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

   1 from __future__ import unicode_literals
   2
   3 import json
   4
   5 from .common import InfoExtractor
   6 from ..utils import (
   7     float_or_none,
   8     int_or_none,
   9     sanitized_Request,
  10 )
  11
  12
  13 class CollegeRamaIE(InfoExtractor):
  14     _VALID_URL = r'https?://collegerama\.tudelft\.nl/Mediasite/Play/(?P<id>[\da-f]+)'
  15     _TESTS = [
  16         {
  17             'url': 'https://collegerama.tudelft.nl/Mediasite/Play/585a43626e544bdd97aeb71a0ec907a01d',
  18             'md5': '481fda1c11f67588c0d9d8fbdced4e39',
  19             'info_dict': {
  20                 'id': '585a43626e544bdd97aeb71a0ec907a01d',
  21                 'ext': 'mp4',
  22                 'title': 'Een nieuwe wereld: waarden, bewustzijn en techniek van de mensheid 2.0.',
  23                 'description': '',
  24                 'thumbnail': r're:^https?://.*\.jpg(?:\?.*?)?$',
  25                 'duration': 7713.088,
  26                 'timestamp': 1413309600,
  27                 'upload_date': '20141014',
  28             },
  29         },
  30         {
  31             'url': 'https://collegerama.tudelft.nl/Mediasite/Play/86a9ea9f53e149079fbdb4202b521ed21d?catalog=fd32fd35-6c99-466c-89d4-cd3c431bc8a4',
  32             'md5': 'ef1fdded95bdf19b12c5999949419c92',
  33             'info_dict': {
  34                 'id': '86a9ea9f53e149079fbdb4202b521ed21d',
  35                 'ext': 'wmv',
  36                 'title': '64ste Vakantiecursus: Afvalwater',
  37                 'description': 'md5:7fd774865cc69d972f542b157c328305',
  38                 'thumbnail': r're:^https?://.*\.jpg(?:\?.*?)?$',
  39                 'duration': 10853,
  40                 'timestamp': 1326446400,
  41                 'upload_date': '20120113',
  42             },
  43         },
  44     ]
  45
  46     def _real_extract(self, url):
  47         video_id = self._match_id(url)
  48
  49         player_options_request = {
  50             'getPlayerOptionsRequest': {
  51                 'ResourceId': video_id,
  52                 'QueryString': '',
  53             }
  54         }
  55
  56         request = sanitized_Request(
  57             'http://collegerama.tudelft.nl/Mediasite/PlayerService/PlayerService.svc/json/GetPlayerOptions',
  58             json.dumps(player_options_request))
  59         request.add_header('Content-Type', 'application/json')
  60
  61         player_options = self._download_json(request, video_id)
  62
  63         presentation = player_options['d']['Presentation']
  64         title = presentation['Title']
  65         description = presentation.get('Description')
  66         thumbnail = None
  67         duration = float_or_none(presentation.get('Duration'), 1000)
  68         timestamp = int_or_none(presentation.get('UnixTime'), 1000)
  69
  70         formats = []
  71         for stream in presentation['Streams']:
  72             for video in stream['VideoUrls']:
  73                 thumbnail_url = stream.get('ThumbnailUrl')
  74                 if thumbnail_url:
  75                     thumbnail = 'http://collegerama.tudelft.nl' + thumbnail_url
  76                 format_id = video['MediaType']
  77                 if format_id == 'SS':
  78                     continue
  79                 formats.append({
  80                     'url': video['Location'],
  81                     'format_id': format_id,
  82                 })
  83         self._sort_formats(formats)
  84
  85         return {
  86             'id': video_id,
  87             'title': title,
  88             'description': description,
  89             'thumbnail': thumbnail,
  90             'duration': duration,
  91             'timestamp': timestamp,
  92             'formats': formats,
  93         }