[youtube] Hardcode codec metadata for av01 video only formats (closes #21381)
[youtube-dl] / youtube_dl / extractor / generic.py
index 6f48b04da9b1beaee6576eef1970ad6f215750fa..eeb0d25f65a7921b0c4ac30207fe995e1336d1e3 100644 (file)
@@ -89,7 +89,10 @@ from .piksel import PikselIE
 from .videa import VideaIE
 from .twentymin import TwentyMinutenIE
 from .ustream import UstreamIE
-from .openload import OpenloadIE
+from .openload import (
+    OpenloadIE,
+    VerystreamIE,
+)
 from .videopress import VideoPressIE
 from .rutube import RutubeIE
 from .limelight import LimelightBaseIE
@@ -2546,11 +2549,11 @@ class GenericIE(InfoExtractor):
             return self.url_result(mobj.group('url'))
 
         # Look for Ooyala videos
-        mobj = (re.search(r'player\.ooyala\.com/[^"?]+[?#][^"]*?(?:embedCode|ec)=(?P<ec>[^"&]+)', webpage) or
-                re.search(r'OO\.Player\.create\([\'"].*?[\'"],\s*[\'"](?P<ec>.{32})[\'"]', webpage) or
-                re.search(r'OO\.Player\.create\.apply\(\s*OO\.Player\s*,\s*op\(\s*\[\s*[\'"][^\'"]*[\'"]\s*,\s*[\'"](?P<ec>.{32})[\'"]', webpage) or
-                re.search(r'SBN\.VideoLinkset\.ooyala\([\'"](?P<ec>.{32})[\'"]\)', webpage) or
-                re.search(r'data-ooyala-video-id\s*=\s*[\'"](?P<ec>.{32})[\'"]', webpage))
+        mobj = (re.search(r'player\.ooyala\.com/[^"?]+[?#][^"]*?(?:embedCode|ec)=(?P<ec>[^"&]+)', webpage)
+                or re.search(r'OO\.Player\.create\([\'"].*?[\'"],\s*[\'"](?P<ec>.{32})[\'"]', webpage)
+                or re.search(r'OO\.Player\.create\.apply\(\s*OO\.Player\s*,\s*op\(\s*\[\s*[\'"][^\'"]*[\'"]\s*,\s*[\'"](?P<ec>.{32})[\'"]', webpage)
+                or re.search(r'SBN\.VideoLinkset\.ooyala\([\'"](?P<ec>.{32})[\'"]\)', webpage)
+                or re.search(r'data-ooyala-video-id\s*=\s*[\'"](?P<ec>.{32})[\'"]', webpage))
         if mobj is not None:
             embed_token = self._search_regex(
                 r'embedToken[\'"]?\s*:\s*[\'"]([^\'"]+)',
@@ -2580,19 +2583,6 @@ class GenericIE(InfoExtractor):
         if mobj is not None:
             return self.url_result(mobj.group(1), 'Mpora')
 
-        # Look for embedded NovaMov-based player
-        mobj = re.search(
-            r'''(?x)<(?:pagespeed_)?iframe[^>]+?src=(["\'])
-                    (?P<url>http://(?:(?:embed|www)\.)?
-                        (?:novamov\.com|
-                           nowvideo\.(?:ch|sx|eu|at|ag|co)|
-                           videoweed\.(?:es|com)|
-                           movshare\.(?:net|sx|ag)|
-                           divxstage\.(?:eu|net|ch|co|at|ag))
-                        /embed\.php.+?)\1''', webpage)
-        if mobj is not None:
-            return self.url_result(mobj.group('url'))
-
         # Look for embedded Facebook player
         facebook_urls = FacebookIE._extract_urls(webpage)
         if facebook_urls:
@@ -3017,6 +3007,12 @@ class GenericIE(InfoExtractor):
             return self.playlist_from_matches(
                 openload_urls, video_id, video_title, ie=OpenloadIE.ie_key())
 
+        # Look for Verystream embeds
+        verystream_urls = VerystreamIE._extract_urls(webpage)
+        if verystream_urls:
+            return self.playlist_from_matches(
+                verystream_urls, video_id, video_title, ie=VerystreamIE.ie_key())
+
         # Look for VideoPress embeds
         videopress_urls = VideoPressIE._extract_urls(webpage)
         if videopress_urls:
@@ -3212,8 +3208,8 @@ class GenericIE(InfoExtractor):
                 else:
                     formats.append({
                         'url': src,
-                        'ext': (mimetype2ext(src_type) or
-                                ext if ext in KNOWN_EXTENSIONS else 'mp4'),
+                        'ext': (mimetype2ext(src_type)
+                                or ext if ext in KNOWN_EXTENSIONS else 'mp4'),
                     })
             if formats:
                 self._sort_formats(formats)