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': 'Artyom Loskutnikov',
  28             'uploader_id': 'artyom.loskutnikov',
  29             'view_count': int,
  30             'like_count': int,
  31             'repost_count': int,
  32             'age_limit': 0,
  33         },
  34     }, {
  35         'url': 'http://c-cdn.coub.com/fb-player.swf?bot_type=vk&coubID=7w5a4',
  36         'only_matching': True,
  37     }, {
  38         'url': 'coub:5u5n1',
  39         'only_matching': True,
  40     }, {
  41         # longer video id
  42         'url': 'http://coub.com/view/237d5l5h',
  43         'only_matching': True,
  44     }]
  45
  46     def _real_extract(self, url):
  47         video_id = self._match_id(url)
  48
  49         coub = self._download_json(
  50             'http://coub.com/api/v2/coubs/%s.json' % video_id, video_id)
  51
  52         if coub.get('error'):
  53             raise ExtractorError(
  54                 '%s said: %s' % (self.IE_NAME, coub['error']), expected=True)
  55
  56         title = coub['title']
  57
  58         file_versions = coub['file_versions']
  59
  60         QUALITIES = ('low', 'med', 'high')
  61
  62         MOBILE = 'mobile'
  63         IPHONE = 'iphone'
  64         HTML5 = 'html5'
  65
  66         SOURCE_PREFERENCE = (MOBILE, IPHONE, HTML5)
  67
  68         quality_key = qualities(QUALITIES)
  69         preference_key = qualities(SOURCE_PREFERENCE)
  70
  71         formats = []
  72
  73         for kind, items in file_versions.get(HTML5, {}).items():
  74             if kind not in ('video', 'audio'):
  75                 continue
  76             if not isinstance(items, dict):
  77                 continue
  78             for quality, item in items.items():
  79                 if not isinstance(item, dict):
  80                     continue
  81                 item_url = item.get('url')
  82                 if not item_url:
  83                     continue
  84                 formats.append({
  85                     'url': item_url,
  86                     'format_id': '%s-%s-%s' % (HTML5, kind, quality),
  87                     'filesize': int_or_none(item.get('size')),
  88                     'vcodec': 'none' if kind == 'audio' else None,
  89                     'quality': quality_key(quality),
  90                     'preference': preference_key(HTML5),
  91                 })
  92
  93         iphone_url = file_versions.get(IPHONE, {}).get('url')
  94         if iphone_url:
  95             formats.append({
  96                 'url': iphone_url,
  97                 'format_id': IPHONE,
  98                 'preference': preference_key(IPHONE),
  99             })
 100
 101         mobile_url = file_versions.get(MOBILE, {}).get('audio_url')
 102         if mobile_url:
 103             formats.append({
 104                 'url': mobile_url,
 105                 'format_id': '%s-audio' % MOBILE,
 106                 'preference': preference_key(MOBILE),
 107             })
 108
 109         self._sort_formats(formats)
 110
 111         thumbnail = coub.get('picture')
 112         duration = float_or_none(coub.get('duration'))
 113         timestamp = parse_iso8601(coub.get('published_at') or coub.get('created_at'))
 114         uploader = coub.get('channel', {}).get('title')
 115         uploader_id = coub.get('channel', {}).get('permalink')
 116
 117         view_count = int_or_none(coub.get('views_count') or coub.get('views_increase_count'))
 118         like_count = int_or_none(coub.get('likes_count'))
 119         repost_count = int_or_none(coub.get('recoubs_count'))
 120
 121         age_restricted = coub.get('age_restricted', coub.get('age_restricted_by_admin'))
 122         if age_restricted is not None:
 123             age_limit = 18 if age_restricted is True else 0
 124         else:
 125             age_limit = None
 126
 127         return {
 128             'id': video_id,
 129             'title': title,
 130             'thumbnail': thumbnail,
 131             'duration': duration,
 132             'timestamp': timestamp,
 133             'uploader': uploader,
 134             'uploader_id': uploader_id,
 135             'view_count': view_count,
 136             'like_count': like_count,
 137             'repost_count': repost_count,
 138             'age_limit': age_limit,
 139             'formats': formats,
 140         }