from ..utils import (
int_or_none,
parse_duration,
+ url_basename,
)
class CNNIE(InfoExtractor):
_VALID_URL = r'''(?x)https?://((edition|www)\.)?cnn\.com/video/(data/.+?|\?)/
- (?P<path>.+?/(?P<title>[^/]+?)(?:\.cnn|(?=&)))'''
+ (?P<path>.+?/(?P<title>[^/]+?)(?:\.cnn(-ap)?|(?=&)))'''
_TESTS = [{
'url': 'http://edition.cnn.com/video/?/video/sports/2013/06/09/nadal-1-on-1.cnn',
self._sort_formats(formats)
- 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]
+ thumbnails = [{
+ 'height': int(t.attrib['height']),
+ 'width': int(t.attrib['width']),
+ 'url': t.text,
+ } for t in info.findall('images/image')]
metas_el = info.find('metas')
upload_date = (
'id': info.attrib['id'],
'title': info.find('headline').text,
'formats': formats,
- 'thumbnail': thumbnails[-1][1],
- 'thumbnails': thumbs_dict,
+ 'thumbnails': thumbnails,
'description': info.find('description').text,
'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(),
+ }