2 from __future__ import unicode_literals
5 from .common import InfoExtractor
6 from ..utils import str_to_int
9 class PressTVIE(InfoExtractor):
10 _VALID_URL = r'https?://(?:www\.)?presstv\.ir/Video/(?P<y>[0-9]+)/(?P<m>[0-9]+)/(?P<d>[0-9]+)/(?P<id>[0-9]+)/'
13 'url': 'http://www.presstv.ir/Video/2015/10/04/431915/Max-Igan-Press-TV-Face-to-Face',
14 'md5': 'e95736ac75088b5f1e5bbb68f248f90d',
18 'title': 'Press TV’s full interview with Max Igan',
19 'upload_date': '20151004',
20 'thumbnail': 'http://217.218.67.233/photo/20151004/d5c333ad-98f9-4bd3-bc3e-a1ad6a192803.jpg',
21 'description': ('Watch Press TV’s full interview with Max Igan, a radio talk show host and political '
22 'commentator.\nThe interview, conducted on Press TV’s Face '
23 'to Face program, was aired on October 3, 2015.')
27 def _real_extract(self, url):
28 video_id = self._match_id(url)
29 webpage = self._download_webpage(url, video_id)
31 # extract video URL from webpage
32 video_url = self._html_search_regex(r'<input type="hidden" id="inpPlayback" value="([^"]+)" />', webpage,
35 # build list of available formats
36 # specified in http://www.presstv.ir/Scripts/playback.js
37 base_url = 'http://192.99.219.222:82/presstv'
40 'url': base_url + video_url,
41 'format': '1080p mp4',
44 'url': base_url + video_url.replace(".mp4", "_low800.mp4"),
48 'url': base_url + video_url.replace(".mp4", "_low400.mp4"),
52 'url': base_url + video_url.replace(".mp4", "_low200.mp4"),
59 # extract video metadata
60 title = self._html_search_meta('title', webpage, 'Title', True)
61 title = title.partition(' - ')[2]
63 description = self._html_search_regex(r'<div class="media-text nano-content">(.*?)</div>', webpage,
64 'Description', flags=re.DOTALL)
66 thumbnail = self._html_search_meta('og:image', webpage, 'Thumbnail', True)
68 year = str_to_int(self._search_regex(PressTVIE._VALID_URL, url, 'Upload year', group='y'))
69 month = str_to_int(self._search_regex(PressTVIE._VALID_URL, url, 'Upload month', group='m'))
70 day = str_to_int(self._search_regex(PressTVIE._VALID_URL, url, 'Upload day', group='d'))
71 upload_date = '%04d%02d%02d' % (year, month, day)
77 'thumbnail': thumbnail,
78 'upload_date': upload_date,
79 'description': description