prism-design 2.13.0

package/scripts/enrich-catalog.ts

@@ -0,0 +1,212 @@
+/**
+ * Prism — Catalog Enrichment
+ *
+ * Reads catalog/index.json and enriches each entry with:
+ *   - category     : brand category classification
+ *   - dark         : dark-mode site flag
+ *   - hasFrontmatter: DESIGN.md has YAML ---…--- block
+ *   - hasScreenshots: screenshots/ folder has files
+ *   - font         : primary font extracted
+ *   - completeness : 0-100 score from YAML frontmatter
+ *   - extractedAt  : ISO date from YAML frontmatter
+ *
+ * Usage: npx tsx scripts/enrich-catalog.ts
+ */
+
+import { readFile, writeFile } from 'fs/promises';
+import { existsSync, readdirSync, statSync } from 'fs';
+import { join } from 'path';
+
+const ROOT = join(process.cwd());
+const EXTRACTIONS = join(ROOT, 'extractions');
+const CATALOG_PATH = join(ROOT, 'catalog', 'index.json');
+
+// ── Category classifier ─────────────────────────────────────────────
+
+const CATEGORY_RULES: Array<{ pattern: RegExp; label: string }> = [
+  { pattern: /binance|deribit|hyperliquid|coinbase|kraken|bybit|okx|ftx|bitfinex|gemini/i, label: 'Crypto / Trading' },
+  { pattern: /stripe|paypal|adyen|braintree|square|klarna|checkout|revolut|wise|mercury/i, label: 'Fintech / Payments' },
+  { pattern: /airbnb|booking|expedia|hotels|tripadvisor|vrbo|hostelworld/i, label: 'Travel & Hospitality' },
+  { pattern: /acnestudios|addicts|nike|adidas|zara|hm\.|farfetch|ssense|mrporter|matchesfashion|sneaks/i, label: 'Fashion & Retail' },
+  { pattern: /shopify|commerce|magento|woocommerce|dimension|myshopify/i, label: 'E-commerce' },
+  { pattern: /linear|notion|akiflow|clickup|basecamp|asana|monday|todoist|craft\.do|cron\./i, label: 'Productivity & PM' },
+  { pattern: /figma|framer|sketch|webflow|canva|pitch|miro|loom|comet/i, label: 'Design & Creative Tools' },
+  { pattern: /cursor|bolt\.new|replit|github|gitlab|vercel|netlify|railway|render|supabase|neon/i, label: 'Dev Tools & Cloud' },
+  { pattern: /arc\.net|browser|firefox|chrome/i, label: 'Browser / Platform' },
+  { pattern: /clerk|auth0|okta|jumpcloud|1password|lastpass/i, label: 'Auth & Security' },
+  { pattern: /fabric|attio|hubspot|salesforce|intercom|zendesk|pipedrive/i, label: 'CRM & Business' },
+  { pattern: /ps-apps|ps-tools|aurax|dimension\.ps|lola/i, label: 'PS-Tools Projects' },
+  { pattern: /example\.com/i, label: 'Test / Example' },
+];
+
+function classifyBrand(domain: string): string {
+  for (const { pattern, label } of CATEGORY_RULES) {
+    if (pattern.test(domain)) return label;
+  }
+  return 'SaaS / Product';
+}
+
+// ── Parse YAML frontmatter from DESIGN.md ───────────────────────────
+
+interface FrontmatterData {
+  hasFrontmatter: boolean;
+  completeness: number | null;
+  extractedAt: string | null;
+}
+
+function parseFrontmatter(content: string): FrontmatterData {
+  if (!content.startsWith('---')) {
+    return { hasFrontmatter: false, completeness: null, extractedAt: null };
+  }
+  const end = content.indexOf('\n---\n', 4);
+  if (end === -1) {
+    return { hasFrontmatter: false, completeness: null, extractedAt: null };
+  }
+  const block = content.slice(4, end);
+
+  const completenessMatch = block.match(/^completeness:\s*(\d+)/m);
+  const extractedAtMatch = block.match(/^extracted_at:\s*"?([^"\n]+)"?/m);
+
+  return {
+    hasFrontmatter: true,
+    completeness: completenessMatch ? parseInt(completenessMatch[1]) : null,
+    extractedAt: extractedAtMatch ? extractedAtMatch[1].trim() : null,
+  };
+}
+
+// ── Extract primary font from tokens.json ───────────────────────────
+
+function extractFont(tokens: any): string | null {
+  const fp = tokens?.typography?.fontFamily?.primary;
+  if (fp) return fp;
+  const bodyFont = tokens?.typography?.body?.fontFamily;
+  if (bodyFont) return bodyFont;
+  return null;
+}
+
+// ── Detect dark site from tokens.json ───────────────────────────────
+
+function isDarkSite(tokens: any): boolean {
+  const bg = tokens?.colors?.background?.primary || '';
+  // Parse rgb(r,g,b) or hex
+  const hexMatch = bg.match(/^#([0-9a-f]{6})/i);
+  if (hexMatch) {
+    const r = parseInt(hexMatch[1].slice(0, 2), 16);
+    const g = parseInt(hexMatch[1].slice(2, 4), 16);
+    const b = parseInt(hexMatch[1].slice(4, 6), 16);
+    const lum = (0.299 * r + 0.587 * g + 0.114 * b) / 255;
+    return lum < 0.3;
+  }
+  const rgbMatch = bg.match(/rgb\(\s*(\d+)\s*,\s*(\d+)\s*,\s*(\d+)/);
+  if (rgbMatch) {
+    const [, r, g, b] = rgbMatch.map(Number);
+    const lum = (0.299 * r + 0.587 * g + 0.114 * b) / 255;
+    return lum < 0.3;
+  }
+  return false;
+}
+
+// ── Count screenshots ────────────────────────────────────────────────
+
+function hasScreenshotsDir(domain: string): boolean {
+  const screenshotsDir = join(EXTRACTIONS, domain, 'screenshots');
+  if (!existsSync(screenshotsDir)) return false;
+  try {
+    const files = readdirSync(screenshotsDir).filter(f => /\.(png|jpg|jpeg|webp)$/i.test(f));
+    return files.length > 0;
+  } catch {
+    return false;
+  }
+}
+
+// ── Main ─────────────────────────────────────────────────────────────
+
+async function main() {
+  console.log('🔧 Enriching catalog/index.json...\n');
+
+  const raw = await readFile(CATALOG_PATH, 'utf-8');
+  const catalog = JSON.parse(raw);
+
+  let enriched = 0;
+  let skipped = 0;
+  let withScore = 0;
+  let withFrontmatter = 0;
+
+  for (const brand of catalog.brands) {
+    const domain: string = brand.domain;
+    const extractionDir = join(EXTRACTIONS, domain);
+    const designMdPath = join(extractionDir, 'DESIGN.md');
+    const tokensPath = join(extractionDir, 'tokens.json');
+
+    // Category (always computable)
+    brand.category = classifyBrand(domain);
+
+    // Screenshots
+    brand.hasScreenshots = hasScreenshotsDir(domain);
+
+    // Tokens-derived fields
+    if (existsSync(tokensPath)) {
+      try {
+        const tokens = JSON.parse(await readFile(tokensPath, 'utf-8'));
+        brand.dark = isDarkSite(tokens);
+        const font = extractFont(tokens);
+        if (font) brand.font = font;
+      } catch {
+        // tokens.json malformed — skip token fields
+      }
+    }
+
+    // DESIGN.md frontmatter fields
+    if (existsSync(designMdPath)) {
+      try {
+        const content = await readFile(designMdPath, 'utf-8');
+        const fm = parseFrontmatter(content);
+        brand.hasFrontmatter = fm.hasFrontmatter;
+        if (fm.hasFrontmatter) withFrontmatter++;
+
+        if (fm.completeness !== null) {
+          brand.completeness = fm.completeness;
+          withScore++;
+        }
+        if (fm.extractedAt) {
+          brand.extractedAt = fm.extractedAt;
+        }
+      } catch {
+        brand.hasFrontmatter = false;
+      }
+      enriched++;
+    } else {
+      skipped++;
+    }
+  }
+
+  // Update metadata
+  catalog.enrichedAt = new Date().toISOString();
+  catalog.version = bumpVersion(catalog.version || '1.0.0');
+
+  await writeFile(CATALOG_PATH, JSON.stringify(catalog, null, 2));
+
+  console.log(`✅ Catalog enriched and saved to catalog/index.json`);
+  console.log(`   ${enriched} brands processed | ${skipped} skipped (no DESIGN.md)`);
+  console.log(`   ${withFrontmatter} have YAML frontmatter | ${withScore} have completeness scores`);
+  console.log(`   Category distribution:`);
+
+  const catCounts = new Map<string, number>();
+  for (const b of catalog.brands) {
+    catCounts.set(b.category, (catCounts.get(b.category) || 0) + 1);
+  }
+  for (const [cat, count] of [...catCounts.entries()].sort((a, b) => b[1] - a[1])) {
+    console.log(`     ${cat}: ${count}`);
+  }
+}
+
+function bumpVersion(v: string): string {
+  const parts = v.split('.').map(Number);
+  parts[2] = (parts[2] || 0) + 1;
+  return parts.join('.');
+}
+
+main().catch(err => {
+  console.error('Error enriching catalog:', err);
+  process.exit(1);
+});