@detiq/crawler 1.1.9 → 1.1.11

package/dist/extractors/artifact-envelope.js

@@ -1,62 +1,63 @@
-import crypto from 'node:crypto';
-export function sha256(input) {
-    return crypto.createHash('sha256').update(input).digest('hex');
+import crypto from "node:crypto";
+function sha256(input) {
+  return crypto.createHash("sha256").update(input).digest("hex");
 }
-export function buildFitMarkdown(markdown, keywords = []) {
-    const lines = markdown
-        .split('\n')
-        .map((line) => line.trimEnd())
-        .filter((line) => line.trim().length > 0);
-    const lowerKeywords = keywords.map((k) => k.toLowerCase()).filter(Boolean);
-    const filtered = lines.filter((line) => {
-        const plain = line.replace(/[#*_`>\-[\]()]/g, '').trim();
-        if (plain.length < 3)
-            return false;
-        if (/^(cookie|privacy|terms|copyright|all rights reserved|subscribe|newsletter)$/i.test(plain))
-            return false;
-        if (lowerKeywords.length === 0)
-            return true;
-        const lower = plain.toLowerCase();
-        return lowerKeywords.some((kw) => lower.includes(kw)) || /^#{1,4}\s/.test(line);
-    });
-    const candidate = filtered.join('\n');
-    return candidate.length >= 200 ? candidate : markdown;
+function buildFitMarkdown(markdown, keywords = []) {
+  const lines = markdown.split("\n").map((line) => line.trimEnd()).filter((line) => line.trim().length > 0);
+  const lowerKeywords = keywords.map((k) => k.toLowerCase()).filter(Boolean);
+  const filtered = lines.filter((line) => {
+    const plain = line.replace(/[#*_`>\-[\]()]/g, "").trim();
+    if (plain.length < 3) return false;
+    if (/^(cookie|privacy|terms|copyright|all rights reserved|subscribe|newsletter)$/i.test(plain)) return false;
+    if (lowerKeywords.length === 0) return true;
+    const lower = plain.toLowerCase();
+    return lowerKeywords.some((kw) => lower.includes(kw)) || /^#{1,4}\s/.test(line);
+  });
+  const candidate = filtered.join("\n");
+  return candidate.length >= 200 ? candidate : markdown;
 }
-export function summarizeStructuredElements(elements) {
-    const forms = elements.filter((e) => ['email', 'password', 'text', 'tel', 'date', 'number', 'select', 'textarea'].includes(e.role));
-    const actions = elements.filter((e) => e.role === 'button' || e.role === 'submit' || e.role === 'link' || e.role === 'a');
-    return {
-        elementCount: elements.length,
-        formFieldCount: forms.length,
-        actionCount: actions.length,
-        fields: forms.slice(0, 50).map((e) => ({ label: e.label, role: e.role, expectedData: e.expectedData })),
-        actions: actions.slice(0, 50).map((e) => ({ label: e.label, role: e.role })),
-    };
+function summarizeStructuredElements(elements) {
+  const forms = elements.filter((e) => ["email", "password", "text", "tel", "date", "number", "select", "textarea"].includes(e.role));
+  const actions = elements.filter((e) => e.role === "button" || e.role === "submit" || e.role === "link" || e.role === "a");
+  return {
+    elementCount: elements.length,
+    formFieldCount: forms.length,
+    actionCount: actions.length,
+    fields: forms.slice(0, 50).map((e) => ({ label: e.label, role: e.role, expectedData: e.expectedData })),
+    actions: actions.slice(0, 50).map((e) => ({ label: e.label, role: e.role }))
+  };
 }
-export function buildCrawlArtifactEnvelope(input) {
-    const accessibilityTree = input.accessibilityElements ? JSON.stringify(input.accessibilityElements) : undefined;
-    return {
-        tenantId: input.tenantId,
-        projectId: input.projectId,
-        jobId: input.jobId,
-        url: input.url,
-        canonicalUrl: input.canonicalUrl,
-        screenId: input.screenId,
-        status: 'ok',
-        outputs: {
-            screenshot: input.screenshotPath,
-            html: input.domPath,
-            markdown: input.markdown,
-            fitMarkdown: input.fitMarkdown,
-            accessibilityTree,
-            structuredJson: input.structuredJson,
-        },
-        fingerprints: {
-            textHash: input.markdown ? sha256(input.markdown) : undefined,
-            domHash: input.domContent ? sha256(input.domContent) : undefined,
-            accessibilityHash: accessibilityTree ? sha256(accessibilityTree) : undefined,
-            urlCanonicalHash: sha256(input.canonicalUrl),
-        },
-        evidence: input.evidence ?? [],
-    };
+function buildCrawlArtifactEnvelope(input) {
+  const accessibilityTree = input.accessibilityElements ? JSON.stringify(input.accessibilityElements) : void 0;
+  return {
+    tenantId: input.tenantId,
+    projectId: input.projectId,
+    jobId: input.jobId,
+    url: input.url,
+    canonicalUrl: input.canonicalUrl,
+    screenId: input.screenId,
+    status: "ok",
+    outputs: {
+      screenshot: input.screenshotPath,
+      html: input.domPath,
+      markdown: input.markdown,
+      fitMarkdown: input.fitMarkdown,
+      accessibilityTree,
+      structuredJson: input.structuredJson
+    },
+    fingerprints: {
+      textHash: input.markdown ? sha256(input.markdown) : void 0,
+      domHash: input.domContent ? sha256(input.domContent) : void 0,
+      accessibilityHash: accessibilityTree ? sha256(accessibilityTree) : void 0,
+      urlCanonicalHash: sha256(input.canonicalUrl)
+    },
+    evidence: input.evidence ?? []
+  };
 }
+export {
+  buildCrawlArtifactEnvelope,
+  buildFitMarkdown,
+  sha256,
+  summarizeStructuredElements
+};
+//# sourceMappingURL=artifact-envelope.js.map

package/dist/failure-analysis-agent.js

@@ -1,37 +1,41 @@
-import { readFileSync } from 'node:fs';
-import { askTrackedWithImages, askTracked, deterministicSeed } from '@detiq/agents';
-import { AppBrain } from '@detiq/app-brain';
-export async function analyzeFailure(ctx) {
-    try {
-        let llmCfg;
-        if (ctx.tenantId) {
-            llmCfg = await AppBrain.getTaskLLMConfig(ctx.tenantId, 'testRunner').catch(() => undefined) ?? undefined;
-        }
-        if (!llmCfg)
-            return null;
-        const consoleSection = ctx.consoleLogs?.length
-            ? `\n\nBrowser console (last ${ctx.consoleLogs.length} entries):\n${ctx.consoleLogs.map(l => `[${l.type}] ${l.text}`).join('\n')}`
-            : '';
-        const networkSection = ctx.networkErrors?.length
-            ? `\n\nFailed network requests:\n${ctx.networkErrors.map(n => `${n.method} ${n.url} → ${n.status}`).join('\n')}`
-            : '';
-        const userText = `A Playwright automated test failed. Identify the root cause in 2-3 sentences. Be specific — name the failing element, API endpoint, or error type. Do not suggest fixes.
+import { readFileSync } from "node:fs";
+import { askTrackedWithImages, askTracked, deterministicSeed } from "@detiq/agents";
+import { AppBrain } from "@detiq/app-brain";
+async function analyzeFailure(ctx) {
+  try {
+    let llmCfg;
+    if (ctx.tenantId) {
+      llmCfg = await AppBrain.getTaskLLMConfig(ctx.tenantId, "testRunner").catch(() => void 0) ?? void 0;
+    }
+    if (!llmCfg) return null;
+    const consoleSection = ctx.consoleLogs?.length ? `
+
+Browser console (last ${ctx.consoleLogs.length} entries):
+${ctx.consoleLogs.map((l) => `[${l.type}] ${l.text}`).join("\n")}` : "";
+    const networkSection = ctx.networkErrors?.length ? `
 
-Test step: ${ctx.stepDescription ?? 'unknown'}
+Failed network requests:
+${ctx.networkErrors.map((n) => `${n.method} ${n.url} \u2192 ${n.status}`).join("\n")}` : "";
+    const userText = `A Playwright automated test failed. Identify the root cause in 2-3 sentences. Be specific \u2014 name the failing element, API endpoint, or error type. Do not suggest fixes.
+
+Test step: ${ctx.stepDescription ?? "unknown"}
 Error: ${ctx.error}${consoleSection}${networkSection}`;
-        if (ctx.screenshotPath) {
-            try {
-                const img = readFileSync(ctx.screenshotPath);
-                const result = await askTrackedWithImages('', userText, [{ base64: img.toString('base64'), mediaType: 'image/png' }], llmCfg);
-                return result.text || null;
-            }
-            catch { /* fall through to text-only */ }
-        }
-        const failureSeed = deterministicSeed(`${ctx.tenantId ?? ''}:failure:${(ctx.stepDescription ?? '').slice(0, 100)}:${ctx.error.slice(0, 200)}`);
-        const result = await askTracked('', userText, llmCfg, 256, undefined, { temperature: 0, seed: failureSeed });
-        return result.text || null;
-    }
-    catch {
-        return null;
+    if (ctx.screenshotPath) {
+      try {
+        const img = readFileSync(ctx.screenshotPath);
+        const result2 = await askTrackedWithImages("", userText, [{ base64: img.toString("base64"), mediaType: "image/png" }], llmCfg);
+        return result2.text || null;
+      } catch {
+      }
     }
+    const failureSeed = deterministicSeed(`${ctx.tenantId ?? ""}:failure:${(ctx.stepDescription ?? "").slice(0, 100)}:${ctx.error.slice(0, 200)}`);
+    const result = await askTracked("", userText, llmCfg, 256, void 0, { temperature: 0, seed: failureSeed });
+    return result.text || null;
+  } catch {
+    return null;
+  }
 }
+export {
+  analyzeFailure
+};
+//# sourceMappingURL=failure-analysis-agent.js.map

package/dist/feature-flag-detector.js

@@ -1,70 +1,55 @@
-/**
- * Feature flag + A/B test detector.
- *
- * Detects popular feature flag / experiment SDKs loaded on the page:
- * LaunchDarkly, Optimizely, Split.io, Unleash, GrowthBook, Statsig, Flagsmith, Amplitude Experiment.
- *
- * Also detects common A/B test patterns: data-testid attributes, gtm.js experiment layers,
- * Optimizely variation IDs, VWO, etc.
- *
- * Why: ZeTa needs to know when a page shows different content to different users
- * so test generation can account for variants and not generate flaky tests.
- */
-// SDK detectors — checked against window object + script URLs
 const FLAG_SDK_CHECKS = [
-    { sdk: 'LaunchDarkly', windowKeys: ['LDClient', 'LaunchDarkly', 'ldClient'], scriptPattern: /launchdarkly/i },
-    { sdk: 'Optimizely', windowKeys: ['optimizely', 'optimizelySdk'], scriptPattern: /optimizely/i },
-    { sdk: 'Split.io', windowKeys: ['splitio', 'SplitFactory'], scriptPattern: /split\.io/i },
-    { sdk: 'GrowthBook', windowKeys: ['growthbook', 'GrowthBook'], scriptPattern: /growthbook/i },
-    { sdk: 'Statsig', windowKeys: ['statsig', 'StatsigClient'], scriptPattern: /statsig/i },
-    { sdk: 'Flagsmith', windowKeys: ['flagsmith'], scriptPattern: /flagsmith/i },
-    { sdk: 'Unleash', windowKeys: ['unleash', 'UnleashClient'], scriptPattern: /unleash/i },
-    { sdk: 'VWO', windowKeys: ['vwo', '_vwo'], scriptPattern: /vwo\.com/i },
-    { sdk: 'Google Optimize', windowKeys: ['gaData', 'google_optimize'], scriptPattern: /googleoptimize/i },
-    { sdk: 'AB Tasty', windowKeys: ['ABTasty'], scriptPattern: /abtasty/i },
+  { sdk: "LaunchDarkly", windowKeys: ["LDClient", "LaunchDarkly", "ldClient"], scriptPattern: /launchdarkly/i },
+  { sdk: "Optimizely", windowKeys: ["optimizely", "optimizelySdk"], scriptPattern: /optimizely/i },
+  { sdk: "Split.io", windowKeys: ["splitio", "SplitFactory"], scriptPattern: /split\.io/i },
+  { sdk: "GrowthBook", windowKeys: ["growthbook", "GrowthBook"], scriptPattern: /growthbook/i },
+  { sdk: "Statsig", windowKeys: ["statsig", "StatsigClient"], scriptPattern: /statsig/i },
+  { sdk: "Flagsmith", windowKeys: ["flagsmith"], scriptPattern: /flagsmith/i },
+  { sdk: "Unleash", windowKeys: ["unleash", "UnleashClient"], scriptPattern: /unleash/i },
+  { sdk: "VWO", windowKeys: ["vwo", "_vwo"], scriptPattern: /vwo\.com/i },
+  { sdk: "Google Optimize", windowKeys: ["gaData", "google_optimize"], scriptPattern: /googleoptimize/i },
+  { sdk: "AB Tasty", windowKeys: ["ABTasty"], scriptPattern: /abtasty/i }
 ];
-export async function detectFeatureFlags(page, url) {
-    const detectedAt = new Date().toISOString();
-    const result = await page.evaluate((checks) => {
-        const flagSdks = [];
-        const scripts = Array.from(document.querySelectorAll('script[src]')).map((s) => s.src ?? '');
-        for (const check of checks) {
-            let found = false;
-            let confidence = 'low';
-            // Check window keys
-            if (check.windowKeys?.some((k) => k in window)) {
-                found = true;
-                confidence = 'high';
-            }
-            // Check script URLs
-            if (!found && check.scriptPattern) {
-                const re = new RegExp(check.scriptPattern.source, 'i');
-                const matchingSrc = scripts.find(s => re.test(s));
-                if (matchingSrc) {
-                    found = true;
-                    confidence = 'medium';
-                }
-            }
-            if (found)
-                flagSdks.push({ sdk: check.sdk, confidence, details: undefined });
+async function detectFeatureFlags(page, url) {
+  const detectedAt = (/* @__PURE__ */ new Date()).toISOString();
+  const result = await page.evaluate((checks) => {
+    const flagSdks = [];
+    const scripts = Array.from(document.querySelectorAll("script[src]")).map((s) => s.src ?? "");
+    for (const check of checks) {
+      let found = false;
+      let confidence = "low";
+      if (check.windowKeys?.some((k) => k in window)) {
+        found = true;
+        confidence = "high";
+      }
+      if (!found && check.scriptPattern) {
+        const re = new RegExp(check.scriptPattern.source, "i");
+        const matchingSrc = scripts.find((s) => re.test(s));
+        if (matchingSrc) {
+          found = true;
+          confidence = "medium";
         }
-        // A/B tools
-        const abTools = [];
-        if ('_vwo_code' in window || '_vis_opt_queue' in window)
-            abTools.push({ tool: 'VWO', experiments: [] });
-        if ('dataLayer' in window) {
-            const dl = window.dataLayer ?? [];
-            const expEvents = dl.filter((e) => e?.event?.includes('experiment') || e?.experimentId);
-            if (expEvents.length > 0)
-                abTools.push({ tool: 'Google Tag Manager Experiments', experiments: expEvents.slice(0, 5).map((e) => e.experimentId ?? '') });
-        }
-        return { flagSdks, abTools };
-    }, FLAG_SDK_CHECKS).catch(() => ({ flagSdks: [], abTools: [] }));
-    return {
-        url,
-        flagSdks: result.flagSdks,
-        abTools: result.abTools,
-        hasAnyFlags: result.flagSdks.length > 0 || result.abTools.length > 0,
-        detectedAt,
-    };
+      }
+      if (found) flagSdks.push({ sdk: check.sdk, confidence, details: void 0 });
+    }
+    const abTools = [];
+    if ("_vwo_code" in window || "_vis_opt_queue" in window) abTools.push({ tool: "VWO", experiments: [] });
+    if ("dataLayer" in window) {
+      const dl = window.dataLayer ?? [];
+      const expEvents = dl.filter((e) => e?.event?.includes("experiment") || e?.experimentId);
+      if (expEvents.length > 0) abTools.push({ tool: "Google Tag Manager Experiments", experiments: expEvents.slice(0, 5).map((e) => e.experimentId ?? "") });
+    }
+    return { flagSdks, abTools };
+  }, FLAG_SDK_CHECKS).catch(() => ({ flagSdks: [], abTools: [] }));
+  return {
+    url,
+    flagSdks: result.flagSdks,
+    abTools: result.abTools,
+    hasAnyFlags: result.flagSdks.length > 0 || result.abTools.length > 0,
+    detectedAt
+  };
 }
+export {
+  detectFeatureFlags
+};
+//# sourceMappingURL=feature-flag-detector.js.map

package/dist/form-detector.js

@@ -1,78 +1,58 @@
-/**
- * Form detector — extract form metadata from crawled pages.
- *
- * For each form found on a page, captures: action URL, method, all input fields
- * with their types/names/required/validation constraints, and submit button text.
- *
- * This data feeds ZeTa's test generator to create form validation test cases
- * automatically (happy path + boundary/error cases per field type).
- */
 function guessPurpose(fields, action) {
-    const names = fields.map(f => f.name?.toLowerCase() + (f.label ?? '').toLowerCase()).join(' ');
-    if (/password/.test(names) && /email|username|user/.test(names) && fields.length <= 4)
-        return 'login';
-    if (/password/.test(names) && /confirm|repeat/.test(names))
-        return 'registration';
-    if (/search|query|q\b/.test(names) && fields.length <= 3)
-        return 'search';
-    if (/card|payment|cvv|expir/.test(names))
-        return 'checkout';
-    if (/message|subject|contact/.test(names))
-        return 'contact';
-    if (/setting|preference|profile/.test(names))
-        return 'settings';
-    return 'unknown';
+  const names = fields.map((f) => f.name?.toLowerCase() + (f.label ?? "").toLowerCase()).join(" ");
+  if (/password/.test(names) && /email|username|user/.test(names) && fields.length <= 4) return "login";
+  if (/password/.test(names) && /confirm|repeat/.test(names)) return "registration";
+  if (/search|query|q\b/.test(names) && fields.length <= 3) return "search";
+  if (/card|payment|cvv|expir/.test(names)) return "checkout";
+  if (/message|subject|contact/.test(names)) return "contact";
+  if (/setting|preference|profile/.test(names)) return "settings";
+  return "unknown";
 }
-export async function detectForms(page) {
-    try {
-        return await page.$$eval('form', (forms) => forms.map((form, index) => {
-            const inputs = Array.from(form.querySelectorAll('input,select,textarea,button[type="submit"]'));
-            const fields = [];
-            for (const input of inputs) {
-                const tag = input.tagName.toLowerCase();
-                const type = tag === 'button' ? 'submit' : (input.type || tag);
-                if (['submit', 'reset', 'button', 'hidden'].includes(type))
-                    continue;
-                // Find associated label
-                let label = '';
-                if (input.id) {
-                    const lbl = form.querySelector(`label[for="${input.id}"]`);
-                    if (lbl)
-                        label = (lbl.textContent ?? '').trim();
-                }
-                if (!label) {
-                    const parent = input.closest('label');
-                    if (parent)
-                        label = (parent.textContent ?? '').replace(input.value ?? '', '').trim().slice(0, 80);
-                }
-                const options = tag === 'select'
-                    ? Array.from(input.options ?? []).map((o) => o.text?.trim()).filter(Boolean).slice(0, 20)
-                    : tag === 'input' && type === 'radio'
-                        ? [input.value]
-                        : undefined;
-                fields.push({
-                    name: input.name || input.id || '',
-                    type,
-                    label: label || undefined,
-                    required: !!input.required,
-                    placeholder: input.placeholder || undefined,
-                    minLength: input.minLength > 0 ? input.minLength : undefined,
-                    maxLength: input.maxLength > 0 && input.maxLength < 524288 ? input.maxLength : undefined,
-                    min: input.min || undefined,
-                    max: input.max || undefined,
-                    pattern: input.pattern || undefined,
-                    options,
-                    autocomplete: input.autocomplete || undefined,
-                });
-            }
-            const submitBtn = form.querySelector('button[type="submit"],input[type="submit"]');
-            const hasFileUpload = !!form.querySelector('input[type="file"]');
-            const action = form.action || undefined;
-            const method = (form.method || 'get').toUpperCase();
-            return { index, action, method, fields, submitText: submitBtn?.textContent?.trim() || submitBtn?.value || undefined, hasFileUpload };
-        })).then(forms => forms.map((f) => ({ ...f, purpose: guessPurpose(f.fields, f.action) })));
-    }
-    catch {
-        return [];
-    }
+async function detectForms(page) {
+  try {
+    return await page.$$eval("form", (forms) => forms.map((form, index) => {
+      const inputs = Array.from(form.querySelectorAll('input,select,textarea,button[type="submit"]'));
+      const fields = [];
+      for (const input of inputs) {
+        const tag = input.tagName.toLowerCase();
+        const type = tag === "button" ? "submit" : input.type || tag;
+        if (["submit", "reset", "button", "hidden"].includes(type)) continue;
+        let label = "";
+        if (input.id) {
+          const lbl = form.querySelector(`label[for="${input.id}"]`);
+          if (lbl) label = (lbl.textContent ?? "").trim();
+        }
+        if (!label) {
+          const parent = input.closest("label");
+          if (parent) label = (parent.textContent ?? "").replace(input.value ?? "", "").trim().slice(0, 80);
+        }
+        const options = tag === "select" ? Array.from(input.options ?? []).map((o) => o.text?.trim()).filter(Boolean).slice(0, 20) : tag === "input" && type === "radio" ? [input.value] : void 0;
+        fields.push({
+          name: input.name || input.id || "",
+          type,
+          label: label || void 0,
+          required: !!input.required,
+          placeholder: input.placeholder || void 0,
+          minLength: input.minLength > 0 ? input.minLength : void 0,
+          maxLength: input.maxLength > 0 && input.maxLength < 524288 ? input.maxLength : void 0,
+          min: input.min || void 0,
+          max: input.max || void 0,
+          pattern: input.pattern || void 0,
+          options,
+          autocomplete: input.autocomplete || void 0
+        });
+      }
+      const submitBtn = form.querySelector('button[type="submit"],input[type="submit"]');
+      const hasFileUpload = !!form.querySelector('input[type="file"]');
+      const action = form.action || void 0;
+      const method = (form.method || "get").toUpperCase();
+      return { index, action, method, fields, submitText: submitBtn?.textContent?.trim() || submitBtn?.value || void 0, hasFileUpload };
+    })).then((forms) => forms.map((f) => ({ ...f, purpose: guessPurpose(f.fields, f.action) })));
+  } catch {
+    return [];
+  }
 }
+export {
+  detectForms
+};
+//# sourceMappingURL=form-detector.js.map