From bc04400bdc5e93679a7337185fdc1aeddbd5cb4a Mon Sep 17 00:00:00 2001
From: bhushan4work <bhushan.agrawal4work@gmail.com>
Date: Fri, 6 Nov 2026 22:12:50 +0530
Subject: [PATCH] fix(meta-fetcher): normalize whitespace in page titles (#393)

---
 src/lib/meta-fetcher.ts        |  2 +-
 tests/lib/meta-fetcher.test.ts | 11 +++++++++++
 2 files changed, 12 insertions(+), 1 deletion(-)

diff --git a/src/lib/meta-fetcher.ts b/src/lib/meta-fetcher.ts
index 59d7775..dbe16f8 100644
--- a/src/lib/meta-fetcher.ts
+++ b/src/lib/meta-fetcher.ts
@@ -52,7 +52,7 @@ export function parseMetaTags(html: string, baseUrl: string): MetaTags {
 
   // Extract <title>
   const titleMatch = html.match(/<title[^>]*>(.*?)<\/title>/is);
-  if (titleMatch) tags.title = decodeEntities(titleMatch[1].trim());
+  if (titleMatch) tags.title = decodeEntities(titleMatch[1]).replace(/[ \t\n\r\f]+/g, ' ').trim();
 
   // Extract meta tags
   for (const tag of findTags(html, 'meta')) {
diff --git a/tests/lib/meta-fetcher.test.ts b/tests/lib/meta-fetcher.test.ts
index 5e90da0..c110c2f 100644
--- a/tests/lib/meta-fetcher.test.ts
+++ b/tests/lib/meta-fetcher.test.ts
@@ -119,6 +119,17 @@ describe('parseMetaTags', () => {
     expect(tags.title).toBe('<Tom & Jerry "Show">');
   });
 
+  it('collapses multiline indentation, internal tabs, and double spaces in title while preserving non-ASCII whitespace', () => {
+    const tags = parseMetaTags(
+      `<title>
+             Cutting\t\tcold  starts in&nbsp;half with Fluid   Compute
+             | Acme Blog
+     </title>`,
+      base,
+    );
+    expect(tags.title).toBe('Cutting cold starts in\u00a0half with Fluid Compute | Acme Blog');
+  });
+
   it('does not decode uppercase or mixed-case entities that are not valid in HTML', () => {
     const tags = parseMetaTags(
       '<meta name="description" content="&APOS; &NBSP; &Amp;">',