X-Git-Url: http://git.bitcoin.ninja/index.cgi?a=blobdiff_plain;f=youtube_dl%2Fextractor%2Fcnn.py;h=b32cb898010a0ad0e02e12f3b3a55c3769cc3979;hb=f92259c0260880ac8e5f285b4b267183b97761c0;hp=ecac5e0e9f952539545ba1ea4a4cad7a67a3a43a;hpb=5f263296eaa72ddca232d734a2625bcd85771908;p=youtube-dl diff --git a/youtube_dl/extractor/cnn.py b/youtube_dl/extractor/cnn.py index ecac5e0e9..b32cb8980 100644 --- a/youtube_dl/extractor/cnn.py +++ b/youtube_dl/extractor/cnn.py @@ -1,9 +1,12 @@ +from __future__ import unicode_literals + import re from .common import InfoExtractor from ..utils import ( int_or_none, parse_duration, + url_basename, ) @@ -12,23 +15,24 @@ class CNNIE(InfoExtractor): (?P.+?/(?P[^/]+?)(?:\.cnn|(?=&)))''' _TESTS = [{ - u'url': u'http://edition.cnn.com/video/?/video/sports/2013/06/09/nadal-1-on-1.cnn', - u'file': u'sports_2013_06_09_nadal-1-on-1.cnn.mp4', - u'md5': u'3e6121ea48df7e2259fe73a0628605c4', - u'info_dict': { - u'title': u'Nadal wins 8th French Open title', - u'description': u'World Sport\'s Amanda Davies chats with 2013 French Open champion Rafael Nadal.', - u'duration': 135, - u'upload_date': u'20130609', + 'url': 'http://edition.cnn.com/video/?/video/sports/2013/06/09/nadal-1-on-1.cnn', + 'file': 'sports_2013_06_09_nadal-1-on-1.cnn.mp4', + 'md5': '3e6121ea48df7e2259fe73a0628605c4', + 'info_dict': { + 'title': 'Nadal wins 8th French Open title', + 'description': 'World Sport\'s Amanda Davies chats with 2013 French Open champion Rafael Nadal.', + 'duration': 135, + 'upload_date': '20130609', }, }, { - u"url": u"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", - u"file": u"us_2013_08_21_sot-student-gives-epic-speech.georgia-institute-of-technology.mp4", - u"md5": u"b5cc60c60a3477d185af8f19a2a26f4e", - u"info_dict": { - u"title": "Student's epic speech stuns new freshmen", - u"description": "A Georgia Tech student welcomes the incoming freshmen with an epic speech backed by music from \"2001: A Space Odyssey.\"" + "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", + "file": "us_2013_08_21_sot-student-gives-epic-speech.georgia-institute-of-technology.mp4", + "md5": "b5cc60c60a3477d185af8f19a2a26f4e", + "info_dict": { + "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", } }] @@ -36,7 +40,7 @@ class CNNIE(InfoExtractor): mobj = re.match(self._VALID_URL, url) path = mobj.group('path') page_title = mobj.group('title') - info_url = u'http://cnn.com/video/data/3.0/%s/index.xml' % path + info_url = 'http://cnn.com/video/data/3.0/%s/index.xml' % path info = self._download_xml(info_url, page_title) formats = [] @@ -95,3 +99,28 @@ class CNNIE(InfoExtractor): 'duration': duration, 'upload_date': upload_date, } + + +class CNNBlogsIE(InfoExtractor): + _VALID_URL = r'https?://[^\.]+\.blogs\.cnn\.com/.+' + _TEST = { + 'url': 'http://reliablesources.blogs.cnn.com/2014/02/09/criminalizing-journalism/', + 'md5': '3e56f97b0b6ffb4b79f4ea0749551084', + 'info_dict': { + 'id': 'bestoftv/2014/02/09/criminalizing-journalism.cnn', + 'ext': 'mp4', + 'title': 'Criminalizing journalism?', + 'description': 'Glenn Greenwald responds to comments made this week on Capitol Hill that journalists could be criminal accessories.', + 'upload_date': '20140209', + }, + 'add_ie': ['CNN'], + } + + def _real_extract(self, url): + webpage = self._download_webpage(url, url_basename(url)) + cnn_url = self._html_search_regex(r'data-url="(.+?)"', webpage, 'cnn url') + return { + '_type': 'url', + 'url': cnn_url, + 'ie_key': CNNIE.ie_key(), + }