package granary

  1. Overview
  2. Docs
package granary
Legend:
Page
Library
Module
Module type
Parameter
Class
Class type
Source

Module Granary_sql.Fts_tokenizerSource

FTS word tokenizer.

Splits text on non-alphanumeric boundaries, lowercases each token, and assigns monotonically-increasing position numbers per column. High-byte UTF-8 sequences (>127) are treated as word characters so that multi-byte Unicode letters are not split on byte boundaries.

Sourcetype token = {
  1. term : string;
    (*

    lowercased word

    *)
  2. col : int;
    (*

    which input column (0-indexed)

    *)
  3. pos : int;
    (*

    word position within this column (0-indexed)

    *)
  4. start_byte : int;
    (*

    byte offset of first byte of raw token in source string

    *)
  5. end_byte : int;
    (*

    byte offset one past last byte of raw token

    *)
}
Sourceval tokenize_string : col:int -> string -> token list

tokenize_string ~col text tokenizes a single text string as column col. Tokens are numbered starting at 0 within this column.

Sourceval tokenize : (int * string) list -> token list

tokenize col_texts tokenizes a list of (col_index, text) pairs. Tokens for each column are numbered starting at 0.