[arte:future] Fix extraction

[arte] Add support for more "Arte Future" uri
This commit is contained in:
François Charlier 2016-01-21 18:47:43 +01:00 committed by Sergey M․
parent 220ee33f2b
commit 24114fee74
1 changed files with 18 additions and 17 deletions

View File

@ -199,25 +199,26 @@ class ArteTVCreativeIE(ArteTVPlus7IE):
class ArteTVFutureIE(ArteTVPlus7IE):
IE_NAME = 'arte.tv:future'
_VALID_URL = r'https?://future\.arte\.tv/(?P<lang>fr|de)/(thema|sujet)/.*?#article-anchor-(?P<id>\d+)'
_VALID_URL = r'https?://future\.arte\.tv/(?P<lang>fr|de)/(?P<id>.+)'
_TEST = {
'url': 'http://future.arte.tv/fr/sujet/info-sciences#article-anchor-7081',
_TESTS = [
{
'url': 'http://future.arte.tv/fr/info-sciences/les-ecrevisses-aussi-sont-anxieuses',
'info_dict': {
'id': '5201',
'id': '050940-028-A',
'ext': 'mp4',
'title': 'Les champignons au secours de la planète',
'upload_date': '20131101',
'title': 'Les écrevisses aussi peuvent être anxieuses',
},
},
{
'url': 'http://future.arte.tv/fr/la-science-est-elle-responsable',
'info_dict': {
'id': '061982-002-A',
'ext': 'mp4',
'title': 'Brian P. Schmidt - Prix Nobel de physique 2011',
},
}
def _real_extract(self, url):
anchor_id, lang = self._extract_url_info(url)
webpage = self._download_webpage(url, anchor_id)
row = self._search_regex(
r'(?s)id="%s"[^>]*>.+?(<div[^>]*arte_vp_url[^>]*>)' % anchor_id,
webpage, 'row')
return self._extract_from_webpage(row, anchor_id, lang)
]
class ArteTVDDCIE(ArteTVPlus7IE):