assert!(column_map.contains_key(&("numbers".to_string(), ColumnType::U64)));
assert!(column_map.contains_key(&("numbers".to_string(), ColumnTypeCategory::Numerical)));
}
#[test]
@@ -51,24 +41,24 @@ fn test_column_coercion_to_i64() {
let columnar2 = make_columnar("numbers", &[2u64]);
let columnars = &[&columnar1, &columnar2];
let merge_order = StackMergeOrder::stack(columnars).into();
let column_map: BTreeMap<(String, ColumnType), Vec<Option<DynamicColumn>>> =
let column_map: BTreeMap<(String, ColumnTypeCategory), GroupedColumnsHandle> =
group_columns_for_merge(columnars, &[], &merge_order).unwrap();
assert_eq!(column_map.len(), 1);
assert!(column_map.contains_key(&("numbers".to_string(), ColumnType::I64)));
assert!(column_map.contains_key(&("numbers".to_string(), ColumnTypeCategory::Numerical)));
}
#[test]
fn test_impossible_coercion_returns_an_error() {
let columnar1 = make_columnar("numbers", &[u64::MAX]);
let merge_order = StackMergeOrder::stack(&[&columnar1]).into();
let group_error = group_columns_for_merge(
&[&columnar1],
&[("numbers".to_string(), ColumnType::I64)],
&merge_order,
)
.unwrap_err();
assert_eq!(group_error.kind(), io::ErrorKind::InvalidInput);
}
//#[test]
// fn test_impossible_coercion_returns_an_error() {
// let columnar1 = make_columnar("numbers", &[u64::MAX]);
// let merge_order = StackMergeOrder::stack(&[&columnar1]).into();
// let group_error = group_columns_for_merge_iter(
//&[&columnar1],
//&[("numbers".to_string(), ColumnType::I64)],
//&merge_order,
//)
//.unwrap_err();
// assert_eq!(group_error.kind(), io::ErrorKind::InvalidInput);
//}
#[test]
fn test_group_columns_with_required_column() {
@@ -76,7 +66,7 @@ fn test_group_columns_with_required_column() {
let columnar2 = make_columnar("numbers", &[2u64]);
let columnars = &[&columnar1, &columnar2];
let merge_order = StackMergeOrder::stack(columnars).into();
let column_map: BTreeMap<(String, ColumnType), Vec<Option<DynamicColumn>>> =
let column_map: BTreeMap<(String, ColumnTypeCategory), GroupedColumnsHandle> =
group_columns_for_merge(
&[&columnar1, &columnar2],
&[("numbers".to_string(), ColumnType::U64)],
@@ -84,7 +74,7 @@ fn test_group_columns_with_required_column() {
)
.unwrap();
assert_eq!(column_map.len(), 1);
assert!(column_map.contains_key(&("numbers".to_string(), ColumnType::U64)));
assert!(column_map.contains_key(&("numbers".to_string(), ColumnTypeCategory::Numerical)));
}
#[test]
@@ -93,17 +83,17 @@ fn test_group_columns_required_column_with_no_existing_columns() {
let columnar2 = make_columnar("numbers", &[2u64]);
let columnars = &[&columnar1, &columnar2];
let merge_order = StackMergeOrder::stack(columnars).into();
let column_map: BTreeMap<(String, ColumnType), Vec<Option<DynamicColumn>>> =
group_columns_for_merge(
columnars,
&[("required_col".to_string(), ColumnType::Str)],
&merge_order,
)
.unwrap();
let column_map: BTreeMap<_, _> = group_columns_for_merge(
columnars,
&[("required_col".to_string(), ColumnType::Str)],
&merge_order,
)
.unwrap();
assert_eq!(column_map.len(), 2);
let columns = column_map
.get(&("required_col".to_string(), ColumnType::Str))
.unwrap();
let columns = &column_map
.get(&("required_col".to_string(), ColumnTypeCategory::Str))
.unwrap()
.columns;
assert_eq!(columns.len(), 2);
assert!(columns[0].is_none());
assert!(columns[1].is_none());
@@ -115,7 +105,7 @@ fn test_group_columns_required_column_is_above_all_columns_have_the_same_type_ru
let columnar2 = make_columnar("numbers", &[2i64]);
let columnars = &[&columnar1, &columnar2];
let merge_order = StackMergeOrder::stack(columnars).into();
let column_map: BTreeMap<(String, ColumnType), Vec<Option<DynamicColumn>>> =
let column_map: BTreeMap<(String, ColumnTypeCategory), GroupedColumnsHandle> =
group_columns_for_merge(
columnars,
&[("numbers".to_string(), ColumnType::U64)],
@@ -123,7 +113,7 @@ fn test_group_columns_required_column_is_above_all_columns_have_the_same_type_ru
)
.unwrap();
assert_eq!(column_map.len(), 1);
assert!(column_map.contains_key(&("numbers".to_string(), ColumnType::U64)));
assert!(column_map.contains_key(&("numbers".to_string(), ColumnTypeCategory::Numerical)));
}
#[test]
@@ -132,21 +122,23 @@ fn test_missing_column() {
let columnar2 = make_columnar("numbers2", &[2u64]);
let columnars = &[&columnar1, &columnar2];
let merge_order = StackMergeOrder::stack(columnars).into();
let column_map: BTreeMap<(String, ColumnType), Vec<Option<DynamicColumn>>> =
let column_map: BTreeMap<(String, ColumnTypeCategory), GroupedColumnsHandle> =
group_columns_for_merge(columnars, &[], &merge_order).unwrap();
assert_eq!(column_map.len(), 2);
assert!(column_map.contains_key(&("numbers".to_string(), ColumnType::I64)));
assert!(column_map.contains_key(&("numbers".to_string(), ColumnTypeCategory::Numerical)));
{
let columns = column_map
.get(&("numbers".to_string(), ColumnType::I64))
.unwrap();
let columns = &column_map
.get(&("numbers".to_string(), ColumnTypeCategory::Numerical))
.unwrap()
.columns;
assert!(columns[0].is_some());
assert!(columns[1].is_none());
}
{
let columns = column_map
.get(&("numbers2".to_string(), ColumnType::U64))
.unwrap();
let columns = &column_map
.get(&("numbers2".to_string(), ColumnTypeCategory::Numerical))
.unwrap()
.columns;
assert!(columns[0].is_none());
assert!(columns[1].is_some());
}

53
columnar/src/columnar/reader/mod.rs
View File

@@ -102,30 +102,41 @@ impl ColumnarReader {
pub fn num_rows(&self) -> RowId {
self.num_rows
}
// Iterate over the columns in a sorted way
pub fn iter_columns(
&self,
) -> io::Result<impl Iterator<Item = (String, DynamicColumnHandle)> + '_> {
let mut stream = self.column_dictionary.stream()?;
Ok(std::iter::from_fn(move || {
if stream.advance() {
let key_bytes: &[u8] = stream.key();
let column_code: u8 = key_bytes.last().cloned().unwrap();
// TODO Error Handling. The API gets quite ugly when returning the error here, so
// instead we could just check the first N columns upfront.
let column_type: ColumnType = ColumnType::try_from_code(column_code)
.map_err(|_| io_invalid_data(format!("Unknown column code `{column_code}`")))
.unwrap();
let range = stream.value().clone();
let column_name =
// The last two bytes are respectively the 0u8 separator and the column_type.
String::from_utf8_lossy(&key_bytes[..key_bytes.len() - 2]).to_string();
let file_slice = self
.column_data
.slice(range.start as usize..range.end as usize);
let column_handle = DynamicColumnHandle {
file_slice,
column_type,
};
Some((column_name, column_handle))
} else {
None
}
}))
}
// TODO Add unit tests
pub fn list_columns(&self) -> io::Result<Vec<(String, DynamicColumnHandle)>> {
let mut stream = self.column_dictionary.stream()?;
let mut results = Vec::new();
while stream.advance() {
let key_bytes: &[u8] = stream.key();
let column_code: u8 = key_bytes.last().cloned().unwrap();
let column_type: ColumnType = ColumnType::try_from_code(column_code)
.map_err(|_| io_invalid_data(format!("Unknown column code `{column_code}`")))?;
let range = stream.value().clone();
let column_name =
// The last two bytes are respectively the 0u8 separator and the column_type.
String::from_utf8_lossy(&key_bytes[..key_bytes.len() - 2]).to_string();
let file_slice = self
.column_data
.slice(range.start as usize..range.end as usize);
let column_handle = DynamicColumnHandle {
file_slice,
column_type,
};
results.push((column_name, column_handle));
}
Ok(results)
Ok(self.iter_columns()?.collect())
}
fn stream_for_column_range(&self, column_name: &str) -> sstable::StreamerBuilder<RangeSSTable> {

10
columnar/src/columnar/writer/mod.rs
View File

@@ -79,7 +79,6 @@ fn mutate_or_create_column<V, TMutator>(
impl ColumnarWriter {
pub fn mem_usage(&self) -> usize {
// TODO add dictionary builders.
self.arena.mem_usage()
+ self.numerical_field_hash_map.mem_usage()
+ self.bool_field_hash_map.mem_usage()
@@ -87,6 +86,11 @@ impl ColumnarWriter {
+ self.str_field_hash_map.mem_usage()
+ self.ip_addr_field_hash_map.mem_usage()
+ self.datetime_field_hash_map.mem_usage()
+ self
.dictionaries
.iter()
.map(|dict| dict.mem_usage())
.sum::<usize>()
}
/// Returns the list of doc ids from 0..num_docs sorted by the `sort_field`
@@ -101,6 +105,10 @@ impl ColumnarWriter {
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();
};

7
columnar/src/dictionary.rs
View File

@@ -32,6 +32,7 @@ pub struct OrderedId(pub u32);
#[derive(Default)]
pub(crate) struct DictionaryBuilder {
dict: FnvHashMap<Vec<u8>, UnorderedId>,
memory_consumption: usize,
}
impl DictionaryBuilder {
@@ -43,6 +44,8 @@ impl DictionaryBuilder {
}
let new_id = UnorderedId(self.dict.len() as u32);
self.dict.insert(term.to_vec(), new_id);
self.memory_consumption += term.len();
self.memory_consumption += 40; // Term Metadata + HashMap overhead
new_id
}
@@ -63,6 +66,10 @@ impl DictionaryBuilder {
sstable_builder.finish()?;
Ok(TermIdMapping { unordered_to_ord })
}
pub(crate) fn mem_usage(&self) -> usize {
self.memory_consumption
}
}
#[cfg(test)]

11
columnar/src/dynamic_column.rs
View File

@@ -228,7 +228,7 @@ static_dynamic_conversions!(StrColumn, Str);
static_dynamic_conversions!(BytesColumn, Bytes);
static_dynamic_conversions!(Column<Ipv6Addr>, IpAddr);
#[derive(Clone)]
#[derive(Clone, Debug)]
pub struct DynamicColumnHandle {
pub(crate) file_slice: FileSlice,
pub(crate) column_type: ColumnType,
@@ -247,7 +247,7 @@ impl DynamicColumnHandle {
}
/// Returns the `u64` fast field reader reader associated with `fields` of types
/// Str, u64, i64, f64, or datetime.
/// Str, u64, i64, f64, bool, or datetime.
///
/// If not, the fastfield reader will returns the u64-value associated with the original
/// FastValue.
@@ -258,9 +258,12 @@ impl DynamicColumnHandle {
let column: BytesColumn = crate::column::open_column_bytes(column_bytes)?;
Ok(Some(column.term_ord_column))
}
ColumnType::Bool => Ok(None),
ColumnType::IpAddr => Ok(None),
ColumnType::I64 | ColumnType::U64 | ColumnType::F64 | ColumnType::DateTime => {
ColumnType::Bool
| ColumnType::I64
| ColumnType::U64
| ColumnType::F64
| ColumnType::DateTime => {
let column = crate::column::open_column_u64::<u64>(column_bytes)?;
Ok(Some(column))
}

4
common/Cargo.toml
View File

@@ -1,6 +1,6 @@
[package]
name = "tantivy-common"
version = "0.5.0"
version = "0.6.0"
authors = ["Paul Masurel <paul@quickwit.io>", "Pascal Seitz <pascal@quickwit.io>"]
license = "MIT"
edition = "2021"
@@ -14,7 +14,7 @@ repository = "https://github.com/quickwit-oss/tantivy"
[dependencies]
byteorder = "1.4.3"
ownedbytes = { version= "0.5", path="../ownedbytes" }
ownedbytes = { version= "0.6", path="../ownedbytes" }
async-trait = "0.1"
time = { version = "0.3.10", features = ["serde-well-known"] }
serde = { version = "1.0.136", features = ["derive"] }

81
common/src/file_slice.rs
View File

@@ -1,3 +1,4 @@
use std::fs::File;
use std::ops::{Deref, Range, RangeBounds};
use std::sync::Arc;
use std::{fmt, io};
@@ -32,6 +33,62 @@ pub trait FileHandle: 'static + Send + Sync + HasLen + fmt::Debug {
}
}
#[derive(Debug)]
/// A File with it's length included.
pub struct WrapFile {
file: File,
len: usize,
}
impl WrapFile {
/// Creates a new WrapFile and stores its length.
pub fn new(file: File) -> io::Result<Self> {
let len = file.metadata()?.len() as usize;
Ok(WrapFile { file, len })
}
}
#[async_trait]
impl FileHandle for WrapFile {
fn read_bytes(&self, range: Range<usize>) -> io::Result<OwnedBytes> {
let file_len = self.len();
// Calculate the actual range to read, ensuring it stays within file boundaries
let start = range.start;
let end = range.end.min(file_len);
// Ensure the start is before the end of the range
if start >= end {
return Err(io::Error::new(io::ErrorKind::InvalidInput, "Invalid range"));
}
let mut buffer = vec![0; end - start];
#[cfg(unix)]
{
use std::os::unix::prelude::FileExt;
self.file.read_exact_at(&mut buffer, start as u64)?;
}
#[cfg(not(unix))]
{
use std::io::{Read, Seek};
let mut file = self.file.try_clone()?; // Clone the file to read from it separately
// Seek to the start position in the file
file.seek(io::SeekFrom::Start(start as u64))?;
// Read the data into the buffer
file.read_exact(&mut buffer)?;
}
Ok(OwnedBytes::new(buffer))
}
// todo implement async
}
impl HasLen for WrapFile {
fn len(&self) -> usize {
self.len
}
}
#[async_trait]
impl FileHandle for &'static [u8] {
fn read_bytes(&self, range: Range<usize>) -> io::Result<OwnedBytes> {
@@ -67,6 +124,30 @@ impl fmt::Debug for FileSlice {
}
}
impl FileSlice {
pub fn stream_file_chunks(&self) -> impl Iterator<Item = io::Result<OwnedBytes>> + '_ {
let len = self.range.end;
let mut start = self.range.start;
std::iter::from_fn(move || {
/// Returns chunks of 1MB of data from the FileHandle.
const CHUNK_SIZE: usize = 1024 * 1024; // 1MB
if start < len {
let end = (start + CHUNK_SIZE).min(len);
let range = start..end;
let chunk = self.data.read_bytes(range);
start += CHUNK_SIZE;
match chunk {
Ok(chunk) => Some(Ok(chunk)),
Err(e) => Some(Err(e)),
}
} else {
None
}
})
}
}
/// Takes a range, a `RangeBounds` object, and returns
/// a `Range` that corresponds to the relative application of the
/// `RangeBounds` object to the original `Range`.

26
common/src/group_by.rs
View File

@@ -27,15 +27,15 @@ pub trait GroupByIteratorExtended: Iterator {
where
Self: Sized,
F: FnMut(&Self::Item) -> K,
K: PartialEq + Copy,
Self::Item: Copy,
K: PartialEq + Clone,
Self::Item: Clone,
{
GroupByIterator::new(self, key)
}
}
impl<I: Iterator> GroupByIteratorExtended for I {}
pub struct GroupByIterator<I, F, K: Copy>
pub struct GroupByIterator<I, F, K: Clone>
where
I: Iterator,
F: FnMut(&I::Item) -> K,
@@ -50,7 +50,7 @@ where
inner: Rc<RefCell<GroupByShared<I, F, K>>>,
}
struct GroupByShared<I, F, K: Copy>
struct GroupByShared<I, F, K: Clone>
where
I: Iterator,
F: FnMut(&I::Item) -> K,
@@ -63,7 +63,7 @@ impl<I, F, K> GroupByIterator<I, F, K>
where
I: Iterator,
F: FnMut(&I::Item) -> K,
K: Copy,
K: Clone,
{
fn new(inner: I, group_by_fn: F) -> Self {
let inner = GroupByShared {
@@ -80,28 +80,28 @@ where
impl<I, F, K> Iterator for GroupByIterator<I, F, K>
where
I: Iterator,
I::Item: Copy,
I::Item: Clone,
F: FnMut(&I::Item) -> K,
K: Copy,
K: Clone,
{
type Item = (K, GroupIterator<I, F, K>);
fn next(&mut self) -> Option<Self::Item> {
let mut inner = self.inner.borrow_mut();
let value = *inner.iter.peek()?;
let value = inner.iter.peek()?.clone();
let key = (inner.group_by_fn)(&value);
let inner = self.inner.clone();
let group_iter = GroupIterator {
inner,
group_key: key,
group_key: key.clone(),
};
Some((key, group_iter))
}
}
pub struct GroupIterator<I, F, K: Copy>
pub struct GroupIterator<I, F, K: Clone>
where
I: Iterator,
F: FnMut(&I::Item) -> K,
@@ -110,10 +110,10 @@ where
group_key: K,
}
impl<I, F, K: PartialEq + Copy> Iterator for GroupIterator<I, F, K>
impl<I, F, K: PartialEq + Clone> Iterator for GroupIterator<I, F, K>
where
I: Iterator,
I::Item: Copy,
I::Item: Clone,
F: FnMut(&I::Item) -> K,
{
type Item = I::Item;
@@ -121,7 +121,7 @@ where
fn next(&mut self) -> Option<Self::Item> {
let mut inner = self.inner.borrow_mut();
// peek if next value is in group
let peek_val = *inner.iter.peek()?;
let peek_val = inner.iter.peek()?.clone();
if (inner.group_by_fn)(&peek_val) == self.group_key {
inner.iter.next()
} else {

14
examples/basic_search.rs
View File

@@ -221,5 +221,19 @@ fn main() -> tantivy::Result<()> {
println!("{}", schema.to_json(&retrieved_doc));
}
// We can also get an explanation to understand
// how a found document got its score.
let query = query_parser.parse_query("title:sea^20 body:whale^70")?;
let (_score, doc_address) = searcher
.search(&query, &TopDocs::with_limit(1))?
.into_iter()
.next()
.unwrap();
let explanation = query.explain(&searcher, doc_address)?;
println!("{}", explanation.to_pretty_json());
Ok(())
}

2
examples/warmer.rs
View File

@@ -143,7 +143,7 @@ fn main() -> tantivy::Result<()> {
const SNEAKERS: ProductId = 23222;
let index = Index::create_in_ram(schema);
let mut writer = index.writer_with_num_threads(1, 10_000_000)?;
let mut writer = index.writer_with_num_threads(1, 15_000_000)?;
writer.add_document(doc!(product_id=>OLIVE_OIL, text=>"cooking olive oil from greece"))?;
writer.add_document(doc!(product_id=>GLOVES, text=>"kitchen gloves, perfect for cooking"))?;
writer.add_document(doc!(product_id=>SNEAKERS, text=>"uber sweet sneakers"))?;

2
ownedbytes/Cargo.toml
View File

@@ -1,7 +1,7 @@
[package]
authors = ["Paul Masurel <paul@quickwit.io>", "Pascal Seitz <pascal@quickwit.io>"]
name = "ownedbytes"
version = "0.5.0"
version = "0.6.0"
edition = "2021"
description = "Expose data as static slice"
license = "MIT"

107
ownedbytes/src/lib.rs
View File

@@ -1,7 +1,7 @@
use std::convert::TryInto;
use std::ops::{Deref, Range};
use std::sync::Arc;
use std::{fmt, io, mem};
use std::{fmt, io};
pub use stable_deref_trait::StableDeref;
@@ -26,8 +26,8 @@ impl OwnedBytes {
data_holder: T,
) -> OwnedBytes {
let box_stable_deref = Arc::new(data_holder);
let bytes: &[u8] = box_stable_deref.as_ref();
let data = unsafe { mem::transmute::<_, &'static [u8]>(bytes.deref()) };
let bytes: &[u8] = box_stable_deref.deref();
let data = unsafe { &*(bytes as *const [u8]) };
OwnedBytes {
data,
box_stable_deref,
@@ -57,6 +57,12 @@ impl OwnedBytes {
self.data.len()
}
/// Returns true iff this `OwnedBytes` is empty.
#[inline]
pub fn is_empty(&self) -> bool {
self.data.is_empty()
}
/// Splits the OwnedBytes into two OwnedBytes `(left, right)`.
///
/// Left will hold `split_len` bytes.
@@ -68,13 +74,14 @@ impl OwnedBytes {
#[inline]
#[must_use]
pub fn split(self, split_len: usize) -> (OwnedBytes, OwnedBytes) {
let (left_data, right_data) = self.data.split_at(split_len);
let right_box_stable_deref = self.box_stable_deref.clone();
let left = OwnedBytes {
data: &self.data[..split_len],
data: left_data,
box_stable_deref: self.box_stable_deref,
};
let right = OwnedBytes {
data: &self.data[split_len..],
data: right_data,
box_stable_deref: right_box_stable_deref,
};
(left, right)
@@ -99,55 +106,45 @@ impl OwnedBytes {
///
/// `self` is truncated to `split_len`, left with the remaining bytes.
pub fn split_off(&mut self, split_len: usize) -> OwnedBytes {
let (left, right) = self.data.split_at(split_len);
let right_box_stable_deref = self.box_stable_deref.clone();
let right_piece = OwnedBytes {
data: &self.data[split_len..],
data: right,
box_stable_deref: right_box_stable_deref,
};
self.data = &self.data[..split_len];
self.data = left;
right_piece
}
/// Returns true iff this `OwnedBytes` is empty.
#[inline]
pub fn is_empty(&self) -> bool {
self.as_slice().is_empty()
}
/// Drops the left most `advance_len` bytes.
#[inline]
pub fn advance(&mut self, advance_len: usize) {
self.data = &self.data[advance_len..]
pub fn advance(&mut self, advance_len: usize) -> &[u8] {
let (data, rest) = self.data.split_at(advance_len);
self.data = rest;
data
}
/// Reads an `u8` from the `OwnedBytes` and advance by one byte.
#[inline]
pub fn read_u8(&mut self) -> u8 {
assert!(!self.is_empty());
let byte = self.as_slice()[0];
self.advance(1);
byte
self.advance(1)[0]
}
/// Reads an `u64` encoded as little-endian from the `OwnedBytes` and advance by 8 bytes.
#[inline]
pub fn read_u64(&mut self) -> u64 {
assert!(self.len() > 7);
let octlet: [u8; 8] = self.as_slice()[..8].try_into().unwrap();
self.advance(8);
u64::from_le_bytes(octlet)
fn read_n<const N: usize>(&mut self) -> [u8; N] {
self.advance(N).try_into().unwrap()
}
/// Reads an `u32` encoded as little-endian from the `OwnedBytes` and advance by 4 bytes.
#[inline]
pub fn read_u32(&mut self) -> u32 {
assert!(self.len() > 3);
u32::from_le_bytes(self.read_n())
}
let quad: [u8; 4] = self.as_slice()[..4].try_into().unwrap();
self.advance(4);
u32::from_le_bytes(quad)
/// Reads an `u64` encoded as little-endian from the `OwnedBytes` and advance by 8 bytes.
#[inline]
pub fn read_u64(&mut self) -> u64 {
u64::from_le_bytes(self.read_n())
}
}
@@ -201,32 +198,33 @@ impl Deref for OwnedBytes {
}
}
impl AsRef<[u8]> for OwnedBytes {
#[inline]
fn as_ref(&self) -> &[u8] {
self.as_slice()
}
}
impl io::Read for OwnedBytes {
#[inline]
fn read(&mut self, buf: &mut [u8]) -> io::Result<usize> {
let read_len = {
let data = self.as_slice();
if data.len() >= buf.len() {
let buf_len = buf.len();
buf.copy_from_slice(&data[..buf_len]);
buf.len()
} else {
let data_len = data.len();
buf[..data_len].copy_from_slice(data);
data_len
}
};
self.advance(read_len);
Ok(read_len)
let data_len = self.data.len();
let buf_len = buf.len();
if data_len >= buf_len {
let data = self.advance(buf_len);
buf.copy_from_slice(data);
Ok(buf_len)
} else {
buf[..data_len].copy_from_slice(self.data);
self.data = &[];
Ok(data_len)
}
}
#[inline]
fn read_to_end(&mut self, buf: &mut Vec<u8>) -> io::Result<usize> {
let read_len = {
let data = self.as_slice();
buf.extend(data);
data.len()
};
self.advance(read_len);
buf.extend(self.data);
let read_len = self.data.len();
self.data = &[];
Ok(read_len)
}
#[inline]
@@ -242,13 +240,6 @@ impl io::Read for OwnedBytes {
}
}
impl AsRef<[u8]> for OwnedBytes {
#[inline]
fn as_ref(&self) -> &[u8] {
self.as_slice()
}
}
#[cfg(test)]
mod tests {
use std::io::{self, Read};

6
query-grammar/Cargo.toml
View File

@@ -1,6 +1,6 @@
[package]
name = "tantivy-query-grammar"
version = "0.20.0"
version = "0.21.0"
authors = ["Paul Masurel <paul.masurel@gmail.com>"]
license = "MIT"
categories = ["database-implementations", "data-structures"]
@@ -12,6 +12,4 @@ keywords = ["search", "information", "retrieval"]
edition = "2021"
[dependencies]
combine = {version="4", default-features=false, features=[] }
once_cell = "1.7.2"
regex ={ version = "1.5.4", default-features = false, features = ["std", "unicode"] }
nom = "7"

353
query-grammar/src/infallible.rs Normal file
View File

@@ -0,0 +1,353 @@
//! nom combinators for infallible operations
use std::convert::Infallible;
use nom::{AsChar, IResult, InputLength, InputTakeAtPosition};
pub(crate) type ErrorList = Vec<LenientErrorInternal>;
pub(crate) type JResult<I, O> = IResult<I, (O, ErrorList), Infallible>;
/// An error, with an end-of-string based offset
#[derive(Debug)]
pub(crate) struct LenientErrorInternal {
pub pos: usize,
pub message: String,
}
/// A recoverable error and the position it happened at
#[derive(Debug, PartialEq)]
pub struct LenientError {
pub pos: usize,
pub message: String,
}
impl LenientError {
pub(crate) fn from_internal(internal: LenientErrorInternal, str_len: usize) -> LenientError {
LenientError {
pos: str_len - internal.pos,
message: internal.message,
}
}
}
fn unwrap_infallible<T>(res: Result<T, nom::Err<Infallible>>) -> T {
match res {
Ok(val) => val,
Err(_) => unreachable!(),
}
}
// when rfcs#1733 get stabilized, this can make things clearer
// trait InfallibleParser<I, O> = nom::Parser<I, (O, ErrorList), std::convert::Infallible>;
/// A variant of the classical `opt` parser, except it returns an infallible error type.
///
/// It's less generic than the original to ease type resolution in the rest of the code.
pub(crate) fn opt_i<I: Clone, O, F>(mut f: F) -> impl FnMut(I) -> JResult<I, Option<O>>
where F: nom::Parser<I, O, nom::error::Error<I>> {
move |input: I| {
let i = input.clone();
match f.parse(input) {
Ok((i, o)) => Ok((i, (Some(o), Vec::new()))),
Err(_) => Ok((i, (None, Vec::new()))),
}
}
}
pub(crate) fn opt_i_err<'a, I: Clone + InputLength, O, F>(
mut f: F,
message: impl ToString + 'a,
) -> impl FnMut(I) -> JResult<I, Option<O>> + 'a
where
F: nom::Parser<I, O, nom::error::Error<I>> + 'a,
{
move |input: I| {
let i = input.clone();
match f.parse(input) {
Ok((i, o)) => Ok((i, (Some(o), Vec::new()))),
Err(_) => {
let errs = vec![LenientErrorInternal {
pos: i.input_len(),
message: message.to_string(),
}];
Ok((i, (None, errs)))
}
}
}
}
pub(crate) fn space0_infallible<T>(input: T) -> JResult<T, T>
where
T: InputTakeAtPosition + Clone,
<T as InputTakeAtPosition>::Item: AsChar + Clone,
{
opt_i(nom::character::complete::space0)(input)
.map(|(left, (spaces, errors))| (left, (spaces.expect("space0 can't fail"), errors)))
}
pub(crate) fn space1_infallible<T>(input: T) -> JResult<T, Option<T>>
where
T: InputTakeAtPosition + Clone + InputLength,
<T as InputTakeAtPosition>::Item: AsChar + Clone,
{
opt_i(nom::character::complete::space1)(input).map(|(left, (spaces, mut errors))| {
if spaces.is_none() {
errors.push(LenientErrorInternal {
pos: left.input_len(),
message: "missing space".to_string(),
})
}
(left, (spaces, errors))
})
}
pub(crate) fn fallible<I, O, E: nom::error::ParseError<I>, F>(
mut f: F,
) -> impl FnMut(I) -> IResult<I, O, E>
where F: nom::Parser<I, (O, ErrorList), Infallible> {
use nom::Err;
move |input: I| match f.parse(input) {
Ok((input, (output, _err))) => Ok((input, output)),
Err(Err::Incomplete(needed)) => Err(Err::Incomplete(needed)),
Err(Err::Error(val)) | Err(Err::Failure(val)) => match val {},
}
}
pub(crate) fn delimited_infallible<I, O1, O2, O3, F, G, H>(
mut first: F,
mut second: G,
mut third: H,
) -> impl FnMut(I) -> JResult<I, O2>
where
F: nom::Parser<I, (O1, ErrorList), Infallible>,
G: nom::Parser<I, (O2, ErrorList), Infallible>,
H: nom::Parser<I, (O3, ErrorList), Infallible>,
{
move |input: I| {
let (input, (_, mut err)) = first.parse(input)?;
let (input, (o2, mut err2)) = second.parse(input)?;
err.append(&mut err2);
let (input, (_, mut err3)) = third.parse(input)?;
err.append(&mut err3);
Ok((input, (o2, err)))
}
}
// Parse nothing. Just a lazy way to not implement terminated/preceded and use delimited instead
pub(crate) fn nothing(i: &str) -> JResult<&str, ()> {
Ok((i, ((), Vec::new())))
}
pub(crate) trait TupleInfallible<I, O> {
/// Parses the input and returns a tuple of results of each parser.
fn parse(&mut self, input: I) -> JResult<I, O>;
}
impl<Input, Output, F: nom::Parser<Input, (Output, ErrorList), Infallible>>
TupleInfallible<Input, (Output,)> for (F,)
{
fn parse(&mut self, input: Input) -> JResult<Input, (Output,)> {
self.0.parse(input).map(|(i, (o, e))| (i, ((o,), e)))
}
}
// these macros are heavily copied from nom, with some minor adaptations for our type
macro_rules! tuple_trait(
($name1:ident $ty1:ident, $name2: ident $ty2:ident, $($name:ident $ty:ident),*) => (
tuple_trait!(__impl $name1 $ty1, $name2 $ty2; $($name $ty),*);
);
(__impl $($name:ident $ty: ident),+; $name1:ident $ty1:ident, $($name2:ident $ty2:ident),*) => (
tuple_trait_impl!($($name $ty),+);
tuple_trait!(__impl $($name $ty),+ , $name1 $ty1; $($name2 $ty2),*);
);
(__impl $($name:ident $ty: ident),+; $name1:ident $ty1:ident) => (
tuple_trait_impl!($($name $ty),+);
tuple_trait_impl!($($name $ty),+, $name1 $ty1);
);
);
macro_rules! tuple_trait_impl(
($($name:ident $ty: ident),+) => (
impl<
Input: Clone, $($ty),+ ,
$($name: nom::Parser<Input, ($ty, ErrorList), Infallible>),+
> TupleInfallible<Input, ( $($ty),+ )> for ( $($name),+ ) {
fn parse(&mut self, input: Input) -> JResult<Input, ( $($ty),+ )> {
let mut error_list = Vec::new();
tuple_trait_inner!(0, self, input, (), error_list, $($name)+)
}
}
);
);
macro_rules! tuple_trait_inner(
($it:tt, $self:expr, $input:expr, (), $error_list:expr, $head:ident $($id:ident)+) => ({
let (i, (o, mut err)) = $self.$it.parse($input.clone())?;
$error_list.append(&mut err);
succ!($it, tuple_trait_inner!($self, i, ( o ), $error_list, $($id)+))
});
($it:tt, $self:expr, $input:expr, ($($parsed:tt)*), $error_list:expr, $head:ident $($id:ident)+) => ({
let (i, (o, mut err)) = $self.$it.parse($input.clone())?;
$error_list.append(&mut err);
succ!($it, tuple_trait_inner!($self, i, ($($parsed)* , o), $error_list, $($id)+))
});
($it:tt, $self:expr, $input:expr, ($($parsed:tt)*), $error_list:expr, $head:ident) => ({
let (i, (o, mut err)) = $self.$it.parse($input.clone())?;
$error_list.append(&mut err);
Ok((i, (($($parsed)* , o), $error_list)))
});
);
macro_rules! succ (
(0, $submac:ident ! ($($rest:tt)*)) => ($submac!(1, $($rest)*));
(1, $submac:ident ! ($($rest:tt)*)) => ($submac!(2, $($rest)*));
(2, $submac:ident ! ($($rest:tt)*)) => ($submac!(3, $($rest)*));
(3, $submac:ident ! ($($rest:tt)*)) => ($submac!(4, $($rest)*));
(4, $submac:ident ! ($($rest:tt)*)) => ($submac!(5, $($rest)*));
(5, $submac:ident ! ($($rest:tt)*)) => ($submac!(6, $($rest)*));
(6, $submac:ident ! ($($rest:tt)*)) => ($submac!(7, $($rest)*));
(7, $submac:ident ! ($($rest:tt)*)) => ($submac!(8, $($rest)*));
(8, $submac:ident ! ($($rest:tt)*)) => ($submac!(9, $($rest)*));
(9, $submac:ident ! ($($rest:tt)*)) => ($submac!(10, $($rest)*));
(10, $submac:ident ! ($($rest:tt)*)) => ($submac!(11, $($rest)*));
(11, $submac:ident ! ($($rest:tt)*)) => ($submac!(12, $($rest)*));
(12, $submac:ident ! ($($rest:tt)*)) => ($submac!(13, $($rest)*));
(13, $submac:ident ! ($($rest:tt)*)) => ($submac!(14, $($rest)*));
(14, $submac:ident ! ($($rest:tt)*)) => ($submac!(15, $($rest)*));
(15, $submac:ident ! ($($rest:tt)*)) => ($submac!(16, $($rest)*));
(16, $submac:ident ! ($($rest:tt)*)) => ($submac!(17, $($rest)*));
(17, $submac:ident ! ($($rest:tt)*)) => ($submac!(18, $($rest)*));
(18, $submac:ident ! ($($rest:tt)*)) => ($submac!(19, $($rest)*));
(19, $submac:ident ! ($($rest:tt)*)) => ($submac!(20, $($rest)*));
(20, $submac:ident ! ($($rest:tt)*)) => ($submac!(21, $($rest)*));
);
tuple_trait!(FnA A, FnB B, FnC C, FnD D, FnE E, FnF F, FnG G, FnH H, FnI I, FnJ J, FnK K, FnL L,
FnM M, FnN N, FnO O, FnP P, FnQ Q, FnR R, FnS S, FnT T, FnU U);
// Special case: implement `TupleInfallible` for `()`, the unit type.
// This can come up in macros which accept a variable number of arguments.
// Literally, `()` is an empty tuple, so it should simply parse nothing.
impl<I> TupleInfallible<I, ()> for () {
fn parse(&mut self, input: I) -> JResult<I, ()> {
Ok((input, ((), Vec::new())))
}
}
pub(crate) fn tuple_infallible<I, O, List: TupleInfallible<I, O>>(
mut l: List,
) -> impl FnMut(I) -> JResult<I, O> {
move |i: I| l.parse(i)
}
pub(crate) fn separated_list_infallible<I, O, O2, F, G>(
mut sep: G,
mut f: F,
) -> impl FnMut(I) -> JResult<I, Vec<O>>
where
I: Clone + InputLength,
F: nom::Parser<I, (O, ErrorList), Infallible>,
G: nom::Parser<I, (O2, ErrorList), Infallible>,
{
move |i: I| {
let mut res: Vec<O> = Vec::new();
let mut errors: ErrorList = Vec::new();
let (mut i, (o, mut err)) = unwrap_infallible(f.parse(i.clone()));
errors.append(&mut err);
res.push(o);
loop {
let (i_sep_parsed, (_, mut err_sep)) = unwrap_infallible(sep.parse(i.clone()));
let len_before = i_sep_parsed.input_len();
let (i_elem_parsed, (o, mut err_elem)) =
unwrap_infallible(f.parse(i_sep_parsed.clone()));
// infinite loop check: the parser must always consume
// if we consumed nothing here, don't produce an element.
if i_elem_parsed.input_len() == len_before {
return Ok((i, (res, errors)));
}
res.push(o);
errors.append(&mut err_sep);
errors.append(&mut err_elem);
i = i_elem_parsed;
}
}
}
pub(crate) trait Alt<I, O> {
/// Tests each parser in the tuple and returns the result of the first one that succeeds
fn choice(&mut self, input: I) -> Option<JResult<I, O>>;
}
macro_rules! alt_trait(
($first_cond:ident $first:ident, $($id_cond:ident $id: ident),+) => (
alt_trait!(__impl $first_cond $first; $($id_cond $id),+);
);
(__impl $($current_cond:ident $current:ident),*; $head_cond:ident $head:ident, $($id_cond:ident $id:ident),+) => (
alt_trait_impl!($($current_cond $current),*);
alt_trait!(__impl $($current_cond $current,)* $head_cond $head; $($id_cond $id),+);
);
(__impl $($current_cond:ident $current:ident),*; $head_cond:ident $head:ident) => (
alt_trait_impl!($($current_cond $current),*);
alt_trait_impl!($($current_cond $current,)* $head_cond $head);
);
);
macro_rules! alt_trait_impl(
($($id_cond:ident $id:ident),+) => (
impl<
Input: Clone, Output,
$(
// () are to make things easier on me, but I'm not entirely sure whether we can do better
// with rule E0207
$id_cond: nom::Parser<Input, (), ()>,
$id: nom::Parser<Input, (Output, ErrorList), Infallible>
),+
> Alt<Input, Output> for ( $(($id_cond, $id),)+ ) {
fn choice(&mut self, input: Input) -> Option<JResult<Input, Output>> {
match self.0.0.parse(input.clone()) {
Err(_) => alt_trait_inner!(1, self, input, $($id_cond $id),+),
Ok((input_left, _)) => Some(self.0.1.parse(input_left)),
}
}
}
);
);
macro_rules! alt_trait_inner(
($it:tt, $self:expr, $input:expr, $head_cond:ident $head:ident, $($id_cond:ident $id:ident),+) => (
match $self.$it.0.parse($input.clone()) {
Err(_) => succ!($it, alt_trait_inner!($self, $input, $($id_cond $id),+)),
Ok((input_left, _)) => Some($self.$it.1.parse(input_left)),
}
);
($it:tt, $self:expr, $input:expr, $head_cond:ident $head:ident) => (
None
);
);
alt_trait!(A1 A, B1 B, C1 C, D1 D, E1 E, F1 F, G1 G, H1 H, I1 I, J1 J, K1 K,
L1 L, M1 M, N1 N, O1 O, P1 P, Q1 Q, R1 R, S1 S, T1 T, U1 U);
/// An alt() like combinator. For each branch, it first tries a fallible parser, which commits to
/// this branch, or tells to check next branch, and the execute the infallible parser which follow.
///
/// In case no branch match, the default (fallible) parser is executed.
pub(crate) fn alt_infallible<I: Clone, O, F, List: Alt<I, O>>(
mut l: List,
mut default: F,
) -> impl FnMut(I) -> JResult<I, O>
where
F: nom::Parser<I, (O, ErrorList), Infallible>,
{
move |i: I| l.choice(i.clone()).unwrap_or_else(|| default.parse(i))
}

13
query-grammar/src/lib.rs
View File

@@ -1,19 +1,26 @@
#![allow(clippy::derive_partial_eq_without_eq)]
mod infallible;
mod occur;
mod query_grammar;
mod user_input_ast;
use combine::parser::Parser;
pub use crate::infallible::LenientError;
pub use crate::occur::Occur;
use crate::query_grammar::parse_to_ast;
use crate::query_grammar::{parse_to_ast, parse_to_ast_lenient};
pub use crate::user_input_ast::{
Delimiter, UserInputAst, UserInputBound, UserInputLeaf, UserInputLiteral,
};
pub struct Error;
/// Parse a query
pub fn parse_query(query: &str) -> Result<UserInputAst, Error> {
let (user_input_ast, _remaining) = parse_to_ast().parse(query).map_err(|_| Error)?;
let (_remaining, user_input_ast) = parse_to_ast(query).map_err(|_| Error)?;
Ok(user_input_ast)
}
/// Parse a query, trying to recover from syntax errors, and giving hints toward fixing errors.
pub fn parse_query_lenient(query: &str) -> (UserInputAst, Vec<LenientError>) {
parse_to_ast_lenient(query)
}

1575
query-grammar/src/query_grammar.rs
View File

File diff suppressed because it is too large Load Diff

48
query-grammar/src/user_input_ast.rs
View File

@@ -3,7 +3,7 @@ use std::fmt::{Debug, Formatter};
use crate::Occur;
#[derive(PartialEq)]
#[derive(PartialEq, Clone)]
pub enum UserInputLeaf {
Literal(UserInputLiteral),
All,
@@ -16,6 +16,34 @@ pub enum UserInputLeaf {
field: Option<String>,
elements: Vec<String>,
},
Exists {
field: String,
},
}
impl UserInputLeaf {
pub(crate) fn set_field(self, field: Option<String>) -> Self {
match self {
UserInputLeaf::Literal(mut literal) => {
literal.field_name = field;
UserInputLeaf::Literal(literal)
}
UserInputLeaf::All => UserInputLeaf::All,
UserInputLeaf::Range {
field: _,
lower,
upper,
} => UserInputLeaf::Range {
field,
lower,
upper,
},
UserInputLeaf::Set { field: _, elements } => UserInputLeaf::Set { field, elements },
UserInputLeaf::Exists { field: _ } => UserInputLeaf::Exists {
field: field.expect("Exist query without a field isn't allowed"),
},
}
}
}
impl Debug for UserInputLeaf {
@@ -28,6 +56,7 @@ impl Debug for UserInputLeaf {
ref upper,
} => {
if let Some(ref field) = field {
// TODO properly escape field (in case of \")
write!(formatter, "\"{field}\":")?;
}
lower.display_lower(formatter)?;
@@ -37,6 +66,7 @@ impl Debug for UserInputLeaf {
}
UserInputLeaf::Set { field, elements } => {
if let Some(ref field) = field {
// TODO properly escape field (in case of \")
write!(formatter, "\"{field}\": ")?;
}
write!(formatter, "IN [")?;
@@ -44,11 +74,15 @@ impl Debug for UserInputLeaf {
if i != 0 {
write!(formatter, " ")?;
}
// TODO properly escape element
write!(formatter, "\"{text}\"")?;
}
write!(formatter, "]")
}
UserInputLeaf::All => write!(formatter, "*"),
UserInputLeaf::Exists { field } => {
write!(formatter, "\"{field}\":*")
}
}
}
}
@@ -60,7 +94,7 @@ pub enum Delimiter {
None,
}
#[derive(PartialEq)]
#[derive(PartialEq, Clone)]
pub struct UserInputLiteral {
pub field_name: Option<String>,
pub phrase: String,
@@ -72,16 +106,20 @@ pub struct UserInputLiteral {
impl fmt::Debug for UserInputLiteral {
fn fmt(&self, formatter: &mut fmt::Formatter) -> Result<(), fmt::Error> {
if let Some(ref field) = self.field_name {
// TODO properly escape field (in case of \")
write!(formatter, "\"{field}\":")?;
}
match self.delimiter {
Delimiter::SingleQuotes => {
// TODO properly escape element (in case of \')
write!(formatter, "'{}'", self.phrase)?;
}
Delimiter::DoubleQuotes => {
// TODO properly escape element (in case of \")
write!(formatter, "\"{}\"", self.phrase)?;
}
Delimiter::None => {
// TODO properly escape element
write!(formatter, "{}", self.phrase)?;
}
}
@@ -94,7 +132,7 @@ impl fmt::Debug for UserInputLiteral {
}
}
#[derive(PartialEq)]
#[derive(PartialEq, Debug, Clone)]
pub enum UserInputBound {
Inclusive(String),
Exclusive(String),
@@ -104,6 +142,7 @@ pub enum UserInputBound {
impl UserInputBound {
fn display_lower(&self, formatter: &mut fmt::Formatter) -> Result<(), fmt::Error> {
match *self {
// TODO properly escape word if required
UserInputBound::Inclusive(ref word) => write!(formatter, "[\"{word}\""),
UserInputBound::Exclusive(ref word) => write!(formatter, "{{\"{word}\""),
UserInputBound::Unbounded => write!(formatter, "{{\"*\""),
@@ -112,6 +151,7 @@ impl UserInputBound {
fn display_upper(&self, formatter: &mut fmt::Formatter) -> Result<(), fmt::Error> {
match *self {
// TODO properly escape word if required
UserInputBound::Inclusive(ref word) => write!(formatter, "\"{word}\"]"),
UserInputBound::Exclusive(ref word) => write!(formatter, "\"{word}\"}}"),
UserInputBound::Unbounded => write!(formatter, "\"*\"}}"),
@@ -127,6 +167,7 @@ impl UserInputBound {
}
}
#[derive(PartialEq, Clone)]
pub enum UserInputAst {
Clause(Vec<(Option<Occur>, UserInputAst)>),
Leaf(Box<UserInputLeaf>),
@@ -196,6 +237,7 @@ impl fmt::Debug for UserInputAst {
match *self {
UserInputAst::Clause(ref subqueries) => {
if subqueries.is_empty() {
// TODO this will break ast reserialization, is writing "( )" enought?
write!(formatter, "<emptyclause>")?;
} else {
write!(formatter, "(")?;

139
src/aggregation/agg_limits.rs
View File

@@ -134,3 +134,142 @@ impl Drop for ResourceLimitGuard {
.fetch_sub(self.allocated_with_the_guard, Ordering::Relaxed);
}
}
#[cfg(test)]
mod tests {
use crate::aggregation::tests::exec_request_with_query;
// https://github.com/quickwit-oss/quickwit/issues/3837
#[test]
fn test_agg_limits_with_empty_merge() {
use crate::aggregation::agg_req::Aggregations;
use crate::aggregation::bucket::tests::get_test_index_from_docs;
let docs = vec![
vec![r#"{ "date": "2015-01-02T00:00:00Z", "text": "bbb", "text2": "bbb" }"#],
vec![r#"{ "text": "aaa", "text2": "bbb" }"#],
];
let index = get_test_index_from_docs(false, &docs).unwrap();
{
let elasticsearch_compatible_json = json!(
{
"1": {
"terms": {"field": "text2", "min_doc_count": 0},
"aggs": {
"2":{
"date_histogram": {
"field": "date",
"fixed_interval": "1d",
"extended_bounds": {
"min": "2015-01-01T00:00:00Z",
"max": "2015-01-10T00:00:00Z"
}
}
}
}
}
}
);
let agg_req: Aggregations = serde_json::from_str(
&serde_json::to_string(&elasticsearch_compatible_json).unwrap(),
)
.unwrap();
let res = exec_request_with_query(agg_req, &index, Some(("text", "bbb"))).unwrap();
let expected_res = json!({
"1": {
"buckets": [
{
"2": {
"buckets": [
{ "doc_count": 0, "key": 1420070400000.0, "key_as_string": "2015-01-01T00:00:00Z" },
{ "doc_count": 1, "key": 1420156800000.0, "key_as_string": "2015-01-02T00:00:00Z" },
{ "doc_count": 0, "key": 1420243200000.0, "key_as_string": "2015-01-03T00:00:00Z" },
{ "doc_count": 0, "key": 1420329600000.0, "key_as_string": "2015-01-04T00:00:00Z" },
{ "doc_count": 0, "key": 1420416000000.0, "key_as_string": "2015-01-05T00:00:00Z" },
{ "doc_count": 0, "key": 1420502400000.0, "key_as_string": "2015-01-06T00:00:00Z" },
{ "doc_count": 0, "key": 1420588800000.0, "key_as_string": "2015-01-07T00:00:00Z" },
{ "doc_count": 0, "key": 1420675200000.0, "key_as_string": "2015-01-08T00:00:00Z" },
{ "doc_count": 0, "key": 1420761600000.0, "key_as_string": "2015-01-09T00:00:00Z" },
{ "doc_count": 0, "key": 1420848000000.0, "key_as_string": "2015-01-10T00:00:00Z" }
]
},
"doc_count": 1,
"key": "bbb"
}
],
"doc_count_error_upper_bound": 0,
"sum_other_doc_count": 0
}
});
assert_eq!(res, expected_res);
}
}
// https://github.com/quickwit-oss/quickwit/issues/3837
#[test]
fn test_agg_limits_with_empty_data() {
use crate::aggregation::agg_req::Aggregations;
use crate::aggregation::bucket::tests::get_test_index_from_docs;
let docs = vec![vec![r#"{ "text": "aaa", "text2": "bbb" }"#]];
let index = get_test_index_from_docs(false, &docs).unwrap();
{
// Empty result since there is no doc with dates
let elasticsearch_compatible_json = json!(
{
"1": {
"terms": {"field": "text2", "min_doc_count": 0},
"aggs": {
"2":{
"date_histogram": {
"field": "date",
"fixed_interval": "1d",
"extended_bounds": {
"min": "2015-01-01T00:00:00Z",
"max": "2015-01-10T00:00:00Z"
}
}
}
}
}
}
);
let agg_req: Aggregations = serde_json::from_str(
&serde_json::to_string(&elasticsearch_compatible_json).unwrap(),
)
.unwrap();
let res = exec_request_with_query(agg_req, &index, Some(("text", "bbb"))).unwrap();
let expected_res = json!({
"1": {
"buckets": [
{
"2": {
"buckets": [
{ "doc_count": 0, "key": 1420070400000.0, "key_as_string": "2015-01-01T00:00:00Z" },
{ "doc_count": 0, "key": 1420156800000.0, "key_as_string": "2015-01-02T00:00:00Z" },
{ "doc_count": 0, "key": 1420243200000.0, "key_as_string": "2015-01-03T00:00:00Z" },
{ "doc_count": 0, "key": 1420329600000.0, "key_as_string": "2015-01-04T00:00:00Z" },
{ "doc_count": 0, "key": 1420416000000.0, "key_as_string": "2015-01-05T00:00:00Z" },
{ "doc_count": 0, "key": 1420502400000.0, "key_as_string": "2015-01-06T00:00:00Z" },
{ "doc_count": 0, "key": 1420588800000.0, "key_as_string": "2015-01-07T00:00:00Z" },
{ "doc_count": 0, "key": 1420675200000.0, "key_as_string": "2015-01-08T00:00:00Z" },
{ "doc_count": 0, "key": 1420761600000.0, "key_as_string": "2015-01-09T00:00:00Z" },
{ "doc_count": 0, "key": 1420848000000.0, "key_as_string": "2015-01-10T00:00:00Z" }
]
},
"doc_count": 0,
"key": "bbb"
}
],
"doc_count_error_upper_bound": 0,
"sum_other_doc_count": 0
}
});
assert_eq!(res, expected_res);
}
}
}

36
src/aggregation/agg_req.rs
View File

@@ -44,22 +44,49 @@ use super::metric::{
/// The key is the user defined name of the aggregation.
pub type Aggregations = HashMap<String, Aggregation>;
#[derive(Clone, Debug, PartialEq, Serialize, Deserialize)]
/// Aggregation request.
///
/// An aggregation is either a bucket or a metric.
#[derive(Clone, Debug, PartialEq, Serialize, Deserialize)]
#[serde(try_from = "AggregationForDeserialization")]
pub struct Aggregation {
/// The aggregation variant, which can be either a bucket or a metric.
#[serde(flatten)]
pub agg: AggregationVariants,
/// The sub_aggregations, only valid for bucket type aggregations. Each bucket will aggregate
/// on the document set in the bucket.
#[serde(rename = "aggs")]
#[serde(default)]
#[serde(skip_serializing_if = "Aggregations::is_empty")]
pub sub_aggregation: Aggregations,
}
/// In order to display proper error message, we cannot rely on flattening
/// the json enum. Instead we introduce an intermediary struct to separate
/// the aggregation from the subaggregation.
#[derive(Deserialize)]
struct AggregationForDeserialization {
#[serde(flatten)]
pub aggs_remaining_json: serde_json::Value,
#[serde(rename = "aggs")]
#[serde(default)]
pub sub_aggregation: Aggregations,
}
impl TryFrom<AggregationForDeserialization> for Aggregation {
type Error = serde_json::Error;
fn try_from(value: AggregationForDeserialization) -> serde_json::Result<Self> {
let AggregationForDeserialization {
aggs_remaining_json,
sub_aggregation,
} = value;
let agg: AggregationVariants = serde_json::from_value(aggs_remaining_json)?;
Ok(Aggregation {
agg,
sub_aggregation,
})
}
}
impl Aggregation {
pub(crate) fn sub_aggregation(&self) -> &Aggregations {
&self.sub_aggregation
@@ -123,7 +150,8 @@ pub enum AggregationVariants {
}
impl AggregationVariants {
fn get_fast_field_name(&self) -> &str {
/// Returns the name of the field used by the aggregation.
pub fn get_fast_field_name(&self) -> &str {
match self {
AggregationVariants::Terms(terms) => terms.field.as_str(),
AggregationVariants::Range(range) => range.field.as_str(),

263
src/aggregation/agg_req_with_accessor.rs
View File

@@ -13,6 +13,7 @@ use super::metric::{
};
use super::segment_agg_result::AggregationLimits;
use super::VecWithNames;
use crate::aggregation::{f64_to_fastfield_u64, Key};
use crate::SegmentReader;
#[derive(Default)]
@@ -35,96 +36,242 @@ pub struct AggregationWithAccessor {
/// based on search terms. That is not that case currently, but eventually this needs to be
/// Option or moved.
pub(crate) accessor: Column<u64>,
/// Load insert u64 for missing use case
pub(crate) missing_value_for_accessor: Option<u64>,
pub(crate) str_dict_column: Option<StrColumn>,
pub(crate) field_type: ColumnType,
/// In case there are multiple types of fast fields, e.g. string and numeric.
/// Only used for term aggregations currently.
pub(crate) accessor2: Option<(Column<u64>, ColumnType)>,
pub(crate) sub_aggregation: AggregationsWithAccessor,
pub(crate) limits: ResourceLimitGuard,
pub(crate) column_block_accessor: ColumnBlockAccessor<u64>,
/// Used for missing term aggregation, which checks all columns for existence.
/// By convention the missing aggregation is chosen, when this property is set
/// (instead bein set in `agg`).
/// If this needs to used by other aggregations, we need to refactor this.
pub(crate) accessors: Vec<Column<u64>>,
pub(crate) agg: Aggregation,
}
impl AggregationWithAccessor {
/// May return multiple accessors if the aggregation is e.g. on mixed field types.
fn try_from_agg(
agg: &Aggregation,
sub_aggregation: &Aggregations,
reader: &SegmentReader,
limits: AggregationLimits,
) -> crate::Result<AggregationWithAccessor> {
let mut str_dict_column = None;
let mut accessor2 = None;
) -> crate::Result<Vec<AggregationWithAccessor>> {
let add_agg_with_accessor = |accessor: Column<u64>,
column_type: ColumnType,
aggs: &mut Vec<AggregationWithAccessor>|
-> crate::Result<()> {
let res = AggregationWithAccessor {
accessor,
accessors: Vec::new(),
field_type: column_type,
sub_aggregation: get_aggs_with_segment_accessor_and_validate(
sub_aggregation,
reader,
&limits,
)?,
agg: agg.clone(),
limits: limits.new_guard(),
missing_value_for_accessor: None,
str_dict_column: None,
column_block_accessor: Default::default(),
};
aggs.push(res);
Ok(())
};
let mut res: Vec<AggregationWithAccessor> = Vec::new();
use AggregationVariants::*;
let (accessor, field_type) = match &agg.agg {
match &agg.agg {
Range(RangeAggregation {
field: field_name, ..
}) => get_ff_reader(reader, field_name, Some(get_numeric_or_date_column_types()))?,
}) => {
let (accessor, column_type) =
get_ff_reader(reader, field_name, Some(get_numeric_or_date_column_types()))?;
add_agg_with_accessor(accessor, column_type, &mut res)?;
}
Histogram(HistogramAggregation {
field: field_name, ..
}) => get_ff_reader(reader, field_name, Some(get_numeric_or_date_column_types()))?,
}) => {
let (accessor, column_type) =
get_ff_reader(reader, field_name, Some(get_numeric_or_date_column_types()))?;
add_agg_with_accessor(accessor, column_type, &mut res)?;
}
DateHistogram(DateHistogramAggregationReq {
field: field_name, ..
}) => get_ff_reader(reader, field_name, Some(get_numeric_or_date_column_types()))?,
Terms(TermsAggregation {
field: field_name, ..
}) => {
str_dict_column = reader.fast_fields().str(field_name)?;
let (accessor, column_type) =
// Only DateTime is supported for DateHistogram
get_ff_reader(reader, field_name, Some(&[ColumnType::DateTime]))?;
add_agg_with_accessor(accessor, column_type, &mut res)?;
}
Terms(TermsAggregation {
field: field_name,
missing,
..
}) => {
let str_dict_column = reader.fast_fields().str(field_name)?;
let allowed_column_types = [
ColumnType::I64,
ColumnType::U64,
ColumnType::F64,
ColumnType::Str,
ColumnType::DateTime,
// ColumnType::Bytes Unsupported
// ColumnType::Bool Unsupported
// ColumnType::IpAddr Unsupported
// ColumnType::DateTime Unsupported
];
let mut columns =
get_all_ff_reader_or_empty(reader, field_name, Some(&allowed_column_types))?;
let first = columns.pop().unwrap();
accessor2 = columns.pop();
first
}
Average(AverageAggregation { field: field_name })
| Count(CountAggregation { field: field_name })
| Max(MaxAggregation { field: field_name })
| Min(MinAggregation { field: field_name })
| Stats(StatsAggregation { field: field_name })
| Sum(SumAggregation { field: field_name }) => {
let (accessor, field_type) =
get_ff_reader(reader, field_name, Some(get_numeric_or_date_column_types()))?;
(accessor, field_type)
// In case the column is empty we want the shim column to match the missing type
let fallback_type = missing
.as_ref()
.map(|missing| match missing {
Key::Str(_) => ColumnType::Str,
Key::F64(_) => ColumnType::F64,
})
.unwrap_or(ColumnType::U64);
let column_and_types = get_all_ff_reader_or_empty(
reader,
field_name,
Some(&allowed_column_types),
fallback_type,
)?;
let missing_and_more_than_one_col = column_and_types.len() > 1 && missing.is_some();
let text_on_non_text_col = column_and_types.len() == 1
&& column_and_types[0].1.numerical_type().is_some()
&& missing
.as_ref()
.map(|m| matches!(m, Key::Str(_)))
.unwrap_or(false);
// Actually we could convert the text to a number and have the fast path, if it is
// provided in Rfc3339 format. But this use case is probably common
// enough to justify the effort.
let text_on_date_col = column_and_types.len() == 1
&& column_and_types[0].1 == ColumnType::DateTime
&& missing
.as_ref()
.map(|m| matches!(m, Key::Str(_)))
.unwrap_or(false);
let use_special_missing_agg =
missing_and_more_than_one_col || text_on_non_text_col || text_on_date_col;
if use_special_missing_agg {
let column_and_types =
get_all_ff_reader_or_empty(reader, field_name, None, fallback_type)?;
let accessors: Vec<Column> =
column_and_types.iter().map(|(a, _)| a.clone()).collect();
let agg_wit_acc = AggregationWithAccessor {
missing_value_for_accessor: None,
accessor: accessors[0].clone(),
accessors,
field_type: ColumnType::U64,
sub_aggregation: get_aggs_with_segment_accessor_and_validate(
sub_aggregation,
reader,
&limits,
)?,
agg: agg.clone(),
str_dict_column: str_dict_column.clone(),
limits: limits.new_guard(),
column_block_accessor: Default::default(),
};
res.push(agg_wit_acc);
}
for (accessor, column_type) in column_and_types {
let missing_value_term_agg = if use_special_missing_agg {
None
} else {
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_name())?
} else {
None
};
let agg = AggregationWithAccessor {
missing_value_for_accessor,
accessor,
accessors: Vec::new(),
field_type: column_type,
sub_aggregation: get_aggs_with_segment_accessor_and_validate(
sub_aggregation,
reader,
&limits,
)?,
agg: agg.clone(),
str_dict_column: str_dict_column.clone(),
limits: limits.new_guard(),
column_block_accessor: Default::default(),
};
res.push(agg);
}
}
Average(AverageAggregation {
field: field_name, ..
})
| Count(CountAggregation {
field: field_name, ..
})
| Max(MaxAggregation {
field: field_name, ..
})
| Min(MinAggregation {
field: field_name, ..
})
| Stats(StatsAggregation {
field: field_name, ..
})
| Sum(SumAggregation {
field: field_name, ..
}) => {
let (accessor, column_type) =
get_ff_reader(reader, field_name, Some(get_numeric_or_date_column_types()))?;
add_agg_with_accessor(accessor, column_type, &mut res)?;
}
Percentiles(percentiles) => {
let (accessor, field_type) = get_ff_reader(
let (accessor, column_type) = get_ff_reader(
reader,
percentiles.field_name(),
Some(get_numeric_or_date_column_types()),
)?;
(accessor, field_type)
add_agg_with_accessor(accessor, column_type, &mut res)?;
}
};
let sub_aggregation = sub_aggregation.clone();
Ok(AggregationWithAccessor {
accessor,
accessor2,
field_type,
sub_aggregation: get_aggs_with_segment_accessor_and_validate(
&sub_aggregation,
reader,
&limits,
)?,
agg: agg.clone(),
str_dict_column,
limits: limits.new_guard(),
column_block_accessor: Default::default(),
})
Ok(res)
}
}
fn get_missing_val(
column_type: ColumnType,
missing: &Key,
field_name: &str,
) -> crate::Result<Option<u64>> {
let missing_val = match missing {
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::F64(val) if column_type.numerical_type().is_some() => {
f64_to_fastfield_u64(*val, &column_type)
}
_ => {
return Err(crate::TantivyError::InvalidArgument(format!(
"Missing value {:?} for field {} is not supported for column type {:?}",
missing, field_name, column_type
)));
}
};
Ok(missing_val)
}
fn get_numeric_or_date_column_types() -> &'static [ColumnType] {
&[
ColumnType::F64,
@@ -141,15 +288,15 @@ pub(crate) fn get_aggs_with_segment_accessor_and_validate(
) -> crate::Result<AggregationsWithAccessor> {
let mut aggss = Vec::new();
for (key, agg) in aggs.iter() {
aggss.push((
key.to_string(),
AggregationWithAccessor::try_from_agg(
agg,
agg.sub_aggregation(),
reader,
limits.clone(),
)?,
));
let aggs = AggregationWithAccessor::try_from_agg(
agg,
agg.sub_aggregation(),
reader,
limits.clone(),
)?;
for agg in aggs {
aggss.push((key.to_string(), agg));
}
}
Ok(AggregationsWithAccessor::from_data(
VecWithNames::from_entries(aggss),
@@ -181,15 +328,13 @@ fn get_all_ff_reader_or_empty(
reader: &SegmentReader,
field_name: &str,
allowed_column_types: Option<&[ColumnType]>,
fallback_type: ColumnType,
) -> crate::Result<Vec<(columnar::Column<u64>, ColumnType)>> {
let ff_fields = reader.fast_fields();
let mut ff_field_with_type =
ff_fields.u64_lenient_for_type_all(allowed_column_types, field_name)?;
if ff_field_with_type.is_empty() {
ff_field_with_type.push((
Column::build_empty_column(reader.num_docs()),
ColumnType::U64,
));
ff_field_with_type.push((Column::build_empty_column(reader.num_docs()), fallback_type));
}
Ok(ff_field_with_type)
}

8
src/aggregation/agg_tests.rs
View File

@@ -558,10 +558,10 @@ fn test_aggregation_invalid_requests() -> crate::Result<()> {
assert_eq!(agg_req_1.is_err(), true);
// TODO: This should list valid values
assert_eq!(
agg_req_1.unwrap_err().to_string(),
"no variant of enum AggregationVariants found in flattened data"
);
assert!(agg_req_1
.unwrap_err()
.to_string()
.contains("unknown variant `doesnotmatchanyagg`, expected one of"));
// TODO: This should return an error
// let agg_res = avg_on_field("not_exist_field").unwrap_err();

8
src/aggregation/bucket/histogram/date_histogram.rs
View File

@@ -132,6 +132,7 @@ impl DateHistogramAggregationReq {
hard_bounds: self.hard_bounds,
extended_bounds: self.extended_bounds,
keyed: self.keyed,
is_normalized_to_ns: false,
})
}
@@ -243,14 +244,14 @@ fn parse_into_milliseconds(input: &str) -> Result<i64, AggregationError> {
}
#[cfg(test)]
mod tests {
pub mod tests {
use pretty_assertions::assert_eq;
use super::*;
use crate::aggregation::agg_req::Aggregations;
use crate::aggregation::tests::exec_request;
use crate::indexer::NoMergePolicy;
use crate::schema::{Schema, FAST};
use crate::schema::{Schema, FAST, STRING};
use crate::Index;
#[test]
@@ -306,7 +307,8 @@ mod tests {
) -> crate::Result<Index> {
let mut schema_builder = Schema::builder();
schema_builder.add_date_field("date", FAST);
schema_builder.add_text_field("text", FAST);
schema_builder.add_text_field("text", FAST | STRING);
schema_builder.add_text_field("text2", FAST | STRING);
let schema = schema_builder.build();
let index = Index::create_in_ram(schema.clone());
{

39
src/aggregation/bucket/histogram/histogram.rs
View File

@@ -122,11 +122,14 @@ pub struct HistogramAggregation {
/// Whether to return the buckets as a hash map
#[serde(default)]
pub keyed: bool,
/// Whether the values are normalized to ns for date time values. Defaults to false.
#[serde(default)]
pub is_normalized_to_ns: bool,
}
impl HistogramAggregation {
pub(crate) fn normalize(&mut self, column_type: ColumnType) {
if column_type.is_date_time() {
pub(crate) fn normalize_date_time(&mut self) {
if !self.is_normalized_to_ns {
// values are provided in ms, but the fastfield is in nano seconds
self.interval *= 1_000_000.0;
self.offset = self.offset.map(|off| off * 1_000_000.0);
@@ -138,6 +141,7 @@ impl HistogramAggregation {
min: bounds.min * 1_000_000.0,
max: bounds.max * 1_000_000.0,
});
self.is_normalized_to_ns = true;
}
}
@@ -351,6 +355,7 @@ impl SegmentHistogramCollector {
let buckets_mem = self.buckets.memory_consumption();
self_mem + sub_aggs_mem + buckets_mem
}
/// Converts the collector result into a intermediate bucket result.
pub fn into_intermediate_bucket_result(
self,
agg_with_accessor: &AggregationWithAccessor,
@@ -369,7 +374,7 @@ impl SegmentHistogramCollector {
Ok(IntermediateBucketResult::Histogram {
buckets,
column_type: Some(self.column_type),
is_date_agg: self.column_type == ColumnType::DateTime,
})
}
@@ -380,7 +385,9 @@ impl SegmentHistogramCollector {
accessor_idx: usize,
) -> crate::Result<Self> {
req.validate()?;
req.normalize(field_type);
if field_type == ColumnType::DateTime {
req.normalize_date_time();
}
let sub_aggregation_blueprint = if sub_aggregation.is_empty() {
None
@@ -438,6 +445,7 @@ fn intermediate_buckets_to_final_buckets_fill_gaps(
// memory check upfront
let (_, first_bucket_num, last_bucket_num) =
generate_bucket_pos_with_opt_minmax(histogram_req, min_max);
// It's based on user input, so we need to account for overflows
let added_buckets = ((last_bucket_num.saturating_sub(first_bucket_num)).max(0) as u64)
.saturating_sub(buckets.len() as u64);
@@ -453,15 +461,12 @@ fn intermediate_buckets_to_final_buckets_fill_gaps(
let final_buckets: Vec<BucketEntry> = buckets
.into_iter()
.merge_join_by(
fill_gaps_buckets.into_iter(),
|existing_bucket, fill_gaps_bucket| {
existing_bucket
.key
.partial_cmp(fill_gaps_bucket)
.unwrap_or(Ordering::Equal)
},
)
.merge_join_by(fill_gaps_buckets, |existing_bucket, fill_gaps_bucket| {
existing_bucket
.key
.partial_cmp(fill_gaps_bucket)
.unwrap_or(Ordering::Equal)
})
.map(|either| match either {
// Ignore the generated bucket
itertools::EitherOrBoth::Both(existing, _) => existing,
@@ -484,7 +489,7 @@ fn intermediate_buckets_to_final_buckets_fill_gaps(
// Convert to BucketEntry
pub(crate) fn intermediate_histogram_buckets_to_final_buckets(
buckets: Vec<IntermediateHistogramBucketEntry>,
column_type: Option<ColumnType>,
is_date_agg: bool,
histogram_req: &HistogramAggregation,
sub_aggregation: &Aggregations,
limits: &AggregationLimits,
@@ -493,8 +498,8 @@ pub(crate) fn intermediate_histogram_buckets_to_final_buckets(
// The request used in the the call to final is not yet be normalized.
// Normalization is changing the precision from milliseconds to nanoseconds.
let mut histogram_req = histogram_req.clone();
if let Some(column_type) = column_type {
histogram_req.normalize(column_type);
if is_date_agg {
histogram_req.normalize_date_time();
}
let mut buckets = if histogram_req.min_doc_count() == 0 {
// With min_doc_count != 0, we may need to add buckets, so that there are no
@@ -518,7 +523,7 @@ pub(crate) fn intermediate_histogram_buckets_to_final_buckets(
// If we have a date type on the histogram buckets, we add the `key_as_string` field as rfc339
// and normalize from nanoseconds to milliseconds
if column_type == Some(ColumnType::DateTime) {
if is_date_agg {
for bucket in buckets.iter_mut() {
if let crate::aggregation::Key::F64(ref mut val) = bucket.key {
let key_as_string = format_date(*val as i64)?;

4
src/aggregation/bucket/mod.rs
View File

@@ -25,15 +25,15 @@
mod histogram;
mod range;
mod term_agg;
mod term_missing_agg;
use std::collections::HashMap;
pub(crate) use histogram::SegmentHistogramCollector;
pub use histogram::*;
pub(crate) use range::SegmentRangeCollector;
pub use range::*;
use serde::{de, Deserialize, Deserializer, Serialize, Serializer};
pub use term_agg::*;
pub use term_missing_agg::*;
/// Order for buckets in a bucket aggregation.
#[derive(Clone, Copy, Debug, PartialEq, Serialize, Deserialize, Default)]

4
src/aggregation/bucket/range.rs
View File

@@ -262,7 +262,7 @@ impl SegmentRangeCollector {
pub(crate) fn from_req_and_validate(
req: &RangeAggregation,
sub_aggregation: &mut AggregationsWithAccessor,
limits: &mut ResourceLimitGuard,
limits: &ResourceLimitGuard,
field_type: ColumnType,
accessor_idx: usize,
) -> crate::Result<Self> {
@@ -465,7 +465,7 @@ mod tests {
SegmentRangeCollector::from_req_and_validate(
&req,
&mut Default::default(),
&mut AggregationLimits::default().new_guard(),
&AggregationLimits::default().new_guard(),
field_type,
0,
)

589
src/aggregation/bucket/term_agg.rs
View File

@@ -1,6 +1,6 @@
use std::fmt::Debug;
use columnar::ColumnType;
use columnar::{BytesColumn, ColumnType, MonotonicallyMappableToU64, StrColumn};
use rustc_hash::FxHashMap;
use serde::{Deserialize, Serialize};
@@ -9,7 +9,6 @@ use crate::aggregation::agg_limits::MemoryConsumption;
use crate::aggregation::agg_req_with_accessor::{
AggregationWithAccessor, AggregationsWithAccessor,
};
use crate::aggregation::f64_from_fastfield_u64;
use crate::aggregation::intermediate_agg_result::{
IntermediateAggregationResult, IntermediateAggregationResults, IntermediateBucketResult,
IntermediateKey, IntermediateTermBucketEntry, IntermediateTermBucketResult,
@@ -17,6 +16,7 @@ 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::error::DataCorruption;
use crate::TantivyError;
@@ -146,6 +146,28 @@ pub struct TermsAggregation {
/// { "average_price": "asc" }
#[serde(skip_serializing_if = "Option::is_none", default)]
pub order: Option<CustomOrder>,
/// 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:
/// { "missing": "NO_DATA" }
///
/// # Internal
///
/// Internally, `missing` requires some specialized handling in some scenarios.
///
/// Simple Case:
/// In the simplest case, we can just put the missing value in the termmap use that. In case of
/// text we put a special u64::MAX and replace it at the end with the actual missing value,
/// 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.
/// 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.
#[serde(skip_serializing_if = "Option::is_none", default)]
pub missing: Option<Key>,
}
/// Same as TermsAggregation, but with populated defaults.
@@ -176,6 +198,7 @@ pub(crate) struct TermsAggregationInternal {
pub min_doc_count: u64,
pub order: CustomOrder,
pub missing: Option<Key>,
}
impl TermsAggregationInternal {
@@ -195,6 +218,7 @@ impl TermsAggregationInternal {
.unwrap_or_else(|| order == CustomOrder::default()),
min_doc_count: req.min_doc_count.unwrap_or(1),
order,
missing: req.missing.clone(),
}
}
}
@@ -224,110 +248,6 @@ impl TermBuckets {
}
}
/// The composite collector is used, when we have different types under one field, to support a term
/// aggregation on both.
#[derive(Clone, Debug)]
pub struct SegmentTermCollectorComposite {
term_agg1: SegmentTermCollector, // field type 1, e.g. strings
term_agg2: SegmentTermCollector, // field type 2, e.g. u64
accessor_idx: usize,
}
impl SegmentAggregationCollector for SegmentTermCollectorComposite {
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 bucket = self
.term_agg1
.into_intermediate_bucket_result(agg_with_accessor)?;
results.push(
name.to_string(),
IntermediateAggregationResult::Bucket(bucket),
)?;
let bucket = self
.term_agg2
.into_intermediate_bucket_result(agg_with_accessor)?;
results.push(name, IntermediateAggregationResult::Bucket(bucket))?;
Ok(())
}
#[inline]
fn collect(
&mut self,
doc: crate::DocId,
agg_with_accessor: &mut AggregationsWithAccessor,
) -> crate::Result<()> {
self.term_agg1.collect_block(&[doc], agg_with_accessor)?;
self.swap_accessor(&mut agg_with_accessor.aggs.values[self.accessor_idx]);
self.term_agg2.collect_block(&[doc], agg_with_accessor)?;
self.swap_accessor(&mut agg_with_accessor.aggs.values[self.accessor_idx]);
Ok(())
}
#[inline]
fn collect_block(
&mut self,
docs: &[crate::DocId],
agg_with_accessor: &mut AggregationsWithAccessor,
) -> crate::Result<()> {
self.term_agg1.collect_block(docs, agg_with_accessor)?;
self.swap_accessor(&mut agg_with_accessor.aggs.values[self.accessor_idx]);
self.term_agg2.collect_block(docs, agg_with_accessor)?;
self.swap_accessor(&mut agg_with_accessor.aggs.values[self.accessor_idx]);
Ok(())
}
fn flush(&mut self, agg_with_accessor: &mut AggregationsWithAccessor) -> crate::Result<()> {
self.term_agg1.flush(agg_with_accessor)?;
self.swap_accessor(&mut agg_with_accessor.aggs.values[self.accessor_idx]);
self.term_agg2.flush(agg_with_accessor)?;
self.swap_accessor(&mut agg_with_accessor.aggs.values[self.accessor_idx]);
Ok(())
}
}
impl SegmentTermCollectorComposite {
/// Swaps the accessor and field type with the second accessor and field type.
/// This way we can use the same code for both aggregations.
fn swap_accessor(&self, aggregations: &mut AggregationWithAccessor) {
if let Some(accessor) = aggregations.accessor2.as_mut() {
std::mem::swap(&mut accessor.0, &mut aggregations.accessor);
std::mem::swap(&mut accessor.1, &mut aggregations.field_type);
}
}
pub(crate) fn from_req_and_validate(
req: &TermsAggregation,
sub_aggregations: &mut AggregationsWithAccessor,
field_type: ColumnType,
field_type2: ColumnType,
accessor_idx: usize,
) -> crate::Result<Self> {
Ok(Self {
term_agg1: SegmentTermCollector::from_req_and_validate(
req,
sub_aggregations,
field_type,
accessor_idx,
)?,
term_agg2: SegmentTermCollector::from_req_and_validate(
req,
sub_aggregations,
field_type2,
accessor_idx,
)?,
accessor_idx,
})
}
}
/// The collector puts values from the fast field into the correct buckets and does a conversion to
/// the correct datatype.
#[derive(Clone, Debug)]
@@ -379,9 +299,16 @@ impl SegmentAggregationCollector for SegmentTermCollector {
let mem_pre = self.get_memory_consumption();
bucket_agg_accessor
.column_block_accessor
.fetch_block(docs, &bucket_agg_accessor.accessor);
if let Some(missing) = bucket_agg_accessor.missing_value_for_accessor {
bucket_agg_accessor
.column_block_accessor
.fetch_block_with_missing(docs, &bucket_agg_accessor.accessor, missing);
} else {
bucket_agg_accessor
.column_block_accessor
.fetch_block(docs, &bucket_agg_accessor.accessor);
}
for term_id in bucket_agg_accessor.column_block_accessor.iter_vals() {
let entry = self.term_buckets.entries.entry(term_id).or_default();
*entry += 1;
@@ -543,19 +470,42 @@ impl SegmentTermCollector {
let term_dict = agg_with_accessor
.str_dict_column
.as_ref()
.expect("internal error: term dictionary not found for term aggregation");
.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 {
if !term_dict.ord_to_str(term_id, &mut buffer)? {
return Err(TantivyError::InternalError(format!(
"Couldn't find term_id {term_id} in dict"
)));
}
let intermediate_entry = into_intermediate_bucket_entry(term_id, doc_count)?;
dict.insert(IntermediateKey::Str(buffer.to_string()), intermediate_entry);
// 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);
}
}
} 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"
)));
}
dict.insert(IntermediateKey::Str(buffer.to_string()), intermediate_entry);
}
}
if self.req.min_doc_count == 0 {
// TODO: Handle rev streaming for descending sorting by keys
@@ -581,6 +531,13 @@ impl SegmentTermCollector {
});
}
}
} else if self.field_type == ColumnType::DateTime {
for (val, doc_count) in entries {
let intermediate_entry = into_intermediate_bucket_entry(val, doc_count)?;
let val = i64::from_u64(val);
let date = format_date(val)?;
dict.insert(IntermediateKey::Str(date), intermediate_entry);
}
} else {
for (val, doc_count) in entries {
let intermediate_entry = into_intermediate_bucket_entry(val, doc_count)?;
@@ -633,6 +590,9 @@ pub(crate) fn cut_off_buckets<T: GetDocCount + Debug>(
#[cfg(test)]
mod tests {
use common::DateTime;
use time::{Date, Month};
use crate::aggregation::agg_req::Aggregations;
use crate::aggregation::tests::{
exec_request, exec_request_with_query, exec_request_with_query_and_memory_limit,
@@ -1321,6 +1281,7 @@ mod tests {
];
let index = get_test_index_from_terms(false, &terms_per_segment)?;
assert_eq!(index.searchable_segments().unwrap().len(), 2);
let agg_req: Aggregations = serde_json::from_value(json!({
"my_texts": {
@@ -1506,6 +1467,47 @@ mod tests {
Ok(())
}
#[test]
fn terms_empty_json() -> crate::Result<()> {
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 = index.writer_for_tests().unwrap();
// => Segment with empty json
index_writer.add_document(doc!()).unwrap();
index_writer.commit().unwrap();
// => Segment with json, but no field partially_empty
index_writer
.add_document(doc!(json => json!({"different_field": "blue"})))
.unwrap();
index_writer.commit().unwrap();
//// => Segment with field partially_empty
index_writer
.add_document(doc!(json => json!({"partially_empty": "blue"})))
.unwrap();
index_writer.add_document(doc!())?;
index_writer.commit().unwrap();
let agg_req: Aggregations = serde_json::from_value(json!({
"my_texts": {
"terms": {
"field": "json.partially_empty"
},
}
}))
.unwrap();
let res = exec_request_with_query(agg_req, &index, None)?;
assert_eq!(res["my_texts"]["buckets"][0]["key"], "blue");
assert_eq!(res["my_texts"]["buckets"][0]["doc_count"], 1);
assert_eq!(res["my_texts"]["buckets"][1], serde_json::Value::Null);
assert_eq!(res["my_texts"]["sum_other_doc_count"], 0);
assert_eq!(res["my_texts"]["doc_count_error_upper_bound"], 0);
Ok(())
}
#[test]
fn terms_aggregation_bytes() -> crate::Result<()> {
@@ -1543,4 +1545,353 @@ mod tests {
Ok(())
}
#[test]
fn terms_aggregation_missing_multi_value() -> crate::Result<()> {
let mut schema_builder = Schema::builder();
let text_field = schema_builder.add_text_field("text", FAST);
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!(
text_field => "Hello Hello",
text_field => "Hello Hello",
id_field => 1u64,
id_field => 1u64,
))?;
// Missing
index_writer.add_document(doc!())?;
index_writer.add_document(doc!(
text_field => "Hello Hello",
))?;
index_writer.add_document(doc!(
text_field => "Hello Hello",
))?;
index_writer.commit()?;
// Empty segment special case
index_writer.add_document(doc!())?;
index_writer.commit()?;
// Full segment special case
index_writer.add_document(doc!(
text_field => "Hello Hello",
id_field => 1u64,
))?;
index_writer.commit()?;
}
let agg_req: Aggregations = serde_json::from_value(json!({
"my_texts": {
"terms": {
"field": "text",
"missing": "Empty"
},
},
"my_texts2": {
"terms": {
"field": "text",
"missing": 1337
},
},
"my_ids": {
"terms": {
"field": "id",
"missing": 1337
},
}
}))
.unwrap();
let res = exec_request_with_query(agg_req, &index, None)?;
// text field
assert_eq!(res["my_texts"]["buckets"][0]["key"], "Hello Hello");
assert_eq!(res["my_texts"]["buckets"][0]["doc_count"], 5);
assert_eq!(res["my_texts"]["buckets"][1]["key"], "Empty");
assert_eq!(res["my_texts"]["buckets"][1]["doc_count"], 2);
assert_eq!(
res["my_texts"]["buckets"][2]["key"],
serde_json::Value::Null
);
// text field with numner 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);
assert_eq!(res["my_texts2"]["buckets"][1]["doc_count"], 2);
assert_eq!(
res["my_texts2"]["buckets"][2]["key"],
serde_json::Value::Null
);
assert_eq!(res["my_texts"]["sum_other_doc_count"], 0);
assert_eq!(res["my_texts"]["doc_count_error_upper_bound"], 0);
// id field
assert_eq!(res["my_ids"]["buckets"][0]["key"], 1337.0);
assert_eq!(res["my_ids"]["buckets"][0]["doc_count"], 4);
assert_eq!(res["my_ids"]["buckets"][1]["key"], 1.0);
assert_eq!(res["my_ids"]["buckets"][1]["doc_count"], 3);
assert_eq!(res["my_ids"]["buckets"][2]["key"], serde_json::Value::Null);
Ok(())
}
#[test]
fn terms_aggregation_missing_simple_id() -> 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 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 => 1u64,
))?;
// Missing
index_writer.add_document(doc!())?;
index_writer.add_document(doc!())?;
index_writer.commit()?;
}
let agg_req: Aggregations = serde_json::from_value(json!({
"my_ids": {
"terms": {
"field": "id",
"missing": 1337
},
}
}))
.unwrap();
let res = exec_request_with_query(agg_req, &index, None)?;
// id field
assert_eq!(res["my_ids"]["buckets"][0]["key"], 1337.0);
assert_eq!(res["my_ids"]["buckets"][0]["doc_count"], 2);
assert_eq!(res["my_ids"]["buckets"][1]["key"], 1.0);
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();
let text_field = schema_builder.add_text_field("text", FAST);
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!(
text_field => "Hello Hello",
id_field => 1u64,
))?;
// Missing
index_writer.add_document(doc!())?;
index_writer.add_document(doc!(
text_field => "Hello Hello",
))?;
index_writer.add_document(doc!(
text_field => "Hello Hello",
))?;
index_writer.commit()?;
// Empty segment special case
index_writer.add_document(doc!())?;
index_writer.commit()?;
// Full segment special case
index_writer.add_document(doc!(
text_field => "Hello Hello",
id_field => 1u64,
))?;
index_writer.commit()?;
}
let agg_req: Aggregations = serde_json::from_value(json!({
"my_texts": {
"terms": {
"field": "text",
"missing": "Empty"
},
},
"my_texts2": {
"terms": {
"field": "text",
"missing": 1337
},
},
"my_ids": {
"terms": {
"field": "id",
"missing": 1337
},
}
}))
.unwrap();
let res = exec_request_with_query(agg_req, &index, None)?;
// text field
assert_eq!(res["my_texts"]["buckets"][0]["key"], "Hello Hello");
assert_eq!(res["my_texts"]["buckets"][0]["doc_count"], 4);
assert_eq!(res["my_texts"]["buckets"][1]["key"], "Empty");
assert_eq!(res["my_texts"]["buckets"][1]["doc_count"], 2);
assert_eq!(
res["my_texts"]["buckets"][2]["key"],
serde_json::Value::Null
);
// text field with numner 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);
assert_eq!(res["my_texts2"]["buckets"][1]["doc_count"], 2);
assert_eq!(
res["my_texts2"]["buckets"][2]["key"],
serde_json::Value::Null
);
assert_eq!(res["my_texts"]["sum_other_doc_count"], 0);
assert_eq!(res["my_texts"]["doc_count_error_upper_bound"], 0);
// id field
assert_eq!(res["my_ids"]["buckets"][0]["key"], 1337.0);
assert_eq!(res["my_ids"]["buckets"][0]["doc_count"], 4);
assert_eq!(res["my_ids"]["buckets"][1]["key"], 1.0);
assert_eq!(res["my_ids"]["buckets"][1]["doc_count"], 2);
assert_eq!(res["my_ids"]["buckets"][2]["key"], serde_json::Value::Null);
Ok(())
}
#[test]
fn terms_aggregation_missing_empty() -> crate::Result<()> {
let mut schema_builder = Schema::builder();
schema_builder.add_text_field("text", FAST);
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));
// Empty segment special case
index_writer.add_document(doc!())?;
index_writer.commit()?;
}
let agg_req: Aggregations = serde_json::from_value(json!({
"my_texts": {
"terms": {
"field": "text",
"missing": "Empty"
},
},
"my_texts2": {
"terms": {
"field": "text",
"missing": 1337
},
},
"my_ids": {
"terms": {
"field": "id",
"missing": 1337
},
}
}))
.unwrap();
let res = exec_request_with_query(agg_req, &index, None)?;
// text field
assert_eq!(res["my_texts"]["buckets"][0]["key"], "Empty");
assert_eq!(res["my_texts"]["buckets"][0]["doc_count"], 1);
assert_eq!(
res["my_texts"]["buckets"][1]["key"],
serde_json::Value::Null
);
// text field with number as missing fallback
assert_eq!(res["my_texts2"]["buckets"][0]["key"], 1337.0);
assert_eq!(res["my_texts2"]["buckets"][0]["doc_count"], 1);
assert_eq!(
res["my_texts2"]["buckets"][1]["key"],
serde_json::Value::Null
);
assert_eq!(res["my_texts"]["sum_other_doc_count"], 0);
assert_eq!(res["my_texts"]["doc_count_error_upper_bound"], 0);
// id field
assert_eq!(res["my_ids"]["buckets"][0]["key"], 1337.0);
assert_eq!(res["my_ids"]["buckets"][0]["doc_count"], 1);
assert_eq!(res["my_ids"]["buckets"][1]["key"], serde_json::Value::Null);
Ok(())
}
#[test]
fn terms_aggregation_date() -> crate::Result<()> {
let mut schema_builder = Schema::builder();
let date_field = schema_builder.add_date_field("date_field", FAST);
let schema = schema_builder.build();
let index = Index::create_in_ram(schema);
{
let mut writer = index.writer_with_num_threads(1, 15_000_000)?;
writer.add_document(doc!(date_field=>DateTime::from_primitive(Date::from_calendar_date(1982, Month::September, 17)?.with_hms(0, 0, 0)?)))?;
writer.add_document(doc!(date_field=>DateTime::from_primitive(Date::from_calendar_date(1982, Month::September, 17)?.with_hms(0, 0, 0)?)))?;
writer.add_document(doc!(date_field=>DateTime::from_primitive(Date::from_calendar_date(1983, Month::September, 27)?.with_hms(0, 0, 0)?)))?;
writer.commit()?;
}
let agg_req: Aggregations = serde_json::from_value(json!({
"my_date": {
"terms": {
"field": "date_field"
},
}
}))
.unwrap();
let res = exec_request_with_query(agg_req, &index, None)?;
// date_field field
assert_eq!(res["my_date"]["buckets"][0]["key"], "1982-09-17T00:00:00Z");
assert_eq!(res["my_date"]["buckets"][0]["doc_count"], 2);
assert_eq!(res["my_date"]["buckets"][1]["key"], "1983-09-27T00:00:00Z");
assert_eq!(res["my_date"]["buckets"][1]["doc_count"], 1);
assert_eq!(res["my_date"]["buckets"][2]["key"], serde_json::Value::Null);
Ok(())
}
#[test]
fn terms_aggregation_date_missing() -> crate::Result<()> {
let mut schema_builder = Schema::builder();
let date_field = schema_builder.add_date_field("date_field", FAST);
let schema = schema_builder.build();
let index = Index::create_in_ram(schema);
{
let mut writer = index.writer_with_num_threads(1, 15_000_000)?;
writer.add_document(doc!(date_field=>DateTime::from_primitive(Date::from_calendar_date(1982, Month::September, 17)?.with_hms(0, 0, 0)?)))?;
writer.add_document(doc!(date_field=>DateTime::from_primitive(Date::from_calendar_date(1982, Month::September, 17)?.with_hms(0, 0, 0)?)))?;
writer.add_document(doc!(date_field=>DateTime::from_primitive(Date::from_calendar_date(1983, Month::September, 27)?.with_hms(0, 0, 0)?)))?;
writer.add_document(doc!())?;
writer.commit()?;
}
let agg_req: Aggregations = serde_json::from_value(json!({
"my_date": {
"terms": {
"field": "date_field",
"missing": "1982-09-17T00:00:00Z"
},
}
}))
.unwrap();
let res = exec_request_with_query(agg_req, &index, None)?;
// date_field field
assert_eq!(res["my_date"]["buckets"][0]["key"], "1982-09-17T00:00:00Z");
assert_eq!(res["my_date"]["buckets"][0]["doc_count"], 3);
assert_eq!(res["my_date"]["buckets"][1]["key"], "1983-09-27T00:00:00Z");
assert_eq!(res["my_date"]["buckets"][1]["doc_count"], 1);
assert_eq!(res["my_date"]["buckets"][2]["key"], serde_json::Value::Null);
Ok(())
}
}

476
src/aggregation/bucket/term_missing_agg.rs Normal file
View File

@@ -0,0 +1,476 @@
use rustc_hash::FxHashMap;
use crate::aggregation::agg_req_with_accessor::AggregationsWithAccessor;
use crate::aggregation::intermediate_agg_result::{
IntermediateAggregationResult, IntermediateAggregationResults, IntermediateBucketResult,
IntermediateKey, IntermediateTermBucketEntry, IntermediateTermBucketResult,
};
use crate::aggregation::segment_agg_result::{
build_segment_agg_collector, SegmentAggregationCollector,
};
/// The specialized missing term aggregation.
#[derive(Default, Debug, Clone)]
pub struct TermMissingAgg {
missing_count: u32,
accessor_idx: usize,
sub_agg: Option<Box<dyn SegmentAggregationCollector>>,
}
impl TermMissingAgg {
pub(crate) fn new(
accessor_idx: usize,
sub_aggregations: &mut AggregationsWithAccessor,
) -> crate::Result<Self> {
let has_sub_aggregations = !sub_aggregations.is_empty();
let sub_agg = if has_sub_aggregations {
let sub_aggregation = build_segment_agg_collector(sub_aggregations)?;
Some(sub_aggregation)
} else {
None
};
Ok(Self {
accessor_idx,
sub_agg,
..Default::default()
})
}
}
impl SegmentAggregationCollector for TermMissingAgg {
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 term_agg = agg_with_accessor
.agg
.agg
.as_term()
.expect("TermMissingAgg collector must be term agg req");
let missing = term_agg
.missing
.as_ref()
.expect("TermMissingAgg collector, but no missing found in agg req")
.clone();
let mut entries: FxHashMap<IntermediateKey, IntermediateTermBucketEntry> =
Default::default();
let mut missing_entry = IntermediateTermBucketEntry {
doc_count: self.missing_count,
sub_aggregation: Default::default(),
};
if let Some(sub_agg) = self.sub_agg {
let mut res = IntermediateAggregationResults::default();
sub_agg.add_intermediate_aggregation_result(
&agg_with_accessor.sub_aggregation,
&mut res,
)?;
missing_entry.sub_aggregation = res;
}
entries.insert(missing.into(), missing_entry);
let bucket = IntermediateBucketResult::Terms(IntermediateTermBucketResult {
entries,
sum_other_doc_count: 0,
doc_count_error_upper_bound: 0,
});
results.push(name, IntermediateAggregationResult::Bucket(bucket))?;
Ok(())
}
fn collect(
&mut self,
doc: crate::DocId,
agg_with_accessor: &mut AggregationsWithAccessor,
) -> crate::Result<()> {
let agg = &mut agg_with_accessor.aggs.values[self.accessor_idx];
let has_value = agg.accessors.iter().any(|acc| acc.index.has_value(doc));
if !has_value {
self.missing_count += 1;
if let Some(sub_agg) = self.sub_agg.as_mut() {
sub_agg.collect(doc, &mut agg.sub_aggregation)?;
}
}
Ok(())
}
fn collect_block(
&mut self,
docs: &[crate::DocId],
agg_with_accessor: &mut AggregationsWithAccessor,
) -> crate::Result<()> {
for doc in docs {
self.collect(*doc, agg_with_accessor)?;
}
Ok(())
}
}
#[cfg(test)]
mod tests {
use crate::aggregation::agg_req::Aggregations;
use crate::aggregation::tests::exec_request_with_query;
use crate::schema::{Schema, FAST};
use crate::Index;
#[test]
fn terms_aggregation_missing_mixed_type_mult_seg_sub_agg() -> crate::Result<()> {
let mut schema_builder = Schema::builder();
let json = schema_builder.add_json_field("json", FAST);
let score = schema_builder.add_f64_field("score", FAST);
let schema = schema_builder.build();
let index = Index::create_in_ram(schema);
let mut index_writer = index.writer_for_tests().unwrap();
// => Segment with all values numeric
index_writer
.add_document(doc!(score => 1.0, json => json!({"mixed_type": 10.0})))
.unwrap();
index_writer.add_document(doc!(score => 5.0))?;
// index_writer.commit().unwrap();
//// => Segment with all values text
index_writer
.add_document(doc!(score => 1.0, json => json!({"mixed_type": "blue"})))
.unwrap();
index_writer.add_document(doc!(score => 5.0))?;
// index_writer.commit().unwrap();
// => Segment with mixed values
index_writer.add_document(doc!(json => json!({"mixed_type": "red"})))?;
index_writer.add_document(doc!(json => json!({"mixed_type": -20.5})))?;
index_writer.add_document(doc!(json => json!({"mixed_type": true})))?;
index_writer.add_document(doc!(score => 5.0))?;
index_writer.commit().unwrap();
let agg_req: Aggregations = serde_json::from_value(json!({
"replace_null": {
"terms": {
"field": "json.mixed_type",
"missing": "NULL"
},
"aggs": {
"sum_score": {
"sum": {
"field": "score"
}
}
}
},
}))
.unwrap();
let res = exec_request_with_query(agg_req, &index, None)?;
// text field
assert_eq!(res["replace_null"]["buckets"][0]["key"], "NULL");
assert_eq!(res["replace_null"]["buckets"][0]["doc_count"], 3);
assert_eq!(
res["replace_null"]["buckets"][0]["sum_score"]["value"],
15.0
);
assert_eq!(res["replace_null"]["sum_other_doc_count"], 0);
assert_eq!(res["replace_null"]["doc_count_error_upper_bound"], 0);
Ok(())
}
#[test]
fn terms_aggregation_missing_mixed_type_sub_agg_reg1() -> crate::Result<()> {
let mut schema_builder = Schema::builder();
let json = schema_builder.add_json_field("json", FAST);
let score = schema_builder.add_f64_field("score", FAST);
let schema = schema_builder.build();
let index = Index::create_in_ram(schema);
let mut index_writer = index.writer_for_tests().unwrap();
// => Segment with all values numeric
index_writer.add_document(doc!(score => 1.0, json => json!({"mixed_type": 10.0})))?;
index_writer.add_document(doc!(score => 5.0))?;
index_writer.add_document(doc!(score => 5.0))?;
index_writer.commit().unwrap();
let agg_req: Aggregations = serde_json::from_value(json!({
"replace_null": {
"terms": {
"field": "json.mixed_type",
"missing": "NULL"
},
"aggs": {
"sum_score": {
"sum": {
"field": "score"
}
}
}
},
}))
.unwrap();
let res = exec_request_with_query(agg_req, &index, None)?;
// text field
assert_eq!(res["replace_null"]["buckets"][0]["key"], "NULL");
assert_eq!(res["replace_null"]["buckets"][0]["doc_count"], 2);
assert_eq!(
res["replace_null"]["buckets"][0]["sum_score"]["value"],
10.0
);
assert_eq!(res["replace_null"]["sum_other_doc_count"], 0);
assert_eq!(res["replace_null"]["doc_count_error_upper_bound"], 0);
Ok(())
}
#[test]
fn terms_aggregation_missing_mult_seg_empty() -> crate::Result<()> {
let mut schema_builder = Schema::builder();
let score = schema_builder.add_f64_field("score", FAST);
let schema = schema_builder.build();
let index = Index::create_in_ram(schema);
let mut index_writer = index.writer_for_tests().unwrap();
index_writer.add_document(doc!(score => 5.0))?;
index_writer.commit().unwrap();
index_writer.add_document(doc!(score => 5.0))?;
index_writer.commit().unwrap();
index_writer.add_document(doc!(score => 5.0))?;
index_writer.commit().unwrap();
let agg_req: Aggregations = serde_json::from_value(json!({
"replace_null": {
"terms": {
"field": "json.mixed_type",
"missing": "NULL"
},
"aggs": {
"sum_score": {
"sum": {
"field": "score"
}
}
}
},
}))
.unwrap();
let res = exec_request_with_query(agg_req, &index, None)?;
// text field
assert_eq!(res["replace_null"]["buckets"][0]["key"], "NULL");
assert_eq!(res["replace_null"]["buckets"][0]["doc_count"], 3);
assert_eq!(
res["replace_null"]["buckets"][0]["sum_score"]["value"],
15.0
);
assert_eq!(res["replace_null"]["sum_other_doc_count"], 0);
assert_eq!(res["replace_null"]["doc_count_error_upper_bound"], 0);
Ok(())
}
#[test]
fn terms_aggregation_missing_single_seg_empty() -> crate::Result<()> {
let mut schema_builder = Schema::builder();
let score = schema_builder.add_f64_field("score", FAST);
let schema = schema_builder.build();
let index = Index::create_in_ram(schema);
let mut index_writer = index.writer_for_tests().unwrap();
index_writer.add_document(doc!(score => 5.0))?;
index_writer.add_document(doc!(score => 5.0))?;
index_writer.add_document(doc!(score => 5.0))?;
index_writer.commit().unwrap();
let agg_req: Aggregations = serde_json::from_value(json!({
"replace_null": {
"terms": {
"field": "json.mixed_type",
"missing": "NULL"
},
"aggs": {
"sum_score": {
"sum": {
"field": "score"
}
}
}
},
}))
.unwrap();
let res = exec_request_with_query(agg_req, &index, None)?;
// text field
assert_eq!(res["replace_null"]["buckets"][0]["key"], "NULL");
assert_eq!(res["replace_null"]["buckets"][0]["doc_count"], 3);
assert_eq!(
res["replace_null"]["buckets"][0]["sum_score"]["value"],
15.0
);
assert_eq!(res["replace_null"]["sum_other_doc_count"], 0);
assert_eq!(res["replace_null"]["doc_count_error_upper_bound"], 0);
Ok(())
}
#[test]
fn terms_aggregation_missing_mixed_type_mult_seg() -> crate::Result<()> {
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 = index.writer_for_tests().unwrap();
// => Segment with all values numeric
index_writer
.add_document(doc!(json => json!({"mixed_type": 10.0})))
.unwrap();
index_writer.add_document(doc!())?;
index_writer.commit().unwrap();
//// => Segment with all values text
index_writer
.add_document(doc!(json => json!({"mixed_type": "blue"})))
.unwrap();
index_writer.add_document(doc!())?;
index_writer.commit().unwrap();
// => Segment with mixed values
index_writer
.add_document(doc!(json => json!({"mixed_type": "red"})))
.unwrap();
index_writer
.add_document(doc!(json => json!({"mixed_type": -20.5})))
.unwrap();
index_writer
.add_document(doc!(json => json!({"mixed_type": true})))
.unwrap();
index_writer.add_document(doc!())?;
index_writer.commit().unwrap();
let agg_req: Aggregations = serde_json::from_value(json!({
"replace_null": {
"terms": {
"field": "json.mixed_type",
"missing": "NULL"
},
},
"replace_num": {
"terms": {
"field": "json.mixed_type",
"missing": 1337
},
},
}))
.unwrap();
let res = exec_request_with_query(agg_req, &index, None)?;
// text field
assert_eq!(res["replace_null"]["buckets"][0]["key"], "NULL");
assert_eq!(res["replace_null"]["buckets"][0]["doc_count"], 3);
assert_eq!(res["replace_num"]["buckets"][0]["key"], 1337.0);
assert_eq!(res["replace_num"]["buckets"][0]["doc_count"], 3);
assert_eq!(res["replace_null"]["sum_other_doc_count"], 0);
assert_eq!(res["replace_null"]["doc_count_error_upper_bound"], 0);
Ok(())
}
#[test]
fn terms_aggregation_missing_str_on_numeric_field() -> crate::Result<()> {
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 = index.writer_for_tests().unwrap();
// => Segment with all values numeric
index_writer
.add_document(doc!(json => json!({"mixed_type": 10.0})))
.unwrap();
index_writer.add_document(doc!())?;
index_writer.add_document(doc!())?;
index_writer
.add_document(doc!(json => json!({"mixed_type": -20.5})))
.unwrap();
index_writer.add_document(doc!())?;
index_writer.commit().unwrap();
let agg_req: Aggregations = serde_json::from_value(json!({
"replace_null": {
"terms": {
"field": "json.mixed_type",
"missing": "NULL"
},
},
}))
.unwrap();
let res = exec_request_with_query(agg_req, &index, None)?;
// text field
assert_eq!(res["replace_null"]["buckets"][0]["key"], "NULL");
assert_eq!(res["replace_null"]["buckets"][0]["doc_count"], 3);
assert_eq!(res["replace_null"]["sum_other_doc_count"], 0);
assert_eq!(res["replace_null"]["doc_count_error_upper_bound"], 0);
Ok(())
}
#[test]
fn terms_aggregation_missing_mixed_type_one_seg() -> crate::Result<()> {
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 = index.writer_for_tests().unwrap();
// => Segment with all values numeric
index_writer
.add_document(doc!(json => json!({"mixed_type": 10.0})))
.unwrap();
index_writer.add_document(doc!())?;
//// => Segment with all values text
index_writer
.add_document(doc!(json => json!({"mixed_type": "blue"})))
.unwrap();
index_writer.add_document(doc!())?;
// => Segment with mixed values
index_writer
.add_document(doc!(json => json!({"mixed_type": "red"})))
.unwrap();
index_writer
.add_document(doc!(json => json!({"mixed_type": -20.5})))
.unwrap();
index_writer
.add_document(doc!(json => json!({"mixed_type": true})))
.unwrap();
index_writer.add_document(doc!())?;
index_writer.commit().unwrap();
let agg_req: Aggregations = serde_json::from_value(json!({
"replace_null": {
"terms": {
"field": "json.mixed_type",
"missing": "NULL"
},
},
}))
.unwrap();
let res = exec_request_with_query(agg_req, &index, None)?;
// text field
assert_eq!(res["replace_null"]["buckets"][0]["key"], "NULL");
assert_eq!(res["replace_null"]["buckets"][0]["doc_count"], 3);
assert_eq!(res["replace_null"]["sum_other_doc_count"], 0);
assert_eq!(res["replace_null"]["doc_count_error_upper_bound"], 0);
Ok(())
}
}

36
src/aggregation/intermediate_agg_result.rs
View File

@@ -111,9 +111,6 @@ impl IntermediateAggregationResults {
}
/// Convert intermediate result and its aggregation request to the final result.
///
/// Internal function, AggregationsInternal is used instead Aggregations, which is optimized
/// for internal processing, by splitting metric and buckets into separate groups.
pub(crate) fn into_final_result_internal(
self,
req: &Aggregations,
@@ -121,7 +118,14 @@ impl IntermediateAggregationResults {
) -> crate::Result<AggregationResults> {
let mut results: FxHashMap<String, AggregationResult> = FxHashMap::default();
for (key, agg_res) in self.aggs_res.into_iter() {
let req = req.get(key.as_str()).unwrap();
let req = req.get(key.as_str()).unwrap_or_else(|| {
panic!(
"Could not find key {:?} in request keys {:?}. This probably means that \
add_intermediate_aggregation_result passed the wrong agg object.",
key,
req.keys().collect::<Vec<_>>()
)
});
results.insert(key, agg_res.into_final_result(req, limits)?);
}
// Handle empty results
@@ -168,10 +172,16 @@ pub(crate) fn empty_from_req(req: &Aggregation) -> IntermediateAggregationResult
Range(_) => IntermediateAggregationResult::Bucket(IntermediateBucketResult::Range(
Default::default(),
)),
Histogram(_) | DateHistogram(_) => {
Histogram(_) => {
IntermediateAggregationResult::Bucket(IntermediateBucketResult::Histogram {
buckets: Vec::new(),
column_type: None,
is_date_agg: false,
})
}
DateHistogram(_) => {
IntermediateAggregationResult::Bucket(IntermediateBucketResult::Histogram {
buckets: Vec::new(),
is_date_agg: true,
})
}
Average(_) => IntermediateAggregationResult::Metric(IntermediateMetricResult::Average(
@@ -339,8 +349,8 @@ pub enum IntermediateBucketResult {
/// This is the histogram entry for a bucket, which contains a key, count, and optionally
/// sub_aggregations.
Histogram {
/// The column_type of the underlying `Column`
column_type: Option<ColumnType>,
/// The column_type of the underlying `Column` is DateTime
is_date_agg: bool,
/// The buckets
buckets: Vec<IntermediateHistogramBucketEntry>,
},
@@ -395,7 +405,7 @@ impl IntermediateBucketResult {
Ok(BucketResult::Range { buckets })
}
IntermediateBucketResult::Histogram {
column_type,
is_date_agg,
buckets,
} => {
let histogram_req = &req
@@ -404,7 +414,7 @@ impl IntermediateBucketResult {
.expect("unexpected aggregation, expected histogram aggregation");
let buckets = intermediate_histogram_buckets_to_final_buckets(
buckets,
column_type,
is_date_agg,
histogram_req,
req.sub_aggregation(),
limits,
@@ -453,17 +463,17 @@ impl IntermediateBucketResult {
(
IntermediateBucketResult::Histogram {
buckets: buckets_left,
..
is_date_agg: _,
},
IntermediateBucketResult::Histogram {
buckets: buckets_right,
..
is_date_agg: _,
},
) => {
let buckets: Result<Vec<IntermediateHistogramBucketEntry>, TantivyError> =
buckets_left
.drain(..)
.merge_join_by(buckets_right.into_iter(), |left, right| {
.merge_join_by(buckets_right, |left, right| {
left.key.partial_cmp(&right.key).unwrap_or(Ordering::Equal)
})
.map(|either| match either {

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

@@ -20,12 +20,21 @@ use super::{IntermediateStats, SegmentStatsCollector};
pub struct AverageAggregation {
/// The field name to compute the average 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(default)]
pub missing: Option<f64>,
}
impl AverageAggregation {
/// Creates a new [`AverageAggregation`] instance from a field name.
pub fn from_field_name(field_name: String) -> Self {
Self { field: field_name }
Self {
field: field_name,
missing: None,
}
}
/// Returns the field name the aggregation is computed on.
pub fn field_name(&self) -> &str {

15
src/aggregation/metric/count.rs
View File

@@ -18,14 +18,23 @@ use super::{IntermediateStats, SegmentStatsCollector};
/// ```
#[derive(Clone, Debug, PartialEq, Serialize, Deserialize)]
pub struct CountAggregation {
/// The field name to compute the minimum on.
/// The field name to compute the count 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(default)]
pub missing: Option<f64>,
}
impl CountAggregation {
/// Creates a new [`CountAggregation`] instance from a field name.
pub fn from_field_name(field_name: String) -> Self {
Self { field: field_name }
Self {
field: field_name,
missing: None,
}
}
/// Returns the field name the aggregation is computed on.
pub fn field_name(&self) -> &str {
@@ -51,7 +60,7 @@ impl IntermediateCount {
pub fn merge_fruits(&mut self, other: IntermediateCount) {
self.stats.merge_fruits(other.stats);
}
/// Computes the final minimum value.
/// Computes the final count value.
pub fn finalize(&self) -> Option<f64> {
Some(self.stats.finalize().count as f64)
}

63
src/aggregation/metric/max.rs
View File

@@ -20,12 +20,21 @@ use super::{IntermediateStats, SegmentStatsCollector};
pub struct MaxAggregation {
/// The field name to compute the maximum 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(default)]
pub missing: Option<f64>,
}
impl MaxAggregation {
/// Creates a new [`MaxAggregation`] instance from a field name.
pub fn from_field_name(field_name: String) -> Self {
Self { field: field_name }
Self {
field: field_name,
missing: None,
}
}
/// Returns the field name the aggregation is computed on.
pub fn field_name(&self) -> &str {
@@ -56,3 +65,55 @@ impl IntermediateMax {
self.stats.finalize().max
}
}
#[cfg(test)]
mod tests {
use crate::aggregation::agg_req::Aggregations;
use crate::aggregation::tests::exec_request_with_query;
use crate::schema::{Schema, FAST};
use crate::Index;
#[test]
fn test_max_agg_with_missing() -> crate::Result<()> {
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 = index.writer_for_tests().unwrap();
// => Segment with empty json
index_writer.add_document(doc!()).unwrap();
index_writer.commit().unwrap();
// => Segment with json, but no field partially_empty
index_writer
.add_document(doc!(json => json!({"different_field": "blue"})))
.unwrap();
index_writer.commit().unwrap();
//// => Segment with field partially_empty
index_writer
.add_document(doc!(json => json!({"partially_empty": 10.0})))
.unwrap();
index_writer.add_document(doc!())?;
index_writer.commit().unwrap();
let agg_req: Aggregations = serde_json::from_value(json!({
"my_stats": {
"max": {
"field": "json.partially_empty",
"missing": 100.0,
}
}
}))
.unwrap();
let res = exec_request_with_query(agg_req, &index, None)?;
assert_eq!(
res["my_stats"],
json!({
"value": 100.0,
})
);
Ok(())
}
}

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

@@ -20,12 +20,21 @@ use super::{IntermediateStats, SegmentStatsCollector};
pub struct MinAggregation {
/// The field name to compute the minimum 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(default)]
pub missing: Option<f64>,
}
impl MinAggregation {
/// Creates a new [`MinAggregation`] instance from a field name.
pub fn from_field_name(field_name: String) -> Self {
Self { field: field_name }
Self {
field: field_name,
missing: None,
}
}
/// Returns the field name the aggregation is computed on.
pub fn field_name(&self) -> &str {

160
src/aggregation/metric/percentiles.rs
View File

@@ -11,7 +11,7 @@ use crate::aggregation::intermediate_agg_result::{
IntermediateAggregationResult, IntermediateAggregationResults, IntermediateMetricResult,
};
use crate::aggregation::segment_agg_result::SegmentAggregationCollector;
use crate::aggregation::{f64_from_fastfield_u64, AggregationError};
use crate::aggregation::{f64_from_fastfield_u64, f64_to_fastfield_u64, AggregationError};
use crate::{DocId, TantivyError};
/// # Percentiles
@@ -80,6 +80,12 @@ pub struct PercentilesAggregationReq {
/// Whether to return the percentiles as a hash map
#[serde(default = "default_as_true")]
pub keyed: bool,
/// 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<f64>,
}
fn default_percentiles() -> &'static [f64] {
&[1.0, 5.0, 25.0, 50.0, 75.0, 95.0, 99.0]
@@ -95,6 +101,7 @@ impl PercentilesAggregationReq {
field: field_name,
percents: None,
keyed: default_as_true(),
missing: None,
}
}
/// Returns the field name the aggregation is computed on.
@@ -127,6 +134,7 @@ pub(crate) struct SegmentPercentilesCollector {
pub(crate) percentiles: PercentilesCollector,
pub(crate) accessor_idx: usize,
val_cache: Vec<u64>,
missing: Option<u64>,
}
#[derive(Clone, Serialize, Deserialize)]
@@ -227,11 +235,16 @@ impl SegmentPercentilesCollector {
accessor_idx: usize,
) -> crate::Result<Self> {
req.validate()?;
let missing = req
.missing
.and_then(|val| f64_to_fastfield_u64(val, &field_type));
Ok(Self {
field_type,
percentiles: PercentilesCollector::new(),
accessor_idx,
val_cache: Default::default(),
missing,
})
}
#[inline]
@@ -240,9 +253,17 @@ impl SegmentPercentilesCollector {
docs: &[DocId],
agg_accessor: &mut AggregationWithAccessor,
) {
agg_accessor
.column_block_accessor
.fetch_block(docs, &agg_accessor.accessor);
if let Some(missing) = self.missing.as_ref() {
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);
}
for val in agg_accessor.column_block_accessor.iter_vals() {
let val1 = f64_from_fastfield_u64(val, &self.field_type);
@@ -277,9 +298,22 @@ impl SegmentAggregationCollector for SegmentPercentilesCollector {
) -> crate::Result<()> {
let field = &agg_with_accessor.aggs.values[self.accessor_idx].accessor;
for val in field.values_for_doc(doc) {
let val1 = f64_from_fastfield_u64(val, &self.field_type);
self.percentiles.collect(val1);
if let Some(missing) = self.missing {
let mut has_val = false;
for val in field.values_for_doc(doc) {
let val1 = f64_from_fastfield_u64(val, &self.field_type);
self.percentiles.collect(val1);
has_val = true;
}
if !has_val {
self.percentiles
.collect(f64_from_fastfield_u64(missing, &self.field_type));
}
} else {
for val in field.values_for_doc(doc) {
let val1 = f64_from_fastfield_u64(val, &self.field_type);
self.percentiles.collect(val1);
}
}
Ok(())
@@ -309,10 +343,12 @@ mod tests {
use crate::aggregation::agg_req::Aggregations;
use crate::aggregation::agg_result::AggregationResults;
use crate::aggregation::tests::{
get_test_index_from_values, get_test_index_from_values_and_terms,
exec_request_with_query, get_test_index_from_values, get_test_index_from_values_and_terms,
};
use crate::aggregation::AggregationCollector;
use crate::query::AllQuery;
use crate::schema::{Schema, FAST};
use crate::Index;
#[test]
fn test_aggregation_percentiles_empty_index() -> crate::Result<()> {
@@ -463,7 +499,7 @@ mod tests {
fn test_aggregation_percentiles(merge_segments: bool) -> crate::Result<()> {
use rand_distr::Distribution;
let num_values_in_segment = vec![100, 30_000, 8000];
let num_values_in_segment = [100, 30_000, 8000];
let lg_norm = rand_distr::LogNormal::new(2.996f64, 0.979f64).unwrap();
let mut rng = StdRng::from_seed([1u8; 32]);
@@ -545,4 +581,110 @@ mod tests {
Ok(())
}
#[test]
fn test_percentiles_missing_sub_agg() -> crate::Result<()> {
// This test verifies the `collect` method (in contrast to `collect_block`), which is
// called when the sub-aggregations are flushed.
let mut schema_builder = Schema::builder();
let text_field = schema_builder.add_text_field("texts", FAST);
let score_field_f64 = schema_builder.add_f64_field("score", FAST);
let schema = schema_builder.build();
let index = Index::create_in_ram(schema);
{
let mut index_writer = index.writer_for_tests()?;
// writing the segment
index_writer.add_document(doc!(
score_field_f64 => 10.0f64,
text_field => "a"
))?;
index_writer.add_document(doc!(
score_field_f64 => 10.0f64,
text_field => "a"
))?;
index_writer.add_document(doc!(text_field => "a"))?;
index_writer.commit()?;
}
let agg_req: Aggregations = {
serde_json::from_value(json!({
"range_with_stats": {
"terms": {
"field": "texts"
},
"aggs": {
"percentiles": {
"percentiles": {
"field": "score",
"missing": 5.0
}
}
}
}
}))
.unwrap()
};
let res = exec_request_with_query(agg_req, &index, None)?;
assert_eq!(res["range_with_stats"]["buckets"][0]["doc_count"], 3);
assert_eq!(
res["range_with_stats"]["buckets"][0]["percentiles"]["values"]["1.0"],
5.0028295751107414
);
assert_eq!(
res["range_with_stats"]["buckets"][0]["percentiles"]["values"]["99.0"],
10.07469668951144
);
Ok(())
}
#[test]
fn test_percentiles_missing() -> crate::Result<()> {
let mut schema_builder = Schema::builder();
let text_field = schema_builder.add_text_field("texts", FAST);
let score_field_f64 = schema_builder.add_f64_field("score", FAST);
let schema = schema_builder.build();
let index = Index::create_in_ram(schema);
{
let mut index_writer = index.writer_for_tests()?;
// writing the segment
index_writer.add_document(doc!(
score_field_f64 => 10.0f64,
text_field => "a"
))?;
index_writer.add_document(doc!(
score_field_f64 => 10.0f64,
text_field => "a"
))?;
index_writer.add_document(doc!(text_field => "a"))?;
index_writer.commit()?;
}
let agg_req: Aggregations = {
serde_json::from_value(json!({
"percentiles": {
"percentiles": {
"field": "score",
"missing": 5.0
}
}
}))
.unwrap()
};
let res = exec_request_with_query(agg_req, &index, None)?;
assert_eq!(res["percentiles"]["values"]["1.0"], 5.0028295751107414);
assert_eq!(res["percentiles"]["values"]["99.0"], 10.07469668951144);
Ok(())
}
}

215
src/aggregation/metric/stats.rs
View File

@@ -5,11 +5,11 @@ use super::*;
use crate::aggregation::agg_req_with_accessor::{
AggregationWithAccessor, AggregationsWithAccessor,
};
use crate::aggregation::f64_from_fastfield_u64;
use crate::aggregation::intermediate_agg_result::{
IntermediateAggregationResult, IntermediateAggregationResults, IntermediateMetricResult,
};
use crate::aggregation::segment_agg_result::SegmentAggregationCollector;
use crate::aggregation::{f64_from_fastfield_u64, f64_to_fastfield_u64};
use crate::{DocId, TantivyError};
/// A multi-value metric aggregation that computes a collection of statistics on numeric values that
@@ -29,12 +29,21 @@ use crate::{DocId, TantivyError};
pub struct StatsAggregation {
/// The field name to compute the stats 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(default)]
pub missing: Option<f64>,
}
impl StatsAggregation {
/// Creates a new [`StatsAggregation`] instance from a field name.
pub fn from_field_name(field_name: String) -> Self {
StatsAggregation { field: field_name }
StatsAggregation {
field: field_name,
missing: None,
}
}
/// Returns the field name the aggregation is computed on.
pub fn field_name(&self) -> &str {
@@ -153,6 +162,7 @@ pub(crate) enum SegmentStatsType {
#[derive(Clone, Debug, PartialEq)]
pub(crate) struct SegmentStatsCollector {
missing: Option<u64>,
field_type: ColumnType,
pub(crate) collecting_for: SegmentStatsType,
pub(crate) stats: IntermediateStats,
@@ -165,12 +175,15 @@ impl SegmentStatsCollector {
field_type: ColumnType,
collecting_for: SegmentStatsType,
accessor_idx: usize,
missing: Option<f64>,
) -> Self {
let missing = missing.and_then(|val| f64_to_fastfield_u64(val, &field_type));
Self {
field_type,
collecting_for,
stats: IntermediateStats::default(),
accessor_idx,
missing,
val_cache: Default::default(),
}
}
@@ -180,10 +193,17 @@ impl SegmentStatsCollector {
docs: &[DocId],
agg_accessor: &mut AggregationWithAccessor,
) {
agg_accessor
.column_block_accessor
.fetch_block(docs, &agg_accessor.accessor);
if let Some(missing) = self.missing.as_ref() {
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);
}
for val in agg_accessor.column_block_accessor.iter_vals() {
let val1 = f64_from_fastfield_u64(val, &self.field_type);
self.stats.collect(val1);
@@ -234,10 +254,22 @@ impl SegmentAggregationCollector for SegmentStatsCollector {
agg_with_accessor: &mut AggregationsWithAccessor,
) -> crate::Result<()> {
let field = &agg_with_accessor.aggs.values[self.accessor_idx].accessor;
for val in field.values_for_doc(doc) {
let val1 = f64_from_fastfield_u64(val, &self.field_type);
self.stats.collect(val1);
if let Some(missing) = self.missing {
let mut has_val = false;
for val in field.values_for_doc(doc) {
let val1 = f64_from_fastfield_u64(val, &self.field_type);
self.stats.collect(val1);
has_val = true;
}
if !has_val {
self.stats
.collect(f64_from_fastfield_u64(missing, &self.field_type));
}
} else {
for val in field.values_for_doc(doc) {
let val1 = f64_from_fastfield_u64(val, &self.field_type);
self.stats.collect(val1);
}
}
Ok(())
@@ -262,11 +294,13 @@ mod tests {
use crate::aggregation::agg_req::{Aggregation, Aggregations};
use crate::aggregation::agg_result::AggregationResults;
use crate::aggregation::tests::{get_test_index_2_segments, get_test_index_from_values};
use crate::aggregation::tests::{
exec_request_with_query, get_test_index_2_segments, get_test_index_from_values,
};
use crate::aggregation::AggregationCollector;
use crate::query::{AllQuery, TermQuery};
use crate::schema::IndexRecordOption;
use crate::Term;
use crate::schema::{IndexRecordOption, Schema, FAST};
use crate::{Index, Term};
#[test]
fn test_aggregation_stats_empty_index() -> crate::Result<()> {
@@ -453,4 +487,159 @@ mod tests {
Ok(())
}
#[test]
fn test_stats_json() -> crate::Result<()> {
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 = index.writer_for_tests().unwrap();
// => Segment with empty json
index_writer.add_document(doc!()).unwrap();
index_writer.commit().unwrap();
// => Segment with json, but no field partially_empty
index_writer
.add_document(doc!(json => json!({"different_field": "blue"})))
.unwrap();
index_writer.commit().unwrap();
//// => Segment with field partially_empty
index_writer
.add_document(doc!(json => json!({"partially_empty": 10.0})))
.unwrap();
index_writer.add_document(doc!())?;
index_writer.commit().unwrap();
let agg_req: Aggregations = serde_json::from_value(json!({
"my_stats": {
"stats": {
"field": "json.partially_empty"
},
}
}))
.unwrap();
let res = exec_request_with_query(agg_req, &index, None)?;
assert_eq!(
res["my_stats"],
json!({
"avg": 10.0,
"count": 1,
"max": 10.0,
"min": 10.0,
"sum": 10.0
})
);
Ok(())
}
#[test]
fn test_stats_json_missing() -> crate::Result<()> {
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 = index.writer_for_tests().unwrap();
// => Segment with empty json
index_writer.add_document(doc!()).unwrap();
index_writer.commit().unwrap();
// => Segment with json, but no field partially_empty
index_writer
.add_document(doc!(json => json!({"different_field": "blue"})))
.unwrap();
index_writer.commit().unwrap();
//// => Segment with field partially_empty
index_writer
.add_document(doc!(json => json!({"partially_empty": 10.0})))
.unwrap();
index_writer.add_document(doc!())?;
index_writer.commit().unwrap();
let agg_req: Aggregations = serde_json::from_value(json!({
"my_stats": {
"stats": {
"field": "json.partially_empty",
"missing": 0.0
},
}
}))
.unwrap();
let res = exec_request_with_query(agg_req, &index, None)?;
assert_eq!(
res["my_stats"],
json!({
"avg": 2.5,
"count": 4,
"max": 10.0,
"min": 0.0,
"sum": 10.0
})
);
Ok(())
}
#[test]
fn test_stats_json_missing_sub_agg() -> crate::Result<()> {
// This test verifies the `collect` method (in contrast to `collect_block`), which is
// called when the sub-aggregations are flushed.
let mut schema_builder = Schema::builder();
let text_field = schema_builder.add_text_field("texts", FAST);
let score_field_f64 = schema_builder.add_f64_field("score", FAST);
let schema = schema_builder.build();
let index = Index::create_in_ram(schema);
{
let mut index_writer = index.writer_for_tests()?;
// writing the segment
index_writer.add_document(doc!(
score_field_f64 => 10.0f64,
text_field => "a"
))?;
index_writer.add_document(doc!(text_field => "a"))?;
index_writer.commit()?;
}
let agg_req: Aggregations = {
serde_json::from_value(json!({
"range_with_stats": {
"terms": {
"field": "texts"
},
"aggs": {
"my_stats": {
"stats": {
"field": "score",
"missing": 0.0
}
}
}
}
}))
.unwrap()
};
let res = exec_request_with_query(agg_req, &index, None)?;
assert_eq!(
res["range_with_stats"]["buckets"][0]["my_stats"]["count"],
2
);
assert_eq!(
res["range_with_stats"]["buckets"][0]["my_stats"]["min"],
0.0
);
assert_eq!(
res["range_with_stats"]["buckets"][0]["my_stats"]["avg"],
5.0
);
Ok(())
}
}

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

@@ -20,12 +20,21 @@ use super::{IntermediateStats, SegmentStatsCollector};
pub struct SumAggregation {
/// The field name to compute the minimum 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(default)]
pub missing: Option<f64>,
}
impl SumAggregation {
/// Creates a new [`SumAggregation`] instance from a field name.
pub fn from_field_name(field_name: String) -> Self {
Self { field: field_name }
Self {
field: field_name,
missing: None,
}
}
/// Returns the field name the aggregation is computed on.
pub fn field_name(&self) -> &str {

49
src/aggregation/segment_agg_result.rs
View File

@@ -15,7 +15,7 @@ use super::metric::{
SegmentPercentilesCollector, SegmentStatsCollector, SegmentStatsType, StatsAggregation,
SumAggregation,
};
use crate::aggregation::bucket::SegmentTermCollectorComposite;
use crate::aggregation::bucket::TermMissingAgg;
pub(crate) trait SegmentAggregationCollector: CollectorClone + Debug {
fn add_intermediate_aggregation_result(
@@ -82,29 +82,24 @@ pub(crate) fn build_single_agg_segment_collector(
use AggregationVariants::*;
match &req.agg.agg {
Terms(terms_req) => {
if let Some(acc2) = req.accessor2.as_ref() {
Ok(Box::new(
SegmentTermCollectorComposite::from_req_and_validate(
terms_req,
&mut req.sub_aggregation,
req.field_type,
acc2.1,
accessor_idx,
)?,
))
} else {
if req.accessors.is_empty() {
Ok(Box::new(SegmentTermCollector::from_req_and_validate(
terms_req,
&mut req.sub_aggregation,
req.field_type,
accessor_idx,
)?))
} else {
Ok(Box::new(TermMissingAgg::new(
accessor_idx,
&mut req.sub_aggregation,
)?))
}
}
Range(range_req) => Ok(Box::new(SegmentRangeCollector::from_req_and_validate(
range_req,
&mut req.sub_aggregation,
&mut req.limits,
&req.limits,
req.field_type,
accessor_idx,
)?)),
@@ -120,35 +115,43 @@ pub(crate) fn build_single_agg_segment_collector(
req.field_type,
accessor_idx,
)?)),
Average(AverageAggregation { .. }) => Ok(Box::new(SegmentStatsCollector::from_req(
req.field_type,
SegmentStatsType::Average,
accessor_idx,
))),
Count(CountAggregation { .. }) => Ok(Box::new(SegmentStatsCollector::from_req(
Average(AverageAggregation { missing, .. }) => {
Ok(Box::new(SegmentStatsCollector::from_req(
req.field_type,
SegmentStatsType::Average,
accessor_idx,
*missing,
)))
}
Count(CountAggregation { missing, .. }) => Ok(Box::new(SegmentStatsCollector::from_req(
req.field_type,
SegmentStatsType::Count,
accessor_idx,
*missing,
))),
Max(MaxAggregation { .. }) => Ok(Box::new(SegmentStatsCollector::from_req(
Max(MaxAggregation { missing, .. }) => Ok(Box::new(SegmentStatsCollector::from_req(
req.field_type,
SegmentStatsType::Max,
accessor_idx,
*missing,
))),
Min(MinAggregation { .. }) => Ok(Box::new(SegmentStatsCollector::from_req(
Min(MinAggregation { missing, .. }) => Ok(Box::new(SegmentStatsCollector::from_req(
req.field_type,
SegmentStatsType::Min,
accessor_idx,
*missing,
))),
Stats(StatsAggregation { .. }) => Ok(Box::new(SegmentStatsCollector::from_req(
Stats(StatsAggregation { missing, .. }) => Ok(Box::new(SegmentStatsCollector::from_req(
req.field_type,
SegmentStatsType::Stats,
accessor_idx,
*missing,
))),
Sum(SumAggregation { .. }) => Ok(Box::new(SegmentStatsCollector::from_req(
Sum(SumAggregation { missing, .. }) => Ok(Box::new(SegmentStatsCollector::from_req(
req.field_type,
SegmentStatsType::Sum,
accessor_idx,
*missing,
))),
Percentiles(percentiles_req) => Ok(Box::new(
SegmentPercentilesCollector::from_req_and_validate(

2
src/collector/count_collector.rs
View File

@@ -16,7 +16,7 @@ use crate::{DocId, Score, SegmentOrdinal, SegmentReader};
/// let schema = schema_builder.build();
/// let index = Index::create_in_ram(schema);
///
/// let mut index_writer = index.writer(3_000_000).unwrap();
/// let mut index_writer = index.writer(15_000_000).unwrap();
/// index_writer.add_document(doc!(title => "The Name of the Wind")).unwrap();
/// index_writer.add_document(doc!(title => "The Diary of Muadib")).unwrap();
/// index_writer.add_document(doc!(title => "A Dairy Cow")).unwrap();

2
src/collector/facet_collector.rs
View File

@@ -89,7 +89,7 @@ fn facet_depth(facet_bytes: &[u8]) -> usize {
/// let schema = schema_builder.build();
/// let index = Index::create_in_ram(schema);
/// {
/// let mut index_writer = index.writer(3_000_000)?;
/// let mut index_writer = index.writer(15_000_000)?;
/// // a document can be associated with any number of facets
/// index_writer.add_document(doc!(
/// title => "The Name of the Wind",

4
src/collector/filter_collector_wrapper.rs
View File

@@ -38,7 +38,7 @@ use crate::{DocId, Score, SegmentReader, TantivyError};
/// let schema = schema_builder.build();
/// let index = Index::create_in_ram(schema);
///
/// let mut index_writer = index.writer_with_num_threads(1, 10_000_000)?;
/// let mut index_writer = index.writer_with_num_threads(1, 20_000_000)?;
/// index_writer.add_document(doc!(title => "The Name of the Wind", price => 30_200u64))?;
/// index_writer.add_document(doc!(title => "The Diary of Muadib", price => 29_240u64))?;
/// index_writer.add_document(doc!(title => "A Dairy Cow", price => 21_240u64))?;
@@ -216,7 +216,7 @@ where
/// let schema = schema_builder.build();
/// let index = Index::create_in_ram(schema);
///
/// let mut index_writer = index.writer_with_num_threads(1, 10_000_000)?;
/// let mut index_writer = index.writer_with_num_threads(1, 20_000_000)?;
/// index_writer.add_document(doc!(title => "The Name of the Wind", barcode => &b"010101"[..]))?;
/// index_writer.add_document(doc!(title => "The Diary of Muadib", barcode => &b"110011"[..]))?;
/// index_writer.add_document(doc!(title => "A Dairy Cow", barcode => &b"110111"[..]))?;

6
src/collector/histogram_collector.rs
View File

@@ -233,7 +233,7 @@ mod tests {
let val_field = schema_builder.add_i64_field("val_field", FAST);
let schema = schema_builder.build();
let index = Index::create_in_ram(schema);
let mut writer = index.writer_with_num_threads(1, 4_000_000)?;
let mut writer = index.writer_for_tests()?;
writer.add_document(doc!(val_field=>12i64))?;
writer.add_document(doc!(val_field=>-30i64))?;
writer.add_document(doc!(val_field=>-12i64))?;
@@ -255,7 +255,7 @@ mod tests {
let val_field = schema_builder.add_i64_field("val_field", FAST);
let schema = schema_builder.build();
let index = Index::create_in_ram(schema);
let mut writer = index.writer_with_num_threads(1, 4_000_000)?;
let mut writer = index.writer_for_tests()?;
writer.add_document(doc!(val_field=>12i64))?;
writer.commit()?;
writer.add_document(doc!(val_field=>-30i64))?;
@@ -280,7 +280,7 @@ mod tests {
let date_field = schema_builder.add_date_field("date_field", FAST);
let schema = schema_builder.build();
let index = Index::create_in_ram(schema);
let mut writer = index.writer_with_num_threads(1, 4_000_000)?;
let mut writer = index.writer_for_tests()?;
writer.add_document(doc!(date_field=>DateTime::from_primitive(Date::from_calendar_date(1982, Month::September, 17)?.with_hms(0, 0, 0)?)))?;
writer.add_document(
doc!(date_field=>DateTime::from_primitive(Date::from_calendar_date(1986, Month::March, 9)?.with_hms(0, 0, 0)?)),

2
src/collector/mod.rs
View File

@@ -44,7 +44,7 @@
//! # let title = schema_builder.add_text_field("title", TEXT);
//! # let schema = schema_builder.build();
//! # let index = Index::create_in_ram(schema);
//! # let mut index_writer = index.writer(3_000_000)?;
//! # let mut index_writer = index.writer(15_000_000)?;
//! # index_writer.add_document(doc!(
//! # title => "The Name of the Wind",
//! # ))?;

2
src/collector/multi_collector.rs
View File

@@ -120,7 +120,7 @@ impl<TFruit: Fruit> FruitHandle<TFruit> {
/// let title = schema_builder.add_text_field("title", TEXT);
/// let schema = schema_builder.build();
/// let index = Index::create_in_ram(schema);
/// let mut index_writer = index.writer(3_000_000)?;
/// let mut index_writer = index.writer(15_000_000)?;
/// index_writer.add_document(doc!(title => "The Name of the Wind"))?;
/// index_writer.add_document(doc!(title => "The Diary of Muadib"))?;
/// index_writer.add_document(doc!(title => "A Dairy Cow"))?;

2
src/collector/tests.rs
View File

@@ -26,7 +26,7 @@ pub fn test_filter_collector() -> crate::Result<()> {
let schema = schema_builder.build();
let index = Index::create_in_ram(schema);
let mut index_writer = index.writer_with_num_threads(1, 10_000_000)?;
let mut index_writer = index.writer_with_num_threads(1, 20_000_000)?;
index_writer.add_document(doc!(title => "The Name of the Wind", price => 30_200u64, date => DateTime::from_utc(OffsetDateTime::parse("1898-04-09T00:00:00+00:00", &Rfc3339).unwrap())))?;
index_writer.add_document(doc!(title => "The Diary of Muadib", price => 29_240u64, date => DateTime::from_utc(OffsetDateTime::parse("2020-04-09T00:00:00+00:00", &Rfc3339).unwrap())))?;
index_writer.add_document(doc!(title => "The Diary of Anne Frank", price => 18_240u64, date => DateTime::from_utc(OffsetDateTime::parse("2019-04-20T00:00:00+00:00", &Rfc3339).unwrap())))?;

16
src/collector/top_score_collector.rs
View File

@@ -105,7 +105,7 @@ where
/// let schema = schema_builder.build();
/// let index = Index::create_in_ram(schema);
///
/// let mut index_writer = index.writer_with_num_threads(1, 10_000_000)?;
/// let mut index_writer = index.writer_with_num_threads(1, 20_000_000)?;
/// index_writer.add_document(doc!(title => "The Name of the Wind"))?;
/// index_writer.add_document(doc!(title => "The Diary of Muadib"))?;
/// index_writer.add_document(doc!(title => "A Dairy Cow"))?;
@@ -210,7 +210,7 @@ impl TopDocs {
/// let schema = schema_builder.build();
/// let index = Index::create_in_ram(schema);
///
/// let mut index_writer = index.writer_with_num_threads(1, 10_000_000)?;
/// let mut index_writer = index.writer_with_num_threads(1, 20_000_000)?;
/// index_writer.add_document(doc!(title => "The Name of the Wind"))?;
/// index_writer.add_document(doc!(title => "The Diary of Muadib"))?;
/// index_writer.add_document(doc!(title => "A Dairy Cow"))?;
@@ -261,7 +261,7 @@ impl TopDocs {
/// # let schema = schema_builder.build();
/// #
/// # let index = Index::create_in_ram(schema);
/// # let mut index_writer = index.writer_with_num_threads(1, 10_000_000)?;
/// # let mut index_writer = index.writer_with_num_threads(1, 20_000_000)?;
/// # index_writer.add_document(doc!(title => "The Name of the Wind", rating => 92u64))?;
/// # index_writer.add_document(doc!(title => "The Diary of Muadib", rating => 97u64))?;
/// # index_writer.add_document(doc!(title => "A Dairy Cow", rating => 63u64))?;
@@ -349,7 +349,7 @@ impl TopDocs {
/// # let schema = schema_builder.build();
/// #
/// # let index = Index::create_in_ram(schema);
/// # let mut index_writer = index.writer_with_num_threads(1, 10_000_000)?;
/// # let mut index_writer = index.writer_with_num_threads(1, 20_000_000)?;
/// # index_writer.add_document(doc!(title => "MadCow Inc.", revenue => 92_000_000i64))?;
/// # index_writer.add_document(doc!(title => "Zozo Cow KKK", revenue => 119_000_000i64))?;
/// # index_writer.add_document(doc!(title => "Declining Cow", revenue => -63_000_000i64))?;
@@ -449,7 +449,7 @@ impl TopDocs {
/// fn create_index() -> tantivy::Result<Index> {
/// let schema = create_schema();
/// let index = Index::create_in_ram(schema);
/// let mut index_writer = index.writer_with_num_threads(1, 10_000_000)?;
/// let mut index_writer = index.writer_with_num_threads(1, 20_000_000)?;
/// let product_name = index.schema().get_field("product_name").unwrap();
/// let popularity: Field = index.schema().get_field("popularity").unwrap();
/// index_writer.add_document(doc!(product_name => "The Diary of Muadib", popularity => 1u64))?;
@@ -556,7 +556,7 @@ impl TopDocs {
/// # fn main() -> tantivy::Result<()> {
/// # let schema = create_schema();
/// # let index = Index::create_in_ram(schema);
/// # let mut index_writer = index.writer_with_num_threads(1, 10_000_000)?;
/// # let mut index_writer = index.writer_with_num_threads(1, 20_000_000)?;
/// # let product_name = index.schema().get_field("product_name").unwrap();
/// #
/// let popularity: Field = index.schema().get_field("popularity").unwrap();
@@ -752,7 +752,7 @@ mod tests {
let schema = schema_builder.build();
let index = Index::create_in_ram(schema);
// writing the segment
let mut index_writer = index.writer_with_num_threads(1, 10_000_000)?;
let mut index_writer = index.writer_with_num_threads(1, 20_000_000)?;
index_writer.add_document(doc!(text_field=>"Hello happy tax payer."))?;
index_writer.add_document(doc!(text_field=>"Droopy says hello happy tax payer"))?;
index_writer.add_document(doc!(text_field=>"I like Droopy"))?;
@@ -1122,7 +1122,7 @@ mod tests {
mut doc_adder: impl FnMut(&mut IndexWriter),
) -> (Index, Box<dyn Query>) {
let index = Index::create_in_ram(schema);
let mut index_writer = index.writer_with_num_threads(1, 10_000_000).unwrap();
let mut index_writer = index.writer_with_num_threads(1, 15_000_000).unwrap();
doc_adder(&mut index_writer);
index_writer.commit().unwrap();
let query_parser = QueryParser::for_index(&index, vec![query_field]);

24
src/core/index.rs
View File

@@ -16,7 +16,7 @@ use crate::directory::error::OpenReadError;
use crate::directory::MmapDirectory;
use crate::directory::{Directory, ManagedDirectory, RamDirectory, INDEX_WRITER_LOCK};
use crate::error::{DataCorruption, TantivyError};
use crate::indexer::index_writer::{MAX_NUM_THREAD, MEMORY_ARENA_NUM_BYTES_MIN};
use crate::indexer::index_writer::{MAX_NUM_THREAD, MEMORY_BUDGET_NUM_BYTES_MIN};
use crate::indexer::segment_updater::save_metas;
use crate::reader::{IndexReader, IndexReaderBuilder};
use crate::schema::{Field, FieldType, Schema};
@@ -523,9 +523,9 @@ impl Index {
/// - `num_threads` defines the number of indexing workers that
/// should work at the same time.
///
/// - `overall_memory_arena_in_bytes` sets the amount of memory
/// - `overall_memory_budget_in_bytes` sets the amount of memory
/// allocated for all indexing thread.
/// Each thread will receive a budget of `overall_memory_arena_in_bytes / num_threads`.
/// Each thread will receive a budget of `overall_memory_budget_in_bytes / num_threads`.
///
/// # Errors
/// If the lockfile already exists, returns `Error::DirectoryLockBusy` or an `Error::IoError`.
@@ -534,7 +534,7 @@ impl Index {
pub fn writer_with_num_threads(
&self,
num_threads: usize,
overall_memory_arena_in_bytes: usize,
overall_memory_budget_in_bytes: usize,
) -> crate::Result<IndexWriter> {
let directory_lock = self
.directory
@@ -550,7 +550,7 @@ impl Index {
),
)
})?;
let memory_arena_in_bytes_per_thread = overall_memory_arena_in_bytes / num_threads;
let memory_arena_in_bytes_per_thread = overall_memory_budget_in_bytes / num_threads;
IndexWriter::new(
self,
num_threads,
@@ -561,11 +561,11 @@ impl Index {
/// Helper to create an index writer for tests.
///
/// That index writer only simply has a single thread and a memory arena of 10 MB.
/// That index writer only simply has a single thread and a memory budget of 15 MB.
/// Using a single thread gives us a deterministic allocation of DocId.
#[cfg(test)]
pub fn writer_for_tests(&self) -> crate::Result<IndexWriter> {
self.writer_with_num_threads(1, 10_000_000)
self.writer_with_num_threads(1, 15_000_000)
}
/// Creates a multithreaded writer
@@ -579,13 +579,13 @@ impl Index {
/// If the lockfile already exists, returns `Error::FileAlreadyExists`.
/// If the memory arena per thread is too small or too big, returns
/// `TantivyError::InvalidArgument`
pub fn writer(&self, memory_arena_num_bytes: usize) -> crate::Result<IndexWriter> {
pub fn writer(&self, memory_budget_in_bytes: usize) -> crate::Result<IndexWriter> {
let mut num_threads = std::cmp::min(num_cpus::get(), MAX_NUM_THREAD);
let memory_arena_num_bytes_per_thread = memory_arena_num_bytes / num_threads;
if memory_arena_num_bytes_per_thread < MEMORY_ARENA_NUM_BYTES_MIN {
num_threads = (memory_arena_num_bytes / MEMORY_ARENA_NUM_BYTES_MIN).max(1);
let memory_budget_num_bytes_per_thread = memory_budget_in_bytes / num_threads;
if memory_budget_num_bytes_per_thread < MEMORY_BUDGET_NUM_BYTES_MIN {
num_threads = (memory_budget_in_bytes / MEMORY_BUDGET_NUM_BYTES_MIN).max(1);
}
self.writer_with_num_threads(num_threads, memory_arena_num_bytes)
self.writer_with_num_threads(num_threads, memory_budget_in_bytes)
}
/// Accessor to the index settings

22
src/core/index_meta.rs
View File

@@ -410,7 +410,9 @@ mod tests {
use super::IndexMeta;
use crate::core::index_meta::UntrackedIndexMeta;
use crate::schema::{Schema, TEXT};
use crate::store::{Compressor, ZstdCompressor};
use crate::store::Compressor;
#[cfg(feature = "zstd-compression")]
use crate::store::ZstdCompressor;
use crate::{IndexSettings, IndexSortByField, Order};
#[test]
@@ -446,6 +448,7 @@ mod tests {
}
#[test]
#[cfg(feature = "zstd-compression")]
fn test_serialize_metas_zstd_compressor() {
let schema = {
let mut schema_builder = Schema::builder();
@@ -482,13 +485,14 @@ mod tests {
}
#[test]
#[cfg(all(feature = "lz4-compression", feature = "zstd-compression"))]
fn test_serialize_metas_invalid_comp() {
let json = r#"{"index_settings":{"sort_by_field":{"field":"text","order":"Asc"},"docstore_compression":"zsstd","docstore_blocksize":1000000},"segments":[],"schema":[{"name":"text","type":"text","options":{"indexing":{"record":"position","fieldnorms":true,"tokenizer":"default"},"stored":false,"fast":false}}],"opstamp":0}"#;
let err = serde_json::from_str::<UntrackedIndexMeta>(json).unwrap_err();
assert_eq!(
err.to_string(),
"unknown variant `zsstd`, expected one of `none`, `lz4`, `brotli`, `snappy`, `zstd`, \
"unknown variant `zsstd`, expected one of `none`, `lz4`, `zstd`, \
`zstd(compression_level=5)` at line 1 column 96"
.to_string()
);
@@ -502,6 +506,20 @@ mod tests {
);
}
#[test]
#[cfg(not(feature = "zstd-compression"))]
fn test_serialize_metas_unsupported_comp() {
let json = r#"{"index_settings":{"sort_by_field":{"field":"text","order":"Asc"},"docstore_compression":"zstd","docstore_blocksize":1000000},"segments":[],"schema":[{"name":"text","type":"text","options":{"indexing":{"record":"position","fieldnorms":true,"tokenizer":"default"},"stored":false,"fast":false}}],"opstamp":0}"#;
let err = serde_json::from_str::<UntrackedIndexMeta>(json).unwrap_err();
assert_eq!(
err.to_string(),
"unsupported variant `zstd`, please enable Tantivy's `zstd-compression` feature at \
line 1 column 95"
.to_string()
);
}
#[test]
#[cfg(feature = "lz4-compression")]
fn test_index_settings_default() {

12
src/core/json_utils.rs
View File

@@ -60,7 +60,7 @@ impl IndexingPositionsPerPath {
fn get_position(&mut self, term: &Term) -> &mut IndexingPosition {
self.positions_per_path
.entry(murmurhash2(term.serialized_term()))
.or_insert_with(Default::default)
.or_default()
}
}
@@ -619,21 +619,21 @@ mod tests {
#[test]
fn test_split_json_path_escaped_dot() {
let json_path = split_json_path(r#"toto\.titi"#);
let json_path = split_json_path(r"toto\.titi");
assert_eq!(&json_path, &["toto.titi"]);
let json_path_2 = split_json_path(r#"k8s\.container\.name"#);
let json_path_2 = split_json_path(r"k8s\.container\.name");
assert_eq!(&json_path_2, &["k8s.container.name"]);
}
#[test]
fn test_split_json_path_escaped_backslash() {
let json_path = split_json_path(r#"toto\\titi"#);
assert_eq!(&json_path, &[r#"toto\titi"#]);
let json_path = split_json_path(r"toto\\titi");
assert_eq!(&json_path, &[r"toto\titi"]);
}
#[test]
fn test_split_json_path_escaped_normal_letter() {
let json_path = split_json_path(r#"toto\titi"#);
let json_path = split_json_path(r"toto\titi");
assert_eq!(&json_path, &[r#"tototiti"#]);
}
}

2
src/core/tests.rs
View File

@@ -283,7 +283,7 @@ fn test_single_segment_index_writer() -> crate::Result<()> {
let directory = RamDirectory::default();
let mut single_segment_index_writer = Index::builder()
.schema(schema)
.single_segment_index_writer(directory, 10_000_000)?;
.single_segment_index_writer(directory, 15_000_000)?;
for _ in 0..10 {
let doc = doc!(text_field=>"hello");
single_segment_index_writer.add_document(doc)?;

6
src/directory/directory.rs
View File

@@ -73,7 +73,7 @@ impl From<io::Error> for TryAcquireLockError {
fn try_acquire_lock(
filepath: &Path,
directory: &mut dyn Directory,
directory: &dyn Directory,
) -> Result<DirectoryLock, TryAcquireLockError> {
let mut write = directory.open_write(filepath).map_err(|e| match e {
OpenWriteError::FileAlreadyExists(_) => TryAcquireLockError::FileExists,
@@ -191,10 +191,10 @@ pub trait Directory: DirectoryClone + fmt::Debug + Send + Sync + 'static {
///
/// The method is blocking or not depending on the [`Lock`] object.
fn acquire_lock(&self, lock: &Lock) -> Result<DirectoryLock, LockError> {
let mut box_directory = self.box_clone();
let box_directory = self.box_clone();
let mut retry_policy = retry_policy(lock.is_blocking);
loop {
match try_acquire_lock(&lock.filepath, &mut *box_directory) {
match try_acquire_lock(&lock.filepath, &*box_directory) {
Ok(result) => {
return Ok(result);
}

6
src/fastfield/mod.rs
View File

@@ -939,7 +939,7 @@ mod tests {
.unwrap()
.first_or_default_col(0);
let numbers = vec![100, 200, 300];
let numbers = [100, 200, 300];
let test_range = |range: RangeInclusive<u64>| {
let expexted_count = numbers.iter().filter(|num| range.contains(num)).count();
let mut vec = vec![];
@@ -1013,7 +1013,7 @@ mod tests {
.unwrap()
.first_or_default_col(0);
let numbers = vec![1000, 1001, 1003];
let numbers = [1000, 1001, 1003];
let test_range = |range: RangeInclusive<u64>| {
let expexted_count = numbers.iter().filter(|num| range.contains(num)).count();
let mut vec = vec![];
@@ -1098,7 +1098,7 @@ mod tests {
.unwrap()
.is_none());
let column = fast_field_reader
.column_opt::<i64>(r#"json.attr\.age"#)
.column_opt::<i64>(r"json.attr\.age")
.unwrap()
.unwrap();
let vals: Vec<i64> = column.values_for_doc(0u32).collect();

61
src/fastfield/readers.rs
View File

@@ -234,6 +234,22 @@ impl FastFieldReaders {
Ok(dynamic_column_handle_opt)
}
/// Returning all `dynamic_column_handle`.
pub fn dynamic_column_handles(
&self,
field_name: &str,
) -> crate::Result<Vec<DynamicColumnHandle>> {
let Some(resolved_field_name) = self.resolve_field(field_name)? else {
return Ok(Vec::new());
};
let dynamic_column_handles = self
.columnar
.read_columns(&resolved_field_name)?
.into_iter()
.collect();
Ok(dynamic_column_handles)
}
#[doc(hidden)]
pub async fn list_dynamic_column_handles(
&self,
@@ -276,7 +292,7 @@ impl FastFieldReaders {
}
/// Returns the all `u64` column used to represent any `u64`-mapped typed (String/Bytes term
/// ids, i64, u64, f64, DateTime).
/// ids, i64, u64, f64, bool, DateTime).
///
/// In case of JSON, there may be two columns. One for term and one for numerical types. (This
/// may change later to 3 types if JSON handles DateTime)
@@ -338,6 +354,8 @@ impl FastFieldReaders {
#[cfg(test)]
mod tests {
use columnar::ColumnType;
use crate::schema::{JsonObjectOptions, Schema, FAST};
use crate::{Document, Index};
@@ -417,4 +435,45 @@ mod tests {
Some("_dyna\u{1}notinschema\u{1}attr\u{1}color".to_string())
);
}
#[test]
fn test_fast_field_reader_dynamic_column_handles() {
let mut schema_builder = Schema::builder();
let id = schema_builder.add_u64_field("id", FAST);
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 = index.writer_for_tests().unwrap();
index_writer
.add_document(doc!(id=> 1u64, json => json!({"foo": 42})))
.unwrap();
index_writer
.add_document(doc!(id=> 2u64, json => json!({"foo": true})))
.unwrap();
index_writer
.add_document(doc!(id=> 3u64, json => json!({"foo": "bar"})))
.unwrap();
index_writer.commit().unwrap();
let reader = index.reader().unwrap();
let searcher = reader.searcher();
let reader = searcher.segment_reader(0u32);
let fast_fields = reader.fast_fields();
let id_columns = fast_fields.dynamic_column_handles("id").unwrap();
assert_eq!(id_columns.len(), 1);
assert_eq!(id_columns.first().unwrap().column_type(), ColumnType::U64);
let foo_columns = fast_fields.dynamic_column_handles("json.foo").unwrap();
assert_eq!(foo_columns.len(), 3);
assert!(foo_columns
.iter()
.any(|column| column.column_type() == ColumnType::I64));
assert!(foo_columns
.iter()
.any(|column| column.column_type() == ColumnType::Bool));
assert!(foo_columns
.iter()
.any(|column| column.column_type() == ColumnType::Str));
println!("*** {:?}", fast_fields.columnar().list_columns());
}
}

3
src/functional_test.rs
View File

@@ -2,6 +2,7 @@ use std::collections::HashSet;
use rand::{thread_rng, Rng};
use crate::indexer::index_writer::MEMORY_BUDGET_NUM_BYTES_MIN;
use crate::schema::*;
use crate::{doc, schema, Index, IndexSettings, IndexSortByField, Order, Searcher};
@@ -30,7 +31,7 @@ fn test_functional_store() -> crate::Result<()> {
let mut rng = thread_rng();
let mut index_writer = index.writer_with_num_threads(3, 12_000_000)?;
let mut index_writer = index.writer_with_num_threads(3, MEMORY_BUDGET_NUM_BYTES_MIN)?;
let mut doc_set: Vec<u64> = Vec::new();

124
src/indexer/doc_id_mapping.rs
View File

@@ -152,8 +152,11 @@ pub(crate) fn get_doc_id_mapping_from_field(
#[cfg(test)]
mod tests_indexsorting {
use common::DateTime;
use crate::collector::TopDocs;
use crate::indexer::doc_id_mapping::DocIdMapping;
use crate::indexer::NoMergePolicy;
use crate::query::QueryParser;
use crate::schema::{Schema, *};
use crate::{DocAddress, Index, IndexSettings, IndexSortByField, Order};
@@ -444,48 +447,93 @@ mod tests_indexsorting {
Ok(())
}
// #[test]
// fn test_sort_index_fast_field() -> crate::Result<()> {
// let index = create_test_index(
// Some(IndexSettings {
// sort_by_field: Some(IndexSortByField {
// field: "my_number".to_string(),
// order: Order::Asc,
// }),
// ..Default::default()
// }),
// get_text_options(),
// )?;
// assert_eq!(
// index.settings().sort_by_field.as_ref().unwrap().field,
// "my_number".to_string()
// );
#[test]
fn test_sort_index_fast_field() -> crate::Result<()> {
let index = create_test_index(
Some(IndexSettings {
sort_by_field: Some(IndexSortByField {
field: "my_number".to_string(),
order: Order::Asc,
}),
..Default::default()
}),
get_text_options(),
)?;
assert_eq!(
index.settings().sort_by_field.as_ref().unwrap().field,
"my_number".to_string()
);
// let searcher = index.reader()?.searcher();
// assert_eq!(searcher.segment_readers().len(), 1);
// let segment_reader = searcher.segment_reader(0);
// let fast_fields = segment_reader.fast_fields();
// let my_number = index.schema().get_field("my_number").unwrap();
let searcher = index.reader()?.searcher();
assert_eq!(searcher.segment_readers().len(), 1);
let segment_reader = searcher.segment_reader(0);
let fast_fields = segment_reader.fast_fields();
// let fast_field = fast_fields.u64(my_number).unwrap();
// assert_eq!(fast_field.get_val(0), 10u64);
// assert_eq!(fast_field.get_val(1), 20u64);
// assert_eq!(fast_field.get_val(2), 30u64);
let fast_field = fast_fields
.u64("my_number")
.unwrap()
.first_or_default_col(999);
assert_eq!(fast_field.get_val(0), 10u64);
assert_eq!(fast_field.get_val(1), 20u64);
assert_eq!(fast_field.get_val(2), 30u64);
// let multi_numbers = index.schema().get_field("multi_numbers").unwrap();
// let multifield = fast_fields.u64s(multi_numbers).unwrap();
// let mut vals = vec![];
// multifield.get_vals(0u32, &mut vals);
// assert_eq!(vals, &[] as &[u64]);
// let mut vals = vec![];
// multifield.get_vals(1u32, &mut vals);
// assert_eq!(vals, &[5, 6]);
let multifield = fast_fields.u64("multi_numbers").unwrap();
let vals: Vec<u64> = multifield.values_for_doc(0u32).collect();
assert_eq!(vals, &[] as &[u64]);
let vals: Vec<_> = multifield.values_for_doc(1u32).collect();
assert_eq!(vals, &[5, 6]);
// let mut vals = vec![];
// multifield.get_vals(2u32, &mut vals);
// assert_eq!(vals, &[3]);
// Ok(())
// }
let vals: Vec<_> = multifield.values_for_doc(2u32).collect();
assert_eq!(vals, &[3]);
Ok(())
}
#[test]
fn test_with_sort_by_date_field() -> crate::Result<()> {
let mut schema_builder = Schema::builder();
let date_field = schema_builder.add_date_field("date", INDEXED | STORED | FAST);
let schema = schema_builder.build();
let settings = IndexSettings {
sort_by_field: Some(IndexSortByField {
field: "date".to_string(),
order: Order::Desc,
}),
..Default::default()
};
let index = Index::builder()
.schema(schema)
.settings(settings)
.create_in_ram()?;
let mut index_writer = index.writer_for_tests()?;
index_writer.set_merge_policy(Box::new(NoMergePolicy));
index_writer.add_document(doc!(
date_field => DateTime::from_timestamp_secs(1000),
))?;
index_writer.add_document(doc!(
date_field => DateTime::from_timestamp_secs(999),
))?;
index_writer.add_document(doc!(
date_field => DateTime::from_timestamp_secs(1001),
))?;
index_writer.commit()?;
let searcher = index.reader()?.searcher();
assert_eq!(searcher.segment_readers().len(), 1);
let segment_reader = searcher.segment_reader(0);
let fast_fields = segment_reader.fast_fields();
let fast_field = fast_fields
.date("date")
.unwrap()
.first_or_default_col(DateTime::from_timestamp_secs(0));
assert_eq!(fast_field.get_val(0), DateTime::from_timestamp_secs(1001));
assert_eq!(fast_field.get_val(1), DateTime::from_timestamp_secs(1000));
assert_eq!(fast_field.get_val(2), DateTime::from_timestamp_secs(999));
Ok(())
}
#[test]
fn test_doc_mapping() {

88
src/indexer/index_writer.rs
View File

@@ -27,9 +27,9 @@ use crate::{FutureResult, Opstamp};
// in the `memory_arena` goes below MARGIN_IN_BYTES.
pub const MARGIN_IN_BYTES: usize = 1_000_000;
// We impose the memory per thread to be at least 3 MB.
pub const MEMORY_ARENA_NUM_BYTES_MIN: usize = ((MARGIN_IN_BYTES as u32) * 3u32) as usize;
pub const MEMORY_ARENA_NUM_BYTES_MAX: usize = u32::MAX as usize - MARGIN_IN_BYTES;
// We impose the memory per thread to be at least 15 MB, as the baseline consumption is 12MB.
pub const MEMORY_BUDGET_NUM_BYTES_MIN: usize = ((MARGIN_IN_BYTES as u32) * 15u32) as usize;
pub const MEMORY_BUDGET_NUM_BYTES_MAX: usize = u32::MAX as usize - MARGIN_IN_BYTES;
// We impose the number of index writer threads to be at most this.
pub const MAX_NUM_THREAD: usize = 8;
@@ -57,7 +57,8 @@ pub struct IndexWriter {
index: Index,
memory_arena_in_bytes_per_thread: usize,
// The memory budget per thread, after which a commit is triggered.
memory_budget_in_bytes_per_thread: usize,
workers_join_handle: Vec<JoinHandle<crate::Result<()>>>,
@@ -167,7 +168,7 @@ fn index_documents(
memory_budget: usize,
segment: Segment,
grouped_document_iterator: &mut dyn Iterator<Item = AddBatch>,
segment_updater: &mut SegmentUpdater,
segment_updater: &SegmentUpdater,
mut delete_cursor: DeleteCursor,
) -> crate::Result<()> {
let mut segment_writer = SegmentWriter::for_segment(memory_budget, segment.clone())?;
@@ -264,19 +265,19 @@ impl IndexWriter {
pub(crate) fn new(
index: &Index,
num_threads: usize,
memory_arena_in_bytes_per_thread: usize,
memory_budget_in_bytes_per_thread: usize,
directory_lock: DirectoryLock,
) -> crate::Result<IndexWriter> {
if memory_arena_in_bytes_per_thread < MEMORY_ARENA_NUM_BYTES_MIN {
if memory_budget_in_bytes_per_thread < MEMORY_BUDGET_NUM_BYTES_MIN {
let err_msg = format!(
"The memory arena in bytes per thread needs to be at least \
{MEMORY_ARENA_NUM_BYTES_MIN}."
{MEMORY_BUDGET_NUM_BYTES_MIN}."
);
return Err(TantivyError::InvalidArgument(err_msg));
}
if memory_arena_in_bytes_per_thread >= MEMORY_ARENA_NUM_BYTES_MAX {
if memory_budget_in_bytes_per_thread >= MEMORY_BUDGET_NUM_BYTES_MAX {
let err_msg = format!(
"The memory arena in bytes per thread cannot exceed {MEMORY_ARENA_NUM_BYTES_MAX}"
"The memory arena in bytes per thread cannot exceed {MEMORY_BUDGET_NUM_BYTES_MAX}"
);
return Err(TantivyError::InvalidArgument(err_msg));
}
@@ -295,7 +296,7 @@ impl IndexWriter {
let mut index_writer = IndexWriter {
_directory_lock: Some(directory_lock),
memory_arena_in_bytes_per_thread,
memory_budget_in_bytes_per_thread,
index: index.clone(),
index_writer_status: IndexWriterStatus::from(document_receiver),
operation_sender: document_sender,
@@ -392,11 +393,11 @@ impl IndexWriter {
let document_receiver_clone = self.operation_receiver()?;
let index_writer_bomb = self.index_writer_status.create_bomb();
let mut segment_updater = self.segment_updater.clone();
let segment_updater = self.segment_updater.clone();
let mut delete_cursor = self.delete_queue.cursor();
let mem_budget = self.memory_arena_in_bytes_per_thread;
let mem_budget = self.memory_budget_in_bytes_per_thread;
let index = self.index.clone();
let join_handle: JoinHandle<crate::Result<()>> = thread::Builder::new()
.name(format!("thrd-tantivy-index{}", self.worker_id))
@@ -428,7 +429,7 @@ impl IndexWriter {
mem_budget,
index.new_segment(),
&mut document_iterator,
&mut segment_updater,
&segment_updater,
delete_cursor.clone(),
)?;
}
@@ -554,7 +555,7 @@ impl IndexWriter {
let new_index_writer: IndexWriter = IndexWriter::new(
&self.index,
self.num_threads,
self.memory_arena_in_bytes_per_thread,
self.memory_budget_in_bytes_per_thread,
directory_lock,
)?;
@@ -619,7 +620,7 @@ impl IndexWriter {
for worker_handle in former_workers_join_handle {
let indexing_worker_result = worker_handle
.join()
.map_err(|e| TantivyError::ErrorInThread(format!("{e:?}")))?;
.map_err(|e| TantivyError::ErrorInThread(e.to_string()))?;
indexing_worker_result?;
self.add_indexing_worker()?;
}
@@ -810,6 +811,7 @@ mod tests {
use crate::collector::TopDocs;
use crate::directory::error::LockError;
use crate::error::*;
use crate::indexer::index_writer::MEMORY_BUDGET_NUM_BYTES_MIN;
use crate::indexer::NoMergePolicy;
use crate::query::{BooleanQuery, Occur, Query, QueryParser, TermQuery};
use crate::schema::{
@@ -941,7 +943,7 @@ mod tests {
fn test_empty_operations_group() {
let schema_builder = schema::Schema::builder();
let index = Index::create_in_ram(schema_builder.build());
let index_writer = index.writer(3_000_000).unwrap();
let index_writer = index.writer_for_tests().unwrap();
let operations1 = vec![];
let batch_opstamp1 = index_writer.run(operations1).unwrap();
assert_eq!(batch_opstamp1, 0u64);
@@ -954,8 +956,8 @@ mod tests {
fn test_lockfile_stops_duplicates() {
let schema_builder = schema::Schema::builder();
let index = Index::create_in_ram(schema_builder.build());
let _index_writer = index.writer(3_000_000).unwrap();
match index.writer(3_000_000) {
let _index_writer = index.writer_for_tests().unwrap();
match index.writer_for_tests() {
Err(TantivyError::LockFailure(LockError::LockBusy, _)) => {}
_ => panic!("Expected a `LockFailure` error"),
}
@@ -979,7 +981,7 @@ mod tests {
fn test_set_merge_policy() {
let schema_builder = schema::Schema::builder();
let index = Index::create_in_ram(schema_builder.build());
let index_writer = index.writer(3_000_000).unwrap();
let index_writer = index.writer_for_tests().unwrap();
assert_eq!(
format!("{:?}", index_writer.get_merge_policy()),
"LogMergePolicy { min_num_segments: 8, max_docs_before_merge: 10000000, \
@@ -998,11 +1000,11 @@ mod tests {
let schema_builder = schema::Schema::builder();
let index = Index::create_in_ram(schema_builder.build());
{
let _index_writer = index.writer(3_000_000).unwrap();
let _index_writer = index.writer_for_tests().unwrap();
// the lock should be released when the
// index_writer leaves the scope.
}
let _index_writer_two = index.writer(3_000_000).unwrap();
let _index_writer_two = index.writer_for_tests().unwrap();
}
#[test]
@@ -1022,7 +1024,7 @@ mod tests {
{
// writing the segment
let mut index_writer = index.writer(3_000_000)?;
let mut index_writer = index.writer_for_tests()?;
index_writer.add_document(doc!(text_field=>"a"))?;
index_writer.rollback()?;
assert_eq!(index_writer.commit_opstamp(), 0u64);
@@ -1054,7 +1056,7 @@ mod tests {
reader.searcher().doc_freq(&term_a).unwrap()
};
// writing the segment
let mut index_writer = index.writer(12_000_000).unwrap();
let mut index_writer = index.writer_for_tests().unwrap();
index_writer.add_document(doc!(text_field=>"a"))?;
index_writer.commit()?;
// this should create 1 segment
@@ -1094,7 +1096,7 @@ mod tests {
reader.searcher().doc_freq(&term_a).unwrap()
};
// writing the segment
let mut index_writer = index.writer(12_000_000).unwrap();
let mut index_writer = index.writer_for_tests().unwrap();
index_writer.add_document(doc!(text_field=>"a"))?;
index_writer.commit()?;
index_writer.add_document(doc!(text_field=>"a"))?;
@@ -1140,7 +1142,7 @@ mod tests {
reader.searcher().doc_freq(&term_a).unwrap()
};
// writing the segment
let mut index_writer = index.writer(12_000_000).unwrap();
let mut index_writer = index.writer(MEMORY_BUDGET_NUM_BYTES_MIN).unwrap();
// create 8 segments with 100 tiny docs
for _doc in 0..100 {
index_writer.add_document(doc!(text_field=>"a"))?;
@@ -1196,7 +1198,8 @@ mod tests {
{
// writing the segment
let mut index_writer = index.writer_with_num_threads(4, 12_000_000)?;
let mut index_writer =
index.writer_with_num_threads(4, MEMORY_BUDGET_NUM_BYTES_MIN * 4)?;
// create 8 segments with 100 tiny docs
for _doc in 0..100 {
index_writer.add_document(doc!(text_field => "a"))?;
@@ -1245,7 +1248,9 @@ mod tests {
let term = Term::from_field_text(text_field, s);
searcher.doc_freq(&term).unwrap()
};
let mut index_writer = index.writer_with_num_threads(4, 12_000_000).unwrap();
let mut index_writer = index
.writer_with_num_threads(4, MEMORY_BUDGET_NUM_BYTES_MIN * 4)
.unwrap();
let add_tstamp = index_writer.add_document(doc!(text_field => "a")).unwrap();
let commit_tstamp = index_writer.commit().unwrap();
@@ -1262,7 +1267,9 @@ mod tests {
let mut schema_builder = schema::Schema::builder();
let text_field = schema_builder.add_text_field("text", TEXT);
let index = Index::create_in_ram(schema_builder.build());
let mut index_writer = index.writer_with_num_threads(4, 12_000_000).unwrap();
let mut index_writer = index
.writer_with_num_threads(4, MEMORY_BUDGET_NUM_BYTES_MIN * 4)
.unwrap();
let add_tstamp = index_writer.add_document(doc!(text_field => "a")).unwrap();
@@ -1311,7 +1318,9 @@ mod tests {
let text_field = schema_builder.add_text_field("text", TEXT);
let index = Index::create_in_ram(schema_builder.build());
// writing the segment
let mut index_writer = index.writer_with_num_threads(4, 12_000_000).unwrap();
let mut index_writer = index
.writer_with_num_threads(4, MEMORY_BUDGET_NUM_BYTES_MIN * 4)
.unwrap();
let res = index_writer.delete_all_documents();
assert!(res.is_ok());
@@ -1338,7 +1347,9 @@ mod tests {
let mut schema_builder = schema::Schema::builder();
let text_field = schema_builder.add_text_field("text", TEXT);
let index = Index::create_in_ram(schema_builder.build());
let mut index_writer = index.writer_with_num_threads(4, 12_000_000).unwrap();
let mut index_writer = index
.writer_with_num_threads(4, MEMORY_BUDGET_NUM_BYTES_MIN * 4)
.unwrap();
// add one simple doc
assert!(index_writer.add_document(doc!(text_field => "a")).is_ok());
@@ -1371,7 +1382,9 @@ mod tests {
fn test_delete_all_documents_empty_index() {
let schema_builder = schema::Schema::builder();
let index = Index::create_in_ram(schema_builder.build());
let mut index_writer = index.writer_with_num_threads(4, 12_000_000).unwrap();
let mut index_writer = index
.writer_with_num_threads(4, MEMORY_BUDGET_NUM_BYTES_MIN * 4)
.unwrap();
let clear = index_writer.delete_all_documents();
let commit = index_writer.commit();
assert!(clear.is_ok());
@@ -1382,7 +1395,9 @@ mod tests {
fn test_delete_all_documents_index_twice() {
let schema_builder = schema::Schema::builder();
let index = Index::create_in_ram(schema_builder.build());
let mut index_writer = index.writer_with_num_threads(4, 12_000_000).unwrap();
let mut index_writer = index
.writer_with_num_threads(4, MEMORY_BUDGET_NUM_BYTES_MIN * 4)
.unwrap();
let clear = index_writer.delete_all_documents();
let commit = index_writer.commit();
assert!(clear.is_ok());
@@ -2426,6 +2441,13 @@ mod tests {
test_operation_strategy(&ops[..], false, true).unwrap();
}
#[test]
fn test_merge_regression_1() {
use IndexingOp::*;
let ops = &[AddDoc { id: 15 }, Commit, AddDoc { id: 9 }, Commit, Merge];
test_operation_strategy(&ops[..], false, true).unwrap();
}
#[test]
fn test_range_query_bug_1() {
use IndexingOp::*;

2
src/indexer/merger.rs
View File

@@ -178,7 +178,7 @@ impl IndexMerger {
alive_bitset_opt: Vec<Option<AliveBitSet>>,
) -> crate::Result<IndexMerger> {
let mut readers = vec![];
for (segment, new_alive_bitset_opt) in segments.iter().zip(alive_bitset_opt.into_iter()) {
for (segment, new_alive_bitset_opt) in segments.iter().zip(alive_bitset_opt) {
if segment.meta().num_docs() > 0 {
let reader =
SegmentReader::open_with_custom_alive_set(segment, new_alive_bitset_opt)?;

4
src/indexer/mod.rs
View File

@@ -89,7 +89,7 @@ mod tests_mmap {
let parse_query = QueryParser::for_index(&index, Vec::new());
{
let query = parse_query
.parse_query(r#"json.k8s\.container\.name:prometheus"#)
.parse_query(r"json.k8s\.container\.name:prometheus")
.unwrap();
let num_docs = searcher.search(&query, &Count).unwrap();
assert_eq!(num_docs, 1);
@@ -127,7 +127,7 @@ mod tests_mmap {
}
{
let query = parse_query
.parse_query(r#"json.k8s\.container\.name:prometheus"#)
.parse_query(r"json.k8s\.container\.name:prometheus")
.unwrap();
let num_docs = searcher.search(&query, &Count).unwrap();
assert_eq!(num_docs, 1);

5
src/indexer/segment_updater.rs
View File

@@ -596,10 +596,15 @@ impl SegmentUpdater {
);
{
if let Some(after_merge_segment_entry) = after_merge_segment_entry.as_mut() {
// Deletes and commits could have happened as we were merging.
// We need to make sure we are up to date with deletes before accepting the
// segment.
let mut delete_cursor = after_merge_segment_entry.delete_cursor().clone();
if let Some(delete_operation) = delete_cursor.get() {
let committed_opstamp = segment_updater.load_meta().opstamp;
if delete_operation.opstamp < committed_opstamp {
// We are not up to date! Let's create a new tombstone file for our
// freshly create split.
let index = &segment_updater.index;
let segment = index.segment(after_merge_segment_entry.meta().clone());
if let Err(advance_deletes_err) = advance_deletes(

2
src/indexer/segment_writer.rs
View File

@@ -26,6 +26,8 @@ use crate::{DocId, Document, Opstamp, SegmentComponent, TantivyError};
fn compute_initial_table_size(per_thread_memory_budget: usize) -> crate::Result<usize> {
let table_memory_upper_bound = per_thread_memory_budget / 3;
(10..20) // We cap it at 2^19 = 512K capacity.
// TODO: There are cases where this limit causes a
// reallocation in the hashmap. Check if this affects performance.
.map(|power| 1 << power)
.take_while(|capacity| compute_table_memory_size(*capacity) < table_memory_upper_bound)
.last()

2
src/lib.rs
View File

@@ -191,7 +191,7 @@ pub use crate::schema::{DateOptions, DateTimePrecision, Document, Term};
/// Index format version.
const INDEX_FORMAT_VERSION: u32 = 5;
#[cfg(unix)]
#[cfg(all(feature = "mmap", unix))]
pub use memmap2::Advice;
/// Structure version for the index.

2
src/positions/mod.rs
View File

@@ -119,7 +119,7 @@ pub mod tests {
serializer.close_term()?;
serializer.close()?;
let position_delta = OwnedBytes::new(positions_buffer);
let mut output_delta_pos_buffer = vec![0u32; 5];
let mut output_delta_pos_buffer = [0u32; 5];
let mut position_reader = PositionReader::open(position_delta)?;
position_reader.read(0, &mut output_delta_pos_buffer[..]);
assert_eq!(

2
src/postings/mod.rs
View File

@@ -225,7 +225,7 @@ pub mod tests {
{
let mut segment_writer =
SegmentWriter::for_segment(3_000_000, segment.clone()).unwrap();
SegmentWriter::for_segment(15_000_000, segment.clone()).unwrap();
{
// checking that position works if the field has two values
let op = AddOperation {

29
src/query/automaton_weight.rs
View File

@@ -4,6 +4,7 @@ use std::sync::Arc;
use common::BitSet;
use tantivy_fst::Automaton;
use super::phrase_prefix_query::prefix_end;
use crate::core::SegmentReader;
use crate::query::{BitSetDocSet, ConstScorer, Explanation, Scorer, Weight};
use crate::schema::{Field, IndexRecordOption};
@@ -14,6 +15,10 @@ use crate::{DocId, Score, TantivyError};
pub struct AutomatonWeight<A> {
field: Field,
automaton: Arc<A>,
// For JSON fields, the term dictionary include terms from all paths.
// We apply additional filtering based on the given JSON path, when searching within the term
// dictionary. This prevents terms from unrelated paths from matching the search criteria.
json_path_bytes: Option<Box<[u8]>>,
}
impl<A> AutomatonWeight<A>
@@ -26,6 +31,20 @@ where
AutomatonWeight {
field,
automaton: automaton.into(),
json_path_bytes: None,
}
}
/// Create a new AutomationWeight for a json path
pub fn new_for_json_path<IntoArcA: Into<Arc<A>>>(
field: Field,
automaton: IntoArcA,
json_path_bytes: &[u8],
) -> AutomatonWeight<A> {
AutomatonWeight {
field,
automaton: automaton.into(),
json_path_bytes: Some(json_path_bytes.to_vec().into_boxed_slice()),
}
}
@@ -34,7 +53,15 @@ where
term_dict: &'a TermDictionary,
) -> io::Result<TermStreamer<'a, &'a A>> {
let automaton: &A = &self.automaton;
let term_stream_builder = term_dict.search(automaton);
let mut term_stream_builder = term_dict.search(automaton);
if let Some(json_path_bytes) = &self.json_path_bytes {
term_stream_builder = term_stream_builder.ge(json_path_bytes);
if let Some(end) = prefix_end(json_path_bytes) {
term_stream_builder = term_stream_builder.lt(&end);
}
}
term_stream_builder.into_stream()
}
}

2
src/query/boolean_query/boolean_query.rs
View File

@@ -32,7 +32,7 @@ use crate::schema::{IndexRecordOption, Term};
/// let schema = schema_builder.build();
/// let index = Index::create_in_ram(schema);
/// {
/// let mut index_writer = index.writer(3_000_000)?;
/// let mut index_writer = index.writer(15_000_000)?;
/// index_writer.add_document(doc!(
/// title => "The Name of the Wind",
/// ))?;

4
src/query/boolean_query/boolean_weight.rs
View File

@@ -98,7 +98,7 @@ impl<TScoreCombiner: ScoreCombiner> BooleanWeight<TScoreCombiner> {
let sub_scorer: Box<dyn Scorer> = subweight.scorer(reader, boost)?;
per_occur_scorers
.entry(*occur)
.or_insert_with(Vec::new)
.or_default()
.push(sub_scorer);
}
Ok(per_occur_scorers)
@@ -193,7 +193,7 @@ impl<TScoreCombiner: ScoreCombiner + Sync> Weight for BooleanWeight<TScoreCombin
return Ok(Explanation::new("BooleanQuery with no scoring", 1.0));
}
let mut explanation = Explanation::new("BooleanClause. Sum of ...", scorer.score());
let mut explanation = Explanation::new("BooleanClause. sum of ...", scorer.score());
for (occur, subweight) in &self.weights {
if is_positive_occur(*occur) {
if let Ok(child_explanation) = subweight.explain(reader, doc) {

2
src/query/boolean_query/mod.rs
View File

@@ -297,7 +297,7 @@ mod tests {
let text = schema_builder.add_text_field("text", STRING);
let schema = schema_builder.build();
let index = Index::create_in_ram(schema);
let mut index_writer = index.writer_with_num_threads(1, 5_000_000)?;
let mut index_writer = index.writer_for_tests()?;
index_writer.add_document(doc!(text=>"a"))?;
index_writer.add_document(doc!(text=>"b"))?;
index_writer.commit()?;

9
src/query/boost_query.rs
View File

@@ -2,7 +2,6 @@ use std::fmt;
use crate::docset::BUFFER_LEN;
use crate::fastfield::AliveBitSet;
use crate::query::explanation::does_not_match;
use crate::query::{EnableScoring, Explanation, Query, Scorer, Weight};
use crate::{DocId, DocSet, Score, SegmentReader, Term};
@@ -73,13 +72,9 @@ impl Weight for BoostWeight {
}
fn explain(&self, reader: &SegmentReader, doc: u32) -> crate::Result<Explanation> {
let mut scorer = self.scorer(reader, 1.0)?;
if scorer.seek(doc) != doc {
return Err(does_not_match(doc));
}
let mut explanation =
Explanation::new(format!("Boost x{} of ...", self.boost), scorer.score());
let underlying_explanation = self.weight.explain(reader, doc)?;
let score = underlying_explanation.value() * self.boost;
let mut explanation = Explanation::new(format!("Boost x{} of ...", self.boost), score);
explanation.add_detail(underlying_explanation);
Ok(explanation)
}

2
src/query/disjunction_max_query.rs
View File

@@ -23,7 +23,7 @@ use crate::{Score, Term};
/// let schema = schema_builder.build();
/// let index = Index::create_in_ram(schema);
/// {
/// let mut index_writer = index.writer(3_000_000)?;
/// let mut index_writer = index.writer(15_000_000)?;
/// index_writer.add_document(doc!(
/// title => "The Name of Girl",
/// ))?;

345
src/query/exist_query.rs Normal file
View File

@@ -0,0 +1,345 @@
use core::fmt::Debug;
use columnar::{ColumnIndex, DynamicColumn};
use super::{ConstScorer, EmptyScorer};
use crate::core::SegmentReader;
use crate::docset::{DocSet, TERMINATED};
use crate::query::explanation::does_not_match;
use crate::query::{EnableScoring, Explanation, Query, Scorer, Weight};
use crate::{DocId, Score, TantivyError};
/// Query that matches all documents with a non-null value in the specified field.
///
/// All of the matched documents get the score 1.0.
#[derive(Clone, Debug)]
pub struct ExistsQuery {
field_name: String,
}
impl ExistsQuery {
/// Creates a new `ExistQuery` from the given field.
///
/// This query matches all documents with at least one non-null value in the specified field.
/// This constructor never fails, but executing the search with this query will return an
/// error if the specified field doesn't exists or is not a fast field.
pub fn new_exists_query(field: String) -> ExistsQuery {
ExistsQuery { field_name: field }
}
}
impl Query for ExistsQuery {
fn weight(&self, enable_scoring: EnableScoring) -> crate::Result<Box<dyn Weight>> {
let schema = enable_scoring.schema();
let Some((field, _path)) = schema.find_field(&self.field_name) else {
return Err(TantivyError::FieldNotFound(self.field_name.clone()));
};
let field_type = schema.get_field_entry(field).field_type();
if !field_type.is_fast() {
return Err(TantivyError::SchemaError(format!(
"Field {} is not a fast field.",
self.field_name
)));
}
Ok(Box::new(ExistsWeight {
field_name: self.field_name.clone(),
}))
}
}
/// Weight associated with the `ExistsQuery` query.
pub struct ExistsWeight {
field_name: String,
}
impl Weight for ExistsWeight {
fn scorer(&self, reader: &SegmentReader, boost: Score) -> crate::Result<Box<dyn Scorer>> {
let fast_field_reader = reader.fast_fields();
let dynamic_columns: crate::Result<Vec<DynamicColumn>> = fast_field_reader
.dynamic_column_handles(&self.field_name)?
.into_iter()
.map(|handle| handle.open().map_err(|io_error| io_error.into()))
.collect();
let mut non_empty_columns = Vec::new();
for column in dynamic_columns? {
if !matches!(column.column_index(), ColumnIndex::Empty { .. }) {
non_empty_columns.push(column)
}
}
// TODO: we can optimizer more here since in most cases we will have only one index
if !non_empty_columns.is_empty() {
let docset = ExistsDocSet::new(non_empty_columns, reader.max_doc());
Ok(Box::new(ConstScorer::new(docset, boost)))
} else {
Ok(Box::new(EmptyScorer))
}
}
fn explain(&self, reader: &SegmentReader, doc: DocId) -> crate::Result<Explanation> {
let mut scorer = self.scorer(reader, 1.0)?;
if scorer.seek(doc) != doc {
return Err(does_not_match(doc));
}
Ok(Explanation::new("ExistsQuery", 1.0))
}
}
pub(crate) struct ExistsDocSet {
columns: Vec<DynamicColumn>,
doc: DocId,
max_doc: DocId,
}
impl ExistsDocSet {
pub(crate) fn new(columns: Vec<DynamicColumn>, max_doc: DocId) -> Self {
let mut set = Self {
columns,
doc: 0u32,
max_doc,
};
set.find_next();
set
}
fn find_next(&mut self) -> DocId {
while self.doc < self.max_doc {
if self
.columns
.iter()
.any(|col| col.column_index().has_value(self.doc))
{
return self.doc;
}
self.doc += 1;
}
self.doc = TERMINATED;
TERMINATED
}
}
impl DocSet for ExistsDocSet {
fn advance(&mut self) -> DocId {
self.seek(self.doc + 1)
}
fn size_hint(&self) -> u32 {
0
}
fn doc(&self) -> DocId {
self.doc
}
#[inline(always)]
fn seek(&mut self, target: DocId) -> DocId {
self.doc = target;
self.find_next()
}
}
#[cfg(test)]
mod tests {
use std::net::Ipv6Addr;
use std::ops::Bound;
use common::DateTime;
use time::OffsetDateTime;
use crate::collector::Count;
use crate::query::exist_query::ExistsQuery;
use crate::query::{BooleanQuery, RangeQuery};
use crate::schema::{Facet, FacetOptions, Schema, FAST, INDEXED, STRING, TEXT};
use crate::{doc, Index, Searcher};
#[test]
fn test_exists_query_simple() -> crate::Result<()> {
let mut schema_builder = Schema::builder();
let all_field = schema_builder.add_u64_field("all", INDEXED | FAST);
let even_field = schema_builder.add_u64_field("even", INDEXED | FAST);
let odd_field = schema_builder.add_text_field("odd", STRING | FAST);
let multi_field = schema_builder.add_text_field("multi", FAST);
let _never_field = schema_builder.add_u64_field("never", INDEXED | FAST);
let schema = schema_builder.build();
let index = Index::create_in_ram(schema);
{
let mut index_writer = index.writer_for_tests()?;
for i in 0u64..100u64 {
if i % 2 == 0 {
if i % 10 == 0 {
index_writer.add_document(doc!(all_field => i, even_field => i, multi_field => i.to_string(), multi_field => (i + 1).to_string()))?;
} else {
index_writer.add_document(doc!(all_field => i, even_field => i))?;
}
} else {
index_writer.add_document(doc!(all_field => i, odd_field => i.to_string()))?;
}
}
index_writer.commit()?;
}
let reader = index.reader()?;
let searcher = reader.searcher();
assert_eq!(count_existing_fields(&searcher, "all")?, 100);
assert_eq!(count_existing_fields(&searcher, "odd")?, 50);
assert_eq!(count_existing_fields(&searcher, "even")?, 50);
assert_eq!(count_existing_fields(&searcher, "multi")?, 10);
assert_eq!(count_existing_fields(&searcher, "never")?, 0);
// exercise seek
let query = BooleanQuery::intersection(vec![
Box::new(RangeQuery::new_u64_bounds(
"all".to_string(),
Bound::Included(50),
Bound::Unbounded,
)),
Box::new(ExistsQuery::new_exists_query("even".to_string())),
]);
assert_eq!(searcher.search(&query, &Count)?, 25);
let query = BooleanQuery::intersection(vec![
Box::new(RangeQuery::new_u64_bounds(
"all".to_string(),
Bound::Included(0),
Bound::Excluded(50),
)),
Box::new(ExistsQuery::new_exists_query("odd".to_string())),
]);
assert_eq!(searcher.search(&query, &Count)?, 25);
Ok(())
}
#[test]
fn test_exists_query_json() -> crate::Result<()> {
let mut schema_builder = Schema::builder();
let json = schema_builder.add_json_field("json", TEXT | FAST);
let schema = schema_builder.build();
let index = Index::create_in_ram(schema);
{
let mut index_writer = index.writer_for_tests()?;
for i in 0u64..100u64 {
if i % 2 == 0 {
index_writer.add_document(doc!(json => json!({"all": i, "even": true})))?;
} else {
index_writer
.add_document(doc!(json => json!({"all": i.to_string(), "odd": true})))?;
}
}
index_writer.commit()?;
}
let reader = index.reader()?;
let searcher = reader.searcher();
assert_eq!(count_existing_fields(&searcher, "json.all")?, 100);
assert_eq!(count_existing_fields(&searcher, "json.even")?, 50);
assert_eq!(count_existing_fields(&searcher, "json.odd")?, 50);
// Handling of non-existing fields:
assert_eq!(count_existing_fields(&searcher, "json.absent")?, 0);
assert_eq!(
searcher
.search(
&ExistsQuery::new_exists_query("does_not_exists.absent".to_string()),
&Count
)
.unwrap_err()
.to_string(),
"The field does not exist: 'does_not_exists.absent'"
);
Ok(())
}
#[test]
fn test_exists_query_misc_supported_types() -> crate::Result<()> {
let mut schema_builder = Schema::builder();
let bool = schema_builder.add_bool_field("bool", FAST);
let bytes = schema_builder.add_bytes_field("bytes", FAST);
let date = schema_builder.add_date_field("date", FAST);
let f64 = schema_builder.add_f64_field("f64", FAST);
let ip_addr = schema_builder.add_ip_addr_field("ip_addr", FAST);
let facet = schema_builder.add_facet_field("facet", FacetOptions::default());
let schema = schema_builder.build();
let index = Index::create_in_ram(schema);
{
let mut index_writer = index.writer_for_tests()?;
let now = OffsetDateTime::now_utc().unix_timestamp();
for i in 0u8..100u8 {
if i % 2 == 0 {
let date_val = DateTime::from_utc(OffsetDateTime::from_unix_timestamp(
now + i as i64 * 100,
)?);
index_writer.add_document(
doc!(bool => i % 3 == 0, bytes => vec![i, i + 1, i + 2], date => date_val),
)?;
} else {
let ip_addr_v6 = Ipv6Addr::new(0, 0, 0, 0, 0, 0xffff, 0xc00a, i.into());
index_writer
.add_document(doc!(f64 => i as f64 * 0.5, ip_addr => ip_addr_v6, facet => Facet::from("/facet/foo"), facet => Facet::from("/facet/bar")))?;
}
}
index_writer.commit()?;
}
let reader = index.reader()?;
let searcher = reader.searcher();
assert_eq!(count_existing_fields(&searcher, "bool")?, 50);
assert_eq!(count_existing_fields(&searcher, "bytes")?, 50);
assert_eq!(count_existing_fields(&searcher, "date")?, 50);
assert_eq!(count_existing_fields(&searcher, "f64")?, 50);
assert_eq!(count_existing_fields(&searcher, "ip_addr")?, 50);
assert_eq!(count_existing_fields(&searcher, "facet")?, 50);
Ok(())
}
#[test]
fn test_exists_query_unsupported_types() -> crate::Result<()> {
let mut schema_builder = Schema::builder();
let not_fast = schema_builder.add_text_field("not_fast", TEXT);
let schema = schema_builder.build();
let index = Index::create_in_ram(schema);
{
let mut index_writer = index.writer_for_tests()?;
index_writer.add_document(doc!(
not_fast => "slow",
))?;
index_writer.commit()?;
}
let reader = index.reader()?;
let searcher = reader.searcher();
assert_eq!(
searcher
.search(
&ExistsQuery::new_exists_query("not_fast".to_string()),
&Count
)
.unwrap_err()
.to_string(),
"Schema error: 'Field not_fast is not a fast field.'"
);
assert_eq!(
searcher
.search(
&ExistsQuery::new_exists_query("does_not_exists".to_string()),
&Count
)
.unwrap_err()
.to_string(),
"The field does not exist: 'does_not_exists'"
);
Ok(())
}
fn count_existing_fields(searcher: &Searcher, field: &str) -> crate::Result<usize> {
let query = ExistsQuery::new_exists_query(field.to_string());
searcher.search(&query, &Count)
}
}

128
src/query/fuzzy_query.rs
View File

@@ -3,7 +3,7 @@ use once_cell::sync::OnceCell;
use tantivy_fst::Automaton;
use crate::query::{AutomatonWeight, EnableScoring, Query, Weight};
use crate::schema::Term;
use crate::schema::{Term, Type};
use crate::TantivyError::InvalidArgument;
pub(crate) struct DfaWrapper(pub DFA);
@@ -46,7 +46,7 @@ impl Automaton for DfaWrapper {
/// let schema = schema_builder.build();
/// let index = Index::create_in_ram(schema);
/// {
/// let mut index_writer = index.writer(3_000_000)?;
/// let mut index_writer = index.writer(15_000_000)?;
/// index_writer.add_document(doc!(
/// title => "The Name of the Wind",
/// ))?;
@@ -132,18 +132,46 @@ impl FuzzyTermQuery {
});
let term_value = self.term.value();
let term_text = term_value.as_str().ok_or_else(|| {
InvalidArgument("The fuzzy term query requires a string term.".to_string())
})?;
let term_text = if term_value.typ() == Type::Json {
if let Some(json_path_type) = term_value.json_path_type() {
if json_path_type != Type::Str {
return Err(InvalidArgument(format!(
"The fuzzy term query requires a string path type for a json term. Found \
{:?}",
json_path_type
)));
}
}
std::str::from_utf8(self.term.serialized_value_bytes()).map_err(|_| {
InvalidArgument(
"Failed to convert json term value bytes to utf8 string.".to_string(),
)
})?
} else {
term_value.as_str().ok_or_else(|| {
InvalidArgument("The fuzzy term query requires a string term.".to_string())
})?
};
let automaton = if self.prefix {
automaton_builder.build_prefix_dfa(term_text)
} else {
automaton_builder.build_dfa(term_text)
};
Ok(AutomatonWeight::new(
self.term.field(),
DfaWrapper(automaton),
))
if let Some((json_path_bytes, _)) = term_value.as_json() {
Ok(AutomatonWeight::new_for_json_path(
self.term.field(),
DfaWrapper(automaton),
json_path_bytes,
))
} else {
Ok(AutomatonWeight::new(
self.term.field(),
DfaWrapper(automaton),
))
}
}
}
@@ -157,9 +185,89 @@ impl Query for FuzzyTermQuery {
mod test {
use super::FuzzyTermQuery;
use crate::collector::{Count, TopDocs};
use crate::schema::{Schema, TEXT};
use crate::indexer::NoMergePolicy;
use crate::query::QueryParser;
use crate::schema::{Schema, STORED, TEXT};
use crate::{assert_nearly_equals, Index, Term};
#[test]
pub fn test_fuzzy_json_path() -> crate::Result<()> {
// # Defining the schema
let mut schema_builder = Schema::builder();
let attributes = schema_builder.add_json_field("attributes", TEXT | STORED);
let schema = schema_builder.build();
// # Indexing documents
let index = Index::create_in_ram(schema.clone());
let mut index_writer = index.writer_for_tests()?;
index_writer.set_merge_policy(Box::new(NoMergePolicy));
let doc = schema.parse_document(
r#"{
"attributes": {
"a": "japan"
}
}"#,
)?;
index_writer.add_document(doc)?;
let doc = schema.parse_document(
r#"{
"attributes": {
"aa": "japan"
}
}"#,
)?;
index_writer.add_document(doc)?;
index_writer.commit()?;
let reader = index.reader()?;
let searcher = reader.searcher();
// # Fuzzy search
let query_parser = QueryParser::for_index(&index, vec![attributes]);
let get_json_path_term = |query: &str| -> crate::Result<Term> {
let query = query_parser.parse_query(query)?;
let mut terms = Vec::new();
query.query_terms(&mut |term, _| {
terms.push(term.clone());
});
Ok(terms[0].clone())
};
// shall not match the first document due to json path mismatch
{
let term = get_json_path_term("attributes.aa:japan")?;
let fuzzy_query = FuzzyTermQuery::new(term, 2, true);
let top_docs = searcher.search(&fuzzy_query, &TopDocs::with_limit(2))?;
assert_eq!(top_docs.len(), 1, "Expected only 1 document");
assert_eq!(top_docs[0].1.doc_id, 1, "Expected the second document");
}
// shall match the first document because Levenshtein distance is 1 (substitute 'o' with
// 'a')
{
let term = get_json_path_term("attributes.a:japon")?;
let fuzzy_query = FuzzyTermQuery::new(term, 1, true);
let top_docs = searcher.search(&fuzzy_query, &TopDocs::with_limit(2))?;
assert_eq!(top_docs.len(), 1, "Expected only 1 document");
assert_eq!(top_docs[0].1.doc_id, 0, "Expected the first document");
}
// shall not match because non-prefix Levenshtein distance is more than 1 (add 'a' and 'n')
{
let term = get_json_path_term("attributes.a:jap")?;
let fuzzy_query = FuzzyTermQuery::new(term, 1, true);
let top_docs = searcher.search(&fuzzy_query, &TopDocs::with_limit(2))?;
assert_eq!(top_docs.len(), 0, "Expected no document");
}
Ok(())
}
#[test]
pub fn test_fuzzy_term() -> crate::Result<()> {
let mut schema_builder = Schema::builder();

2
src/query/mod.rs
View File

@@ -8,6 +8,7 @@ mod const_score_query;
mod disjunction_max_query;
mod empty_query;
mod exclude;
mod exist_query;
mod explanation;
mod fuzzy_query;
mod intersection;
@@ -41,6 +42,7 @@ pub use self::const_score_query::{ConstScoreQuery, ConstScorer};
pub use self::disjunction_max_query::DisjunctionMaxQuery;
pub use self::empty_query::{EmptyQuery, EmptyScorer, EmptyWeight};
pub use self::exclude::Exclude;
pub use self::exist_query::ExistsQuery;
pub use self::explanation::Explanation;
#[cfg(test)]
pub(crate) use self::fuzzy_query::DfaWrapper;

6
src/query/more_like_this/more_like_this.rs
View File

@@ -23,13 +23,15 @@ impl Eq for ScoreTerm {}
impl PartialOrd for ScoreTerm {
fn partial_cmp(&self, other: &Self) -> Option<std::cmp::Ordering> {
self.score.partial_cmp(&other.score)
Some(self.cmp(other))
}
}
impl Ord for ScoreTerm {
fn cmp(&self, other: &Self) -> std::cmp::Ordering {
self.partial_cmp(other).unwrap_or(std::cmp::Ordering::Equal)
self.score
.partial_cmp(&other.score)
.unwrap_or(std::cmp::Ordering::Equal)
}
}

2
src/query/phrase_prefix_query/mod.rs
View File

@@ -6,7 +6,7 @@ pub use phrase_prefix_query::PhrasePrefixQuery;
pub use phrase_prefix_scorer::PhrasePrefixScorer;
pub use phrase_prefix_weight::PhrasePrefixWeight;
fn prefix_end(prefix_start: &[u8]) -> Option<Vec<u8>> {
pub(crate) fn prefix_end(prefix_start: &[u8]) -> Option<Vec<u8>> {
let mut res = prefix_start.to_owned();
while !res.is_empty() {
let end = res.len() - 1;

2
src/query/phrase_query/mod.rs
View File

@@ -75,7 +75,7 @@ pub mod tests {
let index = create_index(&["a b b d c g c", "a b a b c"])?;
let text_field = index.schema().get_field("text").unwrap();
let searcher = index.reader()?.searcher();
let terms: Vec<Term> = vec!["a", "b", "c"]
let terms: Vec<Term> = ["a", "b", "c"]
.iter()
.map(|text| Term::from_field_text(text_field, text))
.collect();

223
src/query/query_parser/query_parser.rs
View File

@@ -5,6 +5,7 @@ use std::str::{FromStr, ParseBoolError};
use base64::engine::general_purpose::STANDARD as BASE64;
use base64::Engine;
use itertools::Itertools;
use query_grammar::{UserInputAst, UserInputBound, UserInputLeaf, UserInputLiteral};
use rustc_hash::FxHashMap;
@@ -166,7 +167,7 @@ fn trim_ast(logical_ast: LogicalAst) -> Option<LogicalAst> {
/// * phrase terms: Quoted terms become phrase searches on fields that have positions indexed. e.g.,
/// `title:"Barack Obama"` will only find documents that have "barack" immediately followed by
/// "obama". Single quotes can also be used. If the text to be searched contains quotation mark,
/// it is possible to escape them with a \.
/// it is possible to escape them with a `\`.
///
/// * range terms: Range searches can be done by specifying the start and end bound. These can be
/// inclusive or exclusive. e.g., `title:[a TO c}` will find all documents whose title contains a
@@ -227,6 +228,25 @@ fn all_negative(ast: &LogicalAst) -> bool {
}
}
// Make an all-negative ast into a normal ast. Must not be used on an already okay ast.
fn make_non_negative(ast: &mut LogicalAst) {
match ast {
LogicalAst::Leaf(_) => (),
LogicalAst::Boost(ref mut child_ast, _) => make_non_negative(child_ast),
LogicalAst::Clause(children) => children.push((Occur::Should, LogicalLiteral::All.into())),
}
}
/// Similar to the try/? macro, but returns a tuple of (None, Vec<Error>) instead of Err(Error)
macro_rules! try_tuple {
($expr:expr) => {{
match $expr {
Ok(val) => val,
Err(e) => return (None, vec![e.into()]),
}
}};
}
impl QueryParser {
/// Creates a `QueryParser`, given
/// * schema - index Schema
@@ -308,17 +328,24 @@ impl QueryParser {
///
/// Note that `parse_query` returns an error if the input
/// is not a valid query.
///
/// There is currently no lenient mode for the query parser
/// which makes it a bad choice for a public/broad user search engine.
///
/// Implementing a lenient mode for this query parser is tracked
/// in [Issue 5](https://github.com/fulmicoton/tantivy/issues/5)
pub fn parse_query(&self, query: &str) -> Result<Box<dyn Query>, QueryParserError> {
let logical_ast = self.parse_query_to_logical_ast(query)?;
Ok(convert_to_query(&self.fuzzy, logical_ast))
}
/// Parse a query leniently
///
/// This variant parses invalid query on a best effort basis. If some part of the query can't
/// reasonably be executed (range query without field, searching on a non existing field,
/// searching without precising field when no default field is provided...), they may get
/// turned into a "match-nothing" subquery.
///
/// In case it encountered such issues, they are reported as a Vec of errors.
pub fn parse_query_lenient(&self, query: &str) -> (Box<dyn Query>, Vec<QueryParserError>) {
let (logical_ast, errors) = self.parse_query_to_logical_ast_lenient(query);
(convert_to_query(&self.fuzzy, logical_ast), errors)
}
/// Build a query from an already parsed user input AST
///
/// This can be useful if the user input AST parsed using [`query_grammar`]
@@ -328,31 +355,70 @@ impl QueryParser {
&self,
user_input_ast: UserInputAst,
) -> Result<Box<dyn Query>, QueryParserError> {
let logical_ast = self.compute_logical_ast(user_input_ast)?;
let (logical_ast, mut err) = self.compute_logical_ast_lenient(user_input_ast);
if !err.is_empty() {
return Err(err.swap_remove(0));
}
Ok(convert_to_query(&self.fuzzy, logical_ast))
}
/// Build leniently a query from an already parsed user input AST.
///
/// See also [`QueryParser::build_query_from_user_input_ast`]
pub fn build_query_from_user_input_ast_lenient(
&self,
user_input_ast: UserInputAst,
) -> (Box<dyn Query>, Vec<QueryParserError>) {
let (logical_ast, errors) = self.compute_logical_ast_lenient(user_input_ast);
(convert_to_query(&self.fuzzy, logical_ast), errors)
}
/// Parse the user query into an AST.
fn parse_query_to_logical_ast(&self, query: &str) -> Result<LogicalAst, QueryParserError> {
let user_input_ast = query_grammar::parse_query(query)
.map_err(|_| QueryParserError::SyntaxError(query.to_string()))?;
self.compute_logical_ast(user_input_ast)
let (ast, mut err) = self.compute_logical_ast_lenient(user_input_ast);
if !err.is_empty() {
return Err(err.swap_remove(0));
}
Ok(ast)
}
fn compute_logical_ast(
/// Parse the user query into an AST.
fn parse_query_to_logical_ast_lenient(
&self,
query: &str,
) -> (LogicalAst, Vec<QueryParserError>) {
let (user_input_ast, errors) = query_grammar::parse_query_lenient(query);
let mut errors: Vec<_> = errors
.into_iter()
.map(|error| {
QueryParserError::SyntaxError(format!(
"{} at position {}",
error.message, error.pos
))
})
.collect();
let (ast, mut ast_errors) = self.compute_logical_ast_lenient(user_input_ast);
errors.append(&mut ast_errors);
(ast, errors)
}
fn compute_logical_ast_lenient(
&self,
user_input_ast: UserInputAst,
) -> Result<LogicalAst, QueryParserError> {
let ast = self.compute_logical_ast_with_occur(user_input_ast)?;
) -> (LogicalAst, Vec<QueryParserError>) {
let (mut ast, mut err) = self.compute_logical_ast_with_occur_lenient(user_input_ast);
if let LogicalAst::Clause(children) = &ast {
if children.is_empty() {
return Ok(ast);
return (ast, err);
}
}
if all_negative(&ast) {
return Err(QueryParserError::AllButQueryForbidden);
err.push(QueryParserError::AllButQueryForbidden);
make_non_negative(&mut ast);
}
Ok(ast)
(ast, err)
}
fn compute_boundary_term(
@@ -571,26 +637,37 @@ impl QueryParser {
}
}
fn compute_logical_ast_with_occur(
fn compute_logical_ast_with_occur_lenient(
&self,
user_input_ast: UserInputAst,
) -> Result<LogicalAst, QueryParserError> {
) -> (LogicalAst, Vec<QueryParserError>) {
match user_input_ast {
UserInputAst::Clause(sub_queries) => {
let default_occur = self.default_occur();
let mut logical_sub_queries: Vec<(Occur, LogicalAst)> = Vec::new();
let mut errors = Vec::new();
for (occur_opt, sub_ast) in sub_queries {
let sub_ast = self.compute_logical_ast_with_occur(sub_ast)?;
let (sub_ast, mut sub_errors) =
self.compute_logical_ast_with_occur_lenient(sub_ast);
let occur = occur_opt.unwrap_or(default_occur);
logical_sub_queries.push((occur, sub_ast));
errors.append(&mut sub_errors);
}
Ok(LogicalAst::Clause(logical_sub_queries))
(LogicalAst::Clause(logical_sub_queries), errors)
}
UserInputAst::Boost(ast, boost) => {
let ast = self.compute_logical_ast_with_occur(*ast)?;
Ok(ast.boost(boost as Score))
let (ast, errors) = self.compute_logical_ast_with_occur_lenient(*ast);
(ast.boost(boost as Score), errors)
}
UserInputAst::Leaf(leaf) => {
let (ast, errors) = self.compute_logical_ast_from_leaf_lenient(*leaf);
// if the error is not recoverable, replace it with an empty clause. We will end up
// trimming those later
(
ast.unwrap_or_else(|| LogicalAst::Clause(Vec::new())),
errors,
)
}
UserInputAst::Leaf(leaf) => self.compute_logical_ast_from_leaf(*leaf),
}
}
@@ -658,23 +735,31 @@ impl QueryParser {
Ok(triplets)
}
fn compute_logical_ast_from_leaf(
fn compute_logical_ast_from_leaf_lenient(
&self,
leaf: UserInputLeaf,
) -> Result<LogicalAst, QueryParserError> {
) -> (Option<LogicalAst>, Vec<QueryParserError>) {
match leaf {
UserInputLeaf::Literal(literal) => {
let term_phrases: Vec<(Field, &str, &str)> =
self.compute_path_triplets_for_literal(&literal)?;
try_tuple!(self.compute_path_triplets_for_literal(&literal));
let mut asts: Vec<LogicalAst> = Vec::new();
let mut errors: Vec<QueryParserError> = Vec::new();
for (field, json_path, phrase) in term_phrases {
for ast in self.compute_logical_ast_for_leaf(
let unboosted_asts = match self.compute_logical_ast_for_leaf(
field,
json_path,
phrase,
literal.slop,
literal.prefix,
)? {
) {
Ok(asts) => asts,
Err(e) => {
errors.push(e);
continue;
}
};
for ast in unboosted_asts {
// Apply some field specific boost defined at the query parser level.
let boost = self.field_boost(field);
asts.push(LogicalAst::Leaf(Box::new(ast)).boost(boost));
@@ -685,57 +770,89 @@ impl QueryParser {
} else {
LogicalAst::Clause(asts.into_iter().map(|ast| (Occur::Should, ast)).collect())
};
Ok(result_ast)
(Some(result_ast), errors)
}
UserInputLeaf::All => Ok(LogicalAst::Leaf(Box::new(LogicalLiteral::All))),
UserInputLeaf::All => (
Some(LogicalAst::Leaf(Box::new(LogicalLiteral::All))),
Vec::new(),
),
UserInputLeaf::Range {
field: full_field_opt,
lower,
upper,
} => {
let full_path = full_field_opt.ok_or_else(|| {
QueryParserError::UnsupportedQuery(
"Range query need to target a specific field.".to_string(),
)
})?;
let (field, json_path) = self
let Some(full_path) = full_field_opt else {
return (
None,
vec![QueryParserError::UnsupportedQuery(
"Range query need to target a specific field.".to_string(),
)],
);
};
let (field, json_path) = try_tuple!(self
.split_full_path(&full_path)
.ok_or_else(|| QueryParserError::FieldDoesNotExist(full_path.clone()))?;
.ok_or_else(|| QueryParserError::FieldDoesNotExist(full_path.clone())));
let field_entry = self.schema.get_field_entry(field);
let value_type = field_entry.field_type().value_type();
let mut errors = Vec::new();
let lower = match self.resolve_bound(field, json_path, &lower) {
Ok(bound) => bound,
Err(error) => {
errors.push(error);
Bound::Unbounded
}
};
let upper = match self.resolve_bound(field, json_path, &upper) {
Ok(bound) => bound,
Err(error) => {
errors.push(error);
Bound::Unbounded
}
};
if lower == Bound::Unbounded && upper == Bound::Unbounded {
// this range is useless, either because a user requested [* TO *], or because
// we failed to parse something. Either way, there is no point emiting it
return (None, errors);
}
let logical_ast = LogicalAst::Leaf(Box::new(LogicalLiteral::Range {
field: self.schema.get_field_name(field).to_string(),
value_type,
lower: self.resolve_bound(field, json_path, &lower)?,
upper: self.resolve_bound(field, json_path, &upper)?,
lower,
upper,
}));
Ok(logical_ast)
(Some(logical_ast), errors)
}
UserInputLeaf::Set {
field: full_field_opt,
elements,
} => {
let full_path = full_field_opt.ok_or_else(|| {
let full_path = try_tuple!(full_field_opt.ok_or_else(|| {
QueryParserError::UnsupportedQuery(
"Set query need to target a specific field.".to_string(),
"Range query need to target a specific field.".to_string(),
)
})?;
let (field, json_path) = self
}));
let (field, json_path) = try_tuple!(self
.split_full_path(&full_path)
.ok_or_else(|| QueryParserError::FieldDoesNotExist(full_path.clone()))?;
.ok_or_else(|| QueryParserError::FieldDoesNotExist(full_path.clone())));
let field_entry = self.schema.get_field_entry(field);
let value_type = field_entry.field_type().value_type();
let (elements, errors) = elements
.into_iter()
.map(|element| self.compute_boundary_term(field, json_path, &element))
.partition_result();
let logical_ast = LogicalAst::Leaf(Box::new(LogicalLiteral::Set {
elements: elements
.into_iter()
.map(|element| self.compute_boundary_term(field, json_path, &element))
.collect::<Result<Vec<_>, _>>()?,
elements,
field,
value_type,
}));
Ok(logical_ast)
(Some(logical_ast), errors)
}
UserInputLeaf::Exists { .. } => (
None,
vec![QueryParserError::UnsupportedQuery(
"Range query need to target a specific field.".to_string(),
)],
),
}
}
}
@@ -1194,7 +1311,7 @@ mod test {
"k8s\u{1}node\u{1}name\0shello"
);
assert_eq!(
extract_query_term_json_path(r#"json.k8s\.node\.name:hello"#),
extract_query_term_json_path(r"json.k8s\.node\.name:hello"),
"k8s.node.name\0shello"
);
}
@@ -1620,10 +1737,10 @@ mod test {
#[test]
fn test_escaped_field() {
let mut schema_builder = Schema::builder();
schema_builder.add_text_field(r#"a\.b"#, STRING);
schema_builder.add_text_field(r"a\.b", STRING);
let schema = schema_builder.build();
let query_parser = QueryParser::new(schema, Vec::new(), TokenizerManager::default());
let query = query_parser.parse_query(r#"a\.b:hello"#).unwrap();
let query = query_parser.parse_query(r"a\.b:hello").unwrap();
assert_eq!(
format!("{query:?}"),
"TermQuery(Term(field=0, type=Str, \"hello\"))"

2
src/query/range_query/range_query.rs
View File

@@ -48,7 +48,7 @@ use crate::{DateTime, DocId, Score};
/// let schema = schema_builder.build();
///
/// let index = Index::create_in_ram(schema);
/// let mut index_writer = index.writer_with_num_threads(1, 10_000_000)?;
/// let mut index_writer = index.writer_with_num_threads(1, 20_000_000)?;
/// for year in 1950u64..2017u64 {
/// let num_docs_within_year = 10 + (year - 1950) * (year - 1950);
/// for _ in 0..num_docs_within_year {

2
src/query/regex_query.rs
View File

@@ -26,7 +26,7 @@ use crate::schema::Field;
/// let schema = schema_builder.build();
/// let index = Index::create_in_ram(schema);
/// {
/// let mut index_writer = index.writer(3_000_000)?;
/// let mut index_writer = index.writer(15_000_000)?;
/// index_writer.add_document(doc!(
/// title => "The Name of the Wind",
/// ))?;

8
src/query/set_query.rs
View File

@@ -72,6 +72,14 @@ impl Query for TermSetQuery {
fn weight(&self, enable_scoring: EnableScoring<'_>) -> crate::Result<Box<dyn Weight>> {
Ok(Box::new(self.specialized_weight(enable_scoring.schema())?))
}
fn query_terms<'a>(&'a self, visitor: &mut dyn FnMut(&'a Term, bool)) {
for terms in self.terms_map.values() {
for term in terms {
visitor(term, false);
}
}
}
}
struct SetDfaWrapper(Map<Vec<u8>>);

4
src/query/term_query/term_query.rs
View File

@@ -27,7 +27,7 @@ use crate::Term;
/// let schema = schema_builder.build();
/// let index = Index::create_in_ram(schema);
/// {
/// let mut index_writer = index.writer(3_000_000)?;
/// let mut index_writer = index.writer(15_000_000)?;
/// index_writer.add_document(doc!(
/// title => "The Name of the Wind",
/// ))?;
@@ -151,7 +151,7 @@ mod tests {
let ip_addr_2 = Ipv6Addr::from_u128(10);
{
let mut index_writer = index.writer(3_000_000).unwrap();
let mut index_writer = index.writer_for_tests().unwrap();
index_writer
.add_document(doc!(
ip_field => ip_addr_1

2
src/reader/mod.rs
View File

@@ -191,7 +191,7 @@ impl InnerIndexReader {
}
/// Opens the freshest segments [`SegmentReader`].
///
/// This function acquires a lot to prevent GC from removing files
/// This function acquires a lock to prevent GC from removing files
/// as we are opening our index.
fn open_segment_readers(index: &Index) -> crate::Result<Vec<SegmentReader>> {
// Prevents segment files from getting deleted while we are in the process of opening them

6
src/reader/warming.rs
View File

@@ -179,6 +179,7 @@ mod tests {
use super::Warmer;
use crate::core::searcher::SearcherGeneration;
use crate::directory::RamDirectory;
use crate::indexer::index_writer::MEMORY_BUDGET_NUM_BYTES_MIN;
use crate::schema::{Schema, INDEXED};
use crate::{Index, IndexSettings, ReloadPolicy, Searcher, SegmentId};
@@ -255,7 +256,10 @@ mod tests {
let num_writer_threads = 4;
let mut writer = index
.writer_with_num_threads(num_writer_threads, 25_000_000)
.writer_with_num_threads(
num_writer_threads,
MEMORY_BUDGET_NUM_BYTES_MIN * num_writer_threads,
)
.unwrap();
for i in 0u64..1000u64 {

5
src/schema/bytes_options.rs
View File

@@ -79,12 +79,9 @@ impl BytesOptions {
self
}
/// Set the field as a single-valued fast field.
/// Set the field as a fast field.
///
/// Fast fields are designed for random access.
/// Access time are similar to a random lookup in an array.
/// If more than one value is associated with a fast field, only the last one is
/// kept.
#[must_use]
pub fn set_fast(mut self) -> BytesOptions {
self.fast = true;

5
src/schema/date_time_options.rs
View File

@@ -78,12 +78,9 @@ impl DateOptions {
self
}
/// Set the field as a single-valued fast field.
/// Set the field as a fast field.
///
/// Fast fields are designed for random access.
/// Access time are similar to a random lookup in an array.
/// If more than one value is associated with a fast field, only the last one is
/// kept.
#[must_use]
pub fn set_fast(mut self) -> DateOptions {
self.fast = true;

3
src/schema/ip_options.rs
View File

@@ -83,9 +83,6 @@ impl IpAddrOptions {
/// Set the field as a fast field.
///
/// Fast fields are designed for random access.
/// Access time are similar to a random lookup in an array.
/// If more than one value is associated with a fast field, only the last one is
/// kept.
#[must_use]
pub fn set_fast(mut self) -> Self {
self.fast = true;

4
src/schema/json_object_options.rs
View File

@@ -80,12 +80,12 @@ impl JsonObjectOptions {
/// When expand_dots is enabled, json object like
/// `{"k8s.node.id": 5}` is processed as if it was
/// `{"k8s": {"node": {"id": 5}}}`.
/// It option has the merit of allowing users to
/// This option has the merit of allowing users to
/// write queries like `k8s.node.id:5`.
/// On the other, enabling that feature can lead to
/// ambiguity.
///
/// If disabled, the "." need to be escaped:
/// If disabled, the "." needs to be escaped:
/// `k8s\.node\.id:5`.
pub fn is_expand_dots_enabled(&self) -> bool {
self.expand_dots_enabled

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