[youtube] Add support for formats 302 and 303 (Fixes #4060)

[youtube-dl] / youtube_dl / extractor / youtube.py

diff --git a/youtube_dl/extractor/youtube.py b/youtube_dl/extractor/youtube.py
index 61228817e319c6571525d1e821593b5b899b5010..19f81412555c83c8b92d80c48ae46387bb32ab2b 100644 (file)
--- a/youtube_dl/extractor/youtube.py
+++ b/youtube_dl/extractor/youtube.py
@@ -185,14 +185,15 @@ class YoutubeBaseInfoExtractor(InfoExtractor):
 
         self._download_webpage(
             req, None,
-            note='Confirming age', errnote='Unable to confirm age')
-        return True
+            note='Confirming age', errnote='Unable to confirm age',
+            fatal=False)
 
     def _real_initialize(self):
         if self._downloader is None:
             return
-        if not self._set_language():
-            return
+        if self._get_login_info()[0] is not None:
+            if not self._set_language():
+                return
         if not self._login():
             return
         self._confirm_age()
@@ -286,6 +287,7 @@ class YoutubeIE(YoutubeBaseInfoExtractor, SubtitlesInfoExtractor):
         '170': {'ext': 'webm', 'height': 1080, 'width': 1920, 'format_note': 'DASH video', 'acodec': 'none', 'container': 'webm', 'vcodec': 'VP8', 'preference': -40},
         '218': {'ext': 'webm', 'height': 480, 'width': 854, 'format_note': 'DASH video', 'acodec': 'none', 'container': 'webm', 'vcodec': 'VP8', 'preference': -40},
         '219': {'ext': 'webm', 'height': 480, 'width': 854, 'format_note': 'DASH video', 'acodec': 'none', 'container': 'webm', 'vcodec': 'VP8', 'preference': -40},
+        '278': {'ext': 'webm', 'height': 144, 'format_note': 'DASH video', 'acodec': 'none', 'preference': -40, 'container': 'webm', 'vcodec': 'VP9'},
         '242': {'ext': 'webm', 'height': 240, 'format_note': 'DASH video', 'acodec': 'none', 'preference': -40},
         '243': {'ext': 'webm', 'height': 360, 'format_note': 'DASH video', 'acodec': 'none', 'preference': -40},
         '244': {'ext': 'webm', 'height': 480, 'format_note': 'DASH video', 'acodec': 'none', 'preference': -40},
@@ -295,11 +297,17 @@ class YoutubeIE(YoutubeBaseInfoExtractor, SubtitlesInfoExtractor):
         '248': {'ext': 'webm', 'height': 1080, 'format_note': 'DASH video', 'acodec': 'none', 'preference': -40},
         '271': {'ext': 'webm', 'height': 1440, 'format_note': 'DASH video', 'acodec': 'none', 'preference': -40},
         '272': {'ext': 'webm', 'height': 2160, 'format_note': 'DASH video', 'acodec': 'none', 'preference': -40},
+        '302': {'ext': 'webm', 'height': 720, 'format_note': 'DASH video', 'acodec': 'none', 'preference': -40, 'fps': 60, 'vcodec': 'VP9'},
+        '303': {'ext': 'webm', 'height': 1080, 'format_note': 'DASH video', 'acodec': 'none', 'preference': -40, 'fps': 60, 'vcodec': 'VP9'},
 
         # Dash webm audio
         '171': {'ext': 'webm', 'vcodec': 'none', 'format_note': 'DASH audio', 'abr': 128, 'preference': -50},
         '172': {'ext': 'webm', 'vcodec': 'none', 'format_note': 'DASH audio', 'abr': 256, 'preference': -50},
 
+        # Dash mov
+        '298': {'ext': 'mov', 'height': 720, 'format_note': 'DASH video', 'acodec': 'none', 'preference': -40, 'fps': 60, 'vcodec': 'h264'},
+        '299': {'ext': 'mov', 'height': 1080, 'format_note': 'DASH video', 'acodec': 'none', 'preference': -40, 'fps': 60, 'vcodec': 'h264'},
+
         # RTMP (unnamed)
         '_rtmp': {'protocol': 'rtmp'},
     }
@@ -655,9 +663,17 @@ class YoutubeIE(YoutubeBaseInfoExtractor, SubtitlesInfoExtractor):
 
         # Get video webpage
         url = proto + '://www.youtube.com/watch?v=%s&gl=US&hl=en&has_verified=1' % video_id
-        req = compat_urllib_request.Request(url)
-        req.add_header('Cookie', 'PREF=hl=en')
-        video_webpage = self._download_webpage(req, video_id)
+        pref_cookies = [
+            c for c in self._downloader.cookiejar
+            if c.domain == '.youtube.com' and c.name == 'PREF']
+        for pc in pref_cookies:
+            if 'hl=' in pc.value:
+                pc.value = re.sub(r'hl=[^&]+', 'hl=en', pc.value)
+            else:
+                if pc.value:
+                    pc.value += '&'
+                pc.value += 'hl=en'
+        video_webpage = self._download_webpage(url, video_id)
 
         # Attempt to extract SWF player URL
         mobj = re.search(r'swfConfig.*?"(https?:\\/\\/.*?watch.*?-.*?\.swf)"', video_webpage)
@@ -930,7 +946,7 @@ class YoutubeIE(YoutubeBaseInfoExtractor, SubtitlesInfoExtractor):
             raise ExtractorError('no conn, hlsvp or url_encoded_fmt_stream_map information found in video info')
 
         # Look for the DASH manifest
-        if (self._downloader.params.get('youtube_include_dash_manifest', False)):
+        if self._downloader.params.get('youtube_include_dash_manifest', True):
             try:
                 # The DASH manifest used needs to be the one from the original video_webpage.
                 # The one found in get_video_info seems to be using different signatures.
@@ -1047,7 +1063,7 @@ class YoutubePlaylistIE(YoutubeBaseInfoExtractor):
         'note': 'issue #673',
         'url': 'PLBB231211A4F62143',
         'info_dict': {
-            'title': 'Team Fortress 2 (Class-based LP)',
+            'title': '[OLD]Team Fortress 2 (Class-based LP)',
         },
         'playlist_mincount': 26,
     }, {