@detiq/crawler 1.1.7 → 1.1.9

package/dist/extractors/advanced.js

@@ -1,265 +1,282 @@
-import { parseHTML } from "linkedom";
-function extractByCssSelector(html, selector, attr) {
-  try {
-    const { document } = parseHTML(html);
-    const nodes = Array.from(document.querySelectorAll(selector));
-    const matches = nodes.map((n) => {
-      if (attr) return n.getAttribute(attr) ?? "";
-      return n.textContent?.trim() ?? "";
-    }).filter(Boolean);
-    return { matches, count: matches.length, method: "css-selector" };
-  } catch (e) {
-    return { matches: [], count: 0, method: "css-selector-error" };
-  }
-}
-function extractByXPath(html, xpath) {
-  try {
-    const { document, XPathResult } = parseHTML(html);
-    const result = document.evaluate(xpath, document, null, XPathResult?.ANY_TYPE ?? 0, null);
-    const matches = [];
-    let node = result.iterateNext?.();
-    while (node) {
-      const text = node.textContent?.trim() ?? node.nodeValue?.trim() ?? "";
-      if (text) matches.push(text);
-      node = result.iterateNext?.();
+import { parseHTML } from 'linkedom';
+export function extractByCssSelector(html, selector, attr) {
+    try {
+        const { document } = parseHTML(html);
+        const nodes = Array.from(document.querySelectorAll(selector));
+        const matches = nodes.map((n) => {
+            if (attr)
+                return n.getAttribute(attr) ?? '';
+            return n.textContent?.trim() ?? '';
+        }).filter(Boolean);
+        return { matches, count: matches.length, method: 'css-selector' };
     }
-    return { matches, count: matches.length, method: "xpath" };
-  } catch (e) {
-    return { matches: [], count: 0, method: "xpath-error" };
-  }
-}
-function extractByRegex(text, pattern, flags) {
-  try {
-    const re = new RegExp(pattern, (flags ?? "g").includes("g") ? flags : (flags ?? "") + "g");
-    const matches = [];
-    let m;
-    while ((m = re.exec(text)) !== null) {
-      matches.push(m[1] ?? m[0]);
-      if (!re.global) break;
+    catch (e) {
+        return { matches: [], count: 0, method: 'css-selector-error' };
     }
-    return { matches, count: matches.length, method: "regex" };
-  } catch {
-    return { matches: [], count: 0, method: "regex-error" };
-  }
 }
-function extractTables(html) {
-  try {
-    const { document } = parseHTML(html);
-    const tables = Array.from(document.querySelectorAll("table"));
-    return tables.map((table) => {
-      const headers = [];
-      const rows = [];
-      const ths = [...table.querySelectorAll("thead th, thead td, tr:first-child th")];
-      if (ths.length > 0) {
-        ths.forEach((th, i) => headers.push(th.textContent?.trim() || `col${i + 1}`));
-      }
-      const trs = [...table.querySelectorAll("tbody tr, tr")];
-      for (const tr of trs) {
-        const cells = [...tr.querySelectorAll("td, th")];
-        if (cells.length === 0) continue;
-        if (headers.length === 0) {
-          cells.forEach((_, i) => headers.push(`col${i + 1}`));
+export function extractByXPath(html, xpath) {
+    try {
+        const { document, XPathResult } = parseHTML(html);
+        const result = document.evaluate(xpath, document, null, XPathResult?.ANY_TYPE ?? 0, null);
+        const matches = [];
+        let node = result.iterateNext?.();
+        while (node) {
+            const text = node.textContent?.trim() ?? node.nodeValue?.trim() ?? '';
+            if (text)
+                matches.push(text);
+            node = result.iterateNext?.();
         }
-        const row = {};
-        cells.forEach((cell, i) => {
-          const key = headers[i] || `col${i + 1}`;
-          row[key] = cell.textContent?.trim() ?? "";
-        });
-        if (Object.values(row).some((v) => v)) rows.push(row);
-      }
-      return rows;
-    }).filter((t) => t.length > 0);
-  } catch {
-    return [];
-  }
-}
-function extractMetadata(html, url) {
-  const meta = {
-    title: null,
-    description: null,
-    keywords: null,
-    ogTitle: null,
-    ogDescription: null,
-    ogImage: null,
-    ogUrl: null,
-    ogType: null,
-    ogSiteName: null,
-    ogLocale: null,
-    twitterCard: null,
-    twitterTitle: null,
-    twitterDescription: null,
-    twitterImage: null,
-    twitterSite: null,
-    canonical: null,
-    robots: null,
-    hreflang: [],
-    themeColor: null,
-    favicon: null,
-    jsonLd: [],
-    schemaTypes: [],
-    lang: null
-  };
-  try {
-    const { document } = parseHTML(html);
-    meta.title = document.title ?? document.querySelector("title")?.textContent?.trim() ?? null;
-    meta.lang = document.documentElement?.getAttribute("lang") ?? null;
-    const metas = [...document.querySelectorAll("meta")];
-    for (const m of metas) {
-      const name = m.getAttribute("name")?.toLowerCase() ?? "";
-      const prop = m.getAttribute("property")?.toLowerCase() ?? "";
-      const content = m.getAttribute("content") ?? "";
-      if (name === "description") meta.description = content;
-      if (name === "keywords") meta.keywords = content;
-      if (name === "robots") meta.robots = content;
-      if (name === "twitter:card") meta.twitterCard = content;
-      if (name === "twitter:title") meta.twitterTitle = content;
-      if (name === "twitter:description") meta.twitterDescription = content;
-      if (name === "twitter:image") meta.twitterImage = content;
-      if (name === "twitter:site") meta.twitterSite = content;
-      if (name === "theme-color") meta.themeColor = content;
-      if (prop === "og:title") meta.ogTitle = content;
-      if (prop === "og:description") meta.ogDescription = content;
-      if (prop === "og:image") meta.ogImage = content;
-      if (prop === "og:url") meta.ogUrl = content;
-      if (prop === "og:type") meta.ogType = content;
-      if (prop === "og:site_name") meta.ogSiteName = content;
-      if (prop === "og:locale") meta.ogLocale = content;
+        return { matches, count: matches.length, method: 'xpath' };
     }
-    const canonical = document.querySelector('link[rel="canonical"]');
-    if (canonical) meta.canonical = canonical.getAttribute("href") ?? null;
-    const faviconEl = document.querySelector('link[rel="icon"], link[rel="shortcut icon"]');
-    if (faviconEl) meta.favicon = faviconEl.getAttribute("href") ?? null;
-    const hreflangEls = [...document.querySelectorAll('link[rel="alternate"][hreflang]')];
-    for (const el of hreflangEls) {
-      const lang = el.getAttribute("hreflang") ?? "";
-      const href = el.getAttribute("href") ?? "";
-      if (lang && href) meta.hreflang.push({ lang, href });
+    catch (e) {
+        return { matches: [], count: 0, method: 'xpath-error' };
     }
-    const jsonLdScripts = [...document.querySelectorAll('script[type="application/ld+json"]')];
-    for (const s of jsonLdScripts) {
-      try {
-        const parsed = JSON.parse(s.textContent ?? "{}");
-        meta.jsonLd.push(parsed);
-        const types = Array.isArray(parsed) ? parsed.map((p) => p["@type"]).flat() : [parsed["@type"]];
-        meta.schemaTypes.push(...types.filter(Boolean));
-      } catch {
-      }
+}
+export function extractByRegex(text, pattern, flags) {
+    try {
+        const re = new RegExp(pattern, (flags ?? 'g').includes('g') ? flags : (flags ?? '') + 'g');
+        const matches = [];
+        let m;
+        while ((m = re.exec(text)) !== null) {
+            matches.push(m[1] ?? m[0]);
+            if (!re.global)
+                break;
+        }
+        return { matches, count: matches.length, method: 'regex' };
+    }
+    catch {
+        return { matches: [], count: 0, method: 'regex-error' };
     }
-  } catch {
-  }
-  return meta;
 }
-function extractStructuredProduct(jsonLd) {
-  for (const item of jsonLd) {
+export function extractTables(html) {
     try {
-      const type = item["@type"];
-      const isProduct = Array.isArray(type) ? type.some((t) => typeof t === "string" && t.toLowerCase() === "product") : typeof type === "string" && type.toLowerCase() === "product";
-      if (!isProduct) continue;
-      const i = item;
-      const offers = Array.isArray(i.offers) ? i.offers[0] : i.offers;
-      const rawPrice = offers?.price;
-      const price = rawPrice != null ? parseFloat(rawPrice) : NaN;
-      const rawRating = i.aggregateRating?.ratingValue;
-      const rating = rawRating != null ? parseFloat(rawRating) : NaN;
-      const rawReviewCount = i.aggregateRating?.reviewCount;
-      const reviewCount = rawReviewCount != null ? parseInt(rawReviewCount, 10) : NaN;
-      const rawAvailability = offers?.availability;
-      const availability = rawAvailability ? rawAvailability.replace(/^https?:\/\/schema\.org\//, "") : null;
-      const rawImage = i.image;
-      const image = Array.isArray(rawImage) ? rawImage[0] ?? null : rawImage ?? null;
-      const brand = i.brand?.name ?? i.brand ?? null;
-      return {
-        name: i.name ?? null,
-        description: i.description ?? null,
-        brand: typeof brand === "string" ? brand : null,
-        sku: i.sku ?? null,
-        price: isNaN(price) ? null : price,
-        priceCurrency: offers?.priceCurrency ?? null,
-        availability,
-        rating: isNaN(rating) ? null : rating,
-        reviewCount: isNaN(reviewCount) ? null : reviewCount,
-        image,
-        url: i.url ?? null
-      };
-    } catch {
+        const { document } = parseHTML(html);
+        const tables = Array.from(document.querySelectorAll('table'));
+        return tables.map((table) => {
+            const headers = [];
+            const rows = [];
+            const ths = [...table.querySelectorAll('thead th, thead td, tr:first-child th')];
+            if (ths.length > 0) {
+                ths.forEach((th, i) => headers.push(th.textContent?.trim() || `col${i + 1}`));
+            }
+            const trs = [...table.querySelectorAll('tbody tr, tr')];
+            for (const tr of trs) {
+                const cells = [...tr.querySelectorAll('td, th')];
+                if (cells.length === 0)
+                    continue;
+                if (headers.length === 0) {
+                    cells.forEach((_, i) => headers.push(`col${i + 1}`));
+                }
+                const row = {};
+                cells.forEach((cell, i) => {
+                    const key = headers[i] || `col${i + 1}`;
+                    row[key] = cell.textContent?.trim() ?? '';
+                });
+                if (Object.values(row).some(v => v))
+                    rows.push(row);
+            }
+            return rows;
+        }).filter(t => t.length > 0);
+    }
+    catch {
+        return [];
     }
-  }
-  return null;
 }
-function extractImages(html, baseUrl) {
-  try {
-    const { document } = parseHTML(html);
-    const base = new URL(baseUrl);
-    return Array.from(document.querySelectorAll("img")).map((img) => {
-      const src = img.getAttribute("src") ?? img.getAttribute("data-src") ?? "";
-      if (!src) return null;
-      try {
-        return {
-          src: new URL(src, base).href,
-          alt: img.getAttribute("alt") ?? "",
-          width: img.getAttribute("width") ? parseInt(img.getAttribute("width")) : void 0,
-          height: img.getAttribute("height") ? parseInt(img.getAttribute("height")) : void 0
-        };
-      } catch {
-        return null;
-      }
-    }).filter((x) => x !== null);
-  } catch {
-    return [];
-  }
+export function extractMetadata(html, url) {
+    const meta = {
+        title: null, description: null, keywords: null,
+        ogTitle: null, ogDescription: null, ogImage: null, ogUrl: null, ogType: null,
+        ogSiteName: null, ogLocale: null,
+        twitterCard: null, twitterTitle: null, twitterDescription: null,
+        twitterImage: null, twitterSite: null,
+        canonical: null, robots: null,
+        hreflang: [], themeColor: null, favicon: null,
+        jsonLd: [], schemaTypes: [], lang: null,
+    };
+    try {
+        const { document } = parseHTML(html);
+        meta.title = document.title ?? document.querySelector('title')?.textContent?.trim() ?? null;
+        meta.lang = document.documentElement?.getAttribute('lang') ?? null;
+        const metas = [...document.querySelectorAll('meta')];
+        for (const m of metas) {
+            const name = m.getAttribute('name')?.toLowerCase() ?? '';
+            const prop = m.getAttribute('property')?.toLowerCase() ?? '';
+            const content = m.getAttribute('content') ?? '';
+            if (name === 'description')
+                meta.description = content;
+            if (name === 'keywords')
+                meta.keywords = content;
+            if (name === 'robots')
+                meta.robots = content;
+            if (name === 'twitter:card')
+                meta.twitterCard = content;
+            if (name === 'twitter:title')
+                meta.twitterTitle = content;
+            if (name === 'twitter:description')
+                meta.twitterDescription = content;
+            if (name === 'twitter:image')
+                meta.twitterImage = content;
+            if (name === 'twitter:site')
+                meta.twitterSite = content;
+            if (name === 'theme-color')
+                meta.themeColor = content;
+            if (prop === 'og:title')
+                meta.ogTitle = content;
+            if (prop === 'og:description')
+                meta.ogDescription = content;
+            if (prop === 'og:image')
+                meta.ogImage = content;
+            if (prop === 'og:url')
+                meta.ogUrl = content;
+            if (prop === 'og:type')
+                meta.ogType = content;
+            if (prop === 'og:site_name')
+                meta.ogSiteName = content;
+            if (prop === 'og:locale')
+                meta.ogLocale = content;
+        }
+        const canonical = document.querySelector('link[rel="canonical"]');
+        if (canonical)
+            meta.canonical = canonical.getAttribute('href') ?? null;
+        const faviconEl = document.querySelector('link[rel="icon"], link[rel="shortcut icon"]');
+        if (faviconEl)
+            meta.favicon = faviconEl.getAttribute('href') ?? null;
+        const hreflangEls = [...document.querySelectorAll('link[rel="alternate"][hreflang]')];
+        for (const el of hreflangEls) {
+            const lang = el.getAttribute('hreflang') ?? '';
+            const href = el.getAttribute('href') ?? '';
+            if (lang && href)
+                meta.hreflang.push({ lang, href });
+        }
+        const jsonLdScripts = [...document.querySelectorAll('script[type="application/ld+json"]')];
+        for (const s of jsonLdScripts) {
+            try {
+                const parsed = JSON.parse(s.textContent ?? '{}');
+                meta.jsonLd.push(parsed);
+                const types = Array.isArray(parsed) ? parsed.map((p) => p['@type']).flat() : [parsed['@type']];
+                meta.schemaTypes.push(...types.filter(Boolean));
+            }
+            catch { /* skip invalid JSON-LD */ }
+        }
+    }
+    catch { /* non-fatal */ }
+    return meta;
 }
-function extractMediaFiles(html, baseUrl) {
-  const IMAGE_EXT = /\.(png|jpe?g|gif|webp|svg|ico|avif)(\?.*)?$/i;
-  const VIDEO_EXT = /\.(mp4|webm|ogv|mov|avi)(\?.*)?$/i;
-  const AUDIO_EXT = /\.(mp3|ogg|wav|flac|aac)(\?.*)?$/i;
-  const DOC_EXT = /\.(pdf|docx?|xlsx?|pptx?|csv|zip)(\?.*)?$/i;
-  function classify(url, tag) {
-    if (tag === "img" || IMAGE_EXT.test(url)) return "image";
-    if (tag === "video" || VIDEO_EXT.test(url)) return "video";
-    if (tag === "audio" || AUDIO_EXT.test(url)) return "audio";
-    if (DOC_EXT.test(url)) return "document";
-    return "document";
-  }
-  try {
-    const { document } = parseHTML(html);
-    const base = new URL(baseUrl);
-    const results = [];
-    const seen = /* @__PURE__ */ new Set();
-    const selectors = ["img[src],video[src],audio[src],source[src]", "src"];
-    const attrPairs = [
-      ["img", "src"],
-      ["video", "src"],
-      ["audio", "src"],
-      ["source", "src"],
-      ["a", "href"]
-    ];
-    for (const [sel, attr] of attrPairs) {
-      for (const el of [...document.querySelectorAll(sel)]) {
-        const raw = el.getAttribute(attr) ?? "";
-        if (!raw) continue;
+export function extractStructuredProduct(jsonLd) {
+    for (const item of jsonLd) {
         try {
-          const resolved = new URL(raw, base).href;
-          if (seen.has(resolved)) continue;
-          if (sel === "a" && !DOC_EXT.test(resolved) && !IMAGE_EXT.test(resolved) && !VIDEO_EXT.test(resolved) && !AUDIO_EXT.test(resolved)) continue;
-          seen.add(resolved);
-          results.push({ url: resolved, type: classify(resolved, sel), tag: sel });
-        } catch {
+            const type = item['@type'];
+            const isProduct = Array.isArray(type)
+                ? type.some((t) => typeof t === 'string' && t.toLowerCase() === 'product')
+                : typeof type === 'string' && type.toLowerCase() === 'product';
+            if (!isProduct)
+                continue;
+            const i = item;
+            const offers = Array.isArray(i.offers) ? i.offers[0] : i.offers;
+            const rawPrice = offers?.price;
+            const price = rawPrice != null ? parseFloat(rawPrice) : NaN;
+            const rawRating = i.aggregateRating?.ratingValue;
+            const rating = rawRating != null ? parseFloat(rawRating) : NaN;
+            const rawReviewCount = i.aggregateRating?.reviewCount;
+            const reviewCount = rawReviewCount != null ? parseInt(rawReviewCount, 10) : NaN;
+            const rawAvailability = offers?.availability;
+            const availability = rawAvailability
+                ? rawAvailability.replace(/^https?:\/\/schema\.org\//, '')
+                : null;
+            const rawImage = i.image;
+            const image = Array.isArray(rawImage) ? (rawImage[0] ?? null) : (rawImage ?? null);
+            const brand = i.brand?.name ?? i.brand ?? null;
+            return {
+                name: i.name ?? null,
+                description: i.description ?? null,
+                brand: typeof brand === 'string' ? brand : null,
+                sku: i.sku ?? null,
+                price: isNaN(price) ? null : price,
+                priceCurrency: offers?.priceCurrency ?? null,
+                availability,
+                rating: isNaN(rating) ? null : rating,
+                reviewCount: isNaN(reviewCount) ? null : reviewCount,
+                image,
+                url: i.url ?? null,
+            };
+        }
+        catch { /* skip malformed item */ }
+    }
+    return null;
+}
+export function extractImages(html, baseUrl) {
+    try {
+        const { document } = parseHTML(html);
+        const base = new URL(baseUrl);
+        return Array.from(document.querySelectorAll('img'))
+            .map((img) => {
+            const src = img.getAttribute('src') ?? img.getAttribute('data-src') ?? '';
+            if (!src)
+                return null;
+            try {
+                return {
+                    src: new URL(src, base).href,
+                    alt: (img.getAttribute('alt') ?? ''),
+                    width: img.getAttribute('width') ? parseInt(img.getAttribute('width')) : undefined,
+                    height: img.getAttribute('height') ? parseInt(img.getAttribute('height')) : undefined,
+                };
+            }
+            catch {
+                return null;
+            }
+        })
+            .filter((x) => x !== null);
+    }
+    catch {
+        return [];
+    }
+}
+export function extractMediaFiles(html, baseUrl) {
+    const IMAGE_EXT = /\.(png|jpe?g|gif|webp|svg|ico|avif)(\?.*)?$/i;
+    const VIDEO_EXT = /\.(mp4|webm|ogv|mov|avi)(\?.*)?$/i;
+    const AUDIO_EXT = /\.(mp3|ogg|wav|flac|aac)(\?.*)?$/i;
+    const DOC_EXT = /\.(pdf|docx?|xlsx?|pptx?|csv|zip)(\?.*)?$/i;
+    function classify(url, tag) {
+        if (tag === 'img' || IMAGE_EXT.test(url))
+            return 'image';
+        if (tag === 'video' || VIDEO_EXT.test(url))
+            return 'video';
+        if (tag === 'audio' || AUDIO_EXT.test(url))
+            return 'audio';
+        if (DOC_EXT.test(url))
+            return 'document';
+        return 'document';
+    }
+    try {
+        const { document } = parseHTML(html);
+        const base = new URL(baseUrl);
+        const results = [];
+        const seen = new Set();
+        const selectors = ['img[src],video[src],audio[src],source[src]', 'src'];
+        const attrPairs = [
+            ['img', 'src'], ['video', 'src'], ['audio', 'src'], ['source', 'src'],
+            ['a', 'href'],
+        ];
+        for (const [sel, attr] of attrPairs) {
+            for (const el of [...document.querySelectorAll(sel)]) {
+                const raw = el.getAttribute(attr) ?? '';
+                if (!raw)
+                    continue;
+                try {
+                    const resolved = new URL(raw, base).href;
+                    if (seen.has(resolved))
+                        continue;
+                    if (sel === 'a' && !DOC_EXT.test(resolved) && !IMAGE_EXT.test(resolved) && !VIDEO_EXT.test(resolved) && !AUDIO_EXT.test(resolved))
+                        continue;
+                    seen.add(resolved);
+                    results.push({ url: resolved, type: classify(resolved, sel), tag: sel });
+                }
+                catch { /* skip */ }
+            }
         }
-      }
+        return results;
+    }
+    catch {
+        return [];
     }
-    return results;
-  } catch {
-    return [];
-  }
 }
-export {
-  extractByCssSelector,
-  extractByRegex,
-  extractByXPath,
-  extractImages,
-  extractMediaFiles,
-  extractMetadata,
-  extractStructuredProduct,
-  extractTables
-};
-//# sourceMappingURL=advanced.js.map

package/dist/extractors/artifact-envelope.js

@@ -1,63 +1,62 @@
-import crypto from "node:crypto";
-function sha256(input) {
-  return crypto.createHash("sha256").update(input).digest("hex");
+import crypto from 'node:crypto';
+export function sha256(input) {
+    return crypto.createHash('sha256').update(input).digest('hex');
 }
-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 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 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 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 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 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 ?? [],
+    };
 }
-export {
-  buildCrawlArtifactEnvelope,
-  buildFitMarkdown,
-  sha256,
-  summarizeStructuredElements
-};
-//# sourceMappingURL=artifact-envelope.js.map