]>
Commit | Line | Data |
---|---|---|
6e925598 | 1 | # coding: utf-8 |
d2b9f362 CS |
2 | from __future__ import unicode_literals |
3 | ||
c319d1c4 | 4 | |
d2b9f362 | 5 | from .common import InfoExtractor |
c319d1c4 S |
6 | from ..utils import ( |
7 | determine_ext, | |
8 | unescapeHTML, | |
9 | ) | |
d2b9f362 CS |
10 | |
11 | ||
12 | class CJSWIE(InfoExtractor): | |
c319d1c4 | 13 | _VALID_URL = r'https?://(?:www\.)?cjsw\.com/program/(?P<program>[^/]+)/episode/(?P<id>\d+)' |
0d2f0b03 | 14 | _TESTS = [{ |
d2b9f362 CS |
15 | 'url': 'http://cjsw.com/program/freshly-squeezed/episode/20170620', |
16 | 'md5': 'cee14d40f1e9433632c56e3d14977120', | |
17 | 'info_dict': { | |
c319d1c4 | 18 | 'id': '91d9f016-a2e7-46c5-8dcb-7cbcd7437c41', |
d2b9f362 | 19 | 'ext': 'mp3', |
c319d1c4 S |
20 | 'title': 'Freshly Squeezed – Episode June 20, 2017', |
21 | 'description': 'md5:c967d63366c3898a80d0c7b0ff337202', | |
22 | 'series': 'Freshly Squeezed', | |
23 | 'episode_id': '20170620', | |
24 | }, | |
0d2f0b03 S |
25 | }, { |
26 | # no description | |
27 | 'url': 'http://cjsw.com/program/road-pops/episode/20170707/', | |
28 | 'only_matching': True, | |
29 | }] | |
d2b9f362 CS |
30 | |
31 | def _real_extract(self, url): | |
5ad28e7f | 32 | mobj = self._match_valid_url(url) |
c319d1c4 S |
33 | program, episode_id = mobj.group('program', 'id') |
34 | audio_id = '%s/%s' % (program, episode_id) | |
d2b9f362 CS |
35 | |
36 | webpage = self._download_webpage(url, episode_id) | |
37 | ||
c319d1c4 S |
38 | title = unescapeHTML(self._search_regex( |
39 | (r'<h1[^>]+class=["\']episode-header__title["\'][^>]*>(?P<title>[^<]+)', | |
40 | r'data-audio-title=(["\'])(?P<title>(?:(?!\1).)+)\1'), | |
41 | webpage, 'title', group='title')) | |
42 | ||
43 | audio_url = self._search_regex( | |
44 | r'<button[^>]+data-audio-src=(["\'])(?P<url>(?:(?!\1).)+)\1', | |
45 | webpage, 'audio url', group='url') | |
46 | ||
47 | audio_id = self._search_regex( | |
48 | r'/([\da-f]{8}-[\da-f]{4}-[\da-f]{4}-[\da-f]{4}-[\da-f]{12})\.mp3', | |
49 | audio_url, 'audio id', default=audio_id) | |
50 | ||
d2b9f362 | 51 | formats = [{ |
c319d1c4 S |
52 | 'url': audio_url, |
53 | 'ext': determine_ext(audio_url, 'mp3'), | |
d2b9f362 CS |
54 | 'vcodec': 'none', |
55 | }] | |
c319d1c4 S |
56 | |
57 | description = self._html_search_regex( | |
0d2f0b03 S |
58 | r'<p>(?P<description>.+?)</p>', webpage, 'description', |
59 | default=None) | |
c319d1c4 S |
60 | series = self._search_regex( |
61 | r'data-showname=(["\'])(?P<name>(?:(?!\1).)+)\1', webpage, | |
62 | 'series', default=program, group='name') | |
63 | ||
d2b9f362 | 64 | return { |
c319d1c4 | 65 | 'id': audio_id, |
d2b9f362 CS |
66 | 'title': title, |
67 | 'description': description, | |
68 | 'formats': formats, | |
c319d1c4 S |
69 | 'series': series, |
70 | 'episode_id': episode_id, | |
d2b9f362 | 71 | } |