Locale detection

Source: platform/apps/app/src/lib/rag/detect-locale.ts · rendered from main on every deploy — edit in the repo, not here

platform/apps/app/src/lib/rag/detect-locale.ts

/**
 * Detect response language from the visitor's own words. ADR-0121 D1.
 *
 * Puccha serves exactly two response languages (Thai, English), so this is a
 * deliberately tiny dominant-language check, not a general language classifier —
 * no dependency, Workers-safe, and nothing added to the widget bundle (the
 * widget never calls this; the chat route does).
 *
 * We weigh Thai *characters* against English *words*, not character-vs-character.
 * Thai is written without spaces, so its natural unit is the character; English's
 * is the word. A strict char-vs-char count let one long English noun outvote a
 * whole Thai phrase — "สนใจทำ software" (6 Thai chars vs 8 Latin chars) was
 * classified English and got an English refusal (the bug that motivated this
 * refinement of ADR-0121 D1). Counting English by word, "สนใจทำ software" is
 * 6 Thai vs 1 English → Thai, while "What is the ค่ะ policy on refunds?" is
 * 2 Thai vs 8 English → English, both correct.
 *
 * Text with no letters at all (empty, digits / emoji / punctuation only) returns
 * `null`, so the caller falls back to the client-supplied locale.
 *
 * This only ever drives *generation* language (system prompt, refusal, canned
 * replies, handoff copy, query-rewrite, answer-cache key). It is never a
 * retrieval filter — hybridSearch / Vectorize / FTS do not gate on locale and
 * bge-m3 is multilingual, so an English query still retrieves Thai chunks.
 */
export function detectLocale(text: string): 'th' | 'en' | null {
  if (!text) return null;
  let thaiChars = 0;
  let latinWords = 0;
  let inLatinWord = false;
  for (const ch of text) {
    if (ch >= '฀' && ch <= '๿') {
      thaiChars++;
      inLatinWord = false;
    } else if ((ch >= 'a' && ch <= 'z') || (ch >= 'A' && ch <= 'Z')) {
      if (!inLatinWord) latinWords++;
      inLatinWord = true;
    } else {
      inLatinWord = false;
    }
  }
  if (thaiChars === 0 && latinWords === 0) return null; // no decidable script
  return thaiChars >= latinWords ? 'th' : 'en';
}