Skip to main content

Edit this page

wildcard

The wildcard template indexes text for wildcard and prefix matching.

It wraps an inner tokenizer to split the input into terms, then emits boundary-marked character n-grams of each term so that LIKE-style patterns can be answered from the index instead of by a full scan. The grams carry start/end markers (not visible in the token text) so that a leading-anchored prefix is distinguishable from a match in the middle of a term. For plain substring search over code and logs, sparse_ngram is usually the better fit.

Options​

OptionTypeDefaultDescription
NGRAMSIZEinteger3N-gram length used for wildcard/prefix indexing (minimum 2)
TOKENIZER_TEMPLATEstringrequiredTemplate of the nested tokenizer that produces the terms
TOKENIZER_*——Options for the nested tokenizer, each prefixed with TOKENIZER_

How it tokenizes​

The nested tokenizer (selected with TOKENIZER_TEMPLATE and configured through its TOKENIZER_-prefixed options) first splits the input into terms. Each term is then expanded into its character n-grams up to NGRAMSIZE long — the shorter boundary grams at the start and end of the term alongside the full-length interior grams.

Conceptually, with NGRAMSIZE = 3, the term search expands into the grams se, sea, ear, arc, rch, ch, h. Each gram also carries a non-printable start/end marker (so a leading-anchored prefix is distinguishable from a mid-term match), which is why these grams are an internal detail rather than something you inspect with ts_lexize — you query them indirectly through LIKE-style patterns.

Searching​

Index a column with the wildcard dictionary, then match it with ts_like (or ts_starts_with / ts_regexp). The pattern is answered from the indexed grams instead of a full scan. A substring pattern matches anywhere in a term:

Query
CREATE INDEX terms_idx ON terms USING inverted (id, term wildcard_dict);
VACUUM (REFRESH_TABLE) terms;
SELECT id, term FROM terms_idx WHERE term @@ ts_like('%ear%') ORDER BY id;
Result
 id | term----+----------  1 | search  2 | research

A prefix pattern is anchored to the start of a term:

Query
SELECT id, term FROM terms_idx WHERE term @@ ts_like('sea%') ORDER BY id;
Result
 id | term----+---------  1 | search  3 | seabird

See also​