@detiq/crawler 0.1.3

package/dist/openapi-generator.js.map

@@ -0,0 +1,7 @@
+{
+  "version": 3,
+  "sources": ["../src/openapi-generator.ts"],
+  "sourcesContent": ["/**\n * OpenAPI 3.0 spec generator from observed network traffic.\n *\n * Takes the ApiEndpointObservationInput[] collected by the network observer\n * during crawl and generates a minimal but valid OpenAPI 3.0 spec.\n *\n * The spec captures: endpoints, methods, observed response shapes, status codes.\n * It does NOT capture request bodies (we'd need to intercept requests too) \u2014 that's\n * marked as a TODO. But it's enough for generating smoke-test suites.\n */\n\nimport type { ApiEndpointObservationInput } from './network-observer.js';\n\nexport interface OpenApiSpec {\n  openapi: '3.0.3';\n  info: { title: string; version: string; description?: string };\n  servers: Array<{ url: string }>;\n  paths: Record<string, Record<string, OpenApiOperation>>;\n}\n\ninterface OpenApiOperation {\n  summary: string;\n  parameters?: OpenApiParam[];\n  responses: Record<string, { description: string; content?: Record<string, { schema: unknown }> }>;\n  tags?: string[];\n}\n\ninterface OpenApiParam {\n  name: string;\n  in: 'path' | 'query';\n  required: boolean;\n  schema: { type: string };\n}\n\nfunction extractPathParams(urlPattern: string): OpenApiParam[] {\n  const params: OpenApiParam[] = [];\n  const matches = urlPattern.matchAll(/\\{(\\w+)\\}/g);\n  for (const m of matches) {\n    params.push({ name: m[1], in: 'path', required: true, schema: { type: 'string' } });\n  }\n  return params;\n}\n\nfunction methodSummary(method: string, path: string): string {\n  const resource = path.split('/').filter(s => !s.startsWith('{') && s.length > 0).pop() ?? 'resource';\n  const summaries: Record<string, string> = {\n    GET: `Get ${resource}`,\n    POST: `Create ${resource}`,\n    PUT: `Update ${resource}`,\n    PATCH: `Partially update ${resource}`,\n    DELETE: `Delete ${resource}`,\n  };\n  return summaries[method.toUpperCase()] ?? `${method} ${resource}`;\n}\n\nfunction inferTag(path: string): string {\n  const parts = path.split('/').filter(s => s.length > 0 && !s.startsWith('{'));\n  return parts[0] ?? 'default';\n}\n\nexport function generateOpenApiSpec(\n  observations: ApiEndpointObservationInput[],\n  opts: { title?: string; baseUrl?: string; version?: string } = {},\n): OpenApiSpec {\n  const { title = 'ZeTa Discovered API', baseUrl = '/', version = '1.0.0' } = opts;\n\n  const spec: OpenApiSpec = {\n    openapi: '3.0.3',\n    info: { title, version, description: 'Auto-generated by ZeTa crawler from observed network traffic.' },\n    servers: [{ url: baseUrl }],\n    paths: {},\n  };\n\n  for (const obs of observations) {\n    const path = obs.urlPattern;\n    const method = obs.method.toLowerCase();\n\n    if (!spec.paths[path]) spec.paths[path] = {};\n\n    const params = extractPathParams(path);\n    const statusCode = obs.statusCode ?? 200;\n    const responses: OpenApiOperation['responses'] = {\n      [String(statusCode)]: {\n        description: statusCode < 400 ? 'Successful response' : 'Error response',\n        content: obs.responseShape\n          ? { 'application/json': { schema: shapeToJsonSchema(obs.responseShape) } }\n          : undefined,\n      },\n    };\n\n    const op: OpenApiOperation = {\n      summary: methodSummary(obs.method, path),\n      parameters: params.length > 0 ? params : undefined,\n      responses,\n      tags: [inferTag(path)],\n    };\n\n    spec.paths[path][method] = op;\n  }\n\n  return spec;\n}\n\nfunction shapeToJsonSchema(shape: unknown): unknown {\n  if (Array.isArray(shape)) return { type: 'array', items: shape.length > 0 ? shapeToJsonSchema(shape[0]) : {} };\n  if (shape && typeof shape === 'object') {\n    const props: Record<string, unknown> = {};\n    for (const [k, v] of Object.entries(shape as object)) props[k] = shapeToJsonSchema(v);\n    return { type: 'object', properties: props };\n  }\n  return { type: String(shape) };\n}\n\nexport function specToYaml(spec: OpenApiSpec): string {\n  // Minimal YAML serializer \u2014 avoids yaml dep\n  function toYaml(val: unknown, indent = 0): string {\n    const pad = '  '.repeat(indent);\n    if (val === null || val === undefined) return 'null';\n    if (typeof val === 'string') return val.includes('\\n') || val.includes(':') ? `\"${val.replace(/\"/g, '\\\\\"')}\"` : val;\n    if (typeof val === 'number' || typeof val === 'boolean') return String(val);\n    if (Array.isArray(val)) return val.map(v => `\\n${pad}- ${toYaml(v, indent + 1).trimStart()}`).join('') || '[]';\n    if (typeof val === 'object') {\n      const entries = Object.entries(val as object);\n      if (entries.length === 0) return '{}';\n      return entries.map(([k, v]) => {\n        const vStr = toYaml(v, indent + 1);\n        return typeof v === 'object' && v !== null && !Array.isArray(v)\n          ? `\\n${pad}${k}:${vStr}`\n          : `\\n${pad}${k}: ${vStr.trimStart()}`;\n      }).join('');\n    }\n    return String(val);\n  }\n  return `openapi: \"${spec.openapi}\"${toYaml(spec, 0)}`;\n}\n"],
+  "mappings": "AAkCA,SAAS,kBAAkB,YAAoC;AAC7D,QAAM,SAAyB,CAAC;AAChC,QAAM,UAAU,WAAW,SAAS,YAAY;AAChD,aAAW,KAAK,SAAS;AACvB,WAAO,KAAK,EAAE,MAAM,EAAE,CAAC,GAAG,IAAI,QAAQ,UAAU,MAAM,QAAQ,EAAE,MAAM,SAAS,EAAE,CAAC;AAAA,EACpF;AACA,SAAO;AACT;AAEA,SAAS,cAAc,QAAgB,MAAsB;AAC3D,QAAM,WAAW,KAAK,MAAM,GAAG,EAAE,OAAO,OAAK,CAAC,EAAE,WAAW,GAAG,KAAK,EAAE,SAAS,CAAC,EAAE,IAAI,KAAK;AAC1F,QAAM,YAAoC;AAAA,IACxC,KAAK,OAAO,QAAQ;AAAA,IACpB,MAAM,UAAU,QAAQ;AAAA,IACxB,KAAK,UAAU,QAAQ;AAAA,IACvB,OAAO,oBAAoB,QAAQ;AAAA,IACnC,QAAQ,UAAU,QAAQ;AAAA,EAC5B;AACA,SAAO,UAAU,OAAO,YAAY,CAAC,KAAK,GAAG,MAAM,IAAI,QAAQ;AACjE;AAEA,SAAS,SAAS,MAAsB;AACtC,QAAM,QAAQ,KAAK,MAAM,GAAG,EAAE,OAAO,OAAK,EAAE,SAAS,KAAK,CAAC,EAAE,WAAW,GAAG,CAAC;AAC5E,SAAO,MAAM,CAAC,KAAK;AACrB;AAEO,SAAS,oBACd,cACA,OAA+D,CAAC,GACnD;AACb,QAAM,EAAE,QAAQ,uBAAuB,UAAU,KAAK,UAAU,QAAQ,IAAI;AAE5E,QAAM,OAAoB;AAAA,IACxB,SAAS;AAAA,IACT,MAAM,EAAE,OAAO,SAAS,aAAa,gEAAgE;AAAA,IACrG,SAAS,CAAC,EAAE,KAAK,QAAQ,CAAC;AAAA,IAC1B,OAAO,CAAC;AAAA,EACV;AAEA,aAAW,OAAO,cAAc;AAC9B,UAAM,OAAO,IAAI;AACjB,UAAM,SAAS,IAAI,OAAO,YAAY;AAEtC,QAAI,CAAC,KAAK,MAAM,IAAI,EAAG,MAAK,MAAM,IAAI,IAAI,CAAC;AAE3C,UAAM,SAAS,kBAAkB,IAAI;AACrC,UAAM,aAAa,IAAI,cAAc;AACrC,UAAM,YAA2C;AAAA,MAC/C,CAAC,OAAO,UAAU,CAAC,GAAG;AAAA,QACpB,aAAa,aAAa,MAAM,wBAAwB;AAAA,QACxD,SAAS,IAAI,gBACT,EAAE,oBAAoB,EAAE,QAAQ,kBAAkB,IAAI,aAAa,EAAE,EAAE,IACvE;AAAA,MACN;AAAA,IACF;AAEA,UAAM,KAAuB;AAAA,MAC3B,SAAS,cAAc,IAAI,QAAQ,IAAI;AAAA,MACvC,YAAY,OAAO,SAAS,IAAI,SAAS;AAAA,MACzC;AAAA,MACA,MAAM,CAAC,SAAS,IAAI,CAAC;AAAA,IACvB;AAEA,SAAK,MAAM,IAAI,EAAE,MAAM,IAAI;AAAA,EAC7B;AAEA,SAAO;AACT;AAEA,SAAS,kBAAkB,OAAyB;AAClD,MAAI,MAAM,QAAQ,KAAK,EAAG,QAAO,EAAE,MAAM,SAAS,OAAO,MAAM,SAAS,IAAI,kBAAkB,MAAM,CAAC,CAAC,IAAI,CAAC,EAAE;AAC7G,MAAI,SAAS,OAAO,UAAU,UAAU;AACtC,UAAM,QAAiC,CAAC;AACxC,eAAW,CAAC,GAAG,CAAC,KAAK,OAAO,QAAQ,KAAe,EAAG,OAAM,CAAC,IAAI,kBAAkB,CAAC;AACpF,WAAO,EAAE,MAAM,UAAU,YAAY,MAAM;AAAA,EAC7C;AACA,SAAO,EAAE,MAAM,OAAO,KAAK,EAAE;AAC/B;AAEO,SAAS,WAAW,MAA2B;AAEpD,WAAS,OAAO,KAAc,SAAS,GAAW;AAChD,UAAM,MAAM,KAAK,OAAO,MAAM;AAC9B,QAAI,QAAQ,QAAQ,QAAQ,OAAW,QAAO;AAC9C,QAAI,OAAO,QAAQ,SAAU,QAAO,IAAI,SAAS,IAAI,KAAK,IAAI,SAAS,GAAG,IAAI,IAAI,IAAI,QAAQ,MAAM,KAAK,CAAC,MAAM;AAChH,QAAI,OAAO,QAAQ,YAAY,OAAO,QAAQ,UAAW,QAAO,OAAO,GAAG;AAC1E,QAAI,MAAM,QAAQ,GAAG,EAAG,QAAO,IAAI,IAAI,OAAK;AAAA,EAAK,GAAG,KAAK,OAAO,GAAG,SAAS,CAAC,EAAE,UAAU,CAAC,EAAE,EAAE,KAAK,EAAE,KAAK;AAC1G,QAAI,OAAO,QAAQ,UAAU;AAC3B,YAAM,UAAU,OAAO,QAAQ,GAAa;AAC5C,UAAI,QAAQ,WAAW,EAAG,QAAO;AACjC,aAAO,QAAQ,IAAI,CAAC,CAAC,GAAG,CAAC,MAAM;AAC7B,cAAM,OAAO,OAAO,GAAG,SAAS,CAAC;AACjC,eAAO,OAAO,MAAM,YAAY,MAAM,QAAQ,CAAC,MAAM,QAAQ,CAAC,IAC1D;AAAA,EAAK,GAAG,GAAG,CAAC,IAAI,IAAI,KACpB;AAAA,EAAK,GAAG,GAAG,CAAC,KAAK,KAAK,UAAU,CAAC;AAAA,MACvC,CAAC,EAAE,KAAK,EAAE;AAAA,IACZ;AACA,WAAO,OAAO,GAAG;AAAA,EACnB;AACA,SAAO,aAAa,KAAK,OAAO,IAAI,OAAO,MAAM,CAAC,CAAC;AACrD;",
+  "names": []
+}

package/dist/pdf-extractor.d.ts

@@ -0,0 +1,28 @@
+/**
+ * PDF extraction — discovers and extracts text from PDF URLs found during crawl.
+ *
+ * When the HTTP engine or browser engine discovers a link to a .pdf file,
+ * this module fetches the PDF, extracts its text content, and returns it
+ * as a screen-compatible markdown string for test generation.
+ *
+ * Uses pdf-parse (already installed in API package, accessible via dynamic import).
+ */
+export interface PdfScreen {
+    url: string;
+    title: string;
+    markdown: string;
+    pageCount: number;
+    info: Record<string, string>;
+    sizeBytes: number;
+}
+export declare function isPdfUrl(url: string, contentType?: string): boolean;
+export declare function extractPdfFromUrl(url: string, opts?: {
+    userAgent?: string;
+    timeoutMs?: number;
+    maxSizeBytes?: number;
+}): Promise<PdfScreen | null>;
+/**
+ * Filter a list of discovered URLs to find PDF links.
+ * Used by http-engine and browser crawler to decide which links to pass to extractPdfFromUrl.
+ */
+export declare function filterPdfUrls(urls: string[]): string[];

package/dist/pdf-extractor.js

@@ -0,0 +1,81 @@
+const PDF_EXT = /\.pdf(\?.*)?$/i;
+const PDF_MIME = "application/pdf";
+function isPdfUrl(url, contentType) {
+  return PDF_EXT.test(url) || (contentType?.includes(PDF_MIME) ?? false);
+}
+async function extractPdfFromUrl(url, opts = {}) {
+  const { userAgent = "ZeTa-Crawler/1.0", timeoutMs = 3e4, maxSizeBytes = 50 * 1024 * 1024 } = opts;
+  let buffer;
+  let sizeBytes;
+  try {
+    const response = await fetch(url, {
+      headers: { "User-Agent": userAgent, "Accept": "application/pdf,*/*" },
+      signal: AbortSignal.timeout(timeoutMs)
+    });
+    if (!response.ok) return null;
+    const contentType = response.headers.get("content-type") ?? "";
+    if (!contentType.includes("pdf") && !PDF_EXT.test(url)) return null;
+    const arrayBuffer = await response.arrayBuffer();
+    sizeBytes = arrayBuffer.byteLength;
+    if (sizeBytes > maxSizeBytes) {
+      console.warn(`[pdf-extractor] PDF too large (${sizeBytes} bytes), skipping: ${url}`);
+      return null;
+    }
+    buffer = Buffer.from(arrayBuffer);
+  } catch (e) {
+    console.error(`[pdf-extractor] Failed to fetch PDF ${url}: ${e.message}`);
+    return null;
+  }
+  try {
+    const pdfParse = await import("pdf-parse/lib/pdf-parse.js").catch(
+      () => import("pdf-parse")
+    );
+    const parse = pdfParse.default ?? pdfParse;
+    const data = await parse(buffer, { max: 0 });
+    const title = data.info?.Title || extractTitleFromUrl(url);
+    const markdown = pdfTextToMarkdown(data.text ?? "", title);
+    return {
+      url,
+      title,
+      markdown,
+      pageCount: data.numpages ?? 0,
+      info: {
+        author: data.info?.Author ?? "",
+        subject: data.info?.Subject ?? "",
+        keywords: data.info?.Keywords ?? "",
+        creator: data.info?.Creator ?? ""
+      },
+      sizeBytes
+    };
+  } catch (e) {
+    console.error(`[pdf-extractor] Failed to parse PDF ${url}: ${e.message}`);
+    return null;
+  }
+}
+function extractTitleFromUrl(url) {
+  try {
+    const { pathname } = new URL(url);
+    const filename = pathname.split("/").pop() ?? "document";
+    return filename.replace(/\.pdf$/i, "").replace(/[-_]/g, " ").trim();
+  } catch {
+    return "PDF Document";
+  }
+}
+function pdfTextToMarkdown(text, title) {
+  if (!text.trim()) return `# ${title}
+
+*No extractable text content.*`;
+  const cleaned = text.replace(/\f/g, "\n\n---\n\n").replace(/\r\n/g, "\n").replace(/\r/g, "\n").replace(/\n{3,}/g, "\n\n").replace(/[ \t]{2,}/g, " ").trim();
+  return `# ${title}
+
+${cleaned}`;
+}
+function filterPdfUrls(urls) {
+  return urls.filter((u) => PDF_EXT.test(u));
+}
+export {
+  extractPdfFromUrl,
+  filterPdfUrls,
+  isPdfUrl
+};
+//# sourceMappingURL=pdf-extractor.js.map

package/dist/pdf-extractor.js.map

@@ -0,0 +1,7 @@
+{
+  "version": 3,
+  "sources": ["../src/pdf-extractor.ts"],
+  "sourcesContent": ["/**\n * PDF extraction \u2014 discovers and extracts text from PDF URLs found during crawl.\n *\n * When the HTTP engine or browser engine discovers a link to a .pdf file,\n * this module fetches the PDF, extracts its text content, and returns it\n * as a screen-compatible markdown string for test generation.\n *\n * Uses pdf-parse (already installed in API package, accessible via dynamic import).\n */\n\nconst PDF_EXT = /\\.pdf(\\?.*)?$/i;\nconst PDF_MIME = 'application/pdf';\n\nexport interface PdfScreen {\n  url: string;\n  title: string;\n  markdown: string;\n  pageCount: number;\n  info: Record<string, string>;\n  sizeBytes: number;\n}\n\nexport function isPdfUrl(url: string, contentType?: string): boolean {\n  return PDF_EXT.test(url) || (contentType?.includes(PDF_MIME) ?? false);\n}\n\nexport async function extractPdfFromUrl(url: string, opts: {\n  userAgent?: string;\n  timeoutMs?: number;\n  maxSizeBytes?: number;\n} = {}): Promise<PdfScreen | null> {\n  const { userAgent = 'ZeTa-Crawler/1.0', timeoutMs = 30_000, maxSizeBytes = 50 * 1024 * 1024 } = opts;\n\n  let buffer: Buffer;\n  let sizeBytes: number;\n\n  try {\n    const response = await fetch(url, {\n      headers: { 'User-Agent': userAgent, 'Accept': 'application/pdf,*/*' },\n      signal: AbortSignal.timeout(timeoutMs),\n    });\n    if (!response.ok) return null;\n\n    const contentType = response.headers.get('content-type') ?? '';\n    if (!contentType.includes('pdf') && !PDF_EXT.test(url)) return null;\n\n    const arrayBuffer = await response.arrayBuffer();\n    sizeBytes = arrayBuffer.byteLength;\n    if (sizeBytes > maxSizeBytes) {\n      console.warn(`[pdf-extractor] PDF too large (${sizeBytes} bytes), skipping: ${url}`);\n      return null;\n    }\n    buffer = Buffer.from(arrayBuffer);\n  } catch (e: any) {\n    console.error(`[pdf-extractor] Failed to fetch PDF ${url}: ${e.message}`);\n    return null;\n  }\n\n  try {\n    // pdf-parse is installed in the API package \u2014 use dynamic import\n    // so this file compiles without adding pdf-parse to crawler's package.json\n    // eslint-disable-next-line @typescript-eslint/ban-ts-comment\n    // @ts-ignore \u2013 no types for pdf-parse/lib/pdf-parse.js\n    const pdfParse = await import('pdf-parse/lib/pdf-parse.js').catch(\n      () => import('pdf-parse' as any)\n    ) as any;\n    const parse = pdfParse.default ?? pdfParse;\n    const data = await parse(buffer, { max: 0 }); // max: 0 = all pages\n\n    const title = (data.info?.Title as string) || extractTitleFromUrl(url);\n    const markdown = pdfTextToMarkdown(data.text ?? '', title);\n\n    return {\n      url,\n      title,\n      markdown,\n      pageCount: data.numpages ?? 0,\n      info: {\n        author: data.info?.Author ?? '',\n        subject: data.info?.Subject ?? '',\n        keywords: data.info?.Keywords ?? '',\n        creator: data.info?.Creator ?? '',\n      },\n      sizeBytes,\n    };\n  } catch (e: any) {\n    console.error(`[pdf-extractor] Failed to parse PDF ${url}: ${e.message}`);\n    return null;\n  }\n}\n\nfunction extractTitleFromUrl(url: string): string {\n  try {\n    const { pathname } = new URL(url);\n    const filename = pathname.split('/').pop() ?? 'document';\n    return filename.replace(/\\.pdf$/i, '').replace(/[-_]/g, ' ').trim();\n  } catch {\n    return 'PDF Document';\n  }\n}\n\nfunction pdfTextToMarkdown(text: string, title: string): string {\n  if (!text.trim()) return `# ${title}\\n\\n*No extractable text content.*`;\n\n  // Clean up common PDF extraction artifacts\n  const cleaned = text\n    .replace(/\\f/g, '\\n\\n---\\n\\n') // form feed = page break\n    .replace(/\\r\\n/g, '\\n')\n    .replace(/\\r/g, '\\n')\n    .replace(/\\n{3,}/g, '\\n\\n')\n    .replace(/[ \\t]{2,}/g, ' ')\n    .trim();\n\n  return `# ${title}\\n\\n${cleaned}`;\n}\n\n/**\n * Filter a list of discovered URLs to find PDF links.\n * Used by http-engine and browser crawler to decide which links to pass to extractPdfFromUrl.\n */\nexport function filterPdfUrls(urls: string[]): string[] {\n  return urls.filter(u => PDF_EXT.test(u));\n}\n"],
+  "mappings": "AAUA,MAAM,UAAU;AAChB,MAAM,WAAW;AAWV,SAAS,SAAS,KAAa,aAA+B;AACnE,SAAO,QAAQ,KAAK,GAAG,MAAM,aAAa,SAAS,QAAQ,KAAK;AAClE;AAEA,eAAsB,kBAAkB,KAAa,OAIjD,CAAC,GAA8B;AACjC,QAAM,EAAE,YAAY,oBAAoB,YAAY,KAAQ,eAAe,KAAK,OAAO,KAAK,IAAI;AAEhG,MAAI;AACJ,MAAI;AAEJ,MAAI;AACF,UAAM,WAAW,MAAM,MAAM,KAAK;AAAA,MAChC,SAAS,EAAE,cAAc,WAAW,UAAU,sBAAsB;AAAA,MACpE,QAAQ,YAAY,QAAQ,SAAS;AAAA,IACvC,CAAC;AACD,QAAI,CAAC,SAAS,GAAI,QAAO;AAEzB,UAAM,cAAc,SAAS,QAAQ,IAAI,cAAc,KAAK;AAC5D,QAAI,CAAC,YAAY,SAAS,KAAK,KAAK,CAAC,QAAQ,KAAK,GAAG,EAAG,QAAO;AAE/D,UAAM,cAAc,MAAM,SAAS,YAAY;AAC/C,gBAAY,YAAY;AACxB,QAAI,YAAY,cAAc;AAC5B,cAAQ,KAAK,kCAAkC,SAAS,sBAAsB,GAAG,EAAE;AACnF,aAAO;AAAA,IACT;AACA,aAAS,OAAO,KAAK,WAAW;AAAA,EAClC,SAAS,GAAQ;AACf,YAAQ,MAAM,uCAAuC,GAAG,KAAK,EAAE,OAAO,EAAE;AACxE,WAAO;AAAA,EACT;AAEA,MAAI;AAKF,UAAM,WAAW,MAAM,OAAO,4BAA4B,EAAE;AAAA,MAC1D,MAAM,OAAO,WAAkB;AAAA,IACjC;AACA,UAAM,QAAQ,SAAS,WAAW;AAClC,UAAM,OAAO,MAAM,MAAM,QAAQ,EAAE,KAAK,EAAE,CAAC;AAE3C,UAAM,QAAS,KAAK,MAAM,SAAoB,oBAAoB,GAAG;AACrE,UAAM,WAAW,kBAAkB,KAAK,QAAQ,IAAI,KAAK;AAEzD,WAAO;AAAA,MACL;AAAA,MACA;AAAA,MACA;AAAA,MACA,WAAW,KAAK,YAAY;AAAA,MAC5B,MAAM;AAAA,QACJ,QAAQ,KAAK,MAAM,UAAU;AAAA,QAC7B,SAAS,KAAK,MAAM,WAAW;AAAA,QAC/B,UAAU,KAAK,MAAM,YAAY;AAAA,QACjC,SAAS,KAAK,MAAM,WAAW;AAAA,MACjC;AAAA,MACA;AAAA,IACF;AAAA,EACF,SAAS,GAAQ;AACf,YAAQ,MAAM,uCAAuC,GAAG,KAAK,EAAE,OAAO,EAAE;AACxE,WAAO;AAAA,EACT;AACF;AAEA,SAAS,oBAAoB,KAAqB;AAChD,MAAI;AACF,UAAM,EAAE,SAAS,IAAI,IAAI,IAAI,GAAG;AAChC,UAAM,WAAW,SAAS,MAAM,GAAG,EAAE,IAAI,KAAK;AAC9C,WAAO,SAAS,QAAQ,WAAW,EAAE,EAAE,QAAQ,SAAS,GAAG,EAAE,KAAK;AAAA,EACpE,QAAQ;AACN,WAAO;AAAA,EACT;AACF;AAEA,SAAS,kBAAkB,MAAc,OAAuB;AAC9D,MAAI,CAAC,KAAK,KAAK,EAAG,QAAO,KAAK,KAAK;AAAA;AAAA;AAGnC,QAAM,UAAU,KACb,QAAQ,OAAO,aAAa,EAC5B,QAAQ,SAAS,IAAI,EACrB,QAAQ,OAAO,IAAI,EACnB,QAAQ,WAAW,MAAM,EACzB,QAAQ,cAAc,GAAG,EACzB,KAAK;AAER,SAAO,KAAK,KAAK;AAAA;AAAA,EAAO,OAAO;AACjC;AAMO,SAAS,cAAc,MAA0B;AACtD,SAAO,KAAK,OAAO,OAAK,QAAQ,KAAK,CAAC,CAAC;AACzC;",
+  "names": []
+}

package/dist/performance-metrics.d.ts

@@ -0,0 +1,53 @@
+/**
+ * Performance metrics collection via CDP + Playwright page.evaluate().
+ *
+ * Collects Core Web Vitals and resource timing after page navigation.
+ * Stored alongside screen data for test generation context and regression detection.
+ *
+ * Metrics:
+ * - LCP (Largest Contentful Paint) — via PerformanceObserver in page
+ * - CLS (Cumulative Layout Shift) — via PerformanceObserver in page
+ * - TTFB (Time to First Byte) — from navigation timing
+ * - DOMContentLoaded — from navigation timing
+ * - Resource count and size by type
+ */
+import type { Page } from 'playwright';
+export interface CoreWebVitals {
+    lcp?: number;
+    cls?: number;
+    ttfb?: number;
+    fcp?: number;
+    domContentLoaded?: number;
+    loadEventEnd?: number;
+}
+export interface ResourceSummary {
+    scripts: {
+        count: number;
+        totalBytes: number;
+    };
+    styles: {
+        count: number;
+        totalBytes: number;
+    };
+    images: {
+        count: number;
+        totalBytes: number;
+    };
+    fonts: {
+        count: number;
+        totalBytes: number;
+    };
+    other: {
+        count: number;
+        totalBytes: number;
+    };
+}
+export interface PagePerfMetrics {
+    url: string;
+    vitals: CoreWebVitals;
+    resources: ResourceSummary;
+    totalTransferBytes: number;
+    requestCount: number;
+    measuredAt: string;
+}
+export declare function collectPerfMetrics(page: Page, url: string): Promise<PagePerfMetrics>;

package/dist/performance-metrics.js

@@ -0,0 +1,83 @@
+async function collectPerfMetrics(page, url) {
+  const measuredAt = (/* @__PURE__ */ new Date()).toISOString();
+  let vitals = {};
+  try {
+    vitals = await page.evaluate(() => {
+      return new Promise((resolve) => {
+        const result = {};
+        const nav = performance.getEntriesByType("navigation")[0];
+        if (nav) {
+          result.ttfb = Math.round(nav.responseStart - nav.requestStart);
+          result.domContentLoaded = Math.round(nav.domContentLoadedEventEnd);
+          result.loadEventEnd = Math.round(nav.loadEventEnd);
+        }
+        let lcpDone = false;
+        let clsDone = false;
+        const tryResolve = () => {
+          if (lcpDone && clsDone) resolve(result);
+        };
+        try {
+          new PerformanceObserver((list) => {
+            const entries = list.getEntries();
+            if (entries.length > 0) result.lcp = Math.round(entries[entries.length - 1].startTime);
+            lcpDone = true;
+            tryResolve();
+          }).observe({ type: "largest-contentful-paint", buffered: true });
+        } catch {
+          lcpDone = true;
+        }
+        try {
+          let cls = 0;
+          new PerformanceObserver((list) => {
+            for (const e of list.getEntries()) {
+              if (!e.hadRecentInput) cls += e.value;
+            }
+            result.cls = Math.round(cls * 1e3) / 1e3;
+            clsDone = true;
+            tryResolve();
+          }).observe({ type: "layout-shift", buffered: true });
+        } catch {
+          clsDone = true;
+        }
+        try {
+          const fcp = performance.getEntriesByName("first-contentful-paint")[0];
+          if (fcp) result.fcp = Math.round(fcp.startTime);
+        } catch {
+        }
+        setTimeout(() => resolve(result), 3e3);
+      });
+    }).catch(() => ({}));
+  } catch {
+  }
+  let resources = {
+    scripts: { count: 0, totalBytes: 0 },
+    styles: { count: 0, totalBytes: 0 },
+    images: { count: 0, totalBytes: 0 },
+    fonts: { count: 0, totalBytes: 0 },
+    other: { count: 0, totalBytes: 0 }
+  };
+  let totalTransferBytes = 0;
+  let requestCount = 0;
+  try {
+    const entries = await page.evaluate(
+      () => performance.getEntriesByType("resource").map((e) => ({
+        name: e.name,
+        initiatorType: e.initiatorType,
+        transferSize: e.transferSize ?? 0
+      }))
+    ).catch(() => []);
+    for (const e of entries) {
+      requestCount++;
+      totalTransferBytes += e.transferSize;
+      const cat = e.initiatorType === "script" ? "scripts" : e.initiatorType === "css" || e.name?.endsWith(".css") ? "styles" : e.initiatorType === "img" ? "images" : e.name?.match(/\.(woff|woff2|ttf|otf|eot)/) ? "fonts" : "other";
+      resources[cat].count++;
+      resources[cat].totalBytes += e.transferSize;
+    }
+  } catch {
+  }
+  return { url, vitals, resources, totalTransferBytes, requestCount, measuredAt };
+}
+export {
+  collectPerfMetrics
+};
+//# sourceMappingURL=performance-metrics.js.map

package/dist/performance-metrics.js.map

@@ -0,0 +1,7 @@
+{
+  "version": 3,
+  "sources": ["../src/performance-metrics.ts"],
+  "sourcesContent": ["/**\n * Performance metrics collection via CDP + Playwright page.evaluate().\n *\n * Collects Core Web Vitals and resource timing after page navigation.\n * Stored alongside screen data for test generation context and regression detection.\n *\n * Metrics:\n * - LCP (Largest Contentful Paint) \u2014 via PerformanceObserver in page\n * - CLS (Cumulative Layout Shift) \u2014 via PerformanceObserver in page\n * - TTFB (Time to First Byte) \u2014 from navigation timing\n * - DOMContentLoaded \u2014 from navigation timing\n * - Resource count and size by type\n */\n\nimport type { Page } from 'playwright';\n\nexport interface CoreWebVitals {\n  lcp?: number;   // ms\n  cls?: number;   // unitless score\n  ttfb?: number;  // ms\n  fcp?: number;   // First Contentful Paint ms\n  domContentLoaded?: number; // ms\n  loadEventEnd?: number;     // ms\n}\n\nexport interface ResourceSummary {\n  scripts: { count: number; totalBytes: number };\n  styles: { count: number; totalBytes: number };\n  images: { count: number; totalBytes: number };\n  fonts: { count: number; totalBytes: number };\n  other: { count: number; totalBytes: number };\n}\n\nexport interface PagePerfMetrics {\n  url: string;\n  vitals: CoreWebVitals;\n  resources: ResourceSummary;\n  totalTransferBytes: number;\n  requestCount: number;\n  measuredAt: string;\n}\n\nexport async function collectPerfMetrics(page: Page, url: string): Promise<PagePerfMetrics> {\n  const measuredAt = new Date().toISOString();\n\n  // Inject PerformanceObserver for LCP + CLS, wait up to 3s\n  let vitals: CoreWebVitals = {};\n  try {\n    vitals = await page.evaluate(() => {\n      return new Promise<any>((resolve) => {\n        const result: any = {};\n\n        // Navigation timing\n        const nav = performance.getEntriesByType('navigation')[0] as any;\n        if (nav) {\n          result.ttfb = Math.round(nav.responseStart - nav.requestStart);\n          result.domContentLoaded = Math.round(nav.domContentLoadedEventEnd);\n          result.loadEventEnd = Math.round(nav.loadEventEnd);\n        }\n\n        let lcpDone = false;\n        let clsDone = false;\n        const tryResolve = () => { if (lcpDone && clsDone) resolve(result); };\n\n        // LCP\n        try {\n          new PerformanceObserver((list) => {\n            const entries = list.getEntries();\n            if (entries.length > 0) result.lcp = Math.round((entries[entries.length - 1] as any).startTime);\n            lcpDone = true;\n            tryResolve();\n          }).observe({ type: 'largest-contentful-paint', buffered: true });\n        } catch { lcpDone = true; }\n\n        // CLS\n        try {\n          let cls = 0;\n          new PerformanceObserver((list) => {\n            for (const e of list.getEntries() as any[]) {\n              if (!e.hadRecentInput) cls += e.value;\n            }\n            result.cls = Math.round(cls * 1000) / 1000;\n            clsDone = true;\n            tryResolve();\n          }).observe({ type: 'layout-shift', buffered: true });\n        } catch { clsDone = true; }\n\n        // FCP\n        try {\n          const fcp = performance.getEntriesByName('first-contentful-paint')[0] as any;\n          if (fcp) result.fcp = Math.round(fcp.startTime);\n        } catch {}\n\n        // Fallback resolve after 3s\n        setTimeout(() => resolve(result), 3_000);\n      });\n    }).catch(() => ({}));\n  } catch { /* page closed or navigate already */ }\n\n  // Resource timing\n  let resources: ResourceSummary = {\n    scripts: { count: 0, totalBytes: 0 },\n    styles: { count: 0, totalBytes: 0 },\n    images: { count: 0, totalBytes: 0 },\n    fonts: { count: 0, totalBytes: 0 },\n    other: { count: 0, totalBytes: 0 },\n  };\n  let totalTransferBytes = 0;\n  let requestCount = 0;\n\n  try {\n    const entries = await page.evaluate(() =>\n      performance.getEntriesByType('resource').map((e: any) => ({\n        name: e.name,\n        initiatorType: e.initiatorType,\n        transferSize: e.transferSize ?? 0,\n      }))\n    ).catch(() => [] as any[]);\n\n    for (const e of entries) {\n      requestCount++;\n      totalTransferBytes += e.transferSize;\n      const cat = e.initiatorType === 'script' ? 'scripts'\n        : e.initiatorType === 'css' || e.name?.endsWith('.css') ? 'styles'\n        : e.initiatorType === 'img' ? 'images'\n        : e.name?.match(/\\.(woff|woff2|ttf|otf|eot)/) ? 'fonts'\n        : 'other';\n      resources[cat as keyof ResourceSummary].count++;\n      resources[cat as keyof ResourceSummary].totalBytes += e.transferSize;\n    }\n  } catch {}\n\n  return { url, vitals, resources, totalTransferBytes, requestCount, measuredAt };\n}\n"],
+  "mappings": "AA0CA,eAAsB,mBAAmB,MAAY,KAAuC;AAC1F,QAAM,cAAa,oBAAI,KAAK,GAAE,YAAY;AAG1C,MAAI,SAAwB,CAAC;AAC7B,MAAI;AACF,aAAS,MAAM,KAAK,SAAS,MAAM;AACjC,aAAO,IAAI,QAAa,CAAC,YAAY;AACnC,cAAM,SAAc,CAAC;AAGrB,cAAM,MAAM,YAAY,iBAAiB,YAAY,EAAE,CAAC;AACxD,YAAI,KAAK;AACP,iBAAO,OAAO,KAAK,MAAM,IAAI,gBAAgB,IAAI,YAAY;AAC7D,iBAAO,mBAAmB,KAAK,MAAM,IAAI,wBAAwB;AACjE,iBAAO,eAAe,KAAK,MAAM,IAAI,YAAY;AAAA,QACnD;AAEA,YAAI,UAAU;AACd,YAAI,UAAU;AACd,cAAM,aAAa,MAAM;AAAE,cAAI,WAAW,QAAS,SAAQ,MAAM;AAAA,QAAG;AAGpE,YAAI;AACF,cAAI,oBAAoB,CAAC,SAAS;AAChC,kBAAM,UAAU,KAAK,WAAW;AAChC,gBAAI,QAAQ,SAAS,EAAG,QAAO,MAAM,KAAK,MAAO,QAAQ,QAAQ,SAAS,CAAC,EAAU,SAAS;AAC9F,sBAAU;AACV,uBAAW;AAAA,UACb,CAAC,EAAE,QAAQ,EAAE,MAAM,4BAA4B,UAAU,KAAK,CAAC;AAAA,QACjE,QAAQ;AAAE,oBAAU;AAAA,QAAM;AAG1B,YAAI;AACF,cAAI,MAAM;AACV,cAAI,oBAAoB,CAAC,SAAS;AAChC,uBAAW,KAAK,KAAK,WAAW,GAAY;AAC1C,kBAAI,CAAC,EAAE,eAAgB,QAAO,EAAE;AAAA,YAClC;AACA,mBAAO,MAAM,KAAK,MAAM,MAAM,GAAI,IAAI;AACtC,sBAAU;AACV,uBAAW;AAAA,UACb,CAAC,EAAE,QAAQ,EAAE,MAAM,gBAAgB,UAAU,KAAK,CAAC;AAAA,QACrD,QAAQ;AAAE,oBAAU;AAAA,QAAM;AAG1B,YAAI;AACF,gBAAM,MAAM,YAAY,iBAAiB,wBAAwB,EAAE,CAAC;AACpE,cAAI,IAAK,QAAO,MAAM,KAAK,MAAM,IAAI,SAAS;AAAA,QAChD,QAAQ;AAAA,QAAC;AAGT,mBAAW,MAAM,QAAQ,MAAM,GAAG,GAAK;AAAA,MACzC,CAAC;AAAA,IACH,CAAC,EAAE,MAAM,OAAO,CAAC,EAAE;AAAA,EACrB,QAAQ;AAAA,EAAwC;AAGhD,MAAI,YAA6B;AAAA,IAC/B,SAAS,EAAE,OAAO,GAAG,YAAY,EAAE;AAAA,IACnC,QAAQ,EAAE,OAAO,GAAG,YAAY,EAAE;AAAA,IAClC,QAAQ,EAAE,OAAO,GAAG,YAAY,EAAE;AAAA,IAClC,OAAO,EAAE,OAAO,GAAG,YAAY,EAAE;AAAA,IACjC,OAAO,EAAE,OAAO,GAAG,YAAY,EAAE;AAAA,EACnC;AACA,MAAI,qBAAqB;AACzB,MAAI,eAAe;AAEnB,MAAI;AACF,UAAM,UAAU,MAAM,KAAK;AAAA,MAAS,MAClC,YAAY,iBAAiB,UAAU,EAAE,IAAI,CAAC,OAAY;AAAA,QACxD,MAAM,EAAE;AAAA,QACR,eAAe,EAAE;AAAA,QACjB,cAAc,EAAE,gBAAgB;AAAA,MAClC,EAAE;AAAA,IACJ,EAAE,MAAM,MAAM,CAAC,CAAU;AAEzB,eAAW,KAAK,SAAS;AACvB;AACA,4BAAsB,EAAE;AACxB,YAAM,MAAM,EAAE,kBAAkB,WAAW,YACvC,EAAE,kBAAkB,SAAS,EAAE,MAAM,SAAS,MAAM,IAAI,WACxD,EAAE,kBAAkB,QAAQ,WAC5B,EAAE,MAAM,MAAM,4BAA4B,IAAI,UAC9C;AACJ,gBAAU,GAA4B,EAAE;AACxC,gBAAU,GAA4B,EAAE,cAAc,EAAE;AAAA,IAC1D;AAAA,EACF,QAAQ;AAAA,EAAC;AAET,SAAO,EAAE,KAAK,QAAQ,WAAW,oBAAoB,cAAc,WAAW;AAChF;",
+  "names": []
+}

package/dist/phash.d.ts

@@ -0,0 +1,38 @@
+/**
+ * Composites neutral grey (128,128,128) over each ignoredRegion (fractional
+ * 0-1 coordinates) so masked content — timestamps, live counters, ads —
+ * contributes nothing to any downstream pixel comparison. Shared by every
+ * comparison path (whole-image hash, per-cell grid hash, AI vision) so a
+ * region masked once is actually ignored everywhere, not just in the
+ * cheapest check.
+ */
+export declare function applyMask(imagePath: string, ignoredRegions: Array<{
+    x: number;
+    y: number;
+    width: number;
+    height: number;
+}>): Promise<Buffer>;
+export declare function screenshotDiffPercent(pathA: string, pathB: string, ignoredRegions?: Array<{
+    x: number;
+    y: number;
+    width: number;
+    height: number;
+}>): Promise<{
+    pHashDistance: number;
+    changeLevel: 'IDENTICAL' | 'MINOR' | 'MODERATE' | 'SIGNIFICANT';
+}>;
+/**
+ * Grid-based layout analysis: divides screenshots into a 3×3 grid and computes
+ * pHash per cell. Returns which regions changed and where, for the VisualDiff.changes field.
+ * Uses sharp's `extract` to crop each cell before hashing.
+ */
+export declare function gridLayoutDiff(pathA: string, pathB: string, ignoredRegions?: Array<{
+    x: number;
+    y: number;
+    width: number;
+    height: number;
+}>): Promise<Array<{
+    location: string;
+    severity: string;
+    description: string;
+}>>;

package/dist/phash.js

@@ -0,0 +1,115 @@
+import sharp from "sharp";
+async function applyMask(imagePath, ignoredRegions) {
+  const meta = await sharp(imagePath).metadata();
+  const W = meta.width ?? 1200;
+  const H = meta.height ?? 800;
+  const overlays = await Promise.all(
+    ignoredRegions.map(async (r) => {
+      const rW = Math.max(1, Math.round(r.width * W));
+      const rH = Math.max(1, Math.round(r.height * H));
+      const buf = await sharp({
+        create: { width: rW, height: rH, channels: 3, background: { r: 128, g: 128, b: 128 } }
+      }).png().toBuffer();
+      return { input: buf, top: Math.round(r.y * H), left: Math.round(r.x * W) };
+    })
+  );
+  return sharp(imagePath).composite(overlays).png().toBuffer();
+}
+async function hashWithMasks(imagePath, ignoredRegions = []) {
+  const pipeline = ignoredRegions.length > 0 ? sharp(await applyMask(imagePath, ignoredRegions)) : sharp(imagePath);
+  const buf = await pipeline.resize(8, 8, { fit: "fill" }).greyscale().raw().toBuffer();
+  const pixels = Array.from(buf);
+  const mean = pixels.reduce((s, p) => s + p, 0) / pixels.length;
+  return pixels.reduce((h, p, i) => p >= mean ? h | 1n << BigInt(63 - i) : h, 0n);
+}
+function hammingDistance(a, b) {
+  let xor = a ^ b;
+  let count = 0;
+  while (xor !== 0n) {
+    count += Number(xor & 1n);
+    xor >>= 1n;
+  }
+  return count;
+}
+async function screenshotDiffPercent(pathA, pathB, ignoredRegions = []) {
+  const [hashA, hashB] = await Promise.all([
+    hashWithMasks(pathA, ignoredRegions),
+    hashWithMasks(pathB, ignoredRegions)
+  ]);
+  let xor = hashA ^ hashB;
+  let dist = 0;
+  while (xor !== 0n) {
+    dist += Number(xor & 1n);
+    xor >>= 1n;
+  }
+  const pct = parseFloat((dist / 64 * 100).toFixed(1));
+  const changeLevel = pct === 0 ? "IDENTICAL" : pct < 10 ? "MINOR" : pct < 30 ? "MODERATE" : "SIGNIFICANT";
+  return { pHashDistance: pct, changeLevel };
+}
+async function gridLayoutDiff(pathA, pathB, ignoredRegions = []) {
+  const GRID_LABELS = [
+    "top-left",
+    "top-center",
+    "top-right",
+    "mid-left",
+    "mid-center",
+    "mid-right",
+    "bottom-left",
+    "bottom-center",
+    "bottom-right"
+  ];
+  try {
+    const [sourceA, sourceB] = ignoredRegions.length > 0 ? await Promise.all([applyMask(pathA, ignoredRegions), applyMask(pathB, ignoredRegions)]) : [pathA, pathB];
+    const [metaA, metaB] = await Promise.all([
+      sharp(sourceA).metadata(),
+      sharp(sourceB).metadata()
+    ]);
+    const W = Math.min(metaA.width ?? 1200, metaB.width ?? 1200);
+    const H = Math.min(metaA.height ?? 800, metaB.height ?? 800);
+    const cW = Math.floor(W / 3);
+    const cH = Math.floor(H / 3);
+    const changes = [];
+    for (let row = 0; row < 3; row++) {
+      for (let col = 0; col < 3; col++) {
+        const region = { left: col * cW, top: row * cH, width: cW, height: cH };
+        const label = GRID_LABELS[row * 3 + col];
+        const [hashA, hashB] = await Promise.all([
+          sharp(sourceA).extract(region).resize(8, 8, { fit: "fill" }).greyscale().raw().toBuffer().then((buf) => {
+            const pixels = Array.from(buf);
+            const mean = pixels.reduce((s, p) => s + p, 0) / pixels.length;
+            return pixels.reduce((h, p, i) => p >= mean ? h | 1n << BigInt(63 - i) : h, 0n);
+          }),
+          sharp(sourceB).extract(region).resize(8, 8, { fit: "fill" }).greyscale().raw().toBuffer().then((buf) => {
+            const pixels = Array.from(buf);
+            const mean = pixels.reduce((s, p) => s + p, 0) / pixels.length;
+            return pixels.reduce((h, p, i) => p >= mean ? h | 1n << BigInt(63 - i) : h, 0n);
+          })
+        ]);
+        let xor = hashA ^ hashB;
+        let dist = 0;
+        while (xor !== 0n) {
+          dist += Number(xor & 1n);
+          xor >>= 1n;
+        }
+        const pct = dist / 64 * 100;
+        if (pct >= 5) {
+          const severity = pct >= 20 ? "HIGH" : pct >= 10 ? "MEDIUM" : "LOW";
+          changes.push({
+            location: label,
+            severity,
+            description: `${Math.round(pct)}% visual change in ${label.replace("-", " ")} region`
+          });
+        }
+      }
+    }
+    return changes;
+  } catch {
+    return [];
+  }
+}
+export {
+  applyMask,
+  gridLayoutDiff,
+  screenshotDiffPercent
+};
+//# sourceMappingURL=phash.js.map

package/dist/phash.js.map

@@ -0,0 +1,7 @@
+{
+  "version": 3,
+  "sources": ["../src/phash.ts"],
+  "sourcesContent": ["import sharp from 'sharp';\nimport type { OverlayOptions } from 'sharp';\n\n/**\n * Composites neutral grey (128,128,128) over each ignoredRegion (fractional\n * 0-1 coordinates) so masked content \u2014 timestamps, live counters, ads \u2014\n * contributes nothing to any downstream pixel comparison. Shared by every\n * comparison path (whole-image hash, per-cell grid hash, AI vision) so a\n * region masked once is actually ignored everywhere, not just in the\n * cheapest check.\n */\nexport async function applyMask(\n  imagePath: string,\n  ignoredRegions: Array<{ x: number; y: number; width: number; height: number }>,\n): Promise<Buffer> {\n  const meta = await sharp(imagePath).metadata();\n  const W = meta.width ?? 1200;\n  const H = meta.height ?? 800;\n\n  const overlays: OverlayOptions[] = await Promise.all(\n    ignoredRegions.map(async (r) => {\n      const rW = Math.max(1, Math.round(r.width * W));\n      const rH = Math.max(1, Math.round(r.height * H));\n      const buf = await sharp({\n        create: { width: rW, height: rH, channels: 3, background: { r: 128, g: 128, b: 128 } },\n      }).png().toBuffer();\n      return { input: buf, top: Math.round(r.y * H), left: Math.round(r.x * W) } as OverlayOptions;\n    }),\n  );\n\n  return sharp(imagePath).composite(overlays).png().toBuffer();\n}\n\n/**\n * Compute an 8x8 average hash (aHash) of an image file, with optional mask overlays.\n * ignoredRegions are fractional (0-1 range), converted to pixels here.\n * Masked regions are filled with neutral grey (128,128,128) before hashing.\n */\nasync function hashWithMasks(\n  imagePath: string,\n  ignoredRegions: Array<{ x: number; y: number; width: number; height: number }> = [],\n): Promise<bigint> {\n  const pipeline = ignoredRegions.length > 0\n    ? sharp(await applyMask(imagePath, ignoredRegions))\n    : sharp(imagePath);\n\n  const buf = await pipeline.resize(8, 8, { fit: 'fill' }).greyscale().raw().toBuffer();\n  const pixels = Array.from(buf);\n  const mean = pixels.reduce((s, p) => s + p, 0) / pixels.length;\n  return pixels.reduce((h, p, i) => (p >= mean ? h | (1n << BigInt(63 - i)) : h), 0n);\n}\n\nfunction hammingDistance(a: bigint, b: bigint): number {\n  let xor = a ^ b;\n  let count = 0;\n  while (xor !== 0n) { count += Number(xor & 1n); xor >>= 1n; }\n  return count;\n}\n\nexport async function screenshotDiffPercent(\n  pathA: string,\n  pathB: string,\n  ignoredRegions: Array<{ x: number; y: number; width: number; height: number }> = [],\n): Promise<{ pHashDistance: number; changeLevel: 'IDENTICAL' | 'MINOR' | 'MODERATE' | 'SIGNIFICANT' }> {\n  const [hashA, hashB] = await Promise.all([\n    hashWithMasks(pathA, ignoredRegions),\n    hashWithMasks(pathB, ignoredRegions),\n  ]);\n  let xor = hashA ^ hashB;\n  let dist = 0;\n  while (xor !== 0n) { dist += Number(xor & 1n); xor >>= 1n; }\n  const pct = parseFloat(((dist / 64) * 100).toFixed(1));\n  const changeLevel = pct === 0 ? 'IDENTICAL' : pct < 10 ? 'MINOR' : pct < 30 ? 'MODERATE' : 'SIGNIFICANT';\n  return { pHashDistance: pct, changeLevel };\n}\n\n/**\n * Grid-based layout analysis: divides screenshots into a 3\u00D73 grid and computes\n * pHash per cell. Returns which regions changed and where, for the VisualDiff.changes field.\n * Uses sharp's `extract` to crop each cell before hashing.\n */\nexport async function gridLayoutDiff(\n  pathA: string,\n  pathB: string,\n  ignoredRegions: Array<{ x: number; y: number; width: number; height: number }> = [],\n): Promise<Array<{ location: string; severity: string; description: string }>> {\n  const GRID_LABELS = [\n    'top-left',    'top-center',    'top-right',\n    'mid-left',    'mid-center',    'mid-right',\n    'bottom-left', 'bottom-center', 'bottom-right',\n  ];\n\n  try {\n    // Mask before cropping into cells \u2014 a masked region (e.g. a timestamp)\n    // then contributes nothing to any cell's hash, instead of only being\n    // ignored by the separate whole-image hash in screenshotDiffPercent().\n    const [sourceA, sourceB] = ignoredRegions.length > 0\n      ? await Promise.all([applyMask(pathA, ignoredRegions), applyMask(pathB, ignoredRegions)])\n      : [pathA, pathB];\n\n    const [metaA, metaB] = await Promise.all([\n      sharp(sourceA).metadata(),\n      sharp(sourceB).metadata(),\n    ]);\n    const W = Math.min(metaA.width ?? 1200, metaB.width ?? 1200);\n    const H = Math.min(metaA.height ?? 800, metaB.height ?? 800);\n    const cW = Math.floor(W / 3);\n    const cH = Math.floor(H / 3);\n\n    const changes: Array<{ location: string; severity: string; description: string }> = [];\n\n    for (let row = 0; row < 3; row++) {\n      for (let col = 0; col < 3; col++) {\n        const region = { left: col * cW, top: row * cH, width: cW, height: cH };\n        const label = GRID_LABELS[row * 3 + col];\n\n        const [hashA, hashB] = await Promise.all([\n          sharp(sourceA).extract(region).resize(8, 8, { fit: 'fill' }).greyscale().raw().toBuffer()\n            .then((buf) => {\n              const pixels = Array.from(buf);\n              const mean = pixels.reduce((s, p) => s + p, 0) / pixels.length;\n              return pixels.reduce((h, p, i) => p >= mean ? h | (1n << BigInt(63 - i)) : h, 0n);\n            }),\n          sharp(sourceB).extract(region).resize(8, 8, { fit: 'fill' }).greyscale().raw().toBuffer()\n            .then((buf) => {\n              const pixels = Array.from(buf);\n              const mean = pixels.reduce((s, p) => s + p, 0) / pixels.length;\n              return pixels.reduce((h, p, i) => p >= mean ? h | (1n << BigInt(63 - i)) : h, 0n);\n            }),\n        ]);\n\n        let xor = hashA ^ hashB;\n        let dist = 0;\n        while (xor !== 0n) { dist += Number(xor & 1n); xor >>= 1n; }\n        const pct = (dist / 64) * 100;\n\n        if (pct >= 5) {\n          const severity = pct >= 20 ? 'HIGH' : pct >= 10 ? 'MEDIUM' : 'LOW';\n          changes.push({\n            location: label,\n            severity,\n            description: `${Math.round(pct)}% visual change in ${label.replace('-', ' ')} region`,\n          });\n        }\n      }\n    }\n    return changes;\n  } catch {\n    return [];\n  }\n}\n"],
+  "mappings": "AAAA,OAAO,WAAW;AAWlB,eAAsB,UACpB,WACA,gBACiB;AACjB,QAAM,OAAO,MAAM,MAAM,SAAS,EAAE,SAAS;AAC7C,QAAM,IAAI,KAAK,SAAS;AACxB,QAAM,IAAI,KAAK,UAAU;AAEzB,QAAM,WAA6B,MAAM,QAAQ;AAAA,IAC/C,eAAe,IAAI,OAAO,MAAM;AAC9B,YAAM,KAAK,KAAK,IAAI,GAAG,KAAK,MAAM,EAAE,QAAQ,CAAC,CAAC;AAC9C,YAAM,KAAK,KAAK,IAAI,GAAG,KAAK,MAAM,EAAE,SAAS,CAAC,CAAC;AAC/C,YAAM,MAAM,MAAM,MAAM;AAAA,QACtB,QAAQ,EAAE,OAAO,IAAI,QAAQ,IAAI,UAAU,GAAG,YAAY,EAAE,GAAG,KAAK,GAAG,KAAK,GAAG,IAAI,EAAE;AAAA,MACvF,CAAC,EAAE,IAAI,EAAE,SAAS;AAClB,aAAO,EAAE,OAAO,KAAK,KAAK,KAAK,MAAM,EAAE,IAAI,CAAC,GAAG,MAAM,KAAK,MAAM,EAAE,IAAI,CAAC,EAAE;AAAA,IAC3E,CAAC;AAAA,EACH;AAEA,SAAO,MAAM,SAAS,EAAE,UAAU,QAAQ,EAAE,IAAI,EAAE,SAAS;AAC7D;AAOA,eAAe,cACb,WACA,iBAAiF,CAAC,GACjE;AACjB,QAAM,WAAW,eAAe,SAAS,IACrC,MAAM,MAAM,UAAU,WAAW,cAAc,CAAC,IAChD,MAAM,SAAS;AAEnB,QAAM,MAAM,MAAM,SAAS,OAAO,GAAG,GAAG,EAAE,KAAK,OAAO,CAAC,EAAE,UAAU,EAAE,IAAI,EAAE,SAAS;AACpF,QAAM,SAAS,MAAM,KAAK,GAAG;AAC7B,QAAM,OAAO,OAAO,OAAO,CAAC,GAAG,MAAM,IAAI,GAAG,CAAC,IAAI,OAAO;AACxD,SAAO,OAAO,OAAO,CAAC,GAAG,GAAG,MAAO,KAAK,OAAO,IAAK,MAAM,OAAO,KAAK,CAAC,IAAK,GAAI,EAAE;AACpF;AAEA,SAAS,gBAAgB,GAAW,GAAmB;AACrD,MAAI,MAAM,IAAI;AACd,MAAI,QAAQ;AACZ,SAAO,QAAQ,IAAI;AAAE,aAAS,OAAO,MAAM,EAAE;AAAG,YAAQ;AAAA,EAAI;AAC5D,SAAO;AACT;AAEA,eAAsB,sBACpB,OACA,OACA,iBAAiF,CAAC,GACmB;AACrG,QAAM,CAAC,OAAO,KAAK,IAAI,MAAM,QAAQ,IAAI;AAAA,IACvC,cAAc,OAAO,cAAc;AAAA,IACnC,cAAc,OAAO,cAAc;AAAA,EACrC,CAAC;AACD,MAAI,MAAM,QAAQ;AAClB,MAAI,OAAO;AACX,SAAO,QAAQ,IAAI;AAAE,YAAQ,OAAO,MAAM,EAAE;AAAG,YAAQ;AAAA,EAAI;AAC3D,QAAM,MAAM,YAAa,OAAO,KAAM,KAAK,QAAQ,CAAC,CAAC;AACrD,QAAM,cAAc,QAAQ,IAAI,cAAc,MAAM,KAAK,UAAU,MAAM,KAAK,aAAa;AAC3F,SAAO,EAAE,eAAe,KAAK,YAAY;AAC3C;AAOA,eAAsB,eACpB,OACA,OACA,iBAAiF,CAAC,GACL;AAC7E,QAAM,cAAc;AAAA,IAClB;AAAA,IAAe;AAAA,IAAiB;AAAA,IAChC;AAAA,IAAe;AAAA,IAAiB;AAAA,IAChC;AAAA,IAAe;AAAA,IAAiB;AAAA,EAClC;AAEA,MAAI;AAIF,UAAM,CAAC,SAAS,OAAO,IAAI,eAAe,SAAS,IAC/C,MAAM,QAAQ,IAAI,CAAC,UAAU,OAAO,cAAc,GAAG,UAAU,OAAO,cAAc,CAAC,CAAC,IACtF,CAAC,OAAO,KAAK;AAEjB,UAAM,CAAC,OAAO,KAAK,IAAI,MAAM,QAAQ,IAAI;AAAA,MACvC,MAAM,OAAO,EAAE,SAAS;AAAA,MACxB,MAAM,OAAO,EAAE,SAAS;AAAA,IAC1B,CAAC;AACD,UAAM,IAAI,KAAK,IAAI,MAAM,SAAS,MAAM,MAAM,SAAS,IAAI;AAC3D,UAAM,IAAI,KAAK,IAAI,MAAM,UAAU,KAAK,MAAM,UAAU,GAAG;AAC3D,UAAM,KAAK,KAAK,MAAM,IAAI,CAAC;AAC3B,UAAM,KAAK,KAAK,MAAM,IAAI,CAAC;AAE3B,UAAM,UAA8E,CAAC;AAErF,aAAS,MAAM,GAAG,MAAM,GAAG,OAAO;AAChC,eAAS,MAAM,GAAG,MAAM,GAAG,OAAO;AAChC,cAAM,SAAS,EAAE,MAAM,MAAM,IAAI,KAAK,MAAM,IAAI,OAAO,IAAI,QAAQ,GAAG;AACtE,cAAM,QAAQ,YAAY,MAAM,IAAI,GAAG;AAEvC,cAAM,CAAC,OAAO,KAAK,IAAI,MAAM,QAAQ,IAAI;AAAA,UACvC,MAAM,OAAO,EAAE,QAAQ,MAAM,EAAE,OAAO,GAAG,GAAG,EAAE,KAAK,OAAO,CAAC,EAAE,UAAU,EAAE,IAAI,EAAE,SAAS,EACrF,KAAK,CAAC,QAAQ;AACb,kBAAM,SAAS,MAAM,KAAK,GAAG;AAC7B,kBAAM,OAAO,OAAO,OAAO,CAAC,GAAG,MAAM,IAAI,GAAG,CAAC,IAAI,OAAO;AACxD,mBAAO,OAAO,OAAO,CAAC,GAAG,GAAG,MAAM,KAAK,OAAO,IAAK,MAAM,OAAO,KAAK,CAAC,IAAK,GAAG,EAAE;AAAA,UAClF,CAAC;AAAA,UACH,MAAM,OAAO,EAAE,QAAQ,MAAM,EAAE,OAAO,GAAG,GAAG,EAAE,KAAK,OAAO,CAAC,EAAE,UAAU,EAAE,IAAI,EAAE,SAAS,EACrF,KAAK,CAAC,QAAQ;AACb,kBAAM,SAAS,MAAM,KAAK,GAAG;AAC7B,kBAAM,OAAO,OAAO,OAAO,CAAC,GAAG,MAAM,IAAI,GAAG,CAAC,IAAI,OAAO;AACxD,mBAAO,OAAO,OAAO,CAAC,GAAG,GAAG,MAAM,KAAK,OAAO,IAAK,MAAM,OAAO,KAAK,CAAC,IAAK,GAAG,EAAE;AAAA,UAClF,CAAC;AAAA,QACL,CAAC;AAED,YAAI,MAAM,QAAQ;AAClB,YAAI,OAAO;AACX,eAAO,QAAQ,IAAI;AAAE,kBAAQ,OAAO,MAAM,EAAE;AAAG,kBAAQ;AAAA,QAAI;AAC3D,cAAM,MAAO,OAAO,KAAM;AAE1B,YAAI,OAAO,GAAG;AACZ,gBAAM,WAAW,OAAO,KAAK,SAAS,OAAO,KAAK,WAAW;AAC7D,kBAAQ,KAAK;AAAA,YACX,UAAU;AAAA,YACV;AAAA,YACA,aAAa,GAAG,KAAK,MAAM,GAAG,CAAC,sBAAsB,MAAM,QAAQ,KAAK,GAAG,CAAC;AAAA,UAC9E,CAAC;AAAA,QACH;AAAA,MACF;AAAA,IACF;AACA,WAAO;AAAA,EACT,QAAQ;AACN,WAAO,CAAC;AAAA,EACV;AACF;",
+  "names": []
+}

package/dist/pii-scrubber.d.ts

@@ -0,0 +1,21 @@
+/**
+ * PII scrubber — detects and redacts Personally Identifiable Information
+ * from crawled markdown, HTML, and extracted text before storage.
+ *
+ * GDPR Article 4(1): "any information relating to an identified or identifiable natural person"
+ */
+export interface PiiMatch {
+    type: 'email' | 'phone' | 'ssn' | 'credit-card' | 'ip-address' | 'passport' | 'iban' | 'dob';
+    value: string;
+    start: number;
+    end: number;
+}
+export interface ScrubResult {
+    text: string;
+    matches: PiiMatch[];
+    scrubbed: boolean;
+}
+export declare function detectPii(text: string): PiiMatch[];
+export declare function scrubPii(text: string): ScrubResult;
+export declare function hasPii(text: string): boolean;
+export declare function scrubObject<T extends Record<string, unknown>>(obj: T, fields: (keyof T)[]): T;

package/dist/pii-scrubber.js

@@ -0,0 +1,75 @@
+const PATTERNS = [
+  // Email — RFC 5321 simplified
+  { type: "email", re: /\b[a-zA-Z0-9._%+\-]+@[a-zA-Z0-9.\-]+\.[a-zA-Z]{2,}\b/g },
+  // US SSN — 9 digits with dashes or spaces
+  { type: "ssn", re: /\b(?!000|666|9\d\d)\d{3}[-\s](?!00)\d{2}[-\s](?!0000)\d{4}\b/g },
+  // Credit card — Visa/MC/Amex/Discover with optional spaces/dashes
+  { type: "credit-card", re: /\b(?:4\d{3}|5[1-5]\d{2}|3[47]\d{2}|6(?:011|5\d{2}))[-\s]?\d{4}[-\s]?\d{4}[-\s]?\d{3,4}\b/g },
+  // IBAN — international bank account (15-34 chars)
+  { type: "iban", re: /\b[A-Z]{2}\d{2}[A-Z0-9]{1,30}\b/g },
+  // Phone — international or US formats
+  { type: "phone", re: /(?:\+?1[-.\s]?)?\(?\d{3}\)?[-.\s]\d{3}[-.\s]\d{4}\b|\+\d{1,3}[-.\s]\d{2,4}[-.\s]\d{3,4}[-.\s]\d{3,4}/g },
+  // Date of birth — common formats
+  { type: "dob", re: /\b(?:0?[1-9]|1[0-2])\/(?:0?[1-9]|[12]\d|3[01])\/(?:19|20)\d{2}\b/g },
+  // IPv4 private/internal addresses — only flag private-looking ones in content
+  { type: "ip-address", re: /\b(?:192\.168\.\d{1,3}\.\d{1,3}|10\.\d{1,3}\.\d{1,3}\.\d{1,3}|172\.(?:1[6-9]|2\d|3[01])\.\d{1,3}\.\d{1,3})\b/g }
+];
+const REDACT_MAP = {
+  email: "[EMAIL]",
+  phone: "[PHONE]",
+  ssn: "[SSN]",
+  "credit-card": "[CARD]",
+  "ip-address": "[IP]",
+  passport: "[PASSPORT]",
+  iban: "[IBAN]",
+  dob: "[DOB]"
+};
+function detectPii(text) {
+  const matches = [];
+  const seen = /* @__PURE__ */ new Set();
+  for (const { type, re } of PATTERNS) {
+    const r = new RegExp(re.source, re.flags.includes("g") ? re.flags : re.flags + "g");
+    let m;
+    while ((m = r.exec(text)) !== null) {
+      const key = `${type}:${m.index}`;
+      if (!seen.has(key)) {
+        seen.add(key);
+        matches.push({ type, value: m[0], start: m.index, end: m.index + m[0].length });
+      }
+    }
+  }
+  return matches.sort((a, b) => a.start - b.start);
+}
+function scrubPii(text) {
+  const matches = detectPii(text);
+  if (matches.length === 0) return { text, matches: [], scrubbed: false };
+  let scrubbed = text;
+  for (let i = matches.length - 1; i >= 0; i--) {
+    const { type, start, end } = matches[i];
+    scrubbed = scrubbed.slice(0, start) + REDACT_MAP[type] + scrubbed.slice(end);
+  }
+  return { text: scrubbed, matches, scrubbed: true };
+}
+function hasPii(text) {
+  for (const { re } of PATTERNS) {
+    const r = new RegExp(re.source, re.flags);
+    if (r.test(text)) return true;
+  }
+  return false;
+}
+function scrubObject(obj, fields) {
+  const result = { ...obj };
+  for (const field of fields) {
+    if (typeof result[field] === "string") {
+      result[field] = scrubPii(result[field]).text;
+    }
+  }
+  return result;
+}
+export {
+  detectPii,
+  hasPii,
+  scrubObject,
+  scrubPii
+};
+//# sourceMappingURL=pii-scrubber.js.map

package/dist/pii-scrubber.js.map

@@ -0,0 +1,7 @@
+{
+  "version": 3,
+  "sources": ["../src/pii-scrubber.ts"],
+  "sourcesContent": ["/**\n * PII scrubber \u2014 detects and redacts Personally Identifiable Information\n * from crawled markdown, HTML, and extracted text before storage.\n *\n * GDPR Article 4(1): \"any information relating to an identified or identifiable natural person\"\n */\n\nexport interface PiiMatch {\n  type: 'email' | 'phone' | 'ssn' | 'credit-card' | 'ip-address' | 'passport' | 'iban' | 'dob';\n  value: string;\n  start: number;\n  end: number;\n}\n\nexport interface ScrubResult {\n  text: string;\n  matches: PiiMatch[];\n  scrubbed: boolean;\n}\n\n// Patterns \u2014 ordered from most-specific to least-specific\nconst PATTERNS: Array<{ type: PiiMatch['type']; re: RegExp }> = [\n  // Email \u2014 RFC 5321 simplified\n  { type: 'email', re: /\\b[a-zA-Z0-9._%+\\-]+@[a-zA-Z0-9.\\-]+\\.[a-zA-Z]{2,}\\b/g },\n  // US SSN \u2014 9 digits with dashes or spaces\n  { type: 'ssn', re: /\\b(?!000|666|9\\d\\d)\\d{3}[-\\s](?!00)\\d{2}[-\\s](?!0000)\\d{4}\\b/g },\n  // Credit card \u2014 Visa/MC/Amex/Discover with optional spaces/dashes\n  { type: 'credit-card', re: /\\b(?:4\\d{3}|5[1-5]\\d{2}|3[47]\\d{2}|6(?:011|5\\d{2}))[-\\s]?\\d{4}[-\\s]?\\d{4}[-\\s]?\\d{3,4}\\b/g },\n  // IBAN \u2014 international bank account (15-34 chars)\n  { type: 'iban', re: /\\b[A-Z]{2}\\d{2}[A-Z0-9]{1,30}\\b/g },\n  // Phone \u2014 international or US formats\n  { type: 'phone', re: /(?:\\+?1[-.\\s]?)?\\(?\\d{3}\\)?[-.\\s]\\d{3}[-.\\s]\\d{4}\\b|\\+\\d{1,3}[-.\\s]\\d{2,4}[-.\\s]\\d{3,4}[-.\\s]\\d{3,4}/g },\n  // Date of birth \u2014 common formats\n  { type: 'dob', re: /\\b(?:0?[1-9]|1[0-2])\\/(?:0?[1-9]|[12]\\d|3[01])\\/(?:19|20)\\d{2}\\b/g },\n  // IPv4 private/internal addresses \u2014 only flag private-looking ones in content\n  { type: 'ip-address', re: /\\b(?:192\\.168\\.\\d{1,3}\\.\\d{1,3}|10\\.\\d{1,3}\\.\\d{1,3}\\.\\d{1,3}|172\\.(?:1[6-9]|2\\d|3[01])\\.\\d{1,3}\\.\\d{1,3})\\b/g },\n];\n\nconst REDACT_MAP: Record<PiiMatch['type'], string> = {\n  email: '[EMAIL]',\n  phone: '[PHONE]',\n  ssn: '[SSN]',\n  'credit-card': '[CARD]',\n  'ip-address': '[IP]',\n  passport: '[PASSPORT]',\n  iban: '[IBAN]',\n  dob: '[DOB]',\n};\n\nexport function detectPii(text: string): PiiMatch[] {\n  const matches: PiiMatch[] = [];\n  const seen = new Set<string>();\n\n  for (const { type, re } of PATTERNS) {\n    const r = new RegExp(re.source, re.flags.includes('g') ? re.flags : re.flags + 'g');\n    let m: RegExpExecArray | null;\n    while ((m = r.exec(text)) !== null) {\n      const key = `${type}:${m.index}`;\n      if (!seen.has(key)) {\n        seen.add(key);\n        matches.push({ type, value: m[0], start: m.index, end: m.index + m[0].length });\n      }\n    }\n  }\n\n  return matches.sort((a, b) => a.start - b.start);\n}\n\nexport function scrubPii(text: string): ScrubResult {\n  const matches = detectPii(text);\n  if (matches.length === 0) return { text, matches: [], scrubbed: false };\n\n  // Replace from end to start so indices stay valid\n  let scrubbed = text;\n  for (let i = matches.length - 1; i >= 0; i--) {\n    const { type, start, end } = matches[i];\n    scrubbed = scrubbed.slice(0, start) + REDACT_MAP[type] + scrubbed.slice(end);\n  }\n\n  return { text: scrubbed, matches, scrubbed: true };\n}\n\nexport function hasPii(text: string): boolean {\n  for (const { re } of PATTERNS) {\n    const r = new RegExp(re.source, re.flags);\n    if (r.test(text)) return true;\n  }\n  return false;\n}\n\nexport function scrubObject<T extends Record<string, unknown>>(obj: T, fields: (keyof T)[]): T {\n  const result = { ...obj };\n  for (const field of fields) {\n    if (typeof result[field] === 'string') {\n      (result as any)[field] = scrubPii(result[field] as string).text;\n    }\n  }\n  return result;\n}\n"],
+  "mappings": "AAqBA,MAAM,WAA0D;AAAA;AAAA,EAE9D,EAAE,MAAM,SAAS,IAAI,wDAAwD;AAAA;AAAA,EAE7E,EAAE,MAAM,OAAO,IAAI,gEAAgE;AAAA;AAAA,EAEnF,EAAE,MAAM,eAAe,IAAI,4FAA4F;AAAA;AAAA,EAEvH,EAAE,MAAM,QAAQ,IAAI,mCAAmC;AAAA;AAAA,EAEvD,EAAE,MAAM,SAAS,IAAI,wGAAwG;AAAA;AAAA,EAE7H,EAAE,MAAM,OAAO,IAAI,oEAAoE;AAAA;AAAA,EAEvF,EAAE,MAAM,cAAc,IAAI,gHAAgH;AAC5I;AAEA,MAAM,aAA+C;AAAA,EACnD,OAAO;AAAA,EACP,OAAO;AAAA,EACP,KAAK;AAAA,EACL,eAAe;AAAA,EACf,cAAc;AAAA,EACd,UAAU;AAAA,EACV,MAAM;AAAA,EACN,KAAK;AACP;AAEO,SAAS,UAAU,MAA0B;AAClD,QAAM,UAAsB,CAAC;AAC7B,QAAM,OAAO,oBAAI,IAAY;AAE7B,aAAW,EAAE,MAAM,GAAG,KAAK,UAAU;AACnC,UAAM,IAAI,IAAI,OAAO,GAAG,QAAQ,GAAG,MAAM,SAAS,GAAG,IAAI,GAAG,QAAQ,GAAG,QAAQ,GAAG;AAClF,QAAI;AACJ,YAAQ,IAAI,EAAE,KAAK,IAAI,OAAO,MAAM;AAClC,YAAM,MAAM,GAAG,IAAI,IAAI,EAAE,KAAK;AAC9B,UAAI,CAAC,KAAK,IAAI,GAAG,GAAG;AAClB,aAAK,IAAI,GAAG;AACZ,gBAAQ,KAAK,EAAE,MAAM,OAAO,EAAE,CAAC,GAAG,OAAO,EAAE,OAAO,KAAK,EAAE,QAAQ,EAAE,CAAC,EAAE,OAAO,CAAC;AAAA,MAChF;AAAA,IACF;AAAA,EACF;AAEA,SAAO,QAAQ,KAAK,CAAC,GAAG,MAAM,EAAE,QAAQ,EAAE,KAAK;AACjD;AAEO,SAAS,SAAS,MAA2B;AAClD,QAAM,UAAU,UAAU,IAAI;AAC9B,MAAI,QAAQ,WAAW,EAAG,QAAO,EAAE,MAAM,SAAS,CAAC,GAAG,UAAU,MAAM;AAGtE,MAAI,WAAW;AACf,WAAS,IAAI,QAAQ,SAAS,GAAG,KAAK,GAAG,KAAK;AAC5C,UAAM,EAAE,MAAM,OAAO,IAAI,IAAI,QAAQ,CAAC;AACtC,eAAW,SAAS,MAAM,GAAG,KAAK,IAAI,WAAW,IAAI,IAAI,SAAS,MAAM,GAAG;AAAA,EAC7E;AAEA,SAAO,EAAE,MAAM,UAAU,SAAS,UAAU,KAAK;AACnD;AAEO,SAAS,OAAO,MAAuB;AAC5C,aAAW,EAAE,GAAG,KAAK,UAAU;AAC7B,UAAM,IAAI,IAAI,OAAO,GAAG,QAAQ,GAAG,KAAK;AACxC,QAAI,EAAE,KAAK,IAAI,EAAG,QAAO;AAAA,EAC3B;AACA,SAAO;AACT;AAEO,SAAS,YAA+C,KAAQ,QAAwB;AAC7F,QAAM,SAAS,EAAE,GAAG,IAAI;AACxB,aAAW,SAAS,QAAQ;AAC1B,QAAI,OAAO,OAAO,KAAK,MAAM,UAAU;AACrC,MAAC,OAAe,KAAK,IAAI,SAAS,OAAO,KAAK,CAAW,EAAE;AAAA,IAC7D;AAAA,EACF;AACA,SAAO;AACT;",
+  "names": []
+}

package/dist/playwright-recorder.d.ts

@@ -0,0 +1,39 @@
+/**
+ * Playwright test recorder — generate Playwright test code from crawl data.
+ *
+ * Given a list of crawled screens, generates a minimal Playwright test file that:
+ * 1. Navigates to each screen URL
+ * 2. Verifies the page title
+ * 3. Checks key elements visible on the page
+ * 4. Verifies no console errors (optional)
+ *
+ * This is a "smoke test" generator — not a full interaction recorder.
+ * Output: valid TypeScript Playwright test file ready to run with npx playwright test.
+ */
+export interface RecorderScreen {
+    id: string;
+    name: string;
+    url: string;
+    title?: string;
+    heading?: string;
+    markdownWordCount?: number;
+    smartLocator?: {
+        tag: string;
+        text?: string;
+        role?: string;
+        placeholder?: string;
+        nearText?: string;
+    };
+}
+export interface RecorderOpts {
+    baseUrl?: string;
+    includeConsoleCheck?: boolean;
+    testTimeout?: number;
+    projectName?: string;
+    useSmartLocators?: boolean;
+}
+export declare function generatePlaywrightTest(screens: RecorderScreen[], opts?: RecorderOpts): string;
+export declare function generatePlaywrightConfig(opts?: {
+    baseUrl?: string;
+    reportDir?: string;
+}): string;