Search Functions
These functions operate on an inverted index — SereneDB's index for full-text, vector and geospatial search. You build a query (most often with the @@ operator and a TSQUERY), optionally rank matches with a relevance scorer and highlight them. The reference is grouped by purpose:
| Page | Functions |
|---|---|
| Full-Text | The @@ match operator, TSQUERY constructors and operators, PostgreSQL-compatible parsers, convenience predicates and ts_lexize |
| Tokenizers | Every dictionary template as a scalar function — split_text, stem_words, generate_ngrams, split_text_csv and the others — one reference page each, for ad hoc analysis without a dictionary |
| Term Dictionary | Facet counts, distinct values, autocomplete and min/max via the ts_dict_* aggregates read from the index dictionary |
| Vector | l2_distance, cosine_distance, l1_distance and the <-> operator |
| Geo | ST_Intersects, ST_Contains, ST_Distance_Between, ST_Distance_Centroid |
| Scoring | BM25, TFIDF and the other relevance scorers |
| Highlighting | ts_highlight, ts_offsets |
For task-oriented guides see Full-Text Search, Vector Search and Geospatial Search.