Raphaël G. Git Repositories - youtubedl/blob - youtube_dl/extractor/dctp.py

   1 # coding: utf-8
   2 from __future__ import unicode_literals
   3
   4 from .common import InfoExtractor
   5 from ..utils import unified_strdate
   6
   7
   8 class DctpTvIE(InfoExtractor):
   9     _VALID_URL = r'https?://(?:www\.)?dctp\.tv/(#/)?filme/(?P<id>.+?)/$'
  10     _TEST = {
  11         'url': 'http://www.dctp.tv/filme/videoinstallation-fuer-eine-kaufhausfassade/',
  12         'md5': '174dd4a8a6225cf5655952f969cfbe24',
  13         'info_dict': {
  14             'id': '95eaa4f33dad413aa17b4ee613cccc6c',
  15             'display_id': 'videoinstallation-fuer-eine-kaufhausfassade',
  16             'ext': 'mp4',
  17             'title': 'Videoinstallation für eine Kaufhausfassade',
  18             'description': 'Kurzfilm',
  19             'upload_date': '20110407',
  20             'thumbnail': r're:^https?://.*\.jpg$',
  21         },
  22     }
  23
  24     def _real_extract(self, url):
  25         video_id = self._match_id(url)
  26         webpage = self._download_webpage(url, video_id)
  27
  28         object_id = self._html_search_meta('DC.identifier', webpage)
  29
  30         servers_json = self._download_json(
  31             'http://www.dctp.tv/elastic_streaming_client/get_streaming_server/',
  32             video_id, note='Downloading server list')
  33         server = servers_json[0]['server']
  34         m3u8_path = self._search_regex(
  35             r'\'([^\'"]+/playlist\.m3u8)"', webpage, 'm3u8 path')
  36         formats = self._extract_m3u8_formats(
  37             'http://%s%s' % (server, m3u8_path), video_id, ext='mp4',
  38             entry_protocol='m3u8_native')
  39
  40         title = self._og_search_title(webpage)
  41         description = self._html_search_meta('DC.description', webpage)
  42         upload_date = unified_strdate(
  43             self._html_search_meta('DC.date.created', webpage))
  44         thumbnail = self._og_search_thumbnail(webpage)
  45
  46         return {
  47             'id': object_id,
  48             'title': title,
  49             'formats': formats,
  50             'display_id': video_id,
  51             'description': description,
  52             'upload_date': upload_date,
  53             'thumbnail': thumbnail,
  54         }