X-Git-Url: http://git.bitcoin.ninja/index.cgi?a=blobdiff_plain;f=youtube_dl%2Fextractor%2Farte.py;h=6ed855a579479dc868f7854f46ade1f1e11dda11;hb=d800609c62703e4e6edd2891a8432306462e4db3;hp=10301a8eac5c79d5e9fcfb254eeeacd6eb95bf10;hpb=06e4874c997fb523bd9d8e675d1ec0e69ab501ed;p=youtube-dl diff --git a/youtube_dl/extractor/arte.py b/youtube_dl/extractor/arte.py index 10301a8ea..6ed855a57 100644 --- a/youtube_dl/extractor/arte.py +++ b/youtube_dl/extractor/arte.py @@ -13,6 +13,7 @@ from ..utils import ( unified_strdate, get_element_by_attribute, int_or_none, + NO_DEFAULT, qualities, ) @@ -93,9 +94,18 @@ class ArteTVPlus7IE(InfoExtractor): json_url = self._html_search_regex( patterns, webpage, 'json vp url', default=None) if not json_url: - iframe_url = self._html_search_regex( - r']+src=(["\'])(?P.+\bjson_url=.+?)\1', - webpage, 'iframe url', group='url') + def find_iframe_url(webpage, default=NO_DEFAULT): + return self._html_search_regex( + r']+src=(["\'])(?P.+\bjson_url=.+?)\1', + webpage, 'iframe url', group='url', default=default) + + iframe_url = find_iframe_url(webpage, None) + if not iframe_url: + embed_url = self._html_search_regex( + r'arte_vp_url_oembed=\'([^\']+?)\'', webpage, 'embed url') + player = self._download_json( + embed_url, video_id, 'Downloading player page') + iframe_url = find_iframe_url(player['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) @@ -199,25 +209,19 @@ class ArteTVCreativeIE(ArteTVPlus7IE): class ArteTVFutureIE(ArteTVPlus7IE): IE_NAME = 'arte.tv:future' - _VALID_URL = r'https?://future\.arte\.tv/(?Pfr|de)/(thema|sujet)/.*?#article-anchor-(?P\d+)' + _VALID_URL = r'https?://future\.arte\.tv/(?Pfr|de)/(?P.+)' - _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', }, - } - - 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"[^>]*>.+?(]*arte_vp_url[^>]*>)' % anchor_id, - webpage, 'row') - return self._extract_from_webpage(row, anchor_id, lang) + }, { + 'url': 'http://future.arte.tv/fr/la-science-est-elle-responsable', + 'only_matching': True, + }] class ArteTVDDCIE(ArteTVPlus7IE): @@ -255,6 +259,23 @@ class ArteTVConcertIE(ArteTVPlus7IE): } +class ArteTVCinemaIE(ArteTVPlus7IE): + IE_NAME = 'arte.tv:cinema' + _VALID_URL = r'https?://cinema\.arte\.tv/(?Pde|fr)/(?P.+)' + + _TEST = { + 'url': 'http://cinema.arte.tv/de/node/38291', + 'md5': '6b275511a5107c60bacbeeda368c3aa1', + 'info_dict': { + 'id': '055876-000_PWA12025-D', + 'ext': 'mp4', + 'title': 'Tod auf dem Nil', + 'upload_date': '20160122', + 'description': 'md5:7f749bbb77d800ef2be11d54529b96bc', + }, + } + + class ArteTVEmbedIE(ArteTVPlus7IE): IE_NAME = 'arte.tv:embed' _VALID_URL = r'''(?x)