2 from __future__ import unicode_literals
4 from .theplatform import ThePlatformIE
5 from ..utils import int_or_none
8 class CNETIE(ThePlatformIE):
9 _VALID_URL = r'https?://(?:www\.)?cnet\.com/videos/(?P<id>[^/]+)/'
11 'url': 'http://www.cnet.com/videos/hands-on-with-microsofts-windows-8-1-update/',
13 'id': '56f4ea68-bd21-4852-b08c-4de5b8354c60',
15 'title': 'Hands-on with Microsoft Windows 8.1 Update',
16 'description': 'The new update to the Windows 8 OS brings improved performance for mouse and keyboard users.',
17 'uploader_id': '6085384d-619e-11e3-b231-14feb5ca9861',
18 'uploader': 'Sarah Mitroff',
22 'url': 'http://www.cnet.com/videos/whiny-pothole-tweets-at-local-government-when-hit-by-cars-tomorrow-daily-187/',
24 'id': '56527b93-d25d-44e3-b738-f989ce2e49ba',
26 'title': 'Whiny potholes tweet at local government when hit by cars (Tomorrow Daily 187)',
27 'description': 'Khail and Ashley wonder what other civic woes can be solved by self-tweeting objects, investigate a new kind of VR camera and watch an origami robot self-assemble, walk, climb, dig and dissolve. #TDPothole',
28 'uploader_id': 'b163284d-6b73-44fc-b3e6-3da66c392d40',
29 'uploader': 'Ashley Esqueda',
34 def _real_extract(self, url):
35 display_id = self._match_id(url)
36 webpage = self._download_webpage(url, display_id)
38 data_json = self._html_search_regex(
39 r"data-cnet-video(?:-uvp)?-options='([^']+)'",
41 data = self._parse_json(data_json, display_id)
42 vdata = data.get('video') or data['videos'][0]
44 video_id = vdata['id']
45 title = vdata['title']
46 author = vdata.get('author')
48 uploader = '%s %s' % (author['firstName'], author['lastName'])
49 uploader_id = author.get('id')
54 metadata = self.get_metadata('kYEXFC/%s' % list(vdata['files'].values())[0], video_id)
55 description = vdata.get('description') or metadata.get('description')
56 duration = int_or_none(vdata.get('duration')) or metadata.get('duration')
60 for (fkey, vid) in vdata['files'].items():
61 if fkey == 'hls_phone' and 'hls_tablet' in vdata['files']:
63 release_url = 'http://link.theplatform.com/s/kYEXFC/%s?mbr=true' % vid
65 release_url += '&manifest=f4m'
66 tp_formats, tp_subtitles = self._extract_theplatform_smil(release_url, video_id, 'Downloading %s SMIL data' % fkey)
67 formats.extend(tp_formats)
68 subtitles = self._merge_subtitles(subtitles, tp_subtitles)
69 self._sort_formats(formats)
73 'display_id': display_id,
75 'description': description,
76 'thumbnail': metadata.get('thumbnail'),
79 'uploader_id': uploader_id,
80 'subtitles': subtitles,