From 1aca444b229b41d979f310f4856390bdb9a4a39d Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?=E4=B8=8D=E5=81=9A=E4=BA=86=E7=9D=A1=E5=A4=A7=E8=A7=89?=
 <64798754+stakeswky@users.noreply.github.com>
Date: Tue, 24 Feb 2026 01:29:09 +0800
Subject: [PATCH] fix: properly strip HTML tags and resolve entities in feed
 article summaries (#149)

* fix: properly strip HTML tags and resolve entities in feed article summaries

Fixes #146

The parseTextFromHtml function was using document.textContent directly on
the parsed HTML document, which could leave raw HTML tags and unresolved
entities in feed article summaries.

Changes:
- Extract text from body element to avoid document wrapper artifacts
- Collapse multiple whitespace/newlines into single spaces for cleaner output
- Add early return for empty/whitespace-only input
- Use optional chaining for safer null handling

* fix: preserve single line breaks, only collapse 2+ consecutive whitespace

Address review feedback: the previous \s+ regex was too aggressive and
broke text-only summaries with legitimate line breaks.

Now:
- Collapse runs of 2+ non-newline whitespace into a single space
- Collapse 3+ consecutive newlines into double newline (paragraph break)
- Single line breaks are preserved

---------

Co-authored-by: User <user@example.com>
---
 lib/feed.ts | 11 +++++++++--
 1 file changed, 9 insertions(+), 2 deletions(-)

diff --git a/lib/feed.ts b/lib/feed.ts
index 77a2e3a..79e6545 100644
--- a/lib/feed.ts
+++ b/lib/feed.ts
@@ -222,13 +222,20 @@ export async function getUrlInfo(url: string): Promise<{ title: string; htmlBody
 }
 
 export async function parseTextFromHtml(html: string): Promise<string> {
-  let text = '';
+  if (!html || !html.trim()) {
+    return '';
+  }
 
   await initParser();
 
   const document = new DOMParser().parseFromString(html, 'text/html');
 
-  text = document!.textContent;
+  // Extract text from body to avoid any artifacts from the document wrapper
+  const text = (document?.querySelector('body')?.textContent || document?.textContent || '')
+    // Collapse runs of 2+ whitespace/newline characters, preserving single line breaks
+    .replace(/[^\S\n]{2,}/g, ' ')
+    .replace(/\n{3,}/g, '\n\n')
+    .trim();
 
   return text;
 }