_ Git - youtube-dl/blob - youtube_dl/extractor/indavideo.py

   1 # coding: utf-8
   2 from __future__ import unicode_literals
   3
   4 from .. import utils
   5 from .common import InfoExtractor
   6
   7
   8 class IndavideoIE(InfoExtractor):
   9     _VALID_URL = r'https?://(?:www\.)?indavideo\.hu/video/(?P<id>.+)'
  10     _TESTS = [
  11         {
  12             'url': 'http://indavideo.hu/video/Cicatanc',
  13             'md5': 'c8a507a1c7410685f83a06eaeeaafeab',
  14             'info_dict': {
  15                 'id': '1837039',
  16                 'title': 'Cicatánc',
  17                 'ext': 'mp4',
  18                 'display_id': 'Cicatanc',
  19                 'thumbnail': 're:^https?://.*\.jpg$',
  20                 'description': '',
  21                 'uploader': 'cukiajanlo',
  22                 'uploader_id': '83729',
  23                 'duration': 72,
  24                 'age_limit': 0,
  25                 'tags': ['tánc', 'cica', 'cuki', 'cukiajanlo', 'newsroom']
  26             },
  27         },
  28         {
  29             'url': 'http://indavideo.hu/video/Vicces_cica_1',
  30             'md5': '8c82244ba85d2a2310275b318eb51eac',
  31             'info_dict': {
  32                 'id': '1335611',
  33                 'title': 'Vicces cica',
  34                 'ext': 'mp4',
  35                 'display_id': 'Vicces_cica_1',
  36                 'thumbnail': 're:^https?://.*\.jpg$',
  37                 'description': 'Játszik a tablettel. :D',
  38                 'uploader': 'Jet_Pack',
  39                 'uploader_id': '491217',
  40                 'duration': 7,
  41                 'age_limit': 0,
  42                 'tags': ['vicces', 'macska', 'cica', 'ügyes', 'nevetés', 'játszik', 'Cukiság', 'Jet_Pack'],
  43             },
  44         },
  45     ]
  46
  47     def _real_extract(self, url):
  48         video_disp_id = self._match_id(url)
  49         webpage = self._download_webpage(url, video_disp_id)
  50
  51         embed_url = self._html_search_regex(r'<link rel="video_src" href="(.+?)"/>', webpage, 'embed_url')
  52         video_hash = embed_url.split('/')[-1]
  53
  54         payload = self._download_json('http://amfphp.indavideo.hu/SYm0json.php/player.playerHandler.getVideoData/' + video_hash, video_disp_id)
  55         video_info = payload['data']
  56
  57         thumbnails = video_info.get('thumbnails')
  58         if thumbnails:
  59             thumbnails = [{'url': self._proto_relative_url(x)} for x in thumbnails]
  60
  61         tags = video_info.get('tags')
  62         if tags:
  63             tags = [x['title'] for x in tags]
  64
  65         return {
  66             'id': video_info.get('id'),
  67             'title': video_info['title'],
  68             'url': video_info['video_file'],
  69             'ext': 'mp4',
  70             'display_id': video_disp_id,
  71             'thumbnails': thumbnails,
  72             'description': video_info.get('description'),
  73             'uploader': video_info.get('user_name'),
  74             # TODO: upload date (it's in CET/CEST)
  75             'uploader_id': video_info.get('user_id'),
  76             'duration': utils.int_or_none(video_info.get('length')),
  77             'age_limit': utils.int_or_none(video_info.get('age_limit')),
  78             'tags': tags,
  79         }