From d7a646382a3cb477dcdc65be43c756ec5fb448ea Mon Sep 17 00:00:00 2001
From: otdoges <otdoges@proton.me>
Date: Tue, 12 Aug 2025 22:38:02 -0500
Subject: [PATCH] Potential fix for code scanning alert no. 186: Incomplete
 multi-character sanitization

Co-authored-by: Copilot Autofix powered by AI <62310815+github-advanced-security[bot]@users.noreply.github.com>
---
 src/lib/search-service.ts | 17 ++++-------------
 1 file changed, 4 insertions(+), 13 deletions(-)

diff --git a/src/lib/search-service.ts b/src/lib/search-service.ts
index 8287901b..f2f85f86 100644
--- a/src/lib/search-service.ts
+++ b/src/lib/search-service.ts
@@ -316,19 +316,10 @@ export class BraveSearchService {
   }
 
   private extractTextContent(html: string): string {
-    // Repeatedly remove <script> and <style> tags and their content
-    let sanitized = html;
-    let previous;
-    do {
-      previous = sanitized;
-      sanitized = sanitized
-        .replace(/<script[^>]*>.*?<\/script>/gis, '')
-        .replace(/<style[^>]*>.*?<\/style>/gis, '');
-    } while (sanitized !== previous);
-    return sanitized
-      .replace(/<[^>]*>/g, ' ')
-      .replace(/\s+/g, ' ')
-      .trim();
+    // Use cheerio to robustly remove <script> and <style> tags and extract text
+    const $ = cheerio.load(html);
+    $('script, style').remove();
+    return $.root().text().replace(/\s+/g, ' ').trim();
   }
 }