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)/)?(?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):
# en and es URLs produce react-based pages with different layout (e.g.
# http://www.arte.tv/guide/en/053330-002-A/carnival-italy?zone=world)
if not iframe_url:
- embed_html = self._parse_json(
- self._search_regex(
- r'program\s*:\s*({.+?["\']embed_html["\'].+?}),?\s*\n',
- webpage, 'program'),
- video_id)['embed_html']
- iframe_url= find_iframe_url(embed_html)
- json_url = compat_parse_qs(
- compat_urllib_parse_urlparse(iframe_url).query)['json_url'][0]
- return self._extract_from_json_url(json_url, video_id, lang)
-
- def _extract_from_json_url(self, json_url, video_id, lang):
+ program = self._search_regex(
+ r'program\s*:\s*({.+?["\']embed_html["\'].+?}),?\s*\n',
+ webpage, 'program', default=None)
+ if program:
+ embed_html = self._parse_json(program, video_id)
+ if embed_html:
+ iframe_url = find_iframe_url(embed_html['embed_html'])
+ if iframe_url:
+ json_url = compat_parse_qs(
+ compat_urllib_parse_urlparse(iframe_url).query)['json_url'][0]
+ if json_url:
+ 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, title=None):
info = self._download_json(json_url, video_id)
player_info = info['videoJsonPlayer']
upload_date_str = player_info.get('shootingDate')
if not upload_date_str:
- upload_date_str = player_info.get('VDA', '').split(' ')[0]
+ 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)/(?:magazine?/)?(?P<id>[^/?#&]+)'
_TESTS = [{
'url': 'http://creative.arte.tv/de/magazin/agentur-amateur-corporate-design',
class ArteTVFutureIE(ArteTVPlus7IE):
IE_NAME = 'arte.tv:future'
- _VALID_URL = r'https?://future\.arte\.tv/(?P<lang>fr|de|en|es)/(?P<id>.+)'
+ _VALID_URL = r'https?://future\.arte\.tv/(?P<lang>fr|de|en|es)/(?P<id>[^/?#&]+)'
_TESTS = [{
'url': 'http://future.arte.tv/fr/info-sciences/les-ecrevisses-aussi-sont-anxieuses',
'id': '050940-028-A',
'ext': 'mp4',
'title': 'Les écrevisses aussi peuvent être anxieuses',
+ 'upload_date': '20140902',
},
}, {
'url': 'http://future.arte.tv/fr/la-science-est-elle-responsable',
class ArteTVDDCIE(ArteTVPlus7IE):
IE_NAME = 'arte.tv:ddc'
- _VALID_URL = r'https?://ddc\.arte\.tv/(?P<lang>emission|folge)/(?P<id>.+)'
+ _VALID_URL = r'https?://ddc\.arte\.tv/(?P<lang>emission|folge)/(?P<id>[^/?#&]+)'
def _real_extract(self, url):
video_id, lang = self._extract_url_info(url)
class ArteTVConcertIE(ArteTVPlus7IE):
IE_NAME = 'arte.tv:concert'
- _VALID_URL = r'https?://concert\.arte\.tv/(?P<lang>fr|de|en|es)/(?P<id>.+)'
+ _VALID_URL = r'https?://concert\.arte\.tv/(?P<lang>fr|de|en|es)/(?P<id>[^/?#&]+)'
_TEST = {
'url': 'http://concert.arte.tv/de/notwist-im-pariser-konzertclub-divan-du-monde',
}
+class ArteTVMagazineIE(ArteTVPlus7IE):
+ IE_NAME = 'arte.tv:magazine'
+ _VALID_URL = r'https?://(?:www\.)?arte\.tv/magazine/[^/]+/(?P<lang>fr|de|en|es)/(?P<id>[^/?#&]+)'
+
+ _TESTS = [{
+ # Embedded via <iframe src="http://www.arte.tv/arte_vp/index.php?json_url=..."
+ 'url': 'http://www.arte.tv/magazine/trepalium/fr/entretien-avec-le-realisateur-vincent-lannoo-trepalium',
+ 'md5': '2a9369bcccf847d1c741e51416299f25',
+ 'info_dict': {
+ 'id': '065965-000-A',
+ 'ext': 'mp4',
+ 'title': 'Trepalium - Extrait Ep.01',
+ 'upload_date': '20160121',
+ },
+ }, {
+ # Embedded via <iframe src="http://www.arte.tv/guide/fr/embed/054813-004-A/medium"
+ 'url': 'http://www.arte.tv/magazine/trepalium/fr/episode-0406-replay-trepalium',
+ 'md5': 'fedc64fc7a946110fe311634e79782ca',
+ 'info_dict': {
+ 'id': '054813-004_PLUS7-F',
+ 'ext': 'mp4',
+ 'title': 'Trepalium (4/6)',
+ 'description': 'md5:10057003c34d54e95350be4f9b05cb40',
+ 'upload_date': '20160218',
+ },
+ }, {
+ 'url': 'http://www.arte.tv/magazine/metropolis/de/frank-woeste-german-paris-metropolis',
+ 'only_matching': True,
+ }]
+
+
class ArteTVEmbedIE(ArteTVPlus7IE):
IE_NAME = 'arte.tv:embed'
_VALID_URL = r'''(?x)