cajeta.search.fuzzy.Matcher — typo-tolerant key→value lookup: composes an ngram.Index (recall prefilter) with Distance.damerau (scoring) and a length-scaled threshold, returning results ranked closest-first. The index uses bigrams (n = 2): a single adjacent transposition (the common typo Fiel→File) destroys every trigram of a short key but leaves a shared bigram, so bigrams keep recall on exactly the typos damerau scores as distance 1. Ties rank by key, so identical inputs always produce identical output — exact matches (distance 0) rank first.

Matcher<int32> m = heap Matcher<int32>();
m.add("File", 10);
m.add("Edit", 20);
ArrayList<Match<int32>> hits #= m.match("Fiel");  // [File, distance 1]
int32 top = hits.get(0).value();                  // 10

Methods

Signature
`Matcher()` ⚑	Construct an empty matcher (bigram index for transposition recall)
`void add(String key, T value)`	Index `value` under `key`, remembering the key for later scoring
`#ArrayList<Match<T>> match(String query)`	Every stored value whose key is within the length-scaled edit-distance threshold, ranked by ascending distance then key

Each result is a Match<T> — value() (the stored value), key() (the key it was indexed under), distance() (edit distance from the query; 0 = exact). A Match owns a copy of its key, so a result list outlives the Matcher.

⚑ = @EntryPoint

Matcher<T>

Methods

See also