Update app.py

app.py

@@ -39,8 +39,9 @@ HF_REPO_TYPE = "dataset"
 TOTAL_DIVISIONS = 25
 VALID_DIVISIONS = set(range(1, TOTAL_DIVISIONS + 1))
 
-# Local Temporary Storage (for processing before upload)
-TEMP_DIR = tempfile.mkdtemp()
+# Local Temporary Storage (Fixed to use local dir to avoid permission issues)
+TEMP_DIR = os.path.abspath("temp_downloads")
+os.makedirs(TEMP_DIR, exist_ok=True)
 
 # ==================================================
 # ADMIN AUTH
@@ -109,7 +110,8 @@ api = HfApi(token=HF_TOKEN)
 
 def get_hf_url(path_in_repo: str) -> str:
     """Returns the direct download URL for a file in the dataset."""
-    return f"https://huggingface.co/datasets/{HF_REPO_ID}/resolve/main/{path_in_repo}"
+    # Using ?download=true helps force the browser to treat it as a file
+    return f"https://huggingface.co/datasets/{HF_REPO_ID}/resolve/main/{path_in_repo}?download=true"
 
 async def sync_pull_json(filename: str, default: dict) -> dict:
     """Download JSON from HF. Returns default if not found."""
@@ -159,10 +161,16 @@ async def sync_push_json(filename: str, data: dict):
     except Exception as e:
         logging.error(f"Failed to push {filename} to HF: {e}")
 
-async def sync_push_file(local_path: str, remote_path: str):
-    """Upload media file to HF."""
+async def sync_push_file(local_path: str, remote_path: str) -> bool:
+    """Upload media file to HF. Returns True if successful."""
     if not HF_TOKEN: 
-        return
+        return False
+        
+    # Verify file exists and is readable
+    if not os.path.exists(local_path):
+        logging.error(f"Upload failed: Local file not found at {local_path}")
+        return False
+
     try:
         loop = asyncio.get_event_loop()
         await loop.run_in_executor(
@@ -175,8 +183,12 @@ async def sync_push_file(local_path: str, remote_path: str):
                 commit_message=f"Add media {remote_path}"
             )
         )
+        # Small delay to let HF API catch up and prevent 429/File Lock issues
+        await asyncio.sleep(2.0) 
+        return True
     except Exception as e:
         logging.error(f"Failed to push media {remote_path}: {e}")
+        return False
 
 # ==================================================
 # STATE MANAGEMENT
@@ -277,13 +289,20 @@ async def get_asset_urls(base_url: str, guids: list) -> dict:
     return (await post_json("webasseturls", base_url, payload)).get("items", {})
 
 async def download_to_temp(url: str, filename: str) -> str:
+    # Use the explicitly created local temp directory
     path = os.path.join(TEMP_DIR, filename)
+    
+    # Remove if exists to ensure clean slate
+    if os.path.exists(path):
+        os.remove(path)
+
     client = await get_client()
     async with client.stream("GET", url) as r:
         r.raise_for_status()
         with open(path, "wb") as f:
             async for chunk in r.aiter_bytes():
                 f.write(chunk)
+    
     return path
 
 # ==================================================
@@ -295,7 +314,13 @@ async def poll_album(token: str):
         base_url = await get_redirected_base_url(base_url, token)
 
         metadata = await get_metadata(base_url)
+        if not metadata:
+            return # Empty album, skip to prevent errors
+
         guids = [p["photoGuid"] for p in metadata]
+        if not guids:
+            return
+
         assets = await get_asset_urls(base_url, guids)
 
         async with DATA_LOCK:
@@ -326,33 +351,54 @@ async def poll_album(token: str):
             asset = assets[best["checksum"]]
             video_url = f"https://{asset['url_location']}{asset['url_path']}"
             
-            # Download to temp then Push to HF
+            # --- PROCESS VIDEO ---
             temp_vid = await download_to_temp(video_url, f"{vid}.mp4")
             hf_vid_path = f"videos/{token}/{vid}.mp4"
-            await sync_push_file(temp_vid, hf_vid_path)
-            os.remove(temp_vid)
-
-            # Handle Thumbnail
-            hf_thumb_path = ""
+            
+            vid_success = await sync_push_file(temp_vid, hf_vid_path)
+            
+            # Cleanup temp video safely
+            try:
+                if os.path.exists(temp_vid):
+                    os.remove(temp_vid)
+            except Exception as e:
+                logging.warning(f"Cleanup error for {vid}: {e}")
+
+            if not vid_success:
+                continue # Skip entry if video upload failed
+
+            # --- PROCESS THUMBNAIL ---
+            final_thumb_url = ""
             pf = derivatives.get("PosterFrame")
             if pf and pf.get("checksum") in assets:
                 pf_asset = assets[pf["checksum"]]
                 poster_url = f"https://{pf_asset['url_location']}{pf_asset['url_path']}"
+                
                 temp_thumb = await download_to_temp(poster_url, f"{vid}.jpg")
                 hf_thumb_path = f"videos/{token}/{vid}.jpg"
-                await sync_push_file(temp_thumb, hf_thumb_path)
-                os.remove(temp_thumb)
+                
+                thumb_success = await sync_push_file(temp_thumb, hf_thumb_path)
+                
+                # Cleanup temp thumb safely
+                try:
+                    if os.path.exists(temp_thumb):
+                        os.remove(temp_thumb)
+                except Exception as e:
+                    logging.warning(f"Cleanup error for thumb {vid}: {e}")
+
+                if thumb_success:
+                    final_thumb_url = get_hf_url(hf_thumb_path)
 
             new_entries.append({
                 "id": vid,
                 "name": photo.get("caption") or "Untitled",
                 "video_url": get_hf_url(hf_vid_path),
-                "thumbnail": get_hf_url(hf_thumb_path) if hf_thumb_path else "",
+                "thumbnail": final_thumb_url, 
                 "upload_date": photo.get("creationDate") or datetime.now(timezone.utc).isoformat(),
                 "category": category,
                 "publisher": publisher,
                 "source_album": token,
-                "allowed_divs": [] # Default: Allowed for all
+                "allowed_divs": [] 
             })
 
         if new_entries: