[ministrygrid] Fix extraction and modernize
[youtube-dl] / youtube_dl / extractor / ministrygrid.py
1 from __future__ import unicode_literals
2
3 from .common import InfoExtractor
4 from ..utils import (
5     ExtractorError,
6     smuggle_url,
7 )
8
9
10 class MinistryGridIE(InfoExtractor):
11     _VALID_URL = r'https?://www\.ministrygrid.com/([^/?#]*/)*(?P<id>[^/#?]+)/?(?:$|[?#])'
12
13     _TEST = {
14         'url': 'http://www.ministrygrid.com/training-viewer/-/training/t4g-2014-conference/the-gospel-by-numbers-4/the-gospel-by-numbers',
15         'md5': '844be0d2a1340422759c2a9101bab017',
16         'info_dict': {
17             'id': '3453494717001',
18             'ext': 'mp4',
19             'title': 'The Gospel by Numbers',
20         },
21         'params': {
22             # m3u8 download
23             'skip_download': True,
24         },
25     }
26
27     def _real_extract(self, url):
28         video_id = self._match_id(url)
29
30         webpage = self._download_webpage(url, video_id)
31         portlets = self._parse_json(self._search_regex(
32             r'Liferay\.Portlet\.list=(\[.+?\])', webpage, 'portlet list'),
33             video_id)
34         pl_id = self._search_regex(
35             r'getPlid:function\(\){return"(\d+)"}', webpage, 'p_l_id')
36
37         for i, portlet in enumerate(portlets):
38             portlet_url = 'http://www.ministrygrid.com/c/portal/render_portlet?p_l_id=%s&p_p_id=%s' % (pl_id, portlet)
39             portlet_code = self._download_webpage(
40                 portlet_url, video_id,
41                 note='Looking in portlet %s (%d/%d)' % (portlet, i + 1, len(portlets)),
42                 fatal=False)
43             video_iframe_url = self._search_regex(
44                 r'<iframe.*?src="([^"]+)"', portlet_code, 'video iframe',
45                 default=None)
46             if video_iframe_url:
47                 return self.url_result(
48                     smuggle_url(video_iframe_url, {'force_videoid': video_id}),
49                     video_id=video_id)
50
51         raise ExtractorError('Could not find video iframe in any portlets')