fix image

frontend/src/lib/sanitization.ts

@@ -153,7 +153,8 @@ function createDOMPurifyConfig(config: SanitizationConfig) {
   const domPurifyConfig: DOMPurify.Config = {
     ALLOWED_TAGS: allowedTags,
     ALLOWED_ATTR: uniqueAttributes,
-    ALLOWED_URI_REGEXP: /^(?:(?:https?|#|\/):?\/?)[\w.\-#/?=&%]+$/i,
+    // More permissive URL regex to allow complex query strings and tokens
+    ALLOWED_URI_REGEXP: /^(?:(?:https?|data|#|\/):)?[\s\S]*$/i,
     ALLOW_UNKNOWN_PROTOCOLS: false,
     SANITIZE_DOM: true,
     KEEP_CONTENT: true,
@@ -193,30 +194,52 @@ function preprocessFigureTags(html: string): string {
     const doc = parser.parseFromString(html, 'text/html');
     const figures = doc.querySelectorAll('figure');
 
-    figures.forEach(figure => {
+    figures.forEach((figure) => {
-      // Find img tags within the figure
+      // Find img tags anywhere within the figure (deep search)
       const images = figure.querySelectorAll('img');
 
       if (images.length > 0) {
         // Extract the first image
         const img = images[0];
 
-        // Check if there's a figcaption to preserve as alt text
+        // Get the src attribute - it might be in the src attribute or data-src
-        const figcaption = figure.querySelector('figcaption');
+        const imgSrc = img.getAttribute('src') || img.getAttribute('data-src') || img.src || '';
-        if (figcaption && !img.hasAttribute('alt')) {
+
-          const captionText = figcaption.textContent?.trim();
+        if (!imgSrc || imgSrc.trim() === '') {
-          if (captionText) {
+          figure.remove();
-            img.setAttribute('alt', captionText);
+          return;
+        }
+
+        // Create a clean img element with just the essential attributes
+        const cleanImg = doc.createElement('img');
+        cleanImg.setAttribute('src', imgSrc);
+
+        // Preserve alt text
+        const existingAlt = img.getAttribute('alt') || img.alt;
+        if (existingAlt) {
+          cleanImg.setAttribute('alt', existingAlt);
+        } else {
+          // Check if there's a figcaption to use as alt text
+          const figcaption = figure.querySelector('figcaption');
+          if (figcaption) {
+            const captionText = figcaption.textContent?.trim();
+            if (captionText) {
+              cleanImg.setAttribute('alt', captionText);
+            }
           }
         }
 
-        // Replace the figure element with just the img
+        // Preserve other useful attributes if they exist
-        figure.replaceWith(img.cloneNode(true));
+        const width = img.getAttribute('width') || img.width;
-        debug.log('Extracted image from figure tag:', img.src);
+        const height = img.getAttribute('height') || img.height;
+        if (width) cleanImg.setAttribute('width', width.toString());
+        if (height) cleanImg.setAttribute('height', height.toString());
+
+        // Replace the figure element with just the clean img
+        figure.replaceWith(cleanImg);
       } else {
         // No images in figure, remove it entirely
         figure.remove();
-        debug.log('Removed figure tag without images');
       }
     });
 
@@ -300,8 +323,10 @@ export function sanitizeHtmlSync(html: string): string {
       'blockquote', 'cite', 'q', 'hr', 'details', 'summary'
     ],
     ALLOWED_ATTR: [
-      'class', 'style', 'colspan', 'rowspan', 'src', 'alt', 'width', 'height'
+      'class', 'style', 'colspan', 'rowspan', 'src', 'alt', 'width', 'height', 'href', 'title'
     ],
+    // More permissive URL regex to allow complex query strings and tokens
+    ALLOWED_URI_REGEXP: /^(?:(?:https?|data|#|\/):)?[\s\S]*$/i,
     ALLOW_UNKNOWN_PROTOCOLS: false,
     SANITIZE_DOM: true,
     KEEP_CONTENT: true,