X-Git-Url: https://git.rapsys.eu/youtubedl/blobdiff_plain/453698570f26bebd37b39df8537d993b57d77b8b..99b46abf79266c2bf5eabc291dc4f366e61b5dde:/youtube_dl/extractor/cnn.py?ds=sidebyside diff --git a/youtube_dl/extractor/cnn.py b/youtube_dl/extractor/cnn.py index 93e8d0d..53489a1 100644 --- a/youtube_dl/extractor/cnn.py +++ b/youtube_dl/extractor/cnn.py @@ -12,7 +12,7 @@ from ..utils import ( class CNNIE(InfoExtractor): _VALID_URL = r'''(?x)https?://(?:(?:edition|www)\.)?cnn\.com/video/(?:data/.+?|\?)/ - (?P.+?/(?P[^/]+?)(?:\.(?:cnn|hln)(?:-ap)?|(?=&)))''' + (?P<path>.+?/(?P<title>[^/]+?)(?:\.(?:[a-z\-]+)|(?=&)))''' _TESTS = [{ 'url': 'http://edition.cnn.com/video/?/video/sports/2013/06/09/nadal-1-on-1.cnn', @@ -26,14 +26,14 @@ class CNNIE(InfoExtractor): 'upload_date': '20130609', }, }, { - "url": "http://edition.cnn.com/video/?/video/us/2013/08/21/sot-student-gives-epic-speech.georgia-institute-of-technology&utm_source=feedburner&utm_medium=feed&utm_campaign=Feed%3A+rss%2Fcnn_topstories+%28RSS%3A+Top+Stories%29", - "md5": "b5cc60c60a3477d185af8f19a2a26f4e", - "info_dict": { + 'url': 'http://edition.cnn.com/video/?/video/us/2013/08/21/sot-student-gives-epic-speech.georgia-institute-of-technology&utm_source=feedburner&utm_medium=feed&utm_campaign=Feed%3A+rss%2Fcnn_topstories+%28RSS%3A+Top+Stories%29', + 'md5': 'b5cc60c60a3477d185af8f19a2a26f4e', + 'info_dict': { 'id': 'us/2013/08/21/sot-student-gives-epic-speech.georgia-institute-of-technology', 'ext': 'mp4', - "title": "Student's epic speech stuns new freshmen", - "description": "A Georgia Tech student welcomes the incoming freshmen with an epic speech backed by music from \"2001: A Space Odyssey.\"", - "upload_date": "20130821", + 'title': "Student's epic speech stuns new freshmen", + 'description': "A Georgia Tech student welcomes the incoming freshmen with an epic speech backed by music from \"2001: A Space Odyssey.\"", + 'upload_date': '20130821', } }, { 'url': 'http://www.cnn.com/video/data/2.0/video/living/2014/12/22/growing-america-nashville-salemtown-board-episode-1.hln.html', @@ -45,13 +45,19 @@ class CNNIE(InfoExtractor): 'description': 'md5:e7223a503315c9f150acac52e76de086', 'upload_date': '20141222', } + }, { + 'url': 'http://cnn.com/video/?/video/politics/2015/03/27/pkg-arizona-senator-church-attendance-mandatory.ktvk', + 'only_matching': True, + }, { + 'url': 'http://cnn.com/video/?/video/us/2015/04/06/dnt-baker-refuses-anti-gay-order.wkmg', + 'only_matching': True, }] def _real_extract(self, url): mobj = re.match(self._VALID_URL, url) path = mobj.group('path') page_title = mobj.group('title') - info_url = 'http://cnn.com/video/data/3.0/%s/index.xml' % path + info_url = 'http://edition.cnn.com/video/data/3.0/%s/index.xml' % path info = self._download_xml(info_url, page_title) formats = [] @@ -143,13 +149,13 @@ class CNNArticleIE(InfoExtractor): _VALID_URL = r'https?://(?:(?:edition|www)\.)?cnn\.com/(?!video/)' _TEST = { 'url': 'http://www.cnn.com/2014/12/21/politics/obama-north-koreas-hack-not-war-but-cyber-vandalism/', - 'md5': '275b326f85d80dff7592a9820f5dc887', + 'md5': '689034c2a3d9c6dc4aa72d65a81efd01', 'info_dict': { - 'id': 'bestoftv/2014/12/21/sotu-crowley-president-obama-north-korea-not-going-to-be-intimidated.cnn', + 'id': 'bestoftv/2014/12/21/ip-north-korea-obama.cnn', 'ext': 'mp4', - 'title': 'Obama: We\'re not going to be intimidated', - 'description': 'md5:e735586f3dc936075fa654a4d91b21f9', - 'upload_date': '20141220', + 'title': 'Obama: Cyberattack not an act of war', + 'description': 'md5:51ce6750450603795cad0cdfbd7d05c5', + 'upload_date': '20141221', }, 'add_ie': ['CNN'], }