Refactor extract_info in watch.py to improve client flexibility

Introduce primary_client, fallback_client, and last_resort_client variables for better configurability.
Replace hardcoded 'android_vr' with primary_client in fetch_player_response call.

1 files changed, 38 insertions, 28 deletions

diff --git a/youtube/watch.py b/youtube/watch.py
index e2762cb..0274cd0 100644
--- a/youtube/watch.py
+++ b/youtube/watch.py
@@ -367,32 +367,42 @@ def fetch_watch_page_info(video_id, playlist_id, index):
     watch_page = watch_page.decode('utf-8')
     return yt_data_extract.extract_watch_info_from_html(watch_page)
 
+
 def extract_info(video_id, use_invidious, playlist_id=None, index=None):
+    primary_client = 'android_vr'
+    fallback_client = 'ios'
+    last_resort_client = 'tv_embedded'
+
     tasks = (
         # Get video metadata from here
         gevent.spawn(fetch_watch_page_info, video_id, playlist_id, index),
-        gevent.spawn(fetch_player_response, 'android_vr', video_id)
+        gevent.spawn(fetch_player_response, primary_client, video_id)
     )
     gevent.joinall(tasks)
     util.check_gevent_exceptions(*tasks)
-    info, player_response = tasks[0].value, tasks[1].value
+
+    info = tasks[0].value or {}
+    player_response = tasks[1].value or {}
 
     yt_data_extract.update_with_new_urls(info, player_response)
 
-    # Age restricted video, retry
-    if info['age_restricted'] or info['player_urls_missing']:
-        if info['age_restricted']:
-            print('Age restricted video, retrying')
-        else:
-            print('Player urls missing, retrying')
-        player_response = fetch_player_response('tv_embedded', video_id)
+    # Fallback to 'ios' if no valid URLs are found
+    if not info.get('formats') or info.get('player_urls_missing'):
+        print(f"No URLs found in '{primary_client}', attempting with '{fallback_client}'.")
+        player_response = fetch_player_response(fallback_client, video_id) or {}
+        yt_data_extract.update_with_new_urls(info, player_response)
+
+    # Final attempt with 'tv_embedded' if there are still no URLs
+    if not info.get('formats') or info.get('player_urls_missing'):
+        print(f"No URLs found in '{fallback_client}', attempting with '{last_resort_client}'")
+        player_response = fetch_player_response(last_resort_client, video_id) or {}
         yt_data_extract.update_with_new_urls(info, player_response)
 
     # signature decryption
-    decryption_error = decrypt_signatures(info, video_id)
-    if decryption_error:
-        decryption_error = 'Error decrypting url signatures: ' + decryption_error
-        info['playability_error'] = decryption_error
+    if info.get('formats'):
+        decryption_error = decrypt_signatures(info, video_id)
+        if decryption_error:
+            info['playability_error'] = 'Error decrypting url signatures: ' + decryption_error
 
     # check if urls ready (non-live format) in former livestream
     # urls not ready if all of them have no filesize
@@ -406,21 +416,21 @@ def extract_info(video_id, use_invidious, playlist_id=None, index=None):
 
     # livestream urls
     # sometimes only the livestream urls work soon after the livestream is over
-    if (info['hls_manifest_url']
-        and (info['live'] or not info['formats'] or not info['urls_ready'])
-    ):
-        manifest = util.fetch_url(info['hls_manifest_url'],
-            debug_name='hls_manifest.m3u8',
-            report_text='Fetched hls manifest'
-        ).decode('utf-8')
-
-        info['hls_formats'], err = yt_data_extract.extract_hls_formats(manifest)
-        if not err:
-            info['playability_error'] = None
-        for fmt in info['hls_formats']:
-            fmt['video_quality'] = video_quality_string(fmt)
-    else:
-        info['hls_formats'] = []
+    info['hls_formats'] = []
+    if info.get('hls_manifest_url') and (info.get('live') or not info.get('formats') or not info['urls_ready']):
+        try:
+            manifest = util.fetch_url(info['hls_manifest_url'],
+                debug_name='hls_manifest.m3u8',
+                report_text='Fetched hls manifest'
+            ).decode('utf-8')
+            info['hls_formats'], err = yt_data_extract.extract_hls_formats(manifest)
+            if not err:
+                info['playability_error'] = None
+            for fmt in info['hls_formats']:
+                fmt['video_quality'] = video_quality_string(fmt)
+        except Exception as e:
+            print(f"Error obteniendo HLS manifest: {e}")
+            info['hls_formats'] = []
 
     # check for 403. Unnecessary for tor video routing b/c ip address is same
     info['invidious_used'] = False