[extractor/generic] Change twitter:player embeds priority to lowest (Closes #10090)

This commit is contained in:
Sergey M․ 2016-07-16 15:59:43 +07:00
parent 998895dffa
commit 371ddb14fe
No known key found for this signature in database
GPG Key ID: 2C393E0F18A9236D

View File

@ -1249,6 +1249,20 @@ class GenericIE(InfoExtractor):
'uploader': 'www.hudl.com', 'uploader': 'www.hudl.com',
}, },
}, },
# twitter:player:stream embed
{
'url': 'http://www.rtl.be/info/video/589263.aspx?CategoryID=288',
'info_dict': {
'id': 'master',
'ext': 'mp4',
'title': 'Une nouvelle espèce de dinosaure découverte en Argentine',
'uploader': 'www.rtl.be',
},
'params': {
# m3u8 downloads
'skip_download': True,
},
},
# twitter:player embed # twitter:player embed
{ {
'url': 'http://www.theatlantic.com/video/index/484130/what-do-black-holes-sound-like/', 'url': 'http://www.theatlantic.com/video/index/484130/what-do-black-holes-sound-like/',
@ -2184,11 +2198,6 @@ class GenericIE(InfoExtractor):
'uploader': video_uploader, 'uploader': video_uploader,
} }
# https://dev.twitter.com/cards/types/player#On_twitter.com_via_desktop_browser
embed_url = self._html_search_meta('twitter:player', webpage, default=None)
if embed_url:
return self.url_result(embed_url)
# Looking for http://schema.org/VideoObject # Looking for http://schema.org/VideoObject
json_ld = self._search_json_ld( json_ld = self._search_json_ld(
webpage, video_id, default=None, expected_type='VideoObject') webpage, video_id, default=None, expected_type='VideoObject')
@ -2245,6 +2254,9 @@ class GenericIE(InfoExtractor):
r"cinerama\.embedPlayer\(\s*\'[^']+\',\s*'([^']+)'", webpage) r"cinerama\.embedPlayer\(\s*\'[^']+\',\s*'([^']+)'", webpage)
if not found: if not found:
# Try to find twitter cards info # Try to find twitter cards info
# twitter:player:stream should be checked before twitter:player since
# it is expected to contain a raw stream (see
# https://dev.twitter.com/cards/types/player#On_twitter.com_via_desktop_browser)
found = filter_video(re.findall( found = filter_video(re.findall(
r'<meta (?:property|name)="twitter:player:stream" (?:content|value)="(.+?)"', webpage)) r'<meta (?:property|name)="twitter:player:stream" (?:content|value)="(.+?)"', webpage))
if not found: if not found:
@ -2278,6 +2290,15 @@ class GenericIE(InfoExtractor):
'_type': 'url', '_type': 'url',
'url': new_url, 'url': new_url,
} }
if not found:
# twitter:player is a https URL to iframe player that may or may not
# be supported by youtube-dl thus this is checked the very last (see
# https://dev.twitter.com/cards/types/player#On_twitter.com_via_desktop_browser)
embed_url = self._html_search_meta('twitter:player', webpage, default=None)
if embed_url:
return self.url_result(embed_url)
if not found: if not found:
raise UnsupportedError(url) raise UnsupportedError(url)