Skip to content

tokenizeKeywords

function tokenizeKeywords(text): string[];

Shared keyword tokenizer — exported so Fts5KeywordIndex queries are tokenized identically to BM25Index for ranking parity.

Keeps combining marks (\p{M}) as token characters — Indic scripts (Tamil, Sinhala, Hindi, …) encode vowel signs as combining marks, and treating them as separators splits every word mid-syllable.

Parameter Type

text

string

string[]