From 1a582dd49d628914fa6a056b490914738f15c56d Mon Sep 17 00:00:00 2001 From: =?UTF-8?q?Jaime=20Marqui=CC=81nez=20Ferra=CC=81ndiz?= Date: Tue, 27 Aug 2013 11:56:48 +0200 Subject: Add an extractor for CNN (closes #1318) --- youtube_dl/extractor/cnn.py | 47 +++++++++++++++++++++++++++++++++++++++++++++ 1 file changed, 47 insertions(+) create mode 100644 youtube_dl/extractor/cnn.py (limited to 'youtube_dl/extractor/cnn.py') diff --git a/youtube_dl/extractor/cnn.py b/youtube_dl/extractor/cnn.py new file mode 100644 index 000000000..cee78765b --- /dev/null +++ b/youtube_dl/extractor/cnn.py @@ -0,0 +1,47 @@ +import re +import xml.etree.ElementTree + +from .common import InfoExtractor +from ..utils import determine_ext + +class CNNIE(InfoExtractor): + _VALID_URL = r'https?://(edition\.)?cnn\.com/video/(data/.+?|\?)/(?P.+?/(?P[^/]+?)\.cnn)' + + _TEST = { + 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.', + }, + } + + def _real_extract(self, url): + mobj = re.match(self._VALID_URL, url) + path = mobj.group('path') + page_title = mobj.group('title') + info_xml = self._download_webpage( + 'http://cnn.com/video/data/3.0/%s/index.xml' % path, page_title) + info = xml.etree.ElementTree.fromstring(info_xml.encode('utf-8')) + + formats = [] + for f in info.findall('files/file'): + mf = re.match(r'(\d+)x(\d+)(?:_(.*)k)?',f.attrib['bitrate']) + if mf is not None: + formats.append((int(mf.group(1)), int(mf.group(2)), int(mf.group(3) or 0), f.text)) + formats = sorted(formats) + (_,_,_, video_path) = formats[-1] + video_url = 'http://ht.cdn.turner.com/cnn/big%s' % video_path + + thumbnails = sorted([((int(t.attrib['height']),int(t.attrib['width'])), t.text) for t in info.findall('images/image')]) + thumbs_dict = [{'resolution': res, 'url': t_url} for (res, t_url) in thumbnails] + + return {'id': info.attrib['id'], + 'title': info.find('headline').text, + 'url': video_url, + 'ext': determine_ext(video_url), + 'thumbnail': thumbnails[-1][1], + 'thumbnails': thumbs_dict, + 'description': info.find('description').text, + } -- cgit v1.2.3 From 273f603efb2028a54e04cca314b72bc2a9d767ef Mon Sep 17 00:00:00 2001 From: Philipp Hagemeister <phihag@phihag.de> Date: Wed, 28 Aug 2013 00:14:19 +0200 Subject: [cnn] Allow more URLs --- youtube_dl/extractor/cnn.py | 22 +++++++++++++++++----- 1 file changed, 17 insertions(+), 5 deletions(-) (limited to 'youtube_dl/extractor/cnn.py') diff --git a/youtube_dl/extractor/cnn.py b/youtube_dl/extractor/cnn.py index cee78765b..4338bd180 100644 --- a/youtube_dl/extractor/cnn.py +++ b/youtube_dl/extractor/cnn.py @@ -4,10 +4,12 @@ import xml.etree.ElementTree from .common import InfoExtractor from ..utils import determine_ext + class CNNIE(InfoExtractor): - _VALID_URL = r'https?://(edition\.)?cnn\.com/video/(data/.+?|\?)/(?P<path>.+?/(?P<title>[^/]+?)\.cnn)' + _VALID_URL = r'''(?x)https?://(edition\.)?cnn\.com/video/(data/.+?|\?)/ + (?P<path>.+?/(?P<title>[^/]+?)(?:\.cnn|(?=&)))''' - _TEST = { + _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', @@ -15,14 +17,24 @@ class CNNIE(InfoExtractor): 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"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.\"" + } + }] def _real_extract(self, url): mobj = re.match(self._VALID_URL, url) path = mobj.group('path') page_title = mobj.group('title') - info_xml = self._download_webpage( - 'http://cnn.com/video/data/3.0/%s/index.xml' % path, page_title) + info_url = u'http://cnn.com/video/data/3.0/%s/index.xml' % path + print(info_url) + info_xml = self._download_webpage(info_url, page_title) info = xml.etree.ElementTree.fromstring(info_xml.encode('utf-8')) formats = [] -- cgit v1.2.3 From a1bb0f8773e0fff787ffe7bd1729073f3385d2ef Mon Sep 17 00:00:00 2001 From: =?UTF-8?q?Jaime=20Marqui=CC=81nez=20Ferra=CC=81ndiz?= <jaime.marquinez.ferrandiz@gmail.com> Date: Wed, 28 Aug 2013 10:20:37 +0200 Subject: [cnn] remove debug print call. --- youtube_dl/extractor/cnn.py | 1 - 1 file changed, 1 deletion(-) (limited to 'youtube_dl/extractor/cnn.py') diff --git a/youtube_dl/extractor/cnn.py b/youtube_dl/extractor/cnn.py index 4338bd180..a79f881cd 100644 --- a/youtube_dl/extractor/cnn.py +++ b/youtube_dl/extractor/cnn.py @@ -33,7 +33,6 @@ class CNNIE(InfoExtractor): path = mobj.group('path') page_title = mobj.group('title') info_url = u'http://cnn.com/video/data/3.0/%s/index.xml' % path - print(info_url) info_xml = self._download_webpage(info_url, page_title) info = xml.etree.ElementTree.fromstring(info_xml.encode('utf-8')) -- cgit v1.2.3