kusamoji/src/pos-source/index.js

Japanese morphological analyzer for Node.js — Viterbi tokenizer with mmap dict loading and pluggable POS-source strategy

"use strict";

/**
 * POS-source strategy factory.
 *
 * Selects the POS feature lookup implementation based on env vars.
 * When KUSAMOJI_LAZY_POS=1, returns a LazyDiskPosSource that disk-backs
 * the ~1.3 GB tid_pos blob. Otherwise returns null (upstream in-heap).
 */

let { InHeapPosSource } = require("./in-heap");
let { LazyDiskPosSource } = require("./lazy-disk");

/**
 * @param {object} opts
 * @param {object} opts.tokenInfoDictionary
 * @param {string} opts.dictPath
 * @returns {object|null}
 */
function selectPosSource({ tokenInfoDictionary, dictPath }) {
    let lazy = (process.env.KUSAMOJI_LAZY_POS || "").trim();
    if (lazy === "1" || lazy === "true" || lazy === "on") {
        return new LazyDiskPosSource({ tokenInfoDictionary, dictPath });
    }
    return null;
}

module.exports = { selectPosSource, InHeapPosSource, LazyDiskPosSource };