aboutsummaryrefslogtreecommitdiffstats
path: root/streamingvisitors
Commit message (Collapse)AuthorAgeFilesLines
* Use explicit and do not expose nbostream in headerfile.Henning Baldersheim2024-04-221-0/+2
|
* Wire fuzzy prefix matching support through the query stackTor Brede Vekterli2024-04-191-27/+87
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | Adds `prefix:[true|false]` annotation support to the `fuzzy` query operator in the YQL and JSON query languages. Fuzzy prefix matching semantics are wired through to the matcher implementations for both indexed and streaming search. Example usage: {maxEditDistance:1,prefix:true}fuzzy("foo") Will match `foo`, `foobar`, `foxtrot`, `zookeeper` and so on. It can be combined with the existing prefix locking feature: {maxEditDistance:1,prefixLength:2,prefix:true}fuzzy("foo") Which will match `foo`, `foobar`, `foxtrot` etc, but _not_ `zookeeper` since the locked prefix (`fo`) does not match. Due to the complexities involved with extending the legacy binary query stack representation, signalling prefix matching for the fuzzy term is done by pragmatically adding a new, generic "prefix matching" term-level flag. This is currently ignored for everything except fuzzy query items. Modernizing the query stack format to make it more extensible (i.e. move encoding to Protobuf) is on the backlog...!
* - Optimize distance calculation for tensors with single dense subspace.Henning Baldersheim2024-04-161-1/+1
| | | | | - Let EmptySubspace be invalid. - Add noexcept to get_tensor(s).
* Update class comment.Tor Egge2024-03-271-1/+1
|
* Update class comment.Tor Egge2024-03-271-2/+3
|
* Add streaming mode version of tokens document field writer.Tor Egge2024-03-279-0/+283
|
* Move UTF8StringFieldSearcherBase tokenize member function to TokenizeReader.Tor Egge2024-03-278-30/+47
| | | | | Move anonymous normalize_mode funtion to a public static FieldSearchSpecMap::convert_normalize_mode member function.
* Don't create attribute vector for value part of a map of struct.Tor Egge2024-03-201-1/+1
|
* Reapply add virtual fields to index environment for streaming mode.Tor Egge2024-03-208-17/+98
|
* Revert "Add virtual fields to index environment for streaming mode."Henning Baldersheim2024-03-198-98/+17
|
* Less noisy error handling for common user errors.Henning Baldersheim2024-03-191-2/+5
|
* Add virtual fields to index environment for streaming mode.Tor Egge2024-03-198-17/+98
|
* Use multi_index_terms() returning boolean to differentiate betweenTor Egge2024-03-182-7/+6
| | | | | | multi term nodes that all searches the same index (e.g. Phrase) and multi term nodes that searches multiple indexes (e.g. Equiv, SameElement).
* Change parent class of search::streaming::SameElementQueryNode fromTor Egge2024-03-184-10/+10
| | | | search::streaming::AndQueryNode to search:streaming::MultiTerm.
* Use filter settings from rank profiles and query terms in streaming search.Tor Egge2024-03-156-10/+28
|
* If schema name is provided, append it to config id.Henning Baldersheim2024-03-081-8/+20
|
* Rename to reflect that is is a configid, not a search cluster.Henning Baldersheim2024-03-052-13/+13
|
* fix typoArne Juul2024-03-051-1/+1
|
* avoid re-using attributeField for sorting, would overwriteArne Juul2024-03-041-1/+10
|
* Propagate omit summary features from summary result class toTor Egge2024-02-201-0/+1
| | | | docsum state in streaming mode.
* Handle bool type too.Henning Baldersheim2024-02-191-1/+3
|
* Merge pull request #30316 from vespa-engine/balder/use-highlight-termsv8.307.19Henning Baldersheim2024-02-192-12/+23
|\ | | | | Balder/use highlight terms
| * GC comment not adding any value, besides being incorrect.Henning Baldersheim2024-02-191-1/+1
| |
| * Normalize highlight termsHenning Baldersheim2024-02-181-1/+4
| |
| * Propagate highlighterms.Henning Baldersheim2024-02-182-0/+10
| |
| * Modernize key/value interface of fs4::FS4PropertiesHenning Baldersheim2024-02-181-12/+10
| |
* | - Hide recall terms from ranking.Henning Baldersheim2024-02-192-19/+9
|/ | | | - Add noexcept.
* Handle rank-score-drop-limit overrides in the query too.Henning Baldersheim2024-02-152-13/+18
|
* - Add inteface for getting summary features for a single hit too.Henning Baldersheim2024-02-137-6/+64
| | | | - Produce summaryhits for grouping results too.
* Take owenship for the stuff you provide. Do not rely on the caller.Henning Baldersheim2024-02-137-64/+48
|
* Unify code layoutHenning Baldersheim2024-02-131-72/+64
|
* Also test size of heap and number of hits kept.Henning Baldersheim2024-02-131-0/+6
|
* Test that all hits are kept.Henning Baldersheim2024-02-132-0/+20
|
* - Add all hits to the hit collector.Henning Baldersheim2024-02-134-95/+115
| | | | | - Maintain a heap on the side, and keep heap property when producing results and features. - Drop teh pointer to the document once it drops off the heap.
* dumpProcessors must also receive sort information up front.Henning Baldersheim2024-02-123-4/+4
|
* Revert "Revert "- Use explicit given wanted hit count.""Henning Baldersheim2024-02-129-56/+77
|
* Revert "- Use explicit given wanted hit count."Henning Baldersheim2024-02-129-77/+56
|
* vespa.summaryFeatures.cached is always 0.0 (false)Henning Baldersheim2024-02-121-1/+0
|
* Merge branch 'master' into balder/use-wanted-hitcountHenning Baldersheim2024-02-125-46/+43
|\
| * - DocId is an overloaded type already. Just use string as directly for ↵Henning Baldersheim2024-02-105-47/+44
| | | | | | | | | | | | readability. - Add noexcept and and using.
* | It is know up front that if we sort by rank or by sortblob. So instead of ↵Henning Baldersheim2024-02-118-44/+46
| | | | | | | | | | | | detecting by first hit, and hoping the rest are the same, set expectations ahead and assert all hits are correct.
* | Make separate make/push/pop heap function to improve readabilityHenning Baldersheim2024-02-112-18/+34
| |
* | Compute actual value up front.Henning Baldersheim2024-02-111-2/+3
| |
* | - Use explicit given wanted hit count.Henning Baldersheim2024-02-113-4/+6
|/ | | | | - Do not rely on using _hits.size(). There is no contract they are the same. They are so for now so only by implementation.
* Handle search::streaming::EquivQueryNode as a leaf in the query tree.Tor Egge2024-02-095-37/+100
|
* Handle search::streaming::PhraseQueryNode as a leaf in the query tree.Tor Egge2024-02-065-156/+42
|
* Merge pull request #30185 from ↵Henning Baldersheim2024-02-068-33/+31
|\ | | | | | | | | vespa-engine/balder/symmetric-query-processing-docsum-time Balder/symmetric query processing docsum time
| * Wire QueryNormalization in to JuniperQueryAdapter and use it there.Henning Baldersheim2024-02-052-6/+10
| |
| * Move Normalization from search::streaming => searchHenning Baldersheim2024-02-058-27/+21
| |
* | Merge pull request #30186 from ↵Geir Storli2024-02-066-4/+39
|\ \ | | | | | | | | | | | | vespa-engine/toregge/track-oldest-config-generation-used-for-streaming-search Track oldest config generation used for streaming search.