@ferrybig/cgeodata-reader
    Preparing search index...

    A query, precomputed once per search (not once per name): both its raw (case-folded only) and diacritic-folded forms, each with its own tokens, since nameMatchQuality needs both to tell a match that needed folding from one that didn't.

    interface Needle {
        raw: string;
        rawTokens: string[];
        folded: string;
        foldedTokens: string[];
    }
    Index
    raw: string

    The query lowercased, diacritics kept.

    rawTokens: string[]

    raw's words (tokenize).

    folded: string

    The query folded (foldDiacritics).

    foldedTokens: string[]

    folded's words.