From ec08af46402891d47ef0845146b2905e5661b6c0 Mon Sep 17 00:00:00 2001
From: Victor Giers <mail@victorgiers.com>
Date: Mon, 15 Jun 2026 01:15:35 +0200
Subject: [PATCH] Improve change detection and enhance content extraction logic

---
 backend/local_rag.py   | 4 +++-
 backend/websearch.py   | 3 +++
 src/LibraryManager.jsx | 5 ++++-
 3 files changed, 10 insertions(+), 2 deletions(-)

diff --git a/backend/local_rag.py b/backend/local_rag.py
index 33ff4c4..220b831 100644
--- a/backend/local_rag.py
+++ b/backend/local_rag.py
@@ -1388,6 +1388,8 @@ async def refresh_library_website(slug: str, item_id: str):
 
     source_path = Path(str(entry.get("path") or ""))
     previous_hash = str(entry.get("sha256") or "")
+    previous_title = str(entry.get("title") or "")
+    previous_url = str(entry.get("url") or "")
     _write_text_atomic(source_path, str(snapshot["text"]))
     next_hash = _sha256_file(source_path)
     entry.update({
@@ -1407,7 +1409,7 @@ async def refresh_library_website(slug: str, item_id: str):
         entry["name"] = title
 
     job_id = None
-    changed = next_hash != previous_hash
+    changed = next_hash != previous_hash or str(entry.get("title") or "") != previous_title or str(entry.get("url") or "") != previous_url
     if changed:
         _mark_entry_pending(entry)
         job_id = await _save_library_change(slug, data)
diff --git a/backend/websearch.py b/backend/websearch.py
index eeb7fcc..0a11aa8 100644
--- a/backend/websearch.py
+++ b/backend/websearch.py
@@ -263,6 +263,9 @@ async def fetch_website_snapshot(url: str) -> Dict[str, Any]:
     text = clean_text(html)
     if len(text.strip()) < 80:
         raise ValueError("The website did not contain enough readable text to save.")
+    if not title:
+        first_line = next((line.strip() for line in text.splitlines() if line.strip()), "")
+        title = re.sub(r"^[#>*_`\-\s]+", "", first_line).strip()[:240]
 
     return {
         "requested_url": requested_url,
diff --git a/src/LibraryManager.jsx b/src/LibraryManager.jsx
index 4463784..e8b7990 100644
--- a/src/LibraryManager.jsx
+++ b/src/LibraryManager.jsx
@@ -108,7 +108,10 @@ export default function LibraryManager({ apiBase, library, jobs, onRefresh }) {
     })
   }
 
-  useEffect(() => () => clearToasts(), [])
+  useEffect(() => () => {
+    toastTimeoutsRef.current.forEach(timeoutId => clearTimeout(timeoutId))
+    toastTimeoutsRef.current.clear()
+  }, [])
 
   function closeForm() {
     setFormMode(null)