@taodigital/crawler 0.0.0-stage → 1.2.0

package/Dockerfile

@@ -0,0 +1,82 @@
+# ZeTa-AI Crawler — includes Playwright + Chromium
+# Playwright needs a real Linux environment with system deps.
+# This service CANNOT run on Vercel serverless — use Railway/Render.
+# IMPORTANT: ECS Fargate task definition must set sharedMemorySize: 2048 (MB) for Chromium stability
+FROM mcr.microsoft.com/playwright:v1.55.0-jammy AS runner
+WORKDIR /app
+ENV NODE_ENV=production
+# Install browsers to a fixed path — base image ships browsers here and
+# we override after npm ci to match the exact installed playwright revision.
+ENV PLAYWRIGHT_BROWSERS_PATH=/ms-playwright
+
+# Install Node.js 20 (Playwright base image ships with 18)
+RUN apt-get update && apt-get install -y curl build-essential python3 && \
+    curl -fsSL https://deb.nodesource.com/setup_20.x | bash - && \
+    apt-get install -y nodejs && \
+    apt-get clean && rm -rf /var/lib/apt/lists/*
+
+# Workspace manifests (layer cache)
+# Copy all workspace package.json files so npm ci can resolve the full lockfile.
+# Packages not used by the crawler (api, web, etc.) are stubs; their source is not copied later.
+COPY package.json package-lock.json ./
+COPY packages/agent-cli/package.json    packages/agent-cli/
+COPY packages/agents/package.json       packages/agents/
+COPY packages/api/package.json          packages/api/
+COPY packages/app-brain/package.json    packages/app-brain/
+COPY packages/chrome-extension/package.json packages/chrome-extension/
+COPY packages/connectors/package.json   packages/connectors/
+COPY packages/core/package.json         packages/core/
+COPY packages/crawler/package.json      packages/crawler/
+COPY packages/database/package.json     packages/database/
+COPY packages/github-action/package.json packages/github-action/
+COPY packages/reranker/package.json     packages/reranker/
+COPY packages/sdk/package.json          packages/sdk/
+COPY packages/session-receiver/package.json packages/session-receiver/
+COPY packages/vscode-extension/package.json packages/vscode-extension/
+COPY packages/web/package.json          packages/web/
+RUN npm ci --include=dev
+
+# Patch Stagehand's act() schema to add `additionalProperties: false` on the
+# inner anyOf[0] object. OpenAI strict mode (v1/responses) requires it on every
+# nested object; Stagehand ≤3.6 omits it, causing a 400 "invalid_json_schema".
+RUN node -e "const fs=require('fs');['packages/crawler/node_modules/@browserbasehq/stagehand/dist/cjs/lib/inference.js','packages/crawler/node_modules/@browserbasehq/stagehand/dist/esm/lib/inference.js'].forEach(f=>{if(!fs.existsSync(f)){console.log('skip',f);return;}let c=fs.readFileSync(f,'utf8');if(c.includes('.strict()')){console.log('already patched',f);return;}c=c.replace(/\\.nullable\\(\\)\\n(\\s+)\\.describe\\(\"The element to act on/g,'\.strict()\n\$1.nullable()\n\$1.describe(\"The element to act on');fs.writeFileSync(f,c);console.log('patched',f);});"
+
+# Re-install chromium after npm ci so the browser revision matches the
+# exact playwright version resolved by the lockfile (not the base image's
+# bundled version, which may differ if the lockfile was updated).
+RUN npx playwright install chromium
+
+# Source
+COPY tsconfig.base.json ./
+COPY packages/database        packages/database
+COPY packages/reranker        packages/reranker
+COPY packages/core            packages/core
+COPY packages/connectors      packages/connectors
+COPY packages/app-brain       packages/app-brain
+COPY packages/agents          packages/agents
+COPY packages/crawler         packages/crawler
+
+# Build order: database → reranker → core → connectors → app-brain → agents → crawler
+RUN npx prisma generate --schema packages/database/prisma/schema.prisma && \
+    npm run -w @detiq/database build && \
+    npm run -w @detiq/reranker build && \
+    npm run -w @detiq/core build && \
+    npm run -w @detiq/connectors build && \
+    npm run -w @detiq/app-brain build && \
+    npm run -w @detiq/agents build && \
+    npm run -w @detiq/crawler build
+
+# Screenshot cache dir
+RUN mkdir -p /app/.screenshots-cache
+
+RUN addgroup --system appgroup && adduser --system --ingroup appgroup appuser
+RUN chown -R appuser:appgroup /app
+USER appuser
+
+EXPOSE 4001
+# index.js is a side-effect-free library barrel (also imported by @detiq/api) — it holds
+# no listener/timer, so running it directly starts the process, finishes its imports,
+# and exits immediately with nothing ever consuming the job queue. worker.js is the
+# actual long-running entrypoint: it starts the durable job-poll loop (and the SQS
+# consumer when JOB_QUEUE_BACKEND=sqs) and keeps the process alive.
+CMD ["node", "packages/crawler/dist/worker.js"]

package/README.md

@@ -1,3 +1,11 @@
-# Temporary Holding Version
+# @detiq/crawler — The Explorer
 
-This version is a temporary placeholder for this package. An operational version to replace this has been submitted for review and is awaiting a staged release.
+Crawls an app, **gets past the login** (the star of Phase 1), discovers
+screens, captures screenshots, and reads each page's elements into the App
+Brain.
+
+- `auth.ts` — login handling. **Invest here**; it's the competitive edge.
+- `screen-reader.ts` — reads pages as a cheap structured outline (accessibility
+  tree), not pixels. Vision is a later fallback.
+- `crawler.ts` — the discovery loop. Swap in an AI navigation library
+  (Stagehand / Browser Use) to explore more intelligently.

package/dist/accessibility-auditor.d.ts

@@ -0,0 +1,29 @@
+/**
+ * Accessibility auditor — lightweight a11y checks using Playwright's aria snapshot
+ * and manual DOM queries. No axe-core dependency needed.
+ *
+ * Checks:
+ * 1. Images without alt text
+ * 2. Form inputs without associated labels
+ * 3. Buttons/links with no accessible name
+ * 4. Missing lang attribute on html element
+ * 5. Missing page title
+ * 6. Empty headings
+ * 7. Color contrast (limited — detects text with inline style color only)
+ * 8. Keyboard-unfocusable interactive elements (tabIndex=-1 on buttons)
+ */
+import type { Page } from 'playwright';
+export interface A11yViolation {
+    rule: string;
+    severity: 'critical' | 'serious' | 'moderate' | 'minor';
+    count: number;
+    examples: string[];
+}
+export interface A11yAuditResult {
+    url: string;
+    violations: A11yViolation[];
+    passCount: number;
+    score: number;
+    auditedAt: string;
+}
+export declare function auditAccessibility(page: Page, url: string): Promise<A11yAuditResult>;

package/dist/accessibility-auditor.js

@@ -0,0 +1,57 @@
+async function queryAll(page, selector) {
+  try {
+    return await page.$$eval(selector, (els) => els.map((el) => ({
+      outerHTML: el.outerHTML?.slice(0, 200) ?? "",
+      textContent: (el.textContent ?? "").trim().slice(0, 100),
+      tabIndex: el.tabIndex,
+      getAttribute: (a) => el.getAttribute(a)
+    })));
+  } catch {
+    return [];
+  }
+}
+async function auditAccessibility(page, url) {
+  const violations = [];
+  let passCount = 0;
+  const auditedAt = (/* @__PURE__ */ new Date()).toISOString();
+  const imgs = await page.$$eval("img:not([alt])", (els) => els.map((e) => e.outerHTML?.slice(0, 120) ?? ""));
+  if (imgs.length > 0) violations.push({ rule: "image-alt", severity: "critical", count: imgs.length, examples: imgs.slice(0, 3) });
+  else passCount++;
+  const unlabeledInputs = await page.$$eval(
+    'input:not([type="hidden"]):not([type="submit"]):not([type="button"]):not([aria-label]):not([aria-labelledby]):not([id])',
+    (els) => els.map((e) => e.outerHTML?.slice(0, 120) ?? "")
+  );
+  if (unlabeledInputs.length > 0) violations.push({ rule: "label", severity: "serious", count: unlabeledInputs.length, examples: unlabeledInputs.slice(0, 3) });
+  else passCount++;
+  const emptyButtons = await page.$$eval(
+    "button:not([aria-label]):not([aria-labelledby])",
+    (els) => els.filter((e) => !(e.textContent ?? "").trim()).map((e) => e.outerHTML?.slice(0, 120) ?? "")
+  );
+  if (emptyButtons.length > 0) violations.push({ rule: "button-name", severity: "critical", count: emptyButtons.length, examples: emptyButtons.slice(0, 3) });
+  else passCount++;
+  const htmlLang = await page.$eval("html", (el) => el.getAttribute("lang") ?? "").catch(() => "");
+  if (!htmlLang) violations.push({ rule: "html-has-lang", severity: "serious", count: 1, examples: ["<html> missing lang attribute"] });
+  else passCount++;
+  const title = await page.title().catch(() => "");
+  if (!title) violations.push({ rule: "document-title", severity: "serious", count: 1, examples: ["Page has no <title>"] });
+  else passCount++;
+  const emptyHeadings = await page.$$eval(
+    "h1,h2,h3,h4,h5,h6",
+    (els) => els.filter((e) => !(e.textContent ?? "").trim()).map((e) => e.tagName.toLowerCase())
+  );
+  if (emptyHeadings.length > 0) violations.push({ rule: "empty-heading", severity: "moderate", count: emptyHeadings.length, examples: emptyHeadings.slice(0, 3) });
+  else passCount++;
+  const emptyLinks = await page.$$eval(
+    "a[href]:not([aria-label]):not([aria-labelledby])",
+    (els) => els.filter((e) => !(e.textContent ?? "").trim() && !e.querySelector("img[alt]")).map((e) => e.outerHTML?.slice(0, 120) ?? "")
+  );
+  if (emptyLinks.length > 0) violations.push({ rule: "link-name", severity: "serious", count: emptyLinks.length, examples: emptyLinks.slice(0, 3) });
+  else passCount++;
+  const totalChecks = passCount + violations.length;
+  const score = totalChecks > 0 ? Math.round(passCount / totalChecks * 100) : 100;
+  return { url, violations, passCount, score, auditedAt };
+}
+export {
+  auditAccessibility
+};
+//# sourceMappingURL=accessibility-auditor.js.map

package/dist/accessibility-auditor.js.map

@@ -0,0 +1,7 @@
+{
+  "version": 3,
+  "sources": ["../src/accessibility-auditor.ts"],
+  "sourcesContent": ["/**\n * Accessibility auditor \u2014 lightweight a11y checks using Playwright's aria snapshot\n * and manual DOM queries. No axe-core dependency needed.\n *\n * Checks:\n * 1. Images without alt text\n * 2. Form inputs without associated labels\n * 3. Buttons/links with no accessible name\n * 4. Missing lang attribute on html element\n * 5. Missing page title\n * 6. Empty headings\n * 7. Color contrast (limited \u2014 detects text with inline style color only)\n * 8. Keyboard-unfocusable interactive elements (tabIndex=-1 on buttons)\n */\n\nimport type { Page } from 'playwright';\n\nexport interface A11yViolation {\n  rule: string;\n  severity: 'critical' | 'serious' | 'moderate' | 'minor';\n  count: number;\n  examples: string[];\n}\n\nexport interface A11yAuditResult {\n  url: string;\n  violations: A11yViolation[];\n  passCount: number;\n  score: number; // 0-100\n  auditedAt: string;\n}\n\nasync function queryAll(page: Page, selector: string): Promise<any[]> {\n  try {\n    return await page.$$eval(selector, els => els.map((el: any) => ({\n      outerHTML: el.outerHTML?.slice(0, 200) ?? '',\n      textContent: (el.textContent ?? '').trim().slice(0, 100),\n      tabIndex: el.tabIndex,\n      getAttribute: (a: string) => el.getAttribute(a),\n    })));\n  } catch {\n    return [];\n  }\n}\n\nexport async function auditAccessibility(page: Page, url: string): Promise<A11yAuditResult> {\n  const violations: A11yViolation[] = [];\n  let passCount = 0;\n  const auditedAt = new Date().toISOString();\n\n  // 1. Images without alt\n  const imgs = await page.$$eval('img:not([alt])', els => els.map((e: any) => e.outerHTML?.slice(0, 120) ?? ''));\n  if (imgs.length > 0) violations.push({ rule: 'image-alt', severity: 'critical', count: imgs.length, examples: imgs.slice(0, 3) });\n  else passCount++;\n\n  // 2. Inputs without labels\n  const unlabeledInputs = await page.$$eval(\n    'input:not([type=\"hidden\"]):not([type=\"submit\"]):not([type=\"button\"]):not([aria-label]):not([aria-labelledby]):not([id])',\n    els => els.map((e: any) => e.outerHTML?.slice(0, 120) ?? ''),\n  );\n  if (unlabeledInputs.length > 0) violations.push({ rule: 'label', severity: 'serious', count: unlabeledInputs.length, examples: unlabeledInputs.slice(0, 3) });\n  else passCount++;\n\n  // 3. Buttons with no text\n  const emptyButtons = await page.$$eval(\n    'button:not([aria-label]):not([aria-labelledby])',\n    els => els.filter((e: any) => !(e.textContent ?? '').trim()).map((e: any) => e.outerHTML?.slice(0, 120) ?? ''),\n  );\n  if (emptyButtons.length > 0) violations.push({ rule: 'button-name', severity: 'critical', count: emptyButtons.length, examples: emptyButtons.slice(0, 3) });\n  else passCount++;\n\n  // 4. Missing html lang\n  const htmlLang = await page.$eval('html', (el: any) => el.getAttribute('lang') ?? '').catch(() => '');\n  if (!htmlLang) violations.push({ rule: 'html-has-lang', severity: 'serious', count: 1, examples: ['<html> missing lang attribute'] });\n  else passCount++;\n\n  // 5. Missing page title\n  const title = await page.title().catch(() => '');\n  if (!title) violations.push({ rule: 'document-title', severity: 'serious', count: 1, examples: ['Page has no <title>'] });\n  else passCount++;\n\n  // 6. Empty headings\n  const emptyHeadings = await page.$$eval(\n    'h1,h2,h3,h4,h5,h6',\n    els => els.filter((e: any) => !(e.textContent ?? '').trim()).map((e: any) => e.tagName.toLowerCase()),\n  );\n  if (emptyHeadings.length > 0) violations.push({ rule: 'empty-heading', severity: 'moderate', count: emptyHeadings.length, examples: emptyHeadings.slice(0, 3) });\n  else passCount++;\n\n  // 7. Links with no accessible name\n  const emptyLinks = await page.$$eval(\n    'a[href]:not([aria-label]):not([aria-labelledby])',\n    els => els.filter((e: any) => !(e.textContent ?? '').trim() && !e.querySelector('img[alt]')).map((e: any) => e.outerHTML?.slice(0, 120) ?? ''),\n  );\n  if (emptyLinks.length > 0) violations.push({ rule: 'link-name', severity: 'serious', count: emptyLinks.length, examples: emptyLinks.slice(0, 3) });\n  else passCount++;\n\n  const totalChecks = passCount + violations.length;\n  const score = totalChecks > 0 ? Math.round((passCount / totalChecks) * 100) : 100;\n\n  return { url, violations, passCount, score, auditedAt };\n}\n"],
+  "mappings": "AAgCA,eAAe,SAAS,MAAY,UAAkC;AACpE,MAAI;AACF,WAAO,MAAM,KAAK,OAAO,UAAU,SAAO,IAAI,IAAI,CAAC,QAAa;AAAA,MAC9D,WAAW,GAAG,WAAW,MAAM,GAAG,GAAG,KAAK;AAAA,MAC1C,cAAc,GAAG,eAAe,IAAI,KAAK,EAAE,MAAM,GAAG,GAAG;AAAA,MACvD,UAAU,GAAG;AAAA,MACb,cAAc,CAAC,MAAc,GAAG,aAAa,CAAC;AAAA,IAChD,EAAE,CAAC;AAAA,EACL,QAAQ;AACN,WAAO,CAAC;AAAA,EACV;AACF;AAEA,eAAsB,mBAAmB,MAAY,KAAuC;AAC1F,QAAM,aAA8B,CAAC;AACrC,MAAI,YAAY;AAChB,QAAM,aAAY,oBAAI,KAAK,GAAE,YAAY;AAGzC,QAAM,OAAO,MAAM,KAAK,OAAO,kBAAkB,SAAO,IAAI,IAAI,CAAC,MAAW,EAAE,WAAW,MAAM,GAAG,GAAG,KAAK,EAAE,CAAC;AAC7G,MAAI,KAAK,SAAS,EAAG,YAAW,KAAK,EAAE,MAAM,aAAa,UAAU,YAAY,OAAO,KAAK,QAAQ,UAAU,KAAK,MAAM,GAAG,CAAC,EAAE,CAAC;AAAA,MAC3H;AAGL,QAAM,kBAAkB,MAAM,KAAK;AAAA,IACjC;AAAA,IACA,SAAO,IAAI,IAAI,CAAC,MAAW,EAAE,WAAW,MAAM,GAAG,GAAG,KAAK,EAAE;AAAA,EAC7D;AACA,MAAI,gBAAgB,SAAS,EAAG,YAAW,KAAK,EAAE,MAAM,SAAS,UAAU,WAAW,OAAO,gBAAgB,QAAQ,UAAU,gBAAgB,MAAM,GAAG,CAAC,EAAE,CAAC;AAAA,MACvJ;AAGL,QAAM,eAAe,MAAM,KAAK;AAAA,IAC9B;AAAA,IACA,SAAO,IAAI,OAAO,CAAC,MAAW,EAAE,EAAE,eAAe,IAAI,KAAK,CAAC,EAAE,IAAI,CAAC,MAAW,EAAE,WAAW,MAAM,GAAG,GAAG,KAAK,EAAE;AAAA,EAC/G;AACA,MAAI,aAAa,SAAS,EAAG,YAAW,KAAK,EAAE,MAAM,eAAe,UAAU,YAAY,OAAO,aAAa,QAAQ,UAAU,aAAa,MAAM,GAAG,CAAC,EAAE,CAAC;AAAA,MACrJ;AAGL,QAAM,WAAW,MAAM,KAAK,MAAM,QAAQ,CAAC,OAAY,GAAG,aAAa,MAAM,KAAK,EAAE,EAAE,MAAM,MAAM,EAAE;AACpG,MAAI,CAAC,SAAU,YAAW,KAAK,EAAE,MAAM,iBAAiB,UAAU,WAAW,OAAO,GAAG,UAAU,CAAC,+BAA+B,EAAE,CAAC;AAAA,MAC/H;AAGL,QAAM,QAAQ,MAAM,KAAK,MAAM,EAAE,MAAM,MAAM,EAAE;AAC/C,MAAI,CAAC,MAAO,YAAW,KAAK,EAAE,MAAM,kBAAkB,UAAU,WAAW,OAAO,GAAG,UAAU,CAAC,qBAAqB,EAAE,CAAC;AAAA,MACnH;AAGL,QAAM,gBAAgB,MAAM,KAAK;AAAA,IAC/B;AAAA,IACA,SAAO,IAAI,OAAO,CAAC,MAAW,EAAE,EAAE,eAAe,IAAI,KAAK,CAAC,EAAE,IAAI,CAAC,MAAW,EAAE,QAAQ,YAAY,CAAC;AAAA,EACtG;AACA,MAAI,cAAc,SAAS,EAAG,YAAW,KAAK,EAAE,MAAM,iBAAiB,UAAU,YAAY,OAAO,cAAc,QAAQ,UAAU,cAAc,MAAM,GAAG,CAAC,EAAE,CAAC;AAAA,MAC1J;AAGL,QAAM,aAAa,MAAM,KAAK;AAAA,IAC5B;AAAA,IACA,SAAO,IAAI,OAAO,CAAC,MAAW,EAAE,EAAE,eAAe,IAAI,KAAK,KAAK,CAAC,EAAE,cAAc,UAAU,CAAC,EAAE,IAAI,CAAC,MAAW,EAAE,WAAW,MAAM,GAAG,GAAG,KAAK,EAAE;AAAA,EAC/I;AACA,MAAI,WAAW,SAAS,EAAG,YAAW,KAAK,EAAE,MAAM,aAAa,UAAU,WAAW,OAAO,WAAW,QAAQ,UAAU,WAAW,MAAM,GAAG,CAAC,EAAE,CAAC;AAAA,MAC5I;AAEL,QAAM,cAAc,YAAY,WAAW;AAC3C,QAAM,QAAQ,cAAc,IAAI,KAAK,MAAO,YAAY,cAAe,GAAG,IAAI;AAE9E,SAAO,EAAE,KAAK,YAAY,WAAW,OAAO,UAAU;AACxD;",
+  "names": []
+}

package/dist/accessibility-tree-crawler.d.ts

@@ -0,0 +1,44 @@
+/**
+ * Accessibility tree crawler — extract landmark, heading, and ARIA metadata from crawled pages.
+ *
+ * Captures: landmarks with interactive counts, heading hierarchy with violation detection,
+ * interactive elements missing accessible names, live regions, and skip-link presence.
+ *
+ * This data feeds ZeTa's accessibility test generator to surface WCAG 2.1 Level A/AA
+ * failures (missing names, heading skips, absent skip links, unlabeled live regions).
+ */
+export interface LandmarkInfo {
+    role: string;
+    label: string;
+    elementId: string;
+    interactiveCount: number;
+}
+export interface HeadingInfo {
+    level: number;
+    text: string;
+    elementId: string;
+}
+export interface MissingNameEl {
+    tagName: string;
+    role: string;
+    elementId: string;
+    htmlSnippet: string;
+}
+export interface LiveRegionInfo {
+    role: string;
+    ariaLive: string;
+    ariaAtomic: string;
+    currentText: string;
+}
+export interface AccessibilityTreeResult {
+    landmarkCount: number;
+    landmarks: LandmarkInfo[];
+    headings: HeadingInfo[];
+    headingHierarchyViolations: string[];
+    missingNameElements: MissingNameEl[];
+    missingNameCount: number;
+    liveRegions: LiveRegionInfo[];
+    totalInteractiveCount: number;
+    hasSkipLink: boolean;
+}
+export declare function crawlAccessibilityTree(page: any): Promise<AccessibilityTreeResult>;

package/dist/accessibility-tree-crawler.js

@@ -0,0 +1,156 @@
+async function extractLandmarks(page) {
+  try {
+    const raw = await page.evaluate(() => {
+      const LANDMARK_SELECTORS = [
+        { role: "banner", selector: 'header,[role="banner"]' },
+        { role: "navigation", selector: 'nav,[role="navigation"]' },
+        { role: "main", selector: 'main,[role="main"]' },
+        { role: "complementary", selector: 'aside,[role="complementary"]' },
+        { role: "contentinfo", selector: 'footer,[role="contentinfo"]' },
+        { role: "search", selector: '[role="search"]' },
+        { role: "form", selector: 'form[aria-label],[role="form"]' },
+        { role: "region", selector: '[role="region"][aria-label]' }
+      ];
+      return LANDMARK_SELECTORS.flatMap(
+        ({ role, selector }) => [...document.querySelectorAll(selector)].slice(0, 10).map((el) => ({
+          role,
+          label: el.getAttribute("aria-label") || el.getAttribute("aria-labelledby") || "",
+          elementId: el.id || "",
+          interactiveCount: el.querySelectorAll(
+            'button,a[href],input,select,textarea,[role="button"],[tabindex="0"]'
+          ).length
+        }))
+      );
+    });
+    return raw.slice(0, 50);
+  } catch {
+    return [];
+  }
+}
+async function extractHeadings(page) {
+  try {
+    const raw = await page.evaluate(() => {
+      const headingEls = [...document.querySelectorAll('h1,h2,h3,h4,h5,h6,[role="heading"]')];
+      return headingEls.map((el) => ({
+        level: parseInt(el.tagName.slice(1)) || parseInt(el.getAttribute("aria-level") || "0") || 0,
+        text: el.textContent?.trim().slice(0, 100) || "",
+        elementId: el.id || ""
+      })).filter((h) => h.level >= 1 && h.level <= 6);
+    });
+    return raw.slice(0, 100);
+  } catch {
+    return [];
+  }
+}
+function detectHeadingViolations(headings) {
+  const violations = [];
+  if (headings.length === 0) return violations;
+  const h1s = headings.filter((h) => h.level === 1);
+  if (h1s.length === 0) {
+    violations.push("Missing h1 landmark");
+  } else if (h1s.length > 1) {
+    violations.push("Multiple h1 elements found");
+  }
+  let prevLevel = 0;
+  for (const heading of headings) {
+    if (prevLevel > 0 && heading.level > prevLevel + 1) {
+      violations.push(
+        `Heading level skip: h${prevLevel} to h${heading.level}`
+      );
+    }
+    prevLevel = heading.level;
+  }
+  return violations;
+}
+async function findMissingNames(page) {
+  try {
+    return await page.evaluate(() => {
+      const INTERACTIVE_SELECTOR = 'button,a[href],input:not([type="hidden"]),select,textarea,[role="button"],[role="link"],[role="checkbox"],[role="radio"],[role="switch"],[role="menuitem"],[role="tab"]';
+      return [...document.querySelectorAll(INTERACTIVE_SELECTOR)].filter((el) => {
+        const labelledById = el.getAttribute("aria-labelledby") || "";
+        const name = (el.getAttribute("aria-label") || "").trim() || (labelledById ? document.getElementById(labelledById)?.textContent?.trim() ?? "" : "") || (el.textContent || "").trim() || (el.placeholder || "").trim() || (el.title || "").trim() || (el.alt || "").trim();
+        return !name || name.length === 0;
+      }).slice(0, 30).map((el) => ({
+        tagName: el.tagName.toLowerCase(),
+        role: el.getAttribute("role") || "",
+        elementId: el.id || "",
+        htmlSnippet: el.outerHTML.slice(0, 120)
+      }));
+    });
+  } catch {
+    return [];
+  }
+}
+async function extractLiveRegions(page) {
+  try {
+    return await page.evaluate(() => {
+      const els = [
+        ...document.querySelectorAll(
+          '[aria-live],[role="alert"],[role="status"],[role="log"]'
+        )
+      ];
+      return els.slice(0, 20).map((el) => ({
+        role: el.getAttribute("role") || "",
+        ariaLive: el.getAttribute("aria-live") || "",
+        ariaAtomic: el.getAttribute("aria-atomic") || "",
+        currentText: (el.textContent || "").trim().slice(0, 200)
+      }));
+    });
+  } catch {
+    return [];
+  }
+}
+async function checkSkipLink(page) {
+  try {
+    return await page.evaluate(() => {
+      const firstFocusable = document.querySelector('a[href^="#"],a:first-of-type');
+      return !!(firstFocusable && /skip/i.test(
+        (firstFocusable.textContent || "") + (firstFocusable.getAttribute("href") || "")
+      ));
+    });
+  } catch {
+    return false;
+  }
+}
+async function crawlAccessibilityTree(page) {
+  const [
+    landmarksResult,
+    headingsResult,
+    missingNamesResult,
+    liveRegionsResult,
+    skipLinkResult,
+    interactiveCountResult
+  ] = await Promise.allSettled([
+    extractLandmarks(page),
+    extractHeadings(page),
+    findMissingNames(page),
+    extractLiveRegions(page),
+    checkSkipLink(page),
+    page.evaluate(() => {
+      const INTERACTIVE_SELECTOR = 'button,a[href],input:not([type="hidden"]),select,textarea,[role="button"],[role="link"],[role="checkbox"],[role="radio"],[role="switch"],[role="menuitem"],[role="tab"]';
+      return document.querySelectorAll(INTERACTIVE_SELECTOR).length;
+    }).catch(() => 0)
+  ]);
+  const landmarks = landmarksResult.status === "fulfilled" ? landmarksResult.value : [];
+  const headings = headingsResult.status === "fulfilled" ? headingsResult.value : [];
+  const missingNameElements = missingNamesResult.status === "fulfilled" ? missingNamesResult.value : [];
+  const liveRegions = liveRegionsResult.status === "fulfilled" ? liveRegionsResult.value : [];
+  const hasSkipLink = skipLinkResult.status === "fulfilled" ? skipLinkResult.value : false;
+  const totalInteractiveCount = interactiveCountResult.status === "fulfilled" ? interactiveCountResult.value : 0;
+  const headingHierarchyViolations = detectHeadingViolations(headings);
+  return {
+    landmarkCount: landmarks.length,
+    landmarks,
+    headings,
+    headingHierarchyViolations,
+    missingNameElements,
+    missingNameCount: missingNameElements.length,
+    liveRegions,
+    totalInteractiveCount,
+    hasSkipLink
+  };
+}
+export {
+  crawlAccessibilityTree
+};
+//# sourceMappingURL=accessibility-tree-crawler.js.map

package/dist/accessibility-tree-crawler.js.map

@@ -0,0 +1,7 @@
+{
+  "version": 3,
+  "sources": ["../src/accessibility-tree-crawler.ts"],
+  "sourcesContent": ["/**\n * Accessibility tree crawler \u2014 extract landmark, heading, and ARIA metadata from crawled pages.\n *\n * Captures: landmarks with interactive counts, heading hierarchy with violation detection,\n * interactive elements missing accessible names, live regions, and skip-link presence.\n *\n * This data feeds ZeTa's accessibility test generator to surface WCAG 2.1 Level A/AA\n * failures (missing names, heading skips, absent skip links, unlabeled live regions).\n */\n\nimport type { Page } from 'playwright';\n\nexport interface LandmarkInfo {\n  role: string;\n  label: string;\n  elementId: string;\n  interactiveCount: number;\n}\n\nexport interface HeadingInfo {\n  level: number;\n  text: string;\n  elementId: string;\n}\n\nexport interface MissingNameEl {\n  tagName: string;\n  role: string;\n  elementId: string;\n  htmlSnippet: string;\n}\n\nexport interface LiveRegionInfo {\n  role: string;\n  ariaLive: string;\n  ariaAtomic: string;\n  currentText: string;\n}\n\nexport interface AccessibilityTreeResult {\n  landmarkCount: number;\n  landmarks: LandmarkInfo[];\n  headings: HeadingInfo[];\n  headingHierarchyViolations: string[];\n  missingNameElements: MissingNameEl[];\n  missingNameCount: number;\n  liveRegions: LiveRegionInfo[];\n  totalInteractiveCount: number;\n  hasSkipLink: boolean;\n}\n\nasync function extractLandmarks(page: any): Promise<LandmarkInfo[]> {\n  try {\n    const raw = await page.evaluate(() => {\n      const LANDMARK_SELECTORS = [\n        { role: 'banner', selector: 'header,[role=\"banner\"]' },\n        { role: 'navigation', selector: 'nav,[role=\"navigation\"]' },\n        { role: 'main', selector: 'main,[role=\"main\"]' },\n        { role: 'complementary', selector: 'aside,[role=\"complementary\"]' },\n        { role: 'contentinfo', selector: 'footer,[role=\"contentinfo\"]' },\n        { role: 'search', selector: '[role=\"search\"]' },\n        { role: 'form', selector: 'form[aria-label],[role=\"form\"]' },\n        { role: 'region', selector: '[role=\"region\"][aria-label]' },\n      ];\n      return LANDMARK_SELECTORS.flatMap(({ role, selector }) =>\n        [...document.querySelectorAll(selector)].slice(0, 10).map(el => ({\n          role,\n          label: el.getAttribute('aria-label') || el.getAttribute('aria-labelledby') || '',\n          elementId: el.id || '',\n          interactiveCount: el.querySelectorAll(\n            'button,a[href],input,select,textarea,[role=\"button\"],[tabindex=\"0\"]'\n          ).length,\n        }))\n      );\n    });\n    return raw.slice(0, 50);\n  } catch {\n    return [];\n  }\n}\n\nasync function extractHeadings(page: any): Promise<HeadingInfo[]> {\n  try {\n    const raw = await page.evaluate(() => {\n      const headingEls = [...document.querySelectorAll('h1,h2,h3,h4,h5,h6,[role=\"heading\"]')];\n      return headingEls\n        .map(el => ({\n          level:\n            parseInt(el.tagName.slice(1)) ||\n            parseInt(el.getAttribute('aria-level') || '0') ||\n            0,\n          text: el.textContent?.trim().slice(0, 100) || '',\n          elementId: el.id || '',\n        }))\n        .filter(h => h.level >= 1 && h.level <= 6);\n    });\n    return raw.slice(0, 100);\n  } catch {\n    return [];\n  }\n}\n\nfunction detectHeadingViolations(headings: HeadingInfo[]): string[] {\n  const violations: string[] = [];\n  if (headings.length === 0) return violations;\n\n  const h1s = headings.filter(h => h.level === 1);\n  if (h1s.length === 0) {\n    violations.push('Missing h1 landmark');\n  } else if (h1s.length > 1) {\n    violations.push('Multiple h1 elements found');\n  }\n\n  let prevLevel = 0;\n  for (const heading of headings) {\n    if (prevLevel > 0 && heading.level > prevLevel + 1) {\n      violations.push(\n        `Heading level skip: h${prevLevel} to h${heading.level}`\n      );\n    }\n    prevLevel = heading.level;\n  }\n\n  return violations;\n}\n\nasync function findMissingNames(page: any): Promise<MissingNameEl[]> {\n  try {\n    return await page.evaluate(() => {\n      const INTERACTIVE_SELECTOR =\n        'button,a[href],input:not([type=\"hidden\"]),select,textarea,' +\n        '[role=\"button\"],[role=\"link\"],[role=\"checkbox\"],[role=\"radio\"],' +\n        '[role=\"switch\"],[role=\"menuitem\"],[role=\"tab\"]';\n      return [...document.querySelectorAll(INTERACTIVE_SELECTOR)]\n        .filter(el => {\n          const labelledById = el.getAttribute('aria-labelledby') || '';\n          const name =\n            (el.getAttribute('aria-label') || '').trim() ||\n            (labelledById\n              ? document.getElementById(labelledById)?.textContent?.trim() ?? ''\n              : '') ||\n            (el.textContent || '').trim() ||\n            ((el as HTMLInputElement).placeholder || '').trim() ||\n            ((el as HTMLElement).title || '').trim() ||\n            ((el as HTMLImageElement).alt || '').trim();\n          return !name || name.length === 0;\n        })\n        .slice(0, 30)\n        .map(el => ({\n          tagName: el.tagName.toLowerCase(),\n          role: el.getAttribute('role') || '',\n          elementId: el.id || '',\n          htmlSnippet: el.outerHTML.slice(0, 120),\n        }));\n    });\n  } catch {\n    return [];\n  }\n}\n\nasync function extractLiveRegions(page: any): Promise<LiveRegionInfo[]> {\n  try {\n    return await page.evaluate(() => {\n      const els = [\n        ...document.querySelectorAll(\n          '[aria-live],[role=\"alert\"],[role=\"status\"],[role=\"log\"]'\n        ),\n      ];\n      return els.slice(0, 20).map(el => ({\n        role: el.getAttribute('role') || '',\n        ariaLive: el.getAttribute('aria-live') || '',\n        ariaAtomic: el.getAttribute('aria-atomic') || '',\n        currentText: (el.textContent || '').trim().slice(0, 200),\n      }));\n    });\n  } catch {\n    return [];\n  }\n}\n\nasync function checkSkipLink(page: any): Promise<boolean> {\n  try {\n    return await page.evaluate(() => {\n      const firstFocusable = document.querySelector('a[href^=\"#\"],a:first-of-type');\n      return !!(\n        firstFocusable &&\n        /skip/i.test(\n          (firstFocusable.textContent || '') + (firstFocusable.getAttribute('href') || '')\n        )\n      );\n    });\n  } catch {\n    return false;\n  }\n}\n\nexport async function crawlAccessibilityTree(page: any): Promise<AccessibilityTreeResult> {\n  const [\n    landmarksResult,\n    headingsResult,\n    missingNamesResult,\n    liveRegionsResult,\n    skipLinkResult,\n    interactiveCountResult,\n  ] = await Promise.allSettled([\n    extractLandmarks(page),\n    extractHeadings(page),\n    findMissingNames(page),\n    extractLiveRegions(page),\n    checkSkipLink(page),\n    page.evaluate(() => {\n      const INTERACTIVE_SELECTOR =\n        'button,a[href],input:not([type=\"hidden\"]),select,textarea,' +\n        '[role=\"button\"],[role=\"link\"],[role=\"checkbox\"],[role=\"radio\"],' +\n        '[role=\"switch\"],[role=\"menuitem\"],[role=\"tab\"]';\n      return document.querySelectorAll(INTERACTIVE_SELECTOR).length;\n    }).catch(() => 0),\n  ]);\n\n  const landmarks =\n    landmarksResult.status === 'fulfilled' ? landmarksResult.value : [];\n  const headings =\n    headingsResult.status === 'fulfilled' ? headingsResult.value : [];\n  const missingNameElements =\n    missingNamesResult.status === 'fulfilled' ? missingNamesResult.value : [];\n  const liveRegions =\n    liveRegionsResult.status === 'fulfilled' ? liveRegionsResult.value : [];\n  const hasSkipLink =\n    skipLinkResult.status === 'fulfilled' ? skipLinkResult.value : false;\n  const totalInteractiveCount =\n    interactiveCountResult.status === 'fulfilled' ? interactiveCountResult.value : 0;\n\n  const headingHierarchyViolations = detectHeadingViolations(headings);\n\n  return {\n    landmarkCount: landmarks.length,\n    landmarks,\n    headings,\n    headingHierarchyViolations,\n    missingNameElements,\n    missingNameCount: missingNameElements.length,\n    liveRegions,\n    totalInteractiveCount,\n    hasSkipLink,\n  };\n}\n"],
+  "mappings": "AAmDA,eAAe,iBAAiB,MAAoC;AAClE,MAAI;AACF,UAAM,MAAM,MAAM,KAAK,SAAS,MAAM;AACpC,YAAM,qBAAqB;AAAA,QACzB,EAAE,MAAM,UAAU,UAAU,yBAAyB;AAAA,QACrD,EAAE,MAAM,cAAc,UAAU,0BAA0B;AAAA,QAC1D,EAAE,MAAM,QAAQ,UAAU,qBAAqB;AAAA,QAC/C,EAAE,MAAM,iBAAiB,UAAU,+BAA+B;AAAA,QAClE,EAAE,MAAM,eAAe,UAAU,8BAA8B;AAAA,QAC/D,EAAE,MAAM,UAAU,UAAU,kBAAkB;AAAA,QAC9C,EAAE,MAAM,QAAQ,UAAU,iCAAiC;AAAA,QAC3D,EAAE,MAAM,UAAU,UAAU,8BAA8B;AAAA,MAC5D;AACA,aAAO,mBAAmB;AAAA,QAAQ,CAAC,EAAE,MAAM,SAAS,MAClD,CAAC,GAAG,SAAS,iBAAiB,QAAQ,CAAC,EAAE,MAAM,GAAG,EAAE,EAAE,IAAI,SAAO;AAAA,UAC/D;AAAA,UACA,OAAO,GAAG,aAAa,YAAY,KAAK,GAAG,aAAa,iBAAiB,KAAK;AAAA,UAC9E,WAAW,GAAG,MAAM;AAAA,UACpB,kBAAkB,GAAG;AAAA,YACnB;AAAA,UACF,EAAE;AAAA,QACJ,EAAE;AAAA,MACJ;AAAA,IACF,CAAC;AACD,WAAO,IAAI,MAAM,GAAG,EAAE;AAAA,EACxB,QAAQ;AACN,WAAO,CAAC;AAAA,EACV;AACF;AAEA,eAAe,gBAAgB,MAAmC;AAChE,MAAI;AACF,UAAM,MAAM,MAAM,KAAK,SAAS,MAAM;AACpC,YAAM,aAAa,CAAC,GAAG,SAAS,iBAAiB,oCAAoC,CAAC;AACtF,aAAO,WACJ,IAAI,SAAO;AAAA,QACV,OACE,SAAS,GAAG,QAAQ,MAAM,CAAC,CAAC,KAC5B,SAAS,GAAG,aAAa,YAAY,KAAK,GAAG,KAC7C;AAAA,QACF,MAAM,GAAG,aAAa,KAAK,EAAE,MAAM,GAAG,GAAG,KAAK;AAAA,QAC9C,WAAW,GAAG,MAAM;AAAA,MACtB,EAAE,EACD,OAAO,OAAK,EAAE,SAAS,KAAK,EAAE,SAAS,CAAC;AAAA,IAC7C,CAAC;AACD,WAAO,IAAI,MAAM,GAAG,GAAG;AAAA,EACzB,QAAQ;AACN,WAAO,CAAC;AAAA,EACV;AACF;AAEA,SAAS,wBAAwB,UAAmC;AAClE,QAAM,aAAuB,CAAC;AAC9B,MAAI,SAAS,WAAW,EAAG,QAAO;AAElC,QAAM,MAAM,SAAS,OAAO,OAAK,EAAE,UAAU,CAAC;AAC9C,MAAI,IAAI,WAAW,GAAG;AACpB,eAAW,KAAK,qBAAqB;AAAA,EACvC,WAAW,IAAI,SAAS,GAAG;AACzB,eAAW,KAAK,4BAA4B;AAAA,EAC9C;AAEA,MAAI,YAAY;AAChB,aAAW,WAAW,UAAU;AAC9B,QAAI,YAAY,KAAK,QAAQ,QAAQ,YAAY,GAAG;AAClD,iBAAW;AAAA,QACT,wBAAwB,SAAS,QAAQ,QAAQ,KAAK;AAAA,MACxD;AAAA,IACF;AACA,gBAAY,QAAQ;AAAA,EACtB;AAEA,SAAO;AACT;AAEA,eAAe,iBAAiB,MAAqC;AACnE,MAAI;AACF,WAAO,MAAM,KAAK,SAAS,MAAM;AAC/B,YAAM,uBACJ;AAGF,aAAO,CAAC,GAAG,SAAS,iBAAiB,oBAAoB,CAAC,EACvD,OAAO,QAAM;AACZ,cAAM,eAAe,GAAG,aAAa,iBAAiB,KAAK;AAC3D,cAAM,QACH,GAAG,aAAa,YAAY,KAAK,IAAI,KAAK,MAC1C,eACG,SAAS,eAAe,YAAY,GAAG,aAAa,KAAK,KAAK,KAC9D,QACH,GAAG,eAAe,IAAI,KAAK,MAC1B,GAAwB,eAAe,IAAI,KAAK,MAChD,GAAmB,SAAS,IAAI,KAAK,MACrC,GAAwB,OAAO,IAAI,KAAK;AAC5C,eAAO,CAAC,QAAQ,KAAK,WAAW;AAAA,MAClC,CAAC,EACA,MAAM,GAAG,EAAE,EACX,IAAI,SAAO;AAAA,QACV,SAAS,GAAG,QAAQ,YAAY;AAAA,QAChC,MAAM,GAAG,aAAa,MAAM,KAAK;AAAA,QACjC,WAAW,GAAG,MAAM;AAAA,QACpB,aAAa,GAAG,UAAU,MAAM,GAAG,GAAG;AAAA,MACxC,EAAE;AAAA,IACN,CAAC;AAAA,EACH,QAAQ;AACN,WAAO,CAAC;AAAA,EACV;AACF;AAEA,eAAe,mBAAmB,MAAsC;AACtE,MAAI;AACF,WAAO,MAAM,KAAK,SAAS,MAAM;AAC/B,YAAM,MAAM;AAAA,QACV,GAAG,SAAS;AAAA,UACV;AAAA,QACF;AAAA,MACF;AACA,aAAO,IAAI,MAAM,GAAG,EAAE,EAAE,IAAI,SAAO;AAAA,QACjC,MAAM,GAAG,aAAa,MAAM,KAAK;AAAA,QACjC,UAAU,GAAG,aAAa,WAAW,KAAK;AAAA,QAC1C,YAAY,GAAG,aAAa,aAAa,KAAK;AAAA,QAC9C,cAAc,GAAG,eAAe,IAAI,KAAK,EAAE,MAAM,GAAG,GAAG;AAAA,MACzD,EAAE;AAAA,IACJ,CAAC;AAAA,EACH,QAAQ;AACN,WAAO,CAAC;AAAA,EACV;AACF;AAEA,eAAe,cAAc,MAA6B;AACxD,MAAI;AACF,WAAO,MAAM,KAAK,SAAS,MAAM;AAC/B,YAAM,iBAAiB,SAAS,cAAc,8BAA8B;AAC5E,aAAO,CAAC,EACN,kBACA,QAAQ;AAAA,SACL,eAAe,eAAe,OAAO,eAAe,aAAa,MAAM,KAAK;AAAA,MAC/E;AAAA,IAEJ,CAAC;AAAA,EACH,QAAQ;AACN,WAAO;AAAA,EACT;AACF;AAEA,eAAsB,uBAAuB,MAA6C;AACxF,QAAM;AAAA,IACJ;AAAA,IACA;AAAA,IACA;AAAA,IACA;AAAA,IACA;AAAA,IACA;AAAA,EACF,IAAI,MAAM,QAAQ,WAAW;AAAA,IAC3B,iBAAiB,IAAI;AAAA,IACrB,gBAAgB,IAAI;AAAA,IACpB,iBAAiB,IAAI;AAAA,IACrB,mBAAmB,IAAI;AAAA,IACvB,cAAc,IAAI;AAAA,IAClB,KAAK,SAAS,MAAM;AAClB,YAAM,uBACJ;AAGF,aAAO,SAAS,iBAAiB,oBAAoB,EAAE;AAAA,IACzD,CAAC,EAAE,MAAM,MAAM,CAAC;AAAA,EAClB,CAAC;AAED,QAAM,YACJ,gBAAgB,WAAW,cAAc,gBAAgB,QAAQ,CAAC;AACpE,QAAM,WACJ,eAAe,WAAW,cAAc,eAAe,QAAQ,CAAC;AAClE,QAAM,sBACJ,mBAAmB,WAAW,cAAc,mBAAmB,QAAQ,CAAC;AAC1E,QAAM,cACJ,kBAAkB,WAAW,cAAc,kBAAkB,QAAQ,CAAC;AACxE,QAAM,cACJ,eAAe,WAAW,cAAc,eAAe,QAAQ;AACjE,QAAM,wBACJ,uBAAuB,WAAW,cAAc,uBAAuB,QAAQ;AAEjF,QAAM,6BAA6B,wBAAwB,QAAQ;AAEnE,SAAO;AAAA,IACL,eAAe,UAAU;AAAA,IACzB;AAAA,IACA;AAAA,IACA;AAAA,IACA;AAAA,IACA,kBAAkB,oBAAoB;AAAA,IACtC;AAAA,IACA;AAAA,IACA;AAAA,EACF;AACF;",
+  "names": []
+}

package/dist/advanced-ui-probes.d.ts

@@ -0,0 +1,47 @@
+/**
+ * Advanced UI probes — detect and exercise drag-drop, date pickers, sliders,
+ * and typeahead/autocomplete widgets on a crawled page.
+ *
+ * Results feed ZeTa's test generator so it can emit interaction-level test
+ * cases for these widget families instead of falling back to generic clicks.
+ *
+ * Each probe is independently try/caught so a broken widget does not abort
+ * the others. All browser-side code runs inside page.evaluate / page.$$eval
+ * to avoid serialisation issues with DOM handles.
+ */
+import type { Page } from 'playwright';
+export interface DragDropCapture {
+    library: 'html5' | 'dnd-kit' | 'react-beautiful-dnd' | 'sortablejs' | 'pointer-event' | 'none';
+    draggableCount: number;
+    dropZoneCount: number;
+    testAttempted: boolean;
+    testCompleted: boolean;
+}
+export interface DatePickerCapture {
+    fieldLabel: string;
+    pickerType: 'calendar-popup' | 'native-date' | 'native-datetime' | 'none';
+    minDate?: string;
+    maxDate?: string;
+    disabledDatesFound: boolean;
+    sampleSelectedValue?: string;
+}
+export interface SliderCapture {
+    label: string;
+    min: number;
+    max: number;
+    step: number;
+    currentValue: number;
+}
+export interface TypeaheadCapture {
+    label: string;
+    minCharsToTrigger: number;
+    sampleSuggestions: string[];
+    noResultsText: string | null;
+}
+export interface AdvancedProbeResult {
+    dragDrop: DragDropCapture;
+    datePickers: DatePickerCapture[];
+    sliders: SliderCapture[];
+    typeaheads: TypeaheadCapture[];
+}
+export declare function probeAdvancedUI(page: Page): Promise<AdvancedProbeResult>;