* Optimizing top K using Adrien Grand's ideas
https://jpountz.github.io/2025/08/28/compiled-vs-vectorized-search-engine-edition.html
* Suffix-sum pruning for multi-term intersection candidates
After scoring each secondary in Phase 2, check whether remaining
secondaries' block_max scores can still beat the threshold. Skip
to the next candidate early if impossible, avoiding expensive seeks
into later secondaries.
Improves three-term intersection by ~8% on the balanced benchmark
while keeping two-term performance neutral.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
* Claude CR comment
* Removed 16 term scorer limit.
---------
Co-authored-by: Paul Masurel <paul.masurel@datadoghq.com>
Co-authored-by: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
* seek_exact + cost based intersection
Adds `seek_exact` and `cost` to `DocSet` for a more efficient intersection.
Unlike `seek`, `seek_exact` does not require the DocSet to advance to the next hit, if the target does not exist.
`cost` allows to address the different DocSet types and their cost
model and is used to determine the DocSet that drives the intersection.
E.g. fast field range queries may do a full scan. Phrase queries load the positions to check if a we have a hit.
They both have a higher cost than their size_hint would suggest.
Improves `size_hint` estimation for intersection and union, by having a
estimation based on random distribution with a co-location factor.
Refactor range query benchmark.
Closes#2531
*Future Work*
Implement `seek_exact` for BufferedUnionScorer and RangeDocSet (fast field range queries)
Evaluate replacing `seek` with `seek_exact` to reduce code complexity
* Apply suggestions from code review
Co-authored-by: Paul Masurel <paul@quickwit.io>
* add API contract verfication
* impl seek_exact on union
* rename seek_exact
* add mixed AND OR test, fix buffered_union
* Add a proptest of BooleanQuery. (#2690)
* fix build
* Increase the document count.
* fix merge conflict
* fix debug assert
* Fix compilation errors after rebase
- Remove duplicate proptest_boolean_query module
- Remove duplicate cost() method implementations
- Fix TopDocs API usage (add .order_by_score())
- Remove duplicate imports
- Remove unused variable assignments
---------
Co-authored-by: Paul Masurel <paul@quickwit.io>
Co-authored-by: Pascal Seitz <pascal.seitz@datadoghq.com>
Co-authored-by: Stu Hood <stuhood@gmail.com>
* Fixed the range issue.
* Fixed the second all scorer issue
* Improved docs + tests
* Improved code.
* Fixed lint issues.
* Improved tests + logic based on PR comments.
* Fixed lint issues.
* Increase the document count.
* Improved the prop-tests
* Expand the index size, and remove unused parameter.
---------
Co-authored-by: Stu Hood <stuhood@gmail.com>
* Refactoring of the score tweaker into `SortKeyComputer`s to unlock two features.
- Allow lazy evaluation of score. As soon as we identified that a doc won't
reach the topK threshold, we can stop the evaluation.
- Allow for a different segment level score, segment level score and their conversion.
This PR breaks public API, but fixing code is straightforward.
* Bumping tantivy version
---------
Co-authored-by: Paul Masurel <paul.masurel@datadoghq.com>
* Optimization when posting list are saturated.
If a posting list doc freq is the segment reader's
max_doc, and if scoring does not matter, we can replace it
by a AllScorer.
In turn, in a boolean query, we can dismiss all scorers and
empty scorers, to accelerate the request.
* Added range query optimization
* CR comment
* CR comments
* CR comment
---------
Co-authored-by: Paul Masurel <paul.masurel@datadoghq.com>
(a OR b) OR (c OR d) can be simplified to (a OR b OR c OR d)
(a AND b) AND (c AND d) can be simplified to (a AND b AND c AND d)
This directly affects how queries are executed
remove unused SumWithCoordsCombiner
the number of fields is unused and private
* fix windows build (#1)
* Fix windows build
* Add doc traits
* Add field value iter
* Add value and serialization
* Adjust order
* Fix bug
* Correct type
* Fix generic bugs
* Reformat code
* Add generic to index writer which I forgot about
* Fix missing generics on single segment writer
* Add missing type export
* Add default methods for convenience
* Cleanup
* Fix more-like-this query to use standard types
* Update API and fix tests
* Add doc traits
* Add field value iter
* Add value and serialization
* Adjust order
* Fix bug
* Correct type
* Rebase main and fix conflicts
* Reformat code
* Merge upstream
* Fix missing generics on single segment writer
* Add missing type export
* Add default methods for convenience
* Cleanup
* Fix more-like-this query to use standard types
* Update API and fix tests
* Add tokenizer improvements from previous commits
* Add tokenizer improvements from previous commits
* Reformat
* Fix unit tests
* Fix unit tests
* Use enum in changes
* Stage changes
* Add new deserializer logic
* Add serializer integration
* Add document deserializer
* Implement new (de)serialization api for existing types
* Fix bugs and type errors
* Add helper implementations
* Fix errors
* Reformat code
* Add unit tests and some code organisation for serialization
* Add unit tests to deserializer
* Add some small docs
* Add support for deserializing serde values
* Reformat
* Fix typo
* Fix typo
* Change repr of facet
* Remove unused trait methods
* Add child value type
* Resolve comments
* Fix build
* Fix more build errors
* Fix more build errors
* Fix the tests I missed
* Fix examples
* fix numerical order, serialize PreTok Str
* fix coverage
* rename Document to TantivyDocument, rename DocumentAccess to Document
add Binary prefix to binary de/serialization
* fix coverage
---------
Co-authored-by: Pascal Seitz <pascal.seitz@gmail.com>
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
* Added optimisation using block wand for single TermScorer.
A proptest was also added.
* Fix block wand algorithm by taking the last doc id of scores until the pivot scorer (included).
* In block wand, when block max score is lower than the threshold, advance the scorer with best score.
* Fix wrong condition in block_wand_single_scorer and add debug_assert to have an equality check on doc to break the loop.
go_to_first_doc was typically calling seek with a target smaller than
doc.
Since SegmentPostings typically do a linear search on the full block,
regardless of the current position, it could have our segment postings
go backward.
* Do nothing when combining score values of excluded scores.
* Add test case for two excluded.
* Test score for two excluded terms.
* Use TopDocs in test_boolean_query_two_excluded
- Change in the DocSet and Scorer API. (@fulmicoton).
A freshly created DocSet point directly to their first doc. A sentinel value called TERMINATED marks the end of a DocSet.
`.advance()` returns the new DocId. `Scorer::skip(target)` has been replaced by `Scorer::seek(target)` and returns the resulting DocId.
As a result, iterating through DocSet now looks as follows
```rust
let mut doc = docset.doc();
while doc != TERMINATED {
// ...
doc = docset.advance();
}
```
The change made it possible to greatly simplify a lot of the docset's code.
- Misc internal optimization and introduction of the `Scorer::for_each_pruning` function. (@fulmicoton)
* Alternative take on boosted queries
* Fixing unit test
* Added boosting to the query grammar.
* Made BoostQuery public.
* Added support for boosting field in QueryParser
Closes#547
* Split Collector into an overall Collector and a per-segment SegmentCollector. Precursor to cross-segment parallelism, and as a side benefit cleans up any per-segment fields from being Option<T> to just T.
* Attempt to add MultiCollector back
* working. Chained collector is broken though
* Fix chained collector
* Fix test
* Make Weight Send+Sync for parallelization purposes
* Expose parameters of RangeQuery for external usage
* Removed &mut self
* fixing tests
* Restored TestCollectors
* blop
* multicollector working
* chained collector working
* test broken
* fixing unit test
* blop
* blop
* Blop
* simplifying APi
* blop
* better syntax
* Simplifying top_collector
* refactoring
* blop
* Sync with master
* Added multithread search
* Collector refactoring
* Schema::builder
* CR and rustdoc
* CR comments
* blop
* Added an executor
* Sorted the segment readers in the searcher
* Update searcher.rs
* Fixed unit testst
* changed the place where we have the sort-segment-by-count heuristic
* using crossbeam::channel
* inlining
* Comments about panics propagating
* Added unit test for executor panicking
* Readded default
* Removed Default impl
* Added unit test for executor
