mcp-scraper 0.3.16 → 0.3.17

package/dist/bin/mcp-stdio-server.js

@@ -2,11 +2,11 @@
 import {
   HttpMcpToolExecutor,
   buildPaaExtractorMcpServer
-} from "../chunk-7XARJHS6.js";
+} from "../chunk-N33YEY6W.js";
 import "../chunk-M2S27J6Z.js";
 import "../chunk-WN7PBKMV.js";
 import "../chunk-LFATOGDF.js";
-import "../chunk-SOMBZK3V.js";
+import "../chunk-PVWWY5NV.js";
 
 // bin/mcp-stdio-server.ts
 import { readFileSync } from "fs";

package/dist/{chunk-7XARJHS6.js → chunk-N33YEY6W.js}

@@ -11,7 +11,7 @@ import {
 } from "./chunk-LFATOGDF.js";
 import {
   PACKAGE_VERSION
-} from "./chunk-SOMBZK3V.js";
+} from "./chunk-PVWWY5NV.js";
 
 // src/harvest-timeout.ts
 var VERCEL_FUNCTION_MAX_MS = 3e5;
@@ -256,6 +256,99 @@ function buildLinkGraph(pages, startUrl) {
   return { edges, metrics };
 }
 
+// src/api/seo-link-report.ts
+function registrableDomain(host) {
+  const h = host.replace(/^www\./, "").toLowerCase();
+  const parts = h.split(".");
+  return parts.length <= 2 ? h : parts.slice(-2).join(".");
+}
+function buildLinkReport(edges, metrics, siteUrl) {
+  let siteReg = "";
+  try {
+    siteReg = registrableDomain(new URL(siteUrl).hostname);
+  } catch {
+    siteReg = "";
+  }
+  const internalEdges = edges.filter((e) => e.internal);
+  const pages = metrics.length;
+  const orphans = metrics.filter((m) => m.orphan).length;
+  const brokenInternal = internalEdges.filter((e) => e.targetStatus != null && e.targetStatus >= 400).length;
+  const sumInlinks = metrics.reduce((a, m) => a + m.inlinks, 0);
+  const sumOutlinks = metrics.reduce((a, m) => a + m.outlinksInternal + m.outlinksExternal, 0);
+  const distribution = { zero: 0, oneToTwo: 0, threeToTen: 0, elevenPlus: 0 };
+  for (const m of metrics) {
+    if (m.inlinks === 0) distribution.zero++;
+    else if (m.inlinks <= 2) distribution.oneToTwo++;
+    else if (m.inlinks <= 10) distribution.threeToTen++;
+    else distribution.elevenPlus++;
+  }
+  const topByInlinks = [...metrics].sort((a, b) => b.inlinks - a.inlinks).slice(0, 20).map((m) => ({ url: m.url, inlinks: m.inlinks, outlinksInternal: m.outlinksInternal, outlinksExternal: m.outlinksExternal }));
+  const domMap = /* @__PURE__ */ new Map();
+  let externalTotal = 0;
+  for (const e of edges) {
+    let domain;
+    try {
+      domain = registrableDomain(new URL(e.to).hostname);
+    } catch {
+      continue;
+    }
+    if (!domain || domain === siteReg) continue;
+    externalTotal++;
+    const d = domMap.get(domain) ?? { links: 0, nofollow: 0, pages: /* @__PURE__ */ new Set() };
+    d.links++;
+    if (e.nofollow) d.nofollow++;
+    d.pages.add(e.from);
+    domMap.set(domain, d);
+  }
+  const externalDomains = [...domMap.entries()].map(([domain, d]) => ({ domain, links: d.links, nofollow: d.nofollow, pages: d.pages.size })).sort((a, b) => b.links - a.links);
+  const round = (n) => Math.round(n * 10) / 10;
+  return {
+    summary: {
+      internal: {
+        totalLinks: internalEdges.length,
+        pages,
+        orphans,
+        brokenInternal,
+        avgInlinks: pages ? round(sumInlinks / pages) : 0,
+        avgOutlinks: pages ? round(sumOutlinks / pages) : 0,
+        distribution,
+        topByInlinks
+      },
+      external: {
+        totalLinks: externalTotal,
+        uniqueDomains: externalDomains.length,
+        topDomains: externalDomains.slice(0, 20)
+      }
+    },
+    externalDomains
+  };
+}
+function renderLinkReport(r) {
+  const s = r.summary;
+  const lines = [
+    `## Link analysis`,
+    `**Internal:** ${s.internal.totalLinks} links \xB7 ${s.internal.pages} pages \xB7 avg ${s.internal.avgInlinks} inlinks/page \xB7 ${s.internal.orphans} orphans \xB7 ${s.internal.brokenInternal} broken`,
+    `**External:** ${s.external.totalLinks} links to ${s.external.uniqueDomains} domains`,
+    ``,
+    `### Inlink distribution`,
+    `- 0 inlinks (orphans): ${s.internal.distribution.zero}`,
+    `- 1\u20132: ${s.internal.distribution.oneToTwo}`,
+    `- 3\u201310: ${s.internal.distribution.threeToTen}`,
+    `- 11+: ${s.internal.distribution.elevenPlus}`,
+    ``,
+    `### Top 20 internal pages by inlinks`,
+    `| inlinks | out (int/ext) | URL |`,
+    `|---|---|---|`,
+    ...s.internal.topByInlinks.map((p) => `| ${p.inlinks} | ${p.outlinksInternal}/${p.outlinksExternal} | ${p.url} |`),
+    ``,
+    `### Top 20 external domains by links`,
+    `| links | nofollow | from pages | domain |`,
+    `|---|---|---|---|`,
+    ...s.external.topDomains.map((d) => `| ${d.links} | ${d.nofollow} | ${d.pages} | ${d.domain} |`)
+  ];
+  return lines.join("\n");
+}
+
 // src/api/seo-issues.ts
 var THIN_WORDS = 200;
 var TITLE_MAX = 60;
@@ -564,6 +657,9 @@ function saveBulkSite(siteUrl, pages, seo, imageAudit) {
       `- \`pages.jsonl\` \u2014 per-page SEO fields, one JSON object per page`,
       `- \`links.jsonl\` \u2014 internal/external link edges (anchor, rel, target status)`,
       `- \`link-metrics.jsonl\` \u2014 inlinks, crawl depth, orphan flags per page`,
+      `- \`link-report.md\` \u2014 link analysis: top internal pages by inlinks, top external domains, distribution`,
+      `- \`links-summary.json\` \u2014 link totals + top-20 internal pages + top-20 external domains`,
+      `- \`external-domains.json\` \u2014 every external domain linked to, ordered by link count`,
       ...imageAudit ? [
         `- \`images.jsonl\` \u2014 every image with byte size, format, over-100KB and legacy-format flags`,
         `- \`images-summary.json\` \u2014 image totals (count, total weight, over-100KB, legacy)`
@@ -599,7 +695,13 @@ ${indexRows.join("\n")}`
       writeFileSync(reportFile, seo.reportMd + (imageAudit ? `
 
 ${renderImageSection(imageAudit)}` : ""), "utf8");
-      seoFiles = [pagesJsonl, linksJsonl, metricsJsonl, issuesFile, reportFile];
+      const linkReportFile = join(dir, "link-report.md");
+      const linksSummaryFile = join(dir, "links-summary.json");
+      const externalDomainsFile = join(dir, "external-domains.json");
+      writeFileSync(linkReportFile, renderLinkReport(seo.linkReport), "utf8");
+      writeFileSync(linksSummaryFile, JSON.stringify(seo.linkReport.summary, null, 2), "utf8");
+      writeFileSync(externalDomainsFile, JSON.stringify(seo.linkReport.externalDomains, null, 2), "utf8");
+      seoFiles = [pagesJsonl, linksJsonl, metricsJsonl, issuesFile, reportFile, linkReportFile, linksSummaryFile, externalDomainsFile];
       if (imageAudit) {
         const imagesJsonl = join(dir, "images.jsonl");
         const imagesSummary = join(dir, "images-summary.json");
@@ -1051,6 +1153,7 @@ function buildSeoExport(siteUrl, pages, branding) {
     metrics: [...metrics.values()],
     issues,
     reportMd: renderIssueReport(siteUrl, pages, issues, metrics),
+    linkReport: buildLinkReport(edges, [...metrics.values()], siteUrl),
     branding: branding ?? void 0
   };
 }
@@ -1159,6 +1262,9 @@ async function formatAuditSite(raw, input) {
   const topIssues = Object.entries(seo.issues).sort((a, b) => b[1].count - a[1].count).slice(0, 8).map(([k, v]) => `- \`${k}\` \u2014 ${v.count}`).join("\n");
   const img = imageAudit.summary;
   const imgLine = `${img.unique} images \xB7 ${img.totalSize} total \xB7 ${img.over100kb} over 100 KB \xB7 ${img.legacyFormat} legacy format`;
+  const lr = seo.linkReport.summary;
+  const topExt = lr.external.topDomains.slice(0, 3).map((dm) => `${dm.domain} (${dm.links})`).join(", ") || "\u2014";
+  const linkLine = `${lr.internal.totalLinks} internal / ${lr.external.totalLinks} external links \xB7 ${lr.internal.orphans} orphans \xB7 ${lr.internal.brokenInternal} broken \xB7 avg ${lr.internal.avgInlinks} inlinks/page \xB7 top external: ${topExt}`;
   const durationLine = `**${pages.length} pages** \xB7 ${((d.durationMs ?? 0) / 1e3).toFixed(1)}s`;
   const structuredContent = {
     url: input.url,
@@ -1166,7 +1272,8 @@ async function formatAuditSite(raw, input) {
     durationMs: d.durationMs ?? 0,
     bulkFolder: bulk?.dir ?? null,
     issues: Object.fromEntries(Object.entries(seo.issues).map(([k, v]) => [k, v.count])),
-    images: { unique: img.unique, totalBytes: img.totalBytes, over100kb: img.over100kb, legacyFormat: img.legacyFormat }
+    images: { unique: img.unique, totalBytes: img.totalBytes, over100kb: img.over100kb, legacyFormat: img.legacyFormat },
+    links: { internal: lr.internal.totalLinks, external: lr.external.totalLinks, orphans: lr.internal.orphans, brokenInternal: lr.internal.brokenInternal, externalDomains: lr.external.uniqueDomains }
   };
   const location = bulk ? `
 ## \u{1F4C1} Technical audit saved
@@ -1185,6 +1292,9 @@ Report saving is disabled in this environment. Enable \`MCP_SCRAPER_SAVE_REPORTS
 ## Top issues
 ${topIssues || "_none found_"}`,
     `
+## Links
+${linkLine}`,
+    `
 ## Images
 ${imgLine}`
   ].join("\n");
@@ -2715,6 +2825,13 @@ var AuditSiteOutputSchema = {
     totalBytes: z.number().min(0),
     over100kb: z.number().int().min(0),
     legacyFormat: z.number().int().min(0)
+  }),
+  links: z.object({
+    internal: z.number().int().min(0),
+    external: z.number().int().min(0),
+    orphans: z.number().int().min(0),
+    brokenInternal: z.number().int().min(0),
+    externalDomains: z.number().int().min(0)
   })
 };
 var MapsPlaceIntelOutputSchema = {
@@ -3566,7 +3683,7 @@ function registerPaaExtractorMcpTools(server, executor, options = {}) {
   }, async (input) => formatExtractSite(await executor.extractSite(input), input));
   server.registerTool("audit_site", {
     title: "Technical SEO Audit",
-    description: withReportNote("Run a full technical SEO audit (Screaming-Frog-style) on a public website. Crawls up to maxPages and produces a complete on-page + crawl analysis: per-page title/meta lengths, heading breakdown, indexability and canonical status, schema, an internal link graph with inlinks/orphans/crawl-depth, an issues report, and an image audit (byte size, format, over-100KB and legacy-format flags). Writes everything to a local folder (report.md, issues.json, pages.jsonl, links.jsonl, link-metrics.jsonl, images.jsonl) plus per-page content, and returns a headline summary plus the folder path. Use this when the user wants an audit or analysis; use extract_site when they just want page content."),
+    description: withReportNote("Run a full technical SEO audit (Screaming-Frog-style) on a public website. Crawls up to maxPages and produces a complete on-page + crawl analysis: per-page title/meta lengths, heading breakdown, indexability and canonical status, schema, an internal link graph (inlinks/orphans/crawl-depth), a link analysis (top internal pages by inlinks, top external domains by link count, broken/orphan counts), an issues report, and an image audit (byte size, format, over-100KB and legacy-format flags). Writes everything to a local folder (report.md, issues.json, pages.jsonl, links.jsonl, link-metrics.jsonl, link-report.md, links-summary.json, external-domains.json, images.jsonl) plus per-page content, and returns a headline summary plus the folder path. Use this when the user wants an audit or analysis; use extract_site when they just want page content."),
     inputSchema: AuditSiteInputSchema,
     outputSchema: AuditSiteOutputSchema,
     annotations: liveWebToolAnnotations("Technical SEO Audit")
@@ -3978,6 +4095,8 @@ var HttpMcpToolExecutor = class {
 export {
   harvestTimeoutBudget,
   buildLinkGraph,
+  buildLinkReport,
+  renderLinkReport,
   computeIssues,
   renderIssueReport,
   auditImages,
@@ -3996,4 +4115,4 @@ export {
   registerPaaExtractorMcpTools,
   HttpMcpToolExecutor
 };
-//# sourceMappingURL=chunk-7XARJHS6.js.map
+//# sourceMappingURL=chunk-N33YEY6W.js.map