cmpstr/dist/esm/metric/QGram.mjs

CmpStr is a lightweight, fast and well performing package for calculating string similarity

// CmpStr v3.2.2 build-bb61120-260311 by Paul Köhler @komed3 / MIT License
import { Pool } from '../utils/Pool.mjs';
import { MetricRegistry, Metric } from './Metric.mjs';

class QGramSimilarity extends Metric {
  constructor(a, b, opt = {}) {
    super('qGram', a, b, opt, true);
  }
  _qGrams(str, q) {
    const len = Math.max(0, str.length - q + 1);
    const grams = Pool.acquire('set', len);
    for (let i = 0; i < len; i++) grams.add(str.slice(i, i + q));
    return grams;
  }
  compute(a, b) {
    const { q = 2 } = this.options;
    const setA = this._qGrams(a, q),
      setB = this._qGrams(b, q);
    const sizeA = setA.size,
      sizeB = setB.size;
    try {
      let intersection = 0;
      for (const gram of setA) if (setB.has(gram)) intersection++;
      const size = Math.max(sizeA, sizeB);
      return {
        res: size === 0 ? 1 : Metric.clamp(intersection / size),
        raw: { intersection, size }
      };
    } finally {
      Pool.release('set', setA, sizeA);
      Pool.release('set', setB, sizeB);
    }
  }
}
MetricRegistry.add('qGram', QGramSimilarity);

export { QGramSimilarity };