dembrandt 0.12.2 → 0.12.3

package/lib/extractors/logo.js

@@ -146,6 +146,12 @@ export async function extractLogo(page, url) {
       if (imgSrc.toLowerCase().includes(siteDomain) || altText.includes(siteDomain) || className.includes(siteDomain)) score += 40;
       if (className.includes('logo') || el.id?.toLowerCase().includes('logo')) score += 30;
 
+      // SVG with aria-label="Homepage" directly in a home anchor — strong signal
+      if (el.tagName === 'svg' || el.tagName === 'SVG') {
+        const ariaLabel = (el.getAttribute('aria-label') || '').toLowerCase();
+        if (ariaLabel.includes('home') || ariaLabel.includes('logo')) score += 40;
+      }
+
       if (parentLink) {
         const href = linkHref.toLowerCase();
         if (href === '/' || href === baseUrl || href.endsWith('://' + new URL(baseUrl).hostname + '/') || href.endsWith('://' + new URL(baseUrl).hostname)) {
@@ -156,8 +162,19 @@ export async function extractLogo(page, url) {
       if (rect.top < 200) score += 10;
       if (rect.left < 400) score += 10;
 
-      const width = el.naturalWidth || el.width?.baseVal?.value || rect.width;
-      const height = el.naturalHeight || el.height?.baseVal?.value || rect.height;
+      // Penalize images hosted on a different domain (CDN paths on same domain are fine)
+      if (el.tagName === 'IMG') {
+        try {
+          const srcHost = new URL(el.src).hostname.replace('www.', '')
+          const pageHost = new URL(baseUrl).hostname.replace('www.', '')
+          // If neither is a subdomain/CDN of the other, it's third-party
+          if (!srcHost.endsWith(pageHost) && !pageHost.endsWith(srcHost)) score -= 60
+        } catch {}
+      }
+
+      // For SVGs use rendered rect — baseVal reflects viewBox coordinates, not display size
+      const width = el.tagName === 'IMG' ? (el.naturalWidth || rect.width) : rect.width;
+      const height = el.tagName === 'IMG' ? (el.naturalHeight || rect.height) : rect.height;
       if (width < 20 || height < 20) score -= 30;
       if (width > 600 || height > 400) score -= 40;
       if (altText.length > 50) score -= 30;
@@ -247,7 +264,18 @@ export async function extractLogo(page, url) {
           if (rect.width === 0 || rect.height === 0) return;
 
           const className = (typeof el.className === 'string' ? el.className : el.className.baseVal || '').toLowerCase();
-          const attrs = (className + ' ' + (el.id || '') + ' ' + (el.getAttribute('alt') || '')).toLowerCase();
+          const altText = (el.getAttribute('alt') || '').toLowerCase();
+          const attrs = (className + ' ' + (el.id || '') + ' ' + altText).toLowerCase();
+
+          // Disqualify third-party brand logos: alt like "Notion logo" / "Perplexity logo" / "Figma" where the brand isn't our site
+          // These appear in customer/integration/testimonial sections on marketing pages.
+          const altBrandMatch = altText.match(/^([a-z][a-z0-9\-\.]{1,30}?)(?:\s+(?:logo|icon|brand|wordmark))?$/i);
+          if (altBrandMatch) {
+            const altBrand = altBrandMatch[1].replace(/[\s\-\.]/g, '').toLowerCase();
+            if (altBrand && altBrand !== 'logo' && altBrand !== 'brand' && altBrand !== 'icon' && altBrand !== siteDomain && !altBrand.includes(siteDomain) && !siteDomain.includes(altBrand)) {
+              return; // third-party brand, skip entirely
+            }
+          }
 
           let qualifies = attrs.includes('logo') || attrs.includes('brand');
 
@@ -257,6 +285,9 @@ export async function extractLogo(page, url) {
               const href = use.getAttribute('href') || use.getAttribute('xlink:href') || '';
               if (href.toLowerCase().includes('logo') || href.toLowerCase().includes('brand')) { qualifies = true; break; }
             }
+            // aria-label="Homepage" or similar on the SVG itself
+            const ariaLabel = (el.getAttribute('aria-label') || '').toLowerCase();
+            if (ariaLabel.includes('home') || ariaLabel.includes('logo')) qualifies = true;
           }
 
           if (!qualifies) {

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "dembrandt",
-  "version": "0.12.2",
+  "version": "0.12.3",
   "description": "Extract design tokens and publicly visible CSS information from any website",
   "mcpName": "io.github.dembrandt/dembrandt",
   "main": "index.js",