packages/database/src/translate-tracks.ts

/**
 * translate-tracks.ts
 *
 * Translates all T*-FR.json track files into EN, ES, PT using GPT-4o.
 * Preserves all non-text fields (IDs, URLs, dayNumber, weights, thresholds).
 * Leaves audioUrl null so TTS is auto-generated on first lesson delivery.
 *
 * Usage:
 *   OPENAI_API_KEY=sk-... npx tsx packages/database/src/translate-tracks.ts
 *
 * Options:
 *   --tracks T1-FR,T2-FR   Translate specific tracks only (default: all)
 *   --langs EN,ES           Target languages only (default: EN,ES,PT)
 *   --dry-run               Print first day only, no file writes
 */

import fs from 'fs';
import path from 'path';
import OpenAI from 'openai';

const CONTENT_DIR = path.join(__dirname, '../content/tracks');
const TARGET_LANGS = ['EN', 'ES', 'PT'] as const;
type TargetLang = typeof TARGET_LANGS[number];

const LANG_LABELS: Record<TargetLang, string> = {
  EN: 'English',
  ES: 'Spanish (Latin American)',
  PT: 'Portuguese (Brazilian)',
};

const LANG_SUFFIX: Record<TargetLang, string> = {
  EN: 'EN',
  ES: 'ES',
  PT: 'PT',
};

// Fields to translate at TrackDay level
const TEXT_FIELDS_DAY = ['title', 'lessonText', 'exercisePrompt', 'videoCaption'] as const;

// Fields to translate at Track level
const TEXT_FIELDS_TRACK = ['title', 'description'] as const;

interface TranslationBatch {
  key: string;
  text: string;
}

async function translateBatch(
  client: OpenAI,
  items: TranslationBatch[],
  lang: TargetLang,
  context: string,
): Promise<Record<string, string>> {
  if (items.length === 0) return {};

  const itemsJson = JSON.stringify(
    items.map(i => ({ key: i.key, text: i.text })),
    null,
    2,
  );

  const prompt = `You are a professional educational content translator for entrepreneurship training in West Africa.

Translate the following strings from French to ${LANG_LABELS[lang]}.

Context: ${context}

Rules:
- Keep the exact tone: direct, encouraging, coach-like
- Adapt examples to feel natural in ${LANG_LABELS[lang]}-speaking African contexts when appropriate
- Keep all formatting (✅, numbering, line breaks \\n\\n, bold asterisks *word*)
- Keep all placeholder patterns like [PRODUCT], [CLIENT], [LIEU] — translate the word inside brackets
- For "videoCaption" fields (Wolof text starting with emojis): translate to ${LANG_LABELS[lang]} keeping the emoji and structure
- Return ONLY a JSON object mapping each "key" to its translated "text"
- Do not add explanations

Items to translate:
${itemsJson}

Return format:
{"key1": "translated text 1", "key2": "translated text 2", ...}`;

  const response = await client.chat.completions.create({
    model: 'gpt-4o',
    messages: [{ role: 'user', content: prompt }],
    response_format: { type: 'json_object' },
    temperature: 0.3,
  });

  const content = response.choices[0]?.message?.content;
  if (!content) throw new Error('Empty response from GPT-4o');
  return JSON.parse(content);
}

function collectDayTexts(day: any, prefix: string): TranslationBatch[] {
  const items: TranslationBatch[] = [];

  for (const field of TEXT_FIELDS_DAY) {
    if (day[field] && typeof day[field] === 'string') {
      items.push({ key: `${prefix}.${field}`, text: day[field] });
    }
  }

  // exerciseCriteria nested text
  const crit = day.exerciseCriteria;
  if (crit) {
    if (crit.goal) items.push({ key: `${prefix}.criteria.goal`, text: crit.goal });
    if (crit.evaluation?.examples) items.push({ key: `${prefix}.criteria.examples`, text: crit.evaluation.examples });
    if (crit.remediation?.hint) items.push({ key: `${prefix}.criteria.hint`, text: crit.remediation.hint });

    // mustInclude descriptions
    if (Array.isArray(crit.success?.mustInclude)) {
      crit.success.mustInclude.forEach((item: any, idx: number) => {
        if (item.desc) {
          items.push({ key: `${prefix}.criteria.must.${idx}.desc`, text: item.desc });
        }
      });
    }
  }

  // buttonsJson titles (keep IDs)
  if (Array.isArray(day.buttonsJson)) {
    day.buttonsJson.forEach((btn: any, idx: number) => {
      if (btn.title) {
        items.push({ key: `${prefix}.btn.${idx}.title`, text: btn.title });
      }
    });
  }

  return items;
}

function applyTranslations(day: any, prefix: string, translations: Record<string, string>): any {
  const d = { ...day };

  for (const field of TEXT_FIELDS_DAY) {
    const key = `${prefix}.${field}`;
    if (translations[key] !== undefined) d[field] = translations[key];
  }

  // exerciseCriteria
  if (d.exerciseCriteria) {
    d.exerciseCriteria = { ...d.exerciseCriteria };
    if (translations[`${prefix}.criteria.goal`]) {
      d.exerciseCriteria.goal = translations[`${prefix}.criteria.goal`];
    }
    if (d.exerciseCriteria.evaluation) {
      d.exerciseCriteria.evaluation = { ...d.exerciseCriteria.evaluation };
      if (translations[`${prefix}.criteria.examples`]) {
        d.exerciseCriteria.evaluation.examples = translations[`${prefix}.criteria.examples`];
      }
    }
    if (d.exerciseCriteria.remediation) {
      d.exerciseCriteria.remediation = { ...d.exerciseCriteria.remediation };
      if (translations[`${prefix}.criteria.hint`]) {
        d.exerciseCriteria.remediation.hint = translations[`${prefix}.criteria.hint`];
      }
    }
    if (d.exerciseCriteria.success?.mustInclude) {
      d.exerciseCriteria.success = { ...d.exerciseCriteria.success };
      d.exerciseCriteria.success.mustInclude = d.exerciseCriteria.success.mustInclude.map(
        (item: any, idx: number) => {
          const key = `${prefix}.criteria.must.${idx}.desc`;
          return translations[key] ? { ...item, desc: translations[key] } : item;
        },
      );
    }
  }

  // buttonsJson
  if (Array.isArray(d.buttonsJson)) {
    d.buttonsJson = d.buttonsJson.map((btn: any, idx: number) => {
      const key = `${prefix}.btn.${idx}.title`;
      return translations[key] ? { ...btn, title: translations[key] } : btn;
    });
  }

  // Always clear audioUrl — TTS auto-generates on first delivery
  d.audioUrl = null;

  return d;
}

async function translateTrack(
  client: OpenAI,
  sourceFile: string,
  lang: TargetLang,
  dryRun: boolean,
): Promise<void> {
  const sourcePath = path.join(CONTENT_DIR, sourceFile);
  const source = JSON.parse(fs.readFileSync(sourcePath, 'utf-8'));

  const sourceId = source.trackId as string; // e.g. T1-FR
  const targetId = sourceId.replace('-FR', `-${LANG_SUFFIX[lang]}`);
  const targetFile = sourceFile.replace('-FR.json', `-${LANG_SUFFIX[lang]}.json`);
  const targetPath = path.join(CONTENT_DIR, targetFile);

  console.log(`\n[${lang}] ${sourceId} → ${targetId}`);

  // Skip if already exists (resume support)
  if (!dryRun && fs.existsSync(targetPath)) {
    console.log(`  ⏭  ${targetFile} already exists — skipping`);
    return;
  }

  // Collect all text from all days in one big batch
  const allItems: TranslationBatch[] = [];

  // Track-level texts
  for (const field of TEXT_FIELDS_TRACK) {
    if (source[field]) allItems.push({ key: `track.${field}`, text: source[field] });
  }

  // Day-level texts
  source.days.forEach((day: any, idx: number) => {
    const dayItems = collectDayTexts(day, `day.${idx}`);
    allItems.push(...dayItems);
  });

  const context = `Entrepreneurship education track "${source.title}" for micro-entrepreneurs in Africa. Day-by-day coaching program on how to understand, build, and pitch a business.`;

  // Batch in chunks of 40 items to stay within token limits
  const CHUNK = 40;
  const translations: Record<string, string> = {};

  for (let i = 0; i < allItems.length; i += CHUNK) {
    const chunk = allItems.slice(i, i + CHUNK);
    process.stdout.write(`  Translating batch ${Math.floor(i / CHUNK) + 1}/${Math.ceil(allItems.length / CHUNK)}...`);
    const result = await translateBatch(client, chunk, lang, context);
    Object.assign(translations, result);
    process.stdout.write(' ✓\n');
  }

  // Build translated track
  const translated = {
    trackId: targetId,
    title: translations['track.title'] ?? source.title,
    language: lang,
    description: translations['track.description'] ?? source.description,
    totalDays: source.totalDays,
    version: source.version,
    days: source.days.map((day: any, idx: number) =>
      applyTranslations(day, `day.${idx}`, translations),
    ),
  };

  if (dryRun) {
    console.log('\n  [DRY RUN] First day preview:');
    console.log(JSON.stringify(translated.days[0], null, 2));
    return;
  }

  fs.writeFileSync(targetPath, JSON.stringify(translated, null, 2), 'utf-8');
  console.log(`  ✅ Written: ${targetFile}`);
}

async function main() {
  const args = process.argv.slice(2);
  const dryRun = args.includes('--dry-run');

  const tracksArg = args.find(a => a.startsWith('--tracks='))?.split('=')[1];
  const langsArg = args.find(a => a.startsWith('--langs='))?.split('=')[1];

  const targetLangs = (langsArg?.split(',') ?? TARGET_LANGS) as TargetLang[];
  const sourceTracks = tracksArg?.split(',').map(t => `${t}.json`)
    ?? fs.readdirSync(CONTENT_DIR).filter(f => f.endsWith('-FR.json'));

  const apiKey = process.env.OPENAI_API_KEY;
  if (!apiKey) throw new Error('OPENAI_API_KEY is required');

  const client = new OpenAI({ apiKey });

  console.log(`🌍 Translating ${sourceTracks.length} tracks → ${targetLangs.join(', ')}${dryRun ? ' [DRY RUN]' : ''}`);
  console.log(`Source tracks: ${sourceTracks.join(', ')}\n`);

  for (const sourceFile of sourceTracks) {
    for (const lang of targetLangs) {
      await translateTrack(client, sourceFile, lang, dryRun);
    }
  }

  console.log('\n🎉 Translation complete!');
  if (!dryRun) {
    console.log('Next step: run the seeder to load translated tracks into DB');
    console.log('  pnpm --filter @repo/database seed');
  }
}

main().catch(err => {
  console.error('❌ Translation failed:', err.message);
  process.exit(1);
});