Update app.py

app.py

@@ -6,7 +6,7 @@ import logging
 
 app = FastAPI()
 
-# Set up basic logging
+# Set up logging for debugging
 logging.basicConfig(level=logging.INFO)
 
 # Allow all CORS so browser apps can call this API
@@ -25,14 +25,14 @@ def base62_to_int(e):
         t = t * 62 + BASE_62_CHAR_SET.index(char)
     return t
 
-# Construct the base URL for the shared album
+# Construct the base URL for the shared album from the token
 def get_base_url(token: str) -> str:
     first_char = token[0]
     n = base62_to_int(token[1]) if first_char == "A" else base62_to_int(token[1:3])
     base_url = f"https://p{n:02d}-sharedstreams.icloud.com/{token}/sharedstreams/"
     return base_url
 
-# Handle 330 redirect by calling the webstream endpoint with allow_redirects=False
+# Handle 330 redirect from iCloud
 def get_redirected_base_url(base_url: str, token: str) -> str:
     url = base_url + "webstream"
     headers = {
@@ -47,7 +47,7 @@ def get_redirected_base_url(base_url: str, token: str) -> str:
         return new_base_url
     return base_url
 
-# Fetch metadata from the webstream endpoint
+# Fetch metadata (the "webstream" data)
 def get_metadata(base_url: str):
     url = base_url + "webstream"
     headers = {
@@ -58,7 +58,7 @@ def get_metadata(base_url: str):
     response = requests.post(url, headers=headers, data=data)
     return response.json().get("photos", [])
 
-# Fetch asset URLs for a list of photoGuids from the webasseturls endpoint
+# Fetch asset URLs mapping for a list of photoGuids
 def get_asset_urls(base_url: str, guids: list):
     url = base_url + "webasseturls"
     headers = {
@@ -69,7 +69,7 @@ def get_asset_urls(base_url: str, guids: list):
     response = requests.post(url, headers=headers, data=data)
     return response.json().get("items", {})
 
-# Main endpoint to extract videos with extensive fallback checks
+# Main endpoint to extract video URLs
 @app.get("/album/{token}")
 def get_album(token: str):
     try:
@@ -80,73 +80,49 @@ def get_album(token: str):
         asset_urls = get_asset_urls(base_url, guids)
 
         video_list = []
-        # Loop over each photo in the album
+        # Process each photo in the album
         for photo in metadata:
-            best_video = None
-            best_size = 0
-
-            # Check derivatives first
-            derivatives = photo.get("derivatives", {})
-            for key, derivative in derivatives.items():
-                # Log the derivative keys for debugging
-                logging.info(f"Photo {photo.get('photoGuid')} derivative key: {key} | content: {list(derivative.keys())}")
-
-                # Standard checks
-                if (derivative.get("mediaAssetType") == "video" or
-                    derivative.get("type", "").startswith("video") or
-                    derivative.get("filename", "").lower().endswith(".mp4")):
+            # We check if the overall photo is a video
+            if photo.get("mediaAssetType", "").lower() == "video":
+                best_video = None
+                best_size = 0
+                # Look at each derivative
+                derivatives = photo.get("derivatives", {})
+                for key, derivative in derivatives.items():
+                    # Skip the poster frame (which is clearly an image)
+                    if key.lower() == "posterframe":
+                        continue
+                    # Try to read the fileSize as integer
                     try:
                         file_size = int(derivative.get("fileSize", 0))
                     except (ValueError, TypeError):
                         file_size = 0
+                    # Choose the derivative with the highest file size
                     if file_size > best_size:
                         best_size = file_size
                         best_video = derivative
+                # If we found a candidate and its checksum exists in asset_urls, build the video URL
+                if best_video:
+                    checksum = best_video.get("checksum")
+                    if checksum in asset_urls:
+                        url_info = asset_urls[checksum]
+                        video_url = f"https://{url_info['url_location']}{url_info['url_path']}"
+                        video_list.append({
+                            "caption": photo.get("caption", ""),
+                            "url": video_url
+                        })
+                    else:
+                        logging.info(f"Checksum {checksum} not found in asset_urls for photo {photo.get('photoGuid')}")
                 else:
-                    # Fallback: Check if any value in the derivative JSON contains "mp4"
-                    derivative_str = json.dumps(derivative).lower()
-                    if "mp4" in derivative_str:
-                        try:
-                            file_size = int(derivative.get("fileSize", 0))
-                        except (ValueError, TypeError):
-                            file_size = 0
-                        if file_size > best_size:
-                            best_size = file_size
-                            best_video = derivative
-
-            # Fallback to 'movies' field if present
-            if not best_video and "movies" in photo:
-                for movie in photo["movies"]:
-                    logging.info(f"Photo {photo.get('photoGuid')} movie entry keys: {list(movie.keys())}")
-                    if movie.get("filename", "").lower().endswith(".mp4"):
-                        try:
-                            file_size = int(movie.get("fileSize", 0))
-                        except (ValueError, TypeError):
-                            file_size = 0
-                        if file_size > best_size:
-                            best_size = file_size
-                            best_video = movie
-
-            # Build the video URL if found
-            if best_video:
-                checksum = best_video.get("checksum")
-                if checksum in asset_urls:
-                    url_info = asset_urls[checksum]
-                    video_url = f"https://{url_info['url_location']}{url_info['url_path']}"
-                    video_list.append({
-                        "caption": photo.get("caption", ""),
-                        "url": video_url
-                    })
-                else:
-                    logging.info(f"Checksum {checksum} not found in asset URLs for photo {photo.get('photoGuid')}")
+                    logging.info(f"No video derivative found for photo {photo.get('photoGuid')}")
             else:
-                logging.info(f"No video derivative found for photo {photo.get('photoGuid')}")
+                logging.info(f"Photo {photo.get('photoGuid')} is not a video. mediaAssetType: {photo.get('mediaAssetType')}")
         
         return {"videos": video_list}
     except Exception as e:
         return {"error": str(e)}
 
-# Debug endpoint to return raw metadata from iCloud
+# Debug endpoint to return raw metadata for further inspection
 @app.get("/album/{token}/raw")
 def get_album_raw(token: str):
     try: