[mailru] Add support for embed URLs

This commit is contained in:
Hongjie Dong 2017-12-04 00:26:09 -08:00 committed by Sergey M․
parent 3dfa9ec213
commit 25475dfab3
No known key found for this signature in database
GPG key ID: 2C393E0F18A9236D

View file

@ -13,8 +13,7 @@ from ..utils import (
class MailRuIE(InfoExtractor): class MailRuIE(InfoExtractor):
IE_NAME = 'mailru' IE_NAME = 'mailru'
IE_DESC = 'Видео@Mail.Ru' IE_DESC = 'Видео@Mail.Ru'
_VALID_URL = r'https?://(?:(?:www|m)\.)?my\.mail\.ru/(?:video/.*#video=/?(?P<idv1>(?:[^/]+/){3}\d+)|(?:(?P<idv2prefix>(?:[^/]+/){2})video/(?P<idv2suffix>[^/]+/\d+))\.html)' _VALID_URL = r'https?://(?:(?:www|m)\.)?my\.mail\.ru/(?:video/.*#video=/?(?P<idv1>(?:[^/]+/){3}\d+)|(?:(?P<idv2prefix>(?:[^/]+/){2})video/(?P<idv2suffix>[^/]+/\d+))\.html|video/embed/(?P<meta_id>\d+))'
_TESTS = [ _TESTS = [
{ {
'url': 'http://my.mail.ru/video/top#video=/mail/sonypicturesrus/75/76', 'url': 'http://my.mail.ru/video/top#video=/mail/sonypicturesrus/75/76',
@ -65,25 +64,34 @@ class MailRuIE(InfoExtractor):
{ {
'url': 'http://m.my.mail.ru/mail/3sktvtr/video/_myvideo/138.html', 'url': 'http://m.my.mail.ru/mail/3sktvtr/video/_myvideo/138.html',
'only_matching': True, 'only_matching': True,
},
{
'url': 'https://my.mail.ru/video/embed/7949340477499637815',
'only_matching': True,
} }
] ]
def _real_extract(self, url): def _real_extract(self, url):
mobj = re.match(self._VALID_URL, url) mobj = re.match(self._VALID_URL, url)
video_id = mobj.group('idv1') meta_id = mobj.group('meta_id')
page_config = None
if not video_id:
video_id = mobj.group('idv2prefix') + mobj.group('idv2suffix')
webpage = self._download_webpage(url, video_id)
video_data = None video_data = None
video_id = None
if not meta_id:
video_id = mobj.group('idv1')
if not video_id:
video_id = mobj.group('idv2prefix') + mobj.group('idv2suffix')
page_config = self._parse_json(self._search_regex( webpage = self._download_webpage(url, video_id)
r'(?s)<script[^>]+class="sp-video__page-config"[^>]*>(.+?)</script>',
webpage, 'page config', default='{}'), video_id, fatal=False) page_config = self._parse_json(self._search_regex(
if page_config: r'(?s)<script[^>]+class="sp-video__page-config"[^>]*>(.+?)</script>',
meta_url = page_config.get('metaUrl') or page_config.get('video', {}).get('metaUrl') webpage, 'page config', default='{}'), video_id, fatal=False)
if page_config or meta_id:
if page_config:
meta_url = page_config.get('metaUrl') or page_config.get('video', {}).get('metaUrl')
elif meta_id:
meta_url = 'https://my.mail.ru/+/video/meta/' + meta_id
if meta_url: if meta_url:
video_data = self._download_json( video_data = self._download_json(
meta_url, video_id, 'Downloading video meta JSON', fatal=False) meta_url, video_id, 'Downloading video meta JSON', fatal=False)