From: Sergey M․ Date: Tue, 9 Jun 2015 15:39:45 +0000 (+0600) Subject: [cbs] Extract display_id X-Git-Url: https://git.cielonegro.org/gitweb.cgi?a=commitdiff_plain;h=9d581f3d5224140ca35ebd06d614b929e9252cd4;p=youtube-dl.git [cbs] Extract display_id --- diff --git a/youtube_dl/extractor/cbs.py b/youtube_dl/extractor/cbs.py index 89614a3c9..75fffb156 100644 --- a/youtube_dl/extractor/cbs.py +++ b/youtube_dl/extractor/cbs.py @@ -4,12 +4,13 @@ from .common import InfoExtractor class CBSIE(InfoExtractor): - _VALID_URL = r'https?://(?:(?:www\.)?cbs\.com/shows/[^/]+/(?:video|artist)|colbertlateshow\.com/(?:video|podcasts))/(?P[^/]+)/.*' + _VALID_URL = r'https?://(?:www\.)?(?:cbs\.com/shows/[^/]+/(?:video|artist)|colbertlateshow\.com/(?:video|podcasts))/[^/]+/(?P[^/]+)' _TESTS = [{ 'url': 'http://www.cbs.com/shows/garth-brooks/video/_u7W953k6la293J7EPTd9oHkSPs6Xn6_/connect-chat-feat-garth-brooks/', 'info_dict': { 'id': '4JUVEwq3wUT7', + 'display_id': 'connect-chat-feat-garth-brooks', 'ext': 'flv', 'title': 'Connect Chat feat. Garth Brooks', 'description': 'Connect with country music singer Garth Brooks, as he chats with fans on Wednesday November 27, 2013. Be sure to tune in to Garth Brooks: Live from Las Vegas, Friday November 29, at 9/8c on CBS!', @@ -24,6 +25,7 @@ class CBSIE(InfoExtractor): 'url': 'http://www.cbs.com/shows/liveonletterman/artist/221752/st-vincent/', 'info_dict': { 'id': 'WWF_5KqY3PK1', + 'display_id': 'st-vincent', 'ext': 'flv', 'title': 'Live on Letterman - St. Vincent', 'description': 'Live On Letterman: St. Vincent in concert from New York\'s Ed Sullivan Theater on Tuesday, July 16, 2014.', @@ -38,14 +40,19 @@ class CBSIE(InfoExtractor): 'url': 'http://colbertlateshow.com/video/8GmB0oY0McANFvp2aEffk9jZZZ2YyXxy/the-colbeard/', 'only_matching': True, }, { - 'url': 'http://colbertlateshow.com/podcasts/dYSwjqPs_X1tvbV_P2FcPWRa_qT6akTC/in-the-bad-room-with-stephen/', + 'url': 'http://www.colbertlateshow.com/podcasts/dYSwjqPs_X1tvbV_P2FcPWRa_qT6akTC/in-the-bad-room-with-stephen/', 'only_matching': True, }] def _real_extract(self, url): - video_id = self._match_id(url) - webpage = self._download_webpage(url, video_id) + display_id = self._match_id(url) + webpage = self._download_webpage(url, display_id) real_id = self._search_regex( [r"video\.settings\.pid\s*=\s*'([^']+)';", r"cbsplayer\.pid\s*=\s*'([^']+)';"], webpage, 'real video ID') - return self.url_result('theplatform:%s' % real_id) + return { + '_type': 'url_transparent', + 'ie_key': 'ThePlatform', + 'url': 'theplatform:%s' % real_id, + 'display_id': display_id, + }