bhashantarh/lib/detectLanguage.js

Tiny translation wrapper with fallbacks (LibreTranslate -> MyMemory) and detection with franc fallback.

const axios = require('axios');
const getSupportedLanguages = require('./supportedLanguages');
const { iso3ToIso1 } = require('./supportedLanguages');

const FALLBACK_URL = 'https://libretranslate.de';

// Lazy-load franc-min for CJS
let cachedFranc = null;
async function getFranc() {
  if (cachedFranc) return cachedFranc;
  try {
    const mod = await import('franc-min');
    cachedFranc = mod.franc;
    return cachedFranc;
  } catch (e) {
    console.error('Failed to load franc-min:', e.message);
    return null;
  }
}

// Script-based guess
function guessByScript(text) {
  const ranges = {
    hi: /[\u0900-\u097F]/,
    mr: /[\u0900-\u097F]/,
    bn: /[\u0980-\u09FF]/,
    gu: /[\u0A80-\u0AFF]/,
    ta: /[\u0B80-\u0BFF]/,
    te: /[\u0C00-\u0C7F]/,
    ml: /[\u0D00-\u0D7F]/,
    ar: /[\u0600-\u06FF]/,
    ur: /[\u0600-\u06FF]/,
    ru: /[\u0400-\u04FF]/,
    zh: /[\u4E00-\u9FFF]/,
    ja: /[\u3040-\u309F\u30A0-\u30FF]/
  };

  const priority = ['zh', 'ja', 'ru', 'ar', 'ur', 'bn', 'gu', 'ta', 'te', 'ml', 'hi', 'mr'];

  for (const code of priority) {
    if (ranges[code]?.test(text)) return code;
  }
  return null;
}

async function tryLibreDetect(text) {
  const res = await axios.post(`${FALLBACK_URL}/detect`, { q: text }, {
    headers: { 'Content-Type': 'application/json' },
    timeout: 8000
  });
  if (Array.isArray(res.data) && res.data[0]?.language) {
    return res.data[0].language;
  }
  return 'unknown';
}

function isSupported(code) {
  return !!getSupportedLanguages()[code];
}

async function detectLanguage(text) {
  const cleaned = (text || '').replace(/[0-9]/g, '').trim();

  // 1) Try LibreTranslate.de
  try {
    const lt = await tryLibreDetect(cleaned);
    if (lt && isSupported(lt)) return lt;
  } catch (err) {
    console.error('LibreTranslate.de detect error:', err.message);
  }

  // add near the top, below guessByScript()
function guessLatinByCommonWords(text) {
  if (!text) return null;
  const t = text.toLowerCase();

  // Spanish
  if (/\bhola\b/.test(t) || /\bamigo(s)?\b/.test(t) || /\b¿?c(ó|o)mo\b/.test(t)) return 'es';

  // French
  if (/\bbonjour\b/.test(t) || /\bsalut\b/.test(t) || /\bmerci\b/.test(t)) return 'fr';

  // Italian
  if (/\bciao\b/.test(t) || /\bgrazie\b/.test(t)) return 'it';

  // Portuguese
  if (/\bol[áa]\b/.test(t) || /\bobrigado(a)?\b/.test(t)) return 'pt';

  // German
  if (/\bhallo\b/.test(t) || /\bdanke\b/.test(t)) return 'de';

  // English
  if (/\bhello\b/.test(t) || /\bhi\b/.test(t) || /\bthanks\b/.test(t)) return 'en';

  return null;
}


  // 2) Script guess
  const scriptGuess = guessByScript(cleaned);
  if (scriptGuess && isSupported(scriptGuess)) return scriptGuess;

  // 2.5) Latin common-words guess (helps short phrases like "Hola amigo")
const latinGuess = guessLatinByCommonWords(cleaned);
if (latinGuess && isSupported(latinGuess)) return latinGuess;

  // 3) franc-min
  try {
    const franc = await getFranc();
    if (franc) {
      const iso3 = franc(cleaned, { minLength: 1 });
      if (iso3 && iso3 !== 'und') {
        const iso1 = iso3ToIso1(iso3);
        if (iso1 && isSupported(iso1)) return iso1;
      }
    }
  } catch (e) {
    console.error('franc detect error:', e.message);
  }

  return 'unknown';
}

module.exports = detectLanguage;