class ArteTvIE(InfoExtractor):
- _VALID_URL = r'http://videos\.arte\.tv/(?P<lang>fr|de|en|es)/.*-(?P<id>.*?)\.html'
+ _VALID_URL = r'https?://videos\.arte\.tv/(?P<lang>fr|de|en|es)/.*-(?P<id>.*?)\.html'
IE_NAME = 'arte.tv'
def _real_extract(self, url):
class ArteTVPlus7IE(InfoExtractor):
IE_NAME = 'arte.tv:+7'
- _VALID_URL = r'https?://(?:www\.)?arte\.tv/guide/(?P<lang>fr|de|en|es)/(?:(?:sendungen|emissions|embed)/)?(?P<id>[^/]+)/(?P<name>[^/?#&+])'
+ _VALID_URL = r'https?://(?:www\.)?arte\.tv/guide/(?P<lang>fr|de|en|es)/(?:(?:sendungen|emissions|embed)/)?(?P<id>[^/]+)/(?P<name>[^/?#&]+)'
@classmethod
def _extract_url_info(cls, url):
json_url = compat_parse_qs(
compat_urllib_parse_urlparse(iframe_url).query)['json_url'][0]
if json_url:
- return self._extract_from_json_url(json_url, video_id, lang)
- # Differend kind of embed URL (e.g.
+ title = self._search_regex(
+ r'<h3[^>]+title=(["\'])(?P<title>.+?)\1',
+ webpage, 'title', default=None, group='title')
+ return self._extract_from_json_url(json_url, video_id, lang, title=title)
+ # Different kind of embed URL (e.g.
# http://www.arte.tv/magazine/trepalium/fr/episode-0406-replay-trepalium)
embed_url = self._search_regex(
r'<iframe[^>]+src=(["\'])(?P<url>.+?)\1',
webpage, 'embed url', group='url')
return self.url_result(embed_url)
- def _extract_from_json_url(self, json_url, video_id, lang):
+ def _extract_from_json_url(self, json_url, video_id, lang, title=None):
info = self._download_json(json_url, video_id)
player_info = info['videoJsonPlayer']
if not upload_date_str:
upload_date_str = (player_info.get('VRA') or player_info.get('VDA') or '').split(' ')[0]
- title = player_info['VTI'].strip()
+ title = (player_info.get('VTI') or title or player_info['VID']).strip()
subtitle = player_info.get('VSU', '').strip()
if subtitle:
title += ' - %s' % subtitle
# It also uses the arte_vp_url url from the webpage to extract the information
class ArteTVCreativeIE(ArteTVPlus7IE):
IE_NAME = 'arte.tv:creative'
- _VALID_URL = r'https?://creative\.arte\.tv/(?P<lang>fr|de|en|es)/(?:magazine?/)?(?P<id>[^/?#&]+)'
+ _VALID_URL = r'https?://creative\.arte\.tv/(?P<lang>fr|de|en|es)/(?:[^/]+/)*(?P<id>[^/?#&]+)'
_TESTS = [{
'url': 'http://creative.arte.tv/de/magazin/agentur-amateur-corporate-design',
'description': 'Événement ! Quarante-cinq ans après leurs premiers succès, les légendaires Monty Python remontent sur scène.\n',
'upload_date': '20140805',
}
+ }, {
+ 'url': 'http://creative.arte.tv/de/episode/agentur-amateur-4-der-erste-kunde',
+ 'only_matching': True,
}]
+class ArteTVInfoIE(ArteTVPlus7IE):
+ IE_NAME = 'arte.tv:info'
+ _VALID_URL = r'https?://info\.arte\.tv/(?P<lang>fr|de|en|es)/(?:[^/]+/)*(?P<id>[^/?#&]+)'
+
+ _TEST = {
+ 'url': 'http://info.arte.tv/fr/service-civique-un-cache-misere',
+ 'info_dict': {
+ 'id': '067528-000-A',
+ 'ext': 'mp4',
+ 'title': 'Service civique, un cache misère ?',
+ 'upload_date': '20160403',
+ },
+ }
+
+
class ArteTVFutureIE(ArteTVPlus7IE):
IE_NAME = 'arte.tv:future'
_VALID_URL = r'https?://future\.arte\.tv/(?P<lang>fr|de|en|es)/(?P<id>[^/?#&]+)'
IE_NAME = 'arte.tv:embed'
_VALID_URL = r'''(?x)
http://www\.arte\.tv
- /playerv2/embed\.php\?json_url=
+ /(?:playerv2/embed|arte_vp/index)\.php\?json_url=
(?P<json_url>
http://arte\.tv/papi/tvguide/videos/stream/player/
(?P<lang>[^/]+)/(?P<id>[^/]+)[^&]*