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': r'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         # longer video id
  43         'url': 'http://coub.com/view/237d5l5h',
  44         'only_matching': True,
  45     }]
  46
  47     def _real_extract(self, url):
  48         video_id = self._match_id(url)
  49
  50         coub = self._download_json(
  51             'http://coub.com/api/v2/coubs/%s.json' % video_id, video_id)
  52
  53         if coub.get('error'):
  54             raise ExtractorError(
  55                 '%s said: %s' % (self.IE_NAME, coub['error']), expected=True)
  56
  57         title = coub['title']
  58
  59         file_versions = coub['file_versions']
  60
  61         QUALITIES = ('low', 'med', 'high')
  62
  63         MOBILE = 'mobile'
  64         IPHONE = 'iphone'
  65         HTML5 = 'html5'
  66
  67         SOURCE_PREFERENCE = (MOBILE, IPHONE, HTML5)
  68
  69         quality_key = qualities(QUALITIES)
  70         preference_key = qualities(SOURCE_PREFERENCE)
  71
  72         formats = []
  73
  74         for kind, items in file_versions.get(HTML5, {}).items():
  75             if kind not in ('video', 'audio'):
  76                 continue
  77             if not isinstance(items, dict):
  78                 continue
  79             for quality, item in items.items():
  80                 if not isinstance(item, dict):
  81                     continue
  82                 item_url = item.get('url')
  83                 if not item_url:
  84                     continue
  85                 formats.append({
  86                     'url': item_url,
  87                     'format_id': '%s-%s-%s' % (HTML5, kind, quality),
  88                     'filesize': int_or_none(item.get('size')),
  89                     'vcodec': 'none' if kind == 'audio' else None,
  90                     'quality': quality_key(quality),
  91                     'preference': preference_key(HTML5),
  92                 })
  93
  94         iphone_url = file_versions.get(IPHONE, {}).get('url')
  95         if iphone_url:
  96             formats.append({
  97                 'url': iphone_url,
  98                 'format_id': IPHONE,
  99                 'preference': preference_key(IPHONE),
 100             })
 101
 102         mobile_url = file_versions.get(MOBILE, {}).get('audio_url')
 103         if mobile_url:
 104             formats.append({
 105                 'url': mobile_url,
 106                 'format_id': '%s-audio' % MOBILE,
 107                 'preference': preference_key(MOBILE),
 108             })
 109
 110         self._sort_formats(formats)
 111
 112         thumbnail = coub.get('picture')
 113         duration = float_or_none(coub.get('duration'))
 114         timestamp = parse_iso8601(coub.get('published_at') or coub.get('created_at'))
 115         uploader = coub.get('channel', {}).get('title')
 116         uploader_id = coub.get('channel', {}).get('permalink')
 117
 118         view_count = int_or_none(coub.get('views_count') or coub.get('views_increase_count'))
 119         like_count = int_or_none(coub.get('likes_count'))
 120         repost_count = int_or_none(coub.get('recoubs_count'))
 121         comment_count = int_or_none(coub.get('comments_count'))
 122
 123         age_restricted = coub.get('age_restricted', coub.get('age_restricted_by_admin'))
 124         if age_restricted is not None:
 125             age_limit = 18 if age_restricted is True else 0
 126         else:
 127             age_limit = None
 128
 129         return {
 130             'id': video_id,
 131             'title': title,
 132             'thumbnail': thumbnail,
 133             'duration': duration,
 134             'timestamp': timestamp,
 135             'uploader': uploader,
 136             'uploader_id': uploader_id,
 137             'view_count': view_count,
 138             'like_count': like_count,
 139             'repost_count': repost_count,
 140             'comment_count': comment_count,
 141             'age_limit': age_limit,
 142             'formats': formats,
 143         }