]>
Commit | Line | Data |
---|---|---|
cbb12756 OB |
1 | import re |
2 | ||
3 | from .common import InfoExtractor | |
cbb12756 | 4 | from .brightcove import BrightcoveNewIE |
4d058c98 S |
5 | from ..utils import ( |
6 | int_or_none, | |
7 | parse_age_limit, | |
8 | smuggle_url, | |
9 | unescapeHTML, | |
10 | ) | |
cbb12756 OB |
11 | |
12 | ||
13 | class VrakIE(InfoExtractor): | |
4d058c98 | 14 | _VALID_URL = r'https?://(?:www\.)?vrak\.tv/videos\?.*?\btarget=(?P<id>[\d.]+)' |
cbb12756 | 15 | _TEST = { |
4d058c98 | 16 | 'url': 'http://www.vrak.tv/videos?target=1.2306782&filtre=emission&id=1.1806721', |
cbb12756 | 17 | 'info_dict': { |
4d058c98 | 18 | 'id': '5345661243001', |
cbb12756 | 19 | 'ext': 'mp4', |
4d058c98 S |
20 | 'title': 'Obésité, film de hockey et Roseline Filion', |
21 | 'timestamp': 1488492126, | |
22 | 'upload_date': '20170302', | |
cbb12756 | 23 | 'uploader_id': '2890187628001', |
4d058c98 S |
24 | 'creator': 'VRAK.TV', |
25 | 'age_limit': 8, | |
26 | 'series': 'ALT (Actualité Légèrement Tordue)', | |
27 | 'episode': 'Obésité, film de hockey et Roseline Filion', | |
28 | 'tags': list, | |
29 | }, | |
30 | 'params': { | |
31 | 'skip_download': True, | |
32 | }, | |
cbb12756 | 33 | } |
4d058c98 | 34 | BRIGHTCOVE_URL_TEMPLATE = 'http://players.brightcove.net/2890187628001/default_default/index.html?videoId=%s' |
cbb12756 OB |
35 | |
36 | def _real_extract(self, url): | |
4d058c98 | 37 | video_id = self._match_id(url) |
cbb12756 | 38 | |
4d058c98 | 39 | webpage = self._download_webpage(url, video_id) |
cbb12756 | 40 | |
4d058c98 S |
41 | title = self._html_search_regex( |
42 | r'<h\d\b[^>]+\bclass=["\']videoTitle["\'][^>]*>([^<]+)', | |
43 | webpage, 'title', default=None) or self._og_search_title(webpage) | |
cbb12756 | 44 | |
4d058c98 S |
45 | content = self._parse_json( |
46 | self._search_regex( | |
47 | r'data-player-options-content=(["\'])(?P<content>{.+?})\1', | |
48 | webpage, 'content', default='{}', group='content'), | |
49 | video_id, transform_source=unescapeHTML) | |
cbb12756 | 50 | |
4d058c98 S |
51 | ref_id = content.get('refId') or self._search_regex( |
52 | r'refId":"([^&]+)"', webpage, 'ref id') | |
cbb12756 | 53 | |
4d058c98 S |
54 | brightcove_id = self._search_regex( |
55 | r'''(?x) | |
56 | java\.lang\.String\s+value\s*=\s*["']brightcove\.article\.\d+\.%s | |
57 | [^>]* | |
58 | java\.lang\.String\s+value\s*=\s*["'](\d+) | |
59 | ''' % re.escape(ref_id), webpage, 'brightcove id') | |
60 | ||
61 | return { | |
62 | '_type': 'url_transparent', | |
63 | 'ie_key': BrightcoveNewIE.ie_key(), | |
64 | 'url': smuggle_url( | |
65 | self.BRIGHTCOVE_URL_TEMPLATE % brightcove_id, | |
66 | {'geo_countries': ['CA']}), | |
67 | 'id': brightcove_id, | |
68 | 'description': content.get('description'), | |
69 | 'creator': content.get('brand'), | |
70 | 'age_limit': parse_age_limit(content.get('rating')), | |
71 | 'series': content.get('showName') or content.get( | |
72 | 'episodeName'), # this is intentional | |
73 | 'season_number': int_or_none(content.get('seasonNumber')), | |
74 | 'episode': title, | |
75 | 'episode_number': int_or_none(content.get('episodeNumber')), | |
76 | 'tags': content.get('tags', []), | |
77 | } |