[generic,vodplatform] improve embed regex
[youtube-dl] / youtube_dl / extractor / vodplatform.py
1 # coding: utf-8
2 from __future__ import unicode_literals
3
4 from .common import InfoExtractor
5 from ..utils import unescapeHTML
6
7
8 class VODPlatformIE(InfoExtractor):
9     _VALID_URL = r'https?://(?:www\.)?vod-platform\.net/[eE]mbed/(?P<id>[^/?#]+)'
10     _TEST = {
11         # from http://www.lbcgroup.tv/watch/chapter/29143/52844/%D8%A7%D9%84%D9%86%D8%B5%D8%B1%D8%A9-%D9%81%D9%8A-%D8%B6%D9%8A%D8%A7%D9%81%D8%A9-%D8%A7%D9%84%D9%80-cnn/ar
12         'url': 'http://vod-platform.net/embed/RufMcytHDolTH1MuKHY9Fw',
13         'md5': '1db2b7249ce383d6be96499006e951fc',
14         'info_dict': {
15             'id': 'RufMcytHDolTH1MuKHY9Fw',
16             'ext': 'mp4',
17             'title': 'LBCi News_ النصرة في ضيافة الـ "سي.أن.أن"',
18         }
19     }
20
21     def _real_extract(self, url):
22         video_id = self._match_id(url)
23         webpage = self._download_webpage(url, video_id)
24
25         title = unescapeHTML(self._og_search_title(webpage))
26         hidden_inputs = self._hidden_inputs(webpage)
27
28         base_url = self._search_regex(
29             '(.*/)(?:playlist.m3u8|manifest.mpd)',
30             hidden_inputs.get('HiddenmyhHlsLink') or hidden_inputs['HiddenmyDashLink'],
31             'base url')
32         formats = self._extract_m3u8_formats(
33             base_url + 'playlist.m3u8', video_id, 'mp4',
34             'm3u8_native', m3u8_id='hls', fatal=False)
35         formats.extend(self._extract_mpd_formats(
36             base_url + 'manifest.mpd', video_id,
37             mpd_id='dash', fatal=False))
38         rtmp_formats = self._extract_smil_formats(
39             base_url + 'jwplayer.smil', video_id, fatal=False)
40         for rtmp_format in rtmp_formats:
41             rtsp_format = rtmp_format.copy()
42             rtsp_format['url'] = '%s/%s' % (rtmp_format['url'], rtmp_format['play_path'])
43             del rtsp_format['play_path']
44             del rtsp_format['ext']
45             rtsp_format.update({
46                 'url': rtsp_format['url'].replace('rtmp://', 'rtsp://'),
47                 'format_id': rtmp_format['format_id'].replace('rtmp', 'rtsp'),
48                 'protocol': 'rtsp',
49             })
50             formats.extend([rtmp_format, rtsp_format])
51         self._sort_formats(formats)
52
53         return {
54             'id': video_id,
55             'title': title,
56             'thumbnail': hidden_inputs.get('HiddenThumbnail') or self._og_search_thumbnail(webpage),
57             'formats': formats,
58         }