[ie/wrestleuniverse] Avoid partial stream formats (#9800)

Authored by: bashonly
yt-dlp · May 4, 2024 · c485365 · c485365
1 parent ac817bc
commit c485365
Showing 1 changed file with 12 additions and 4 deletions.
diff --git a/yt_dlp/extractor/wrestleuniverse.py b/yt_dlp/extractor/wrestleuniverse.py
@@ -12,6 +12,7 @@
     jwt_decode_hs256,
     traverse_obj,
     try_call,
+    url_basename,
     url_or_none,
     urlencode_postdata,
     variadic,
@@ -194,8 +195,7 @@ def _real_extract(self, url):
 
         return {
             'id': video_id,
-            'formats': self._get_formats(video_data, (
-                (('protocolHls', 'url'), ('chromecastUrls', ...)), {url_or_none}), video_id),
+            'formats': self._get_formats(video_data, ('protocolHls', 'url', {url_or_none}), video_id),
             **traverse_obj(metadata, {
                 'title': ('displayName', {str}),
                 'description': ('description', {str}),
@@ -259,6 +259,10 @@ class WrestleUniversePPVIE(WrestleUniverseBaseIE):
         'params': {
             'skip_download': 'm3u8',
         },
+    }, {
+        'note': 'manifest provides live-a (partial) and live-b (full) streams',
+        'url': 'https://www.wrestle-universe.com/en/lives/umc99R9XsexXrxr9VjTo9g',
+        'only_matching': True,
     }]
 
     _API_PATH = 'events'
@@ -285,12 +289,16 @@ def _real_extract(self, url):
 
         video_data, decrypt = self._call_encrypted_api(
             video_id, ':watchArchive', 'watch archive', data={'method': 1})
-        info['formats'] = self._get_formats(video_data, (
-            ('hls', None), ('urls', 'chromecastUrls'), ..., {url_or_none}), video_id)
+        # 'chromecastUrls' can be only partial videos, avoid
+        info['formats'] = self._get_formats(video_data, ('hls', (('urls', ...), 'url'), {url_or_none}), video_id)
         for f in info['formats']:
             # bitrates are exaggerated in PPV playlists, so avoid wrong/huge filesize_approx values
             if f.get('tbr'):
                 f['tbr'] = int(f['tbr'] / 2.5)
+            # prefer variants with the same basename as the master playlist to avoid partial streams
+            f['format_id'] = url_basename(f['url']).partition('.')[0]
+            if not f['format_id'].startswith(url_basename(f['manifest_url']).partition('.')[0]):
+                f['preference'] = -10
 
         hls_aes_key = traverse_obj(video_data, ('hls', 'key', {decrypt}))
         if hls_aes_key: