[coub] Add extractor (Closes #9609)
[youtube-dl] / youtube_dl / extractor / coub.py
1 # coding: utf-8
2 from __future__ import unicode_literals
3
4 from .common import InfoExtractor
5 from ..utils import (
6     ExtractorError,
7     float_or_none,
8     int_or_none,
9     parse_iso8601,
10     qualities,
11 )
12
13
14 class CoubIE(InfoExtractor):
15     _VALID_URL = r'(?:coub:|https?://(?:coub\.com/(?:view|embed|coubs)/|c-cdn\.coub\.com/fb-player\.swf\?.*\bcoub(?:ID|id)=))(?P<id>[\da-z]+)'
16
17     _TESTS = [{
18         'url': 'http://coub.com/view/5u5n1',
19         'info_dict': {
20             'id': '5u5n1',
21             'ext': 'mp4',
22             'title': 'The Matrix Moonwalk',
23             'thumbnail': 're:^https?://.*\.jpg$',
24             'duration': 4.6,
25             'timestamp': 1428527772,
26             'upload_date': '20150408',
27             'uploader': 'Артём Лоскутников',
28             'uploader_id': 'artyom.loskutnikov',
29             'view_count': int,
30             'like_count': int,
31             'repost_count': int,
32             'comment_count': int,
33             'age_limit': 0,
34         },
35     }, {
36         'url': 'http://c-cdn.coub.com/fb-player.swf?bot_type=vk&coubID=7w5a4',
37         'only_matching': True,
38     }, {
39         'url': 'coub:5u5n1',
40         'only_matching': True,
41     }]
42
43     def _real_extract(self, url):
44         video_id = self._match_id(url)
45
46         coub = self._download_json(
47             'http://coub.com/api/v2/coubs/%s.json' % video_id, video_id)
48
49         if coub.get('error'):
50             raise ExtractorError(
51                 '%s said: %s' % (self.IE_NAME, coub['error']), expected=True)
52
53         title = coub['title']
54
55         file_versions = coub['file_versions']
56
57         QUALITIES = ('low', 'med', 'high')
58
59         MOBILE = 'mobile'
60         IPHONE = 'iphone'
61         HTML5 = 'html5'
62
63         SOURCE_PREFERENCE = (MOBILE, IPHONE, HTML5)
64
65         quality_key = qualities(QUALITIES)
66         preference_key = qualities(SOURCE_PREFERENCE)
67
68         formats = []
69
70         for kind, items in file_versions.get(HTML5, {}).items():
71             if kind not in ('video', 'audio'):
72                 continue
73             if not isinstance(items, dict):
74                 continue
75             for quality, item in items.items():
76                 if not isinstance(item, dict):
77                     continue
78                 item_url = item.get('url')
79                 if not item_url:
80                     continue
81                 formats.append({
82                     'url': item_url,
83                     'format_id': '%s-%s-%s' % (HTML5, kind, quality),
84                     'filesize': int_or_none(item.get('size')),
85                     'vcodec': 'none' if kind == 'audio' else None,
86                     'quality': quality_key(quality),
87                     'preference': preference_key(HTML5),
88                 })
89
90         iphone_url = file_versions.get(IPHONE, {}).get('url')
91         if iphone_url:
92             formats.append({
93                 'url': iphone_url,
94                 'format_id': IPHONE,
95                 'preference': preference_key(IPHONE),
96             })
97
98         mobile_url = file_versions.get(MOBILE, {}).get('audio_url')
99         if mobile_url:
100             formats.append({
101                 'url': mobile_url,
102                 'format_id': '%s-audio' % MOBILE,
103                 'preference': preference_key(MOBILE),
104             })
105
106         self._sort_formats(formats)
107
108         thumbnail = coub.get('picture')
109         duration = float_or_none(coub.get('duration'))
110         timestamp = parse_iso8601(coub.get('published_at') or coub.get('created_at'))
111         uploader = coub.get('channel', {}).get('title')
112         uploader_id = coub.get('channel', {}).get('permalink')
113
114         view_count = int_or_none(coub.get('views_count') or coub.get('views_increase_count'))
115         like_count = int_or_none(coub.get('likes_count'))
116         repost_count = int_or_none(coub.get('recoubs_count'))
117         comment_count = int_or_none(coub.get('comments_count'))
118
119         age_restricted = coub.get('age_restricted', coub.get('age_restricted_by_admin'))
120         if age_restricted is not None:
121             age_limit = 18 if age_restricted is True else 0
122         else:
123             age_limit = None
124
125         return {
126             'id': video_id,
127             'title': title,
128             'thumbnail': thumbnail,
129             'duration': duration,
130             'timestamp': timestamp,
131             'uploader': uploader,
132             'uploader_id': uploader_id,
133             'view_count': view_count,
134             'like_count': like_count,
135             'repost_count': repost_count,
136             'comment_count': comment_count,
137             'age_limit': age_limit,
138             'formats': formats,
139         }