/**
 * Clone Architect — Orchestrateur principal
 *
 * Pipeline complet : URL → Extract → Analyze → Tokenize
 * Usage: npm run clone -- <URL>
 */

import { execSync } from 'child_process';
import { existsSync } from 'fs';
import { join } from 'path';
import { shouldSkipExtraction, writeFingerprint } from './shared/cache.js';

// Phase 2.3 — Parse CLI flags: clone <url> [--force]
const args = process.argv.slice(2);
const forceFlag = args.includes('--force');
const url = args.find(a => !a.startsWith('--'));

if (!url) {
  console.error('Usage: npm run clone -- <URL> [--force]');
  console.error('Example: npm run clone -- https://linear.app');
  console.error('         npm run clone -- https://linear.app --force  (bypass cache)');
  process.exit(1);
}

try {
  new URL(url);
} catch {
  console.error(`Invalid URL: ${url}`);
  process.exit(1);
}

const domain = new URL(url).hostname.replace('www.', '');
const baseDir = join(process.cwd(), 'extractions', domain);

const startTime = Date.now();

console.log('╔══════════════════════════════════════════════════╗');
console.log('║         CLONE ARCHITECT — Full Pipeline         ║');
console.log('╚══════════════════════════════════════════════════╝');
console.log(`\n🎯 Target: ${url}\n`);

// Phase 2.3 — Cache check before expensive Playwright extraction
const cacheResult = await shouldSkipExtraction(url, domain, { force: forceFlag });
if (cacheResult.skip) {
  console.log(`✨ Cache hit: ${cacheResult.reason} — skipping extraction (saved ~140s)`);
  console.log(`   Use --force to bypass cache and re-extract`);
  console.log(`\n📁 Existing extraction at: ${baseDir}\n`);
  process.exit(0);
}
console.log(`🔍 Cache: ${cacheResult.reason}`);

// Step 1 — Extract
console.log('━━━ STEP 1/4 — EXTRACTION ━━━');
try {
  execSync(`npx tsx scripts/extract.ts "${url}"`, {
    stdio: 'inherit',
    cwd: process.cwd(),
    timeout: 120000,
  });
} catch (err) {
  console.error('❌ Extraction failed');
  process.exit(1);
}

// Phase 5.4.1 — Write fingerprint ONLY after extraction validation
// Previously (bug G4): fingerprint was written even if raw-css.json was empty/cassée.
// Result: future runs hit cache and skip a broken extraction forever.
// Now: validate raw-css.json has at least 5 CSS vars AND a body element before writing fingerprint.
if (cacheResult.fingerprint) {
  try {
    const rawCssPath = join(baseDir, 'raw-css.json');
    if (existsSync(rawCssPath)) {
      const rawData = JSON.parse(require('fs').readFileSync(rawCssPath, 'utf-8'));
      const cssVarCount = Object.keys(rawData?.desktop?.cssCustomProperties || {}).length;
      const hasBody = !!rawData?.desktop?.elements?.body;
      const hasColors = (rawData?.desktop?.allColors || []).length >= 3;
      // Threshold: at least 3 colors + a body element. Extractions that fail anti-bot have ~0 vars + no body.
      if (hasBody && hasColors) {
        await writeFingerprint(domain, cacheResult.fingerprint);
      } else {
        console.warn(`  ⚠️  Skipping cache fingerprint (cssVars=${cssVarCount}, hasBody=${hasBody}, colors=${(rawData?.desktop?.allColors || []).length}) — looks like an aborted extraction`);
      }
    }
  } catch (e) {
    console.warn(`  ⚠️  Failed to validate raw-css.json before fingerprint: ${(e as Error).message}`);
  }
}

// Check extraction output
if (!existsSync(join(baseDir, 'raw-css.json'))) {
  console.error('❌ raw-css.json not found after extraction');
  process.exit(1);
}

// Step 2 — Analyze
console.log('\n━━━ STEP 2/4 — LAYOUT ANALYSIS ━━━');
try {
  execSync(`npx tsx scripts/analyze.ts "${domain}"`, {
    stdio: 'inherit',
    cwd: process.cwd(),
    timeout: 30000,
  });
} catch (err) {
  console.error('❌ Analysis failed');
  process.exit(1);
}

// Step 3 — Tokenize
console.log('\n━━━ STEP 3/4 — TOKENIZATION ━━━');
try {
  execSync(`npx tsx scripts/tokenize.ts "${domain}"`, {
    stdio: 'inherit',
    cwd: process.cwd(),
    timeout: 30000,
  });
} catch (err) {
  console.error('❌ Tokenization failed');
  process.exit(1);
}

// Step 4 — Generate DESIGN.md
console.log('\n━━━ STEP 4/5 — DESIGN.MD GENERATION ━━━\n');
try {
  execSync(`npx tsx scripts/generate-design-md.ts "${domain}"`, {
    stdio: 'inherit',
    cwd: process.cwd(),
    timeout: 30000,
  });
} catch (err) {
  console.error('⚠️  DESIGN.md generation failed (non-blocking)');
}

// Step 5 — Generate showcase HTML
console.log('\n━━━ STEP 5/5 — SHOWCASE ━━━\n');
try {
  execSync(`npx tsx scripts/generate-showcase.ts "${domain}"`, {
    stdio: 'inherit',
    cwd: process.cwd(),
    timeout: 30000,
  });
} catch (err) {
  console.error('⚠️  Showcase generation failed (non-blocking)');
}

const duration = ((Date.now() - startTime) / 1000).toFixed(1);

console.log('\n╔══════════════════════════════════════════════════╗');
console.log('║              PIPELINE COMPLETE                   ║');
console.log('╚══════════════════════════════════════════════════╝');
console.log(`\n⏱️  Duration: ${duration}s`);
console.log(`📁 Output: extractions/${domain}/`);
console.log('   ├── screenshots/          — Playwright captures');
console.log('   ├── raw-css.json          — Computed styles brut');
console.log('   ├── extraction-summary.json');
console.log('   ├── layout-analysis.md    — Structure analysée');
console.log('   ├── tokens.json           — Design tokens normalisés');
console.log('   ├── DESIGN.md             — Narratif LLM-optimisé (9 sections)');
console.log('   └── showcase/index.html   — Page de présentation visuelle');
console.log('\n📋 Next steps:');
console.log('   1. Review tokens.json and layout-analysis.md');
console.log('   2. Use tokens to generate faithful code (R16 pipeline)');
console.log('   3. Run: npm run compare -- ' + domain);