| name | full-text-search |
| description | Use when adding keyword, full-text, or FTS5/BM25 search to a Go app with gosqlite — the typed fts.Index[K, V] API, tokenizers (incl. custom Go tokenizers), query builder, ranking, snippet/highlight. |
Full-text search (FTS5)
import (
_ "gosqlite.org"
"gosqlite.org/fts"
)
idx, _ := fts.New[int64, string](ctx, db, "docs", fts.Options{
Tokenizer: fts.Porter{Base: fts.Unicode61{RemoveDiacritics: 2}},
})
idx.Insert(ctx, fts.Attr[int64, string]{Key: 1, Value: "the quick brown fox"})
matches, _ := idx.SearchSlice(ctx, fts.Term("fox"),
fts.WithRanking(),
fts.WithSnippet("value", "[", "]", "…", 8))
- Index type params:
fts.New[K, V] — K is the row key type, V the stored value type.
- Tokenizers:
fts.Porter, fts.Unicode61{RemoveDiacritics: 2}, fts.Trigram. For CJK / stemming / synonyms, register a custom Go tokenizer via (*Conn).RegisterFTS5Tokenizer.
- Query builder:
fts.Term, fts.And/Or/Not, phrase, NEAR, prefix, column filters — compose in Go, no raw MATCH strings.
- Options:
fts.WithRanking(), fts.WithSnippet(...), fts.WithHighlight(...), fts.WithLimit(n) / WithOffset(n), fts.WithFilter(sql, args...).
- Content modes: external-content (default), in-table, or contentless via
fts.Options.
- Vocab / autocomplete:
fts.NewVocab(...) exposes term frequencies + prefix lookups.
Tokenizers are applied symmetrically to both the indexed document and the query, so a reversible transform is invisible to MATCH — inspect indexed terms via fts5vocab if a custom tokenizer behaves unexpectedly. Tokens need iStart < iEnd (no zero-width).
For gorm models, use the tag-driven sidecar (fts5:"tokenize=porter+unicode61") — see the gorm skill. To combine with vector search, see hybrid-search. Full reference: docs/guides/full-text-search.md.
