Function normalizeWithMap

lib/redact-engine.ts:109–151

(input: string)

 * the map exact for the transformations we apply (which are all local).
 */
export function normalizeWithMap(input: string): {
  normalized: string;
  map: number[];
} {
  const out: string[] = [];
  const map: number[] = [];
  let i = 0;
  while (i < input.length) {
    // HTML entity expansion (fixed small set; longest first).
    let matchedEntity = false;
    for (const ent in HTML_ENTITIES) {
      if (input.startsWith(ent, i)) {
        const rep = HTML_ENTITIES[ent];
        for (const ch of rep) {
          out.push(ch);
          map.push(i);
        }
        i += ent.length;
        matchedEntity = true;
        break;
      }
    }
    if (matchedEntity) continue;

    const ch = input[i];
    if (ZERO_WIDTH.test(ch)) {
      ZERO_WIDTH.lastIndex = 0;
      i += 1;
      continue;
    }
    ZERO_WIDTH.lastIndex = 0;

    const norm = ch.normalize("NFKC");
    for (const nch of norm) {
      out.push(nch);
      map.push(i);
    }
    i += 1;
  }
  // Sentinel so an offset == length maps to the original length.
  map.push(input.length);
  return { normalized: out.join(""), map };
}

// ── Offset → line/col on the ORIGINAL text ────────────────────────────────────