From 234f61d066ca0a017dc2e99da141d13805b6c1aa Mon Sep 17 00:00:00 2001
From: "yanlong.wang" <yanlong.wang@naiver.org>
Date: Mon, 20 Jan 2025 11:54:31 +0800
Subject: [PATCH] remove more attrs in readerlm preprocessing

---
 backend/functions/src/services/jsdom.ts | 18 ++++++++++++++++++
 1 file changed, 18 insertions(+)

diff --git a/backend/functions/src/services/jsdom.ts b/backend/functions/src/services/jsdom.ts
index 65124ae..88cadac 100644
--- a/backend/functions/src/services/jsdom.ts
+++ b/backend/functions/src/services/jsdom.ts
@@ -273,6 +273,24 @@ export class JSDomControl extends AsyncService {
             }
             x.removeAttribute('style');
         });
+        const treeWalker = jsdom.window.document.createTreeWalker(
+            jsdom.window.document, // Start from the root document
+            0x80 // Only show comment nodes
+        );
+
+        let currentNode;
+        while ((currentNode = treeWalker.nextNode())) {
+            currentNode.parentNode?.removeChild(currentNode); // Remove each comment node
+        }
+
+        jsdom.window.document.querySelectorAll('*').forEach((x)=> {
+            const attrs = x.getAttributeNames();
+            for (const attr of attrs) {
+                if (attr.startsWith('data-') || attr.startsWith('aria-')) {
+                    x.removeAttribute(attr);
+                }
+            }
+        });
 
         const dt = Date.now() - t0;
         if (dt > 1000) {