[youtube] Fix extraction.
[youtube-dl] / youtube_dl / extractor / minoto.py
1 # coding: utf-8
2 from __future__ import unicode_literals
3
4 import re
5
6 from .common import InfoExtractor
7 from ..utils import (
8     int_or_none,
9     parse_codecs,
10 )
11
12
13 class MinotoIE(InfoExtractor):
14     _VALID_URL = r'(?:minoto:|https?://(?:play|iframe|embed)\.minoto-video\.com/(?P<player_id>[0-9]+)/)(?P<id>[a-zA-Z0-9]+)'
15
16     def _real_extract(self, url):
17         mobj = re.match(self._VALID_URL, url)
18         player_id = mobj.group('player_id') or '1'
19         video_id = mobj.group('id')
20         video_data = self._download_json('http://play.minoto-video.com/%s/%s.js' % (player_id, video_id), video_id)
21         video_metadata = video_data['video-metadata']
22         formats = []
23         for fmt in video_data['video-files']:
24             fmt_url = fmt.get('url')
25             if not fmt_url:
26                 continue
27             container = fmt.get('container')
28             if container == 'hls':
29                 formats.extend(fmt_url, video_id, 'mp4', m3u8_id='hls', fatal=False)
30             else:
31                 fmt_profile = fmt.get('profile') or {}
32                 formats.append({
33                     'format_id': fmt_profile.get('name-short'),
34                     'format_note': fmt_profile.get('name'),
35                     'url': fmt_url,
36                     'container': container,
37                     'tbr': int_or_none(fmt.get('bitrate')),
38                     'filesize': int_or_none(fmt.get('filesize')),
39                     'width': int_or_none(fmt.get('width')),
40                     'height': int_or_none(fmt.get('height')),
41                     'codecs': parse_codecs(fmt.get('codecs')),
42                 })
43         self._sort_formats(formats)
44
45         return {
46             'id': video_id,
47             'title': video_metadata['title'],
48             'description': video_metadata.get('description'),
49             'thumbnail': video_metadata.get('video-poster', {}).get('url'),
50             'formats': formats,
51         }