@adobe/spacecat-shared-html-analyzer 1.2.3 → 1.2.5

package/.releaserc.cjs

@@ -1,8 +1,12 @@
 module.exports = {
   extends: "semantic-release-monorepo",
   plugins: [
-    "@semantic-release/commit-analyzer",
-    "@semantic-release/release-notes-generator",
+    ["@semantic-release/commit-analyzer", {
+      "preset": "conventionalcommits",
+    }],
+    ["@semantic-release/release-notes-generator", {
+      "preset": "conventionalcommits",
+    }],
     ["@semantic-release/changelog", {
       "changelogFile": "CHANGELOG.md",
     }],

package/CHANGELOG.md

@@ -1,3 +1,16 @@
+## [@adobe/spacecat-shared-html-analyzer-v1.2.5](https://github.com/adobe/spacecat-shared/compare/@adobe/spacecat-shared-html-analyzer-v1.2.4...@adobe/spacecat-shared-html-analyzer-v1.2.5) (2026-02-23)
+
+### Bug Fixes
+
+* filtering out digit access related accessibility divs ([#1366](https://github.com/adobe/spacecat-shared/issues/1366)) ([3ad20a5](https://github.com/adobe/spacecat-shared/commit/3ad20a582f6a9d7ec21050fd56afcbeb16a6f096))
+
+# [@adobe/spacecat-shared-html-analyzer-v1.2.4](https://github.com/adobe/spacecat-shared/compare/@adobe/spacecat-shared-html-analyzer-v1.2.3...@adobe/spacecat-shared-html-analyzer-v1.2.4) (2026-02-05)
+
+
+### Bug Fixes
+
+* rcv adding new cookie selectors ([#1323](https://github.com/adobe/spacecat-shared/issues/1323)) ([9c566e6](https://github.com/adobe/spacecat-shared/commit/9c566e6dccac85db2da4915dfa143c6bb7d43f90))
+
 # [@adobe/spacecat-shared-html-analyzer-v1.2.3](https://github.com/adobe/spacecat-shared/compare/@adobe/spacecat-shared-html-analyzer-v1.2.2...@adobe/spacecat-shared-html-analyzer-v1.2.3) (2026-02-04)
 
 

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@adobe/spacecat-shared-html-analyzer",
-  "version": "1.2.3",
+  "version": "1.2.5",
   "description": "Analyze HTML content visibility for AI crawlers and citations - compare static HTML vs fully rendered content",
   "type": "module",
   "engines": {

package/src/html-filter.js

@@ -59,6 +59,7 @@ const COOKIE_BANNER_CLASS_SELECTORS = [
   '.cookie-modal', '.privacy-modal', '.consent-modal', '.gdpr-modal',
   '.cookie-overlay', '.privacy-overlay', '.consent-overlay', '.gdpr-overlay',
   '[class*="syrenis-cookie"]',
+  '.tc-privacy-wrapper',
 ];
 
 const COOKIE_BANNER_ID_SELECTORS = [
@@ -67,6 +68,7 @@ const COOKIE_BANNER_ID_SELECTORS = [
   '#cookie-bar', '#privacy-bar', '#consent-bar', '#gdpr-bar', '#cookiemgmt',
   '#cookie-popup', '#privacy-popup', '#consent-popup', '#gdpr-popup',
   '#onetrust-consent-sdk', '#onetrust-banner-sdk',
+  '#tc-privacy-wrapper',
 ];
 
 const COOKIE_BANNER_ARIA_SELECTORS = [
@@ -80,6 +82,10 @@ const COOKIE_BANNER_ARIA_SELECTORS = [
   '[aria-describedby*="privacy" i]',
 ];
 
+const ACCESSIBILITY_SELECTORS = [
+  '#digiAccess',
+];
+
 /**
  * Validates if an element is likely a cookie banner based on text content
  * Optimized: Set lookup + early exit for common keywords (3x faster)
@@ -119,6 +125,13 @@ function removeCookieBanners(element) {
   });
 }
 
+function removeAccessibilityElements(element) {
+  const elements = element.querySelectorAll(ACCESSIBILITY_SELECTORS);
+  elements.forEach((el) => {
+    el.remove();
+  });
+}
+
 /**
  * Remove navigation and footer elements from DOM element (browser environment)
  * For Chrome extension DOM manipulation use cases
@@ -165,6 +178,12 @@ function removeCookieBannersCheerio($) {
   });
 }
 
+function removeAccessibilityElementsCheerio($) {
+  ACCESSIBILITY_SELECTORS.forEach((selector) => {
+    $(selector).remove();
+  });
+}
+
 /**
  * Remove navigation and footer elements (Node.js environment)
  * Optimized: single cheerio query instead of 35 separate queries (35x performance improvement)
@@ -247,6 +266,9 @@ function filterHtmlBrowser(htmlContent, ignoreNavFooter, returnText, includeNosc
   // Remove consent banners with intelligent detection
   removeCookieBanners(documentElement);
 
+  // Remove accessibility elements
+  removeAccessibilityElements(documentElement);
+
   // Conditionally remove navigation and footer elements
   if (ignoreNavFooter) {
     filterNavigationAndFooterBrowser(documentElement);
@@ -323,6 +345,9 @@ async function filterHtmlNode(htmlContent, ignoreNavFooter, returnText, includeN
   // Remove cookie banners with comprehensive detection
   removeCookieBannersCheerio($);
 
+  // Remove accessibility elements
+  removeAccessibilityElementsCheerio($);
+
   // Conditionally remove navigation and footer elements
   if (ignoreNavFooter) {
     filterNavigationAndFooterCheerio($);

package/test/index.test.js

@@ -183,6 +183,51 @@ describe('HTML Visibility Analyzer', () => {
       expect(text).to.include('Footer');
     });
 
+    it('should remove accessibility widget elements', async () => {
+      const html = `<html><body>
+        <h1>Title</h1>
+        <div id="digiAccess">Accessibility Widget</div>
+        <div id="dAopener">Accessibility Opener</div>
+        <p>Content</p>
+      </body></html>`;
+
+      const text = await stripTagsToText(html, true);
+
+      expect(text).to.include('Title');
+      expect(text).to.include('Content');
+      expect(text).to.not.include('Accessibility Widget');
+      expect(text).to.include('Accessibility Opener');
+    });
+
+    it('should remove cookie banner when selector matches and content indicates consent', async () => {
+      const html = `<html><body>
+        <h1>Title</h1>
+        <div id="onetrust-consent-sdk">We use cookies. Manage consent preferences.</div>
+        <p>Content</p>
+      </body></html>`;
+
+      const text = await stripTagsToText(html, true);
+
+      expect(text).to.include('Title');
+      expect(text).to.include('Content');
+      expect(text).to.not.include('We use cookies');
+      expect(text).to.not.include('Manage consent preferences');
+    });
+
+    it('should not remove cookie-banner selectors when content does not indicate consent', async () => {
+      const html = `<html><body>
+        <h1>Title</h1>
+        <div id="onetrust-consent-sdk">Just a container with neutral text.</div>
+        <p>Content</p>
+      </body></html>`;
+
+      const text = await stripTagsToText(html, true);
+
+      expect(text).to.include('Title');
+      expect(text).to.include('Content');
+      expect(text).to.include('Just a container with neutral text.');
+    });
+
     it('should remove noscript elements by default', async () => {
       const html = '<html><body><h1>Title</h1><noscript>Please enable JavaScript</noscript><p>Content</p></body></html>';
       const text = await stripTagsToText(html);