cajeta.search.ngram.Index — an n-gram inverted index for fuzzy candidate generation. Maps key strings to stored values of type T and answers candidates(query): every value whose key shares at least one n-gram with the query, deduplicated. This is a recall-oriented prefilter — a superset the caller (or fuzzy.Matcher) then scores with an edit distance. Gram extraction is lowercased for recall; a key shorter than n is indexed under itself as a single gram. add is O(len(key)); candidates visits only the postings of the query’s grams (sub-linear in the key count).

Index<int32> idx = heap Index<int32>();           // trigrams (n = 3)
idx.add("apple", 1);
idx.add("zebra", 2);
ArrayList<int32> hits #= idx.candidates("aple");   // [1] — shares "ple"

Methods

Signature
`Index()` ⚑	Construct a trigram index (n = 3)
`Index(int32 gramSize)` ⚑	Construct an index with the given gram size
`void add(String key, T value)`	Index `value` under every n-gram of `key`
`#ArrayList<T> candidates(String query)`	Every value whose key shares at least one n-gram with `query`, deduplicated; empty when nothing shares a gram

⚑ = @EntryPoint

Index<T>

Methods

See also