[wakanim] Add extractor (closes #14374)
[youtube-dl] / youtube_dl / extractor / wakanim.py
1 # coding: utf-8
2 from __future__ import unicode_literals
3
4 from .common import InfoExtractor
5 from ..utils import (
6     merge_dicts,
7     urljoin,
8 )
9
10
11 class WakanimIE(InfoExtractor):
12     _VALID_URL = r'https://(?:www\.)?wakanim\.tv/[^/]+/v2/catalogue/episode/(?P<id>\d+)'
13     _TEST = {
14         'url': 'https://www.wakanim.tv/de/v2/catalogue/episode/2997/the-asterisk-war-omu-staffel-1-episode-02-omu',
15         'info_dict': {
16             'id': '2997',
17             'ext': 'mp4',
18             'title': 'Episode 02',
19             'description': 'md5:2927701ea2f7e901de8bfa8d39b2852d',
20             'series': 'The Asterisk War  (OmU.)',
21             'season_number': 1,
22             'episode': 'Episode 02',
23             'episode_number': 2,
24         },
25         'params': {
26             'format': 'bestvideo',
27             'skip_download': True,
28         },
29     }
30
31     def _real_extract(self, url):
32         video_id = self._match_id(url)
33
34         webpage = self._download_webpage(url, video_id)
35
36         m3u8_url = urljoin(url, self._search_regex(
37             r'file\s*:\s*(["\'])(?P<url>(?:(?!\1).)+)\1', webpage, 'm3u8 url',
38             group='url'))
39
40         formats = self._extract_m3u8_formats(
41             m3u8_url, video_id, 'mp4', entry_protocol='m3u8_native',
42             m3u8_id='hls')
43
44         info = self._search_json_ld(webpage, video_id, default={})
45
46         title = self._search_regex(
47             (r'<h1[^>]+\bclass=["\']episode_h1[^>]+\btitle=(["\'])(?P<title>(?:(?!\1).)+)\1',
48              r'<span[^>]+\bclass=["\']episode_title["\'][^>]*>(?P<title>[^<]+)'),
49             webpage, 'title', default=None, group='title')
50
51         return merge_dicts(info, {
52             'id': video_id,
53             'title': title,
54             'formats': formats,
55         })