From 540c9ec7095c97cd7be9235e7e51dd5cffccfab5 Mon Sep 17 00:00:00 2001
From: Rohit <rohit.rajan031101@gmail.com>
Date: Wed, 23 Apr 2025 17:21:14 +0530
Subject: [PATCH] feat: check parent element url and extract

---
 maxun-core/src/browserSide/scraper.js | 95 +++++++++++++++------------
 1 file changed, 54 insertions(+), 41 deletions(-)

diff --git a/maxun-core/src/browserSide/scraper.js b/maxun-core/src/browserSide/scraper.js
index 6e4bf029..e1f99c1d 100644
--- a/maxun-core/src/browserSide/scraper.js
+++ b/maxun-core/src/browserSide/scraper.js
@@ -523,49 +523,62 @@ function scrapableHeuristics(maxCountPerPage = 50, minArea = 20000, scrolls = 3,
 
     // Enhanced value extraction with context awareness
     function extractValue(element, attribute) {
-      if (!element) return null;
-    
-      // Get context-aware base URL
-      const baseURL = element.ownerDocument?.location?.href || window.location.origin;
+        if (!element) return null;
       
-      // Check shadow root first
-      if (element.shadowRoot) {
-          const shadowContent = element.shadowRoot.textContent;
-          if (shadowContent?.trim()) {
-              return shadowContent.trim();
-          }
+        // Get context-aware base URL
+        const baseURL = element.ownerDocument?.location?.href || window.location.origin;
+        
+        // Check shadow root first
+        if (element.shadowRoot) {
+            const shadowContent = element.shadowRoot.textContent;
+            if (shadowContent?.trim()) {
+                return shadowContent.trim();
+            }
+        }
+      
+        if (attribute === 'innerText') {
+            return element.innerText.trim();
+        } else if (attribute === 'innerHTML') {
+            return element.innerHTML.trim();
+        } else if (attribute === 'src' || attribute === 'href') {
+            if (attribute === 'href' && element.tagName !== 'A') {
+                const parentElement = element.parentElement;
+                if (parentElement && parentElement.tagName === 'A') {
+                    const parentHref = parentElement.getAttribute('href');
+                    if (parentHref) {
+                        try {
+                            return new URL(parentHref, baseURL).href;
+                        } catch (e) {
+                            return parentHref;
+                        }
+                    }
+                }
+            }
+            
+            const attrValue = element.getAttribute(attribute);
+            const dataAttr = attrValue || element.getAttribute('data-' + attribute);
+            
+            if (!dataAttr || dataAttr.trim() === '') {
+                if (attribute === 'src') {
+                    const style = window.getComputedStyle(element);
+                    const bgImage = style.backgroundImage;
+                    if (bgImage && bgImage !== 'none') {
+                        const matches = bgImage.match(/url\(['"]?([^'")]+)['"]?\)/);
+                        return matches ? new URL(matches[1], baseURL).href : null;
+                    }
+                }
+                return null;
+            }
+            
+            try {
+                return new URL(dataAttr, baseURL).href;
+            } catch (e) {
+                console.warn('Error creating URL from', dataAttr, e);
+                return dataAttr; // Return the original value if URL construction fails
+            }
+        }
+        return element.getAttribute(attribute);
       }
-    
-      if (attribute === 'innerText') {
-          return element.innerText.trim();
-      } else if (attribute === 'innerHTML') {
-          return element.innerHTML.trim();
-      } else if (attribute === 'src' || attribute === 'href') {
-          const attrValue = element.getAttribute(attribute);
-          
-          const dataAttr = attrValue || element.getAttribute('data-' + attribute);
-          
-          if (!dataAttr || dataAttr.trim() === '') {
-              if (attribute === 'src') {
-                  const style = window.getComputedStyle(element);
-                  const bgImage = style.backgroundImage;
-                  if (bgImage && bgImage !== 'none') {
-                      const matches = bgImage.match(/url\(['"]?([^'")]+)['"]?\)/);
-                      return matches ? new URL(matches[1], baseURL).href : null;
-                  }
-              }
-              return null;
-          }
-          
-          try {
-              return new URL(dataAttr, baseURL).href;
-          } catch (e) {
-              console.warn('Error creating URL from', dataAttr, e);
-              return dataAttr; // Return the original value if URL construction fails
-          }
-      }
-      return element.getAttribute(attribute);
-    }
 
     // Enhanced table ancestor finding with context support
     function findTableAncestor(element) {