]>
Commit | Line | Data |
---|---|---|
1132eae5 NJ |
1 | # coding: utf-8 |
2 | from __future__ import unicode_literals | |
3 | ||
4 | import re | |
5 | ||
6 | from .common import InfoExtractor | |
7 | ||
8 | ||
9 | class GazetaIE(InfoExtractor): | |
759e37c9 | 10 | _VALID_URL = r'(?P<url>https?://(?:www\.)?gazeta\.ru/(?:[^/]+/)?video/(?:main/)*(?:\d{4}/\d{2}/\d{2}/)?(?P<id>[A-Za-z0-9-_.]+)\.s?html)' |
bdf6eee0 | 11 | _TESTS = [{ |
1132eae5 NJ |
12 | 'url': 'http://www.gazeta.ru/video/main/zadaite_vopros_vladislavu_yurevichu.shtml', |
13 | 'md5': 'd49c9bdc6e5a7888f27475dc215ee789', | |
14 | 'info_dict': { | |
15 | 'id': '205566', | |
16 | 'ext': 'mp4', | |
17 | 'title': '«70–80 процентов гражданских в Донецке на грани голода»', | |
18 | 'description': 'md5:38617526050bd17b234728e7f9620a71', | |
ec85ded8 | 19 | 'thumbnail': r're:^https?://.*\.jpg', |
1132eae5 | 20 | }, |
759e37c9 | 21 | 'skip': 'video not found', |
bdf6eee0 S |
22 | }, { |
23 | 'url': 'http://www.gazeta.ru/lifestyle/video/2015/03/08/master-klass_krasivoi_byt._delaem_vesennii_makiyazh.shtml', | |
24 | 'only_matching': True, | |
759e37c9 YCH |
25 | }, { |
26 | 'url': 'http://www.gazeta.ru/video/main/main/2015/06/22/platit_ili_ne_platit_po_isku_yukosa.shtml', | |
237a4110 | 27 | 'md5': '37f19f78355eb2f4256ee1688359f24c', |
759e37c9 YCH |
28 | 'info_dict': { |
29 | 'id': '252048', | |
30 | 'ext': 'mp4', | |
31 | 'title': '"Если по иску ЮКОСа придется платить, это будет большой удар по бюджету"', | |
32 | }, | |
759e37c9 | 33 | 'add_ie': ['EaglePlatform'], |
bdf6eee0 | 34 | }] |
1132eae5 NJ |
35 | |
36 | def _real_extract(self, url): | |
37 | mobj = re.match(self._VALID_URL, url) | |
38 | ||
39 | display_id = mobj.group('id') | |
40 | embed_url = '%s?p=embed' % mobj.group('url') | |
41 | embed_page = self._download_webpage( | |
42 | embed_url, display_id, 'Downloading embed page') | |
43 | ||
44 | video_id = self._search_regex( | |
45 | r'<div[^>]*?class="eagleplayer"[^>]*?data-id="([^"]+)"', embed_page, 'video id') | |
46 | ||
47 | return self.url_result( | |
48 | 'eagleplatform:gazeta.media.eagleplatform.com:%s' % video_id, 'EaglePlatform') |