aeorank 3.0.0 → 3.0.2

package/dist/browser.d.ts

@@ -298,6 +298,12 @@ interface RawDataSummary {
     crawl_discovered: number;
     crawl_fetched: number;
     crawl_skipped: number;
+    citation_ready_sentences: number;
+    answer_first_ratio: number;
+    evidence_citations_avg: number;
+    entity_disambiguation_ratio: number;
+    extraction_friction_avg: number;
+    image_figure_ratio: number;
 }
 /**
  * Fetches all site data in parallel with HTTPS/HTTP fallback.

package/dist/browser.js

@@ -1824,7 +1824,83 @@ function extractRawDataSummary(data) {
     // Full-crawl stats
     crawl_discovered: data.crawlStats?.discovered ?? 0,
     crawl_fetched: data.crawlStats?.fetched ?? 0,
-    crawl_skipped: data.crawlStats?.skipped ?? 0
+    crawl_skipped: data.crawlStats?.skipped ?? 0,
+    // V2 criteria fields
+    citation_ready_sentences: (() => {
+      const combinedText = text + " " + (data.blogSample?.map((p) => p.text.replace(/<[^>]*>/g, " ")).join(" ") || "");
+      return (combinedText.match(/\b\w+\s+(is\s+(?:a|an)\s|refers\s+to|defined\s+as)\b/gi) || []).length;
+    })(),
+    answer_first_ratio: (() => {
+      const pages = [html, ...data.blogSample?.map((p) => p.text) || []];
+      let answerFirst = 0;
+      for (const pageHtml of pages) {
+        const bodyMatch = pageHtml.match(/<body[^>]*>([\s\S]*)/i);
+        const bodyHtml = bodyMatch ? bodyMatch[1] : pageHtml;
+        const earlyParas = bodyHtml.match(/<p[^>]*>([\s\S]*?)<\/p>/gi)?.slice(0, 5) || [];
+        for (const p of earlyParas) {
+          const pText = p.replace(/<[^>]*>/g, "").trim();
+          const wc = pText.split(/\s+/).length;
+          if (wc >= 40 && wc <= 80) {
+            answerFirst++;
+            break;
+          }
+        }
+      }
+      return pages.length > 0 ? Math.round(answerFirst / pages.length * 100) : 0;
+    })(),
+    evidence_citations_avg: (() => {
+      const allHtml = html + "\n" + (data.blogSample?.map((p) => p.text).join("\n") || "");
+      const paragraphs = allHtml.match(/<p[^>]*>[\s\S]*?<\/p>/gi) || [];
+      let citations = 0;
+      const domainLower = data.domain.replace(/^www\./, "").toLowerCase();
+      for (const p of paragraphs) {
+        const links = p.match(/<a[^>]*href=["'](https?:\/\/[^"']+)["'][^>]*>/gi) || [];
+        for (const link of links) {
+          const href = link.match(/href=["'](https?:\/\/[^"']+)["']/i);
+          if (href) {
+            try {
+              const ld = new URL(href[1]).hostname.replace(/^www\./, "").toLowerCase();
+              if (ld !== domainLower) citations++;
+            } catch {
+            }
+          }
+        }
+      }
+      const pageCount = Math.max(1, 1 + (data.blogSample?.length ?? 0));
+      return Math.round(citations / pageCount * 10) / 10;
+    })(),
+    entity_disambiguation_ratio: (() => {
+      const pages = [html, ...data.blogSample?.map((p) => p.text) || []];
+      let defined = 0;
+      for (const pageHtml of pages) {
+        const h1Match = pageHtml.match(/<h1[^>]*>([\s\S]*?)<\/h1>/i);
+        if (!h1Match) continue;
+        const h1Text = h1Match[1].replace(/<[^>]*>/g, "").trim();
+        const h1Words = h1Text.split(/\s+/).filter((w) => w.length > 3);
+        const primaryNoun = h1Words.sort((a, b) => b.length - a.length)[0] || "";
+        if (!primaryNoun) continue;
+        const pageText = pageHtml.replace(/<[^>]*>/g, " ").replace(/\s+/g, " ").slice(0, 500);
+        if (new RegExp(`\\b${primaryNoun.replace(/[.*+?^${}()|[\]\\]/g, "\\$&")}\\b[^.]*\\b(is|refers|defined|means)\\b`, "i").test(pageText)) {
+          defined++;
+        }
+      }
+      return pages.length > 0 ? Math.round(defined / pages.length * 100) : 0;
+    })(),
+    extraction_friction_avg: (() => {
+      const combinedText = text + " " + (data.blogSample?.map((p) => p.text.replace(/<[^>]*>/g, " ")).join(" ") || "");
+      const sentences = combinedText.split(/[.!?]+/).filter((s) => s.trim().length > 5);
+      if (sentences.length === 0) return 0;
+      const totalWords = sentences.reduce((sum, s) => sum + s.trim().split(/\s+/).length, 0);
+      return Math.round(totalWords / sentences.length * 10) / 10;
+    })(),
+    image_figure_ratio: (() => {
+      const combinedHtml = html + "\n" + (data.blogSample?.map((p) => p.text).join("\n") || "");
+      const allImages = combinedHtml.match(/<img\s[^>]*>/gi) || [];
+      if (allImages.length === 0) return 0;
+      const figureBlocks = combinedHtml.match(/<figure[\s\S]*?<\/figure>/gi) || [];
+      const figuresWithCaption = figureBlocks.filter((f) => /<figcaption/i.test(f));
+      return Math.round(figuresWithCaption.length / allImages.length * 100);
+    })()
   };
 }
 function getPageTopicText(html) {
@@ -4364,8 +4440,8 @@ function extractLinksWithAnchors(html, sourceUrl, domain) {
     if (href.startsWith("//")) {
       fullUrl = `https:${href}`;
     } else if (href.startsWith("/")) {
-      if (href === "/" || href.startsWith("/#")) continue;
-      fullUrl = `https://${domain}${href}`;
+      if (href.startsWith("/#")) continue;
+      fullUrl = href === "/" ? `https://${domain}` : `https://${domain}${href}`;
     } else if (href.startsWith("http")) {
       fullUrl = href;
     } else if (href.startsWith("#") || href.startsWith("?") || href.startsWith("mailto:") || href.startsWith("tel:") || href.startsWith("javascript:")) {
@@ -4379,7 +4455,7 @@ function extractLinksWithAnchors(html, sourceUrl, domain) {
       if (linkDomain !== cleanDomain) continue;
       parsed.hash = "";
       const path = parsed.pathname;
-      if (path === "/" || path === "") continue;
+      if (path === "") continue;
       if (RESOURCE_EXTENSIONS.test(path)) continue;
       if (SKIP_PATH_PATTERNS.test(path)) continue;
       const normalized = normalizeUrl(fullUrl);