Greenlens/server/lib/scanGrounding.js

const clamp = (value, min, max) => {
  return Math.min(max, Math.max(min, value));
};

const normalizeText = (value) => {
  return String(value || '')
    .toLowerCase()
    .normalize('NFD')
    .replace(/[\u0300-\u036f]/g, '')
    .replace(/\s+/g, ' ')
    .trim();
};

const GERMAN_COMMON_NAME_HINTS = [
  'weihnachtsstern',
  'weinachtsstern',
  'einblatt',
  'fensterblatt',
  'korbmarante',
  'glucksfeder',
  'gluecksfeder',
  'efeutute',
  'drachenbaum',
  'gummibaum',
  'geigenfeige',
  'bogenhanf',
  'yucca palme',
  'gluckskastanie',
  'glueckskastanie',
];

// Most of the catalog's German common names are compound nouns ending in one
// of these — a hardcoded name list alone (GERMAN_COMMON_NAME_HINTS) only
// catches names we've already seen slip through, so this suffix check acts
// as a general-purpose backstop for names we haven't hardcoded.
const GERMAN_NAME_SUFFIXES = [
  'blume', 'blueten', 'bluten', 'baum', 'strauch', 'kraut', 'wurz', 'wurzel',
  'farn', 'palme', 'pflanze', 'hanf', 'tute', 'veilchen', 'dorn', 'kaktus',
  'lilie', 'birke', 'weide', 'ranke', 'stern', 'kette', 'schwanz', 'ohren',
  'feige', 'moos', 'zwiebel', 'knolle', 'rohr', 'kirsche',
];

const isLikelyGermanCommonName = (value) => {
  const raw = String(value || '').trim();
  if (!raw) return false;
  if (/[äöüß]/i.test(raw)) return true;

  const normalized = normalizeText(raw).replace(/[^a-z0-9 ]+/g, ' ');
  if (!normalized) return false;
  if (/\b(der|die|das|ein|eine|und|mit|fuer)\b/.test(normalized)) return true;

  const compact = normalized.replace(/\s+/g, '');
  if (GERMAN_NAME_SUFFIXES.some((suffix) => compact.endsWith(suffix))) return true;

  return GERMAN_COMMON_NAME_HINTS.some((hint) => normalized.includes(hint));
};

// Last-resort safety net applied to the FINAL result regardless of whether
// it went through catalog grounding — the AI is instructed to never return a
// German name when English was requested, but doesn't always comply, so this
// catches the ones that slip through and falls back to the botanical name
// (which is language-neutral) rather than showing a mismatched German name
// alongside English description/care text.
const enforceEnglishName = (result, language) => {
  if (!result || language !== 'en') return result;
  if (!isLikelyGermanCommonName(result.name)) return result;
  if (!result.botanicalName) return result;
  return { ...result, name: result.botanicalName };
};

const isLikelyBotanicalName = (value, botanicalName) => {
  const raw = String(value || '').trim();
  const botanicalRaw = String(botanicalName || '').trim();
  if (!raw) return false;

  if (normalizeText(raw) === normalizeText(botanicalRaw)) return true;

  return /^[A-Z][a-z-]+(?:\s[a-z.-]+){1,2}$/.test(raw);
};

const findCatalogMatch = (aiResult, entries) => {
  if (!aiResult || !Array.isArray(entries) || entries.length === 0) return null;
  const aiBotanical = normalizeText(aiResult.botanicalName);
  const aiName = normalizeText(aiResult.name);
  if (!aiBotanical && !aiName) return null;

  const byExactBotanical = entries.find((entry) => normalizeText(entry.botanicalName) === aiBotanical);
  if (byExactBotanical) return byExactBotanical;

  const byExactName = entries.find((entry) => normalizeText(entry.name) === aiName);
  if (byExactName) return byExactName;

  if (aiBotanical) {
    const aiGenus = aiBotanical.split(' ')[0];
    if (aiGenus) {
      const byGenus = entries.find((entry) => normalizeText(entry.botanicalName).startsWith(`${aiGenus} `));
      if (byGenus) return byGenus;
    }
  }

  const byContains = entries.find((entry) => {
    const plantName = normalizeText(entry.name);
    const botanical = normalizeText(entry.botanicalName);
    return (aiName && (plantName.includes(aiName) || aiName.includes(plantName)))
      || (aiBotanical && (botanical.includes(aiBotanical) || aiBotanical.includes(botanical)));
  });
  if (byContains) return byContains;

  return null;
};

const shouldUseCatalogNameOverride = ({ language, aiResult, matchedEntry }) => {
  const catalogName = String(matchedEntry?.name || '').trim();
  if (!catalogName) return false;
  if (language !== 'en') return true;

  if (isLikelyBotanicalName(catalogName, matchedEntry?.botanicalName || aiResult?.botanicalName)) {
    return true;
  }

  if (isLikelyGermanCommonName(catalogName)) {
    return false;
  }

  return true;
};

const applyCatalogGrounding = (aiResult, catalogEntries, language = 'en') => {
  const matchedEntry = findCatalogMatch(aiResult, catalogEntries);
  if (!matchedEntry) {
    return { grounded: false, result: aiResult };
  }

  const useCatalogName = shouldUseCatalogNameOverride({ language, aiResult, matchedEntry });

  return {
    grounded: true,
    result: {
      name: useCatalogName ? matchedEntry.name || aiResult.name : aiResult.name,
      botanicalName: matchedEntry.botanicalName || aiResult.botanicalName,
      confidence: clamp(aiResult.confidence || 0.6, 0.05, 0.99),
      description: aiResult.description || matchedEntry.description || '',
      careInfo: {
        waterIntervalDays: Math.max(1, Number(matchedEntry.careInfo?.waterIntervalDays) || Number(aiResult.careInfo?.waterIntervalDays) || 7),
        light: (matchedEntry.careInfo?.light && matchedEntry.careInfo.light !== 'Unknown') ? matchedEntry.careInfo.light : (aiResult.careInfo?.light || 'Unknown'),
        temp: (matchedEntry.careInfo?.temp && matchedEntry.careInfo.temp !== 'Unknown') ? matchedEntry.careInfo.temp : (aiResult.careInfo?.temp || 'Unknown'),
      },
    },
  };
};

module.exports = {
  applyCatalogGrounding,
  enforceEnglishName,
  findCatalogMatch,
  isLikelyGermanCommonName,
  normalizeText,
  shouldUseCatalogNameOverride,
};