Sirius/youtube-dl
1
0
Fork 0
Code Issues Pull requests Projects Releases Wiki Activity

[kaltura] add support for videos with reference id

Browse source
This commit is contained in:
remitamine 2016-03-11 16:52:07 +01:00
parent f6f6217a98
commit fac7e79277
1 changed files with 42 additions and 12 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

52
youtube_dl/extractor/kaltura.py
View file

@ -8,6 +8,7 @@ from .common import InfoExtractor
from ..compat import ( from ..compat import (
compat_urllib_parse, compat_urllib_parse,
compat_urlparse, compat_urlparse,
compat_parse_qs,
) )
from ..utils import ( from ..utils import (
clean_html, clean_html,
@ -20,21 +21,17 @@ from ..utils import (
class KalturaIE(InfoExtractor): class KalturaIE(InfoExtractor):
_VALID_URL = r'''(?x) _VALID_URL = r'''(?x)
(?: (?:
kaltura:(?P<partner_id_s>\d+):(?P<id_s>[0-9a-z_]+)| kaltura:(?P<partner_id>\d+):(?P<id>[0-9a-z_]+)|
https?:// https?://
(:?(?:www|cdnapi(?:sec)?)\.)?kaltura\.com/ (:?(?:www|cdnapi(?:sec)?)\.)?kaltura\.com/
(?: (?:
(?: (?:
# flash player # flash player
index\.php/kwidget/ index\.php/kwidget|
(?:[^/]+/)*?wid/_(?P<partner_id>\d+)/
(?:[^/]+/)*?entry_id/(?P<id>[0-9a-z_]+)|
# html5 player # html5 player
html5/html5lib/ html5/html5lib/[^/]+/mwEmbedFrame\.php
(?:[^/]+/)*?entry_id/(?P<id_html5>[0-9a-z_]+)
.*\?.*\bwid=_(?P<partner_id_html5>\d+)
)
) )
)(?:/(?P<path>[^?]+))?(?:\?(?P<query>.*))?
) )
''' '''
_API_BASE = 'http://cdnapi.kaltura.com/api_v3/index.php?' _API_BASE = 'http://cdnapi.kaltura.com/api_v3/index.php?'
@ -127,10 +124,43 @@ class KalturaIE(InfoExtractor):
url, smuggled_data = unsmuggle_url(url, {}) url, smuggled_data = unsmuggle_url(url, {})
mobj = re.match(self._VALID_URL, url) mobj = re.match(self._VALID_URL, url)
partner_id = mobj.group('partner_id_s') or mobj.group('partner_id') or mobj.group('partner_id_html5') partner_id, entry_id = mobj.group('partner_id', 'id')
entry_id = mobj.group('id_s') or mobj.group('id') or mobj.group('id_html5') info, flavor_assets = None, None
if partner_id and entry_id:
info, flavor_assets = self._get_video_info(entry_id, partner_id) info, flavor_assets = self._get_video_info(entry_id, partner_id)
else:
path, query = mobj.group('path', 'query')
if not path and not query:
raise ExtractorError('Invalid URL', expected=True)
params = {}
if query:
params = compat_parse_qs(query)
if path:
splitted_path = path.split('/')
if not splitted_path[-1]:
splitted_path = splitted_path[:-1]
for i in range(0, len(splitted_path), 2):
params[splitted_path[i]] = [splitted_path[i + 1]]
if 'wid' in params:
partner_id = params['wid'][0][1:]
elif 'p' in params:
partner_id = params['p'][0]
else:
raise ExtractorError('Invalid URL', expected=True)
if 'entry_id' in params:
entry_id = params['entry_id'][0]
info, flavor_assets = self._get_video_info(entry_id, partner_id)
elif 'uiconf_id' in params and 'flashvars[referenceId]' in params:
reference_id = params['flashvars[referenceId]'][0]
webpage = self._download_webpage(url, reference_id)
entry_data = self._parse_json(self._search_regex(
r'window\.kalturaIframePackageData\s*=\s*({.*});',
webpage, 'kalturaIframePackageData'),
reference_id)['entryResult']
info, flavor_assets = entry_data['meta'], entry_data['contextData']['flavorAssets']
entry_id = info['id']
else:
raise ExtractorError('Invalid URL', expected=True)
source_url = smuggled_data.get('source_url') source_url = smuggled_data.get('source_url')
if source_url: if source_url: