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

   1 # coding: utf-8
   2 from __future__ import unicode_literals
   3
   4 import re
   5 import json
   6
   7 from .common import InfoExtractor
   8 from ..compat import compat_urlparse
   9 from ..utils import (
  10     ExtractorError,
  11     unsmuggle_url,
  12     mimetype2ext,
  13     float_or_none,
  14 )
  15
  16
  17 class MediasiteIE(InfoExtractor):
  18     _VALID_URL = r'''(?xi)
  19         https?://[a-z0-9\-\.:\[\]]+/Mediasite/Play/
  20         (?P<id>[0-9a-f]{32,34})
  21         (?P<QueryString>\?[^#]+|)
  22     '''
  23     _TESTS = [
  24         {
  25             'url': 'https://hitsmediaweb.h-its.org/mediasite/Play/2db6c271681e4f199af3c60d1f82869b1d',
  26             'info_dict': {
  27                 'id': '2db6c271681e4f199af3c60d1f82869b1d',
  28                 'ext': 'mp4',
  29                 'title': 'Lecture: Tuesday, September 20, 2016 - Sir Andrew Wiles',
  30                 'description': 'Sir Andrew Wiles: “Equations in arithmetic”\\n\\nI will describe some of the interactions between modern number theory and the problem of solving equations in rational numbers or integers\\u0027.',
  31                 'timestamp': 1474268400.0,
  32                 'upload_date': '20160919',
  33             },
  34         },
  35         {
  36             'url': 'http://mediasite.uib.no/Mediasite/Play/90bb363295d945d6b548c867d01181361d?catalog=a452b7df-9ae1-46b7-a3ba-aceeb285f3eb',
  37             'info_dict': {
  38                 'id': '90bb363295d945d6b548c867d01181361d',
  39                 'ext': 'mp4',
  40                 'upload_date': '20150429',
  41                 'title': '5) IT-forum 2015-Dag 1  - Dungbeetle -  How and why Rain created a tiny bug tracker for Unity',
  42                 'timestamp': 1430311380.0,
  43             },
  44         },
  45         {
  46             'url': 'https://collegerama.tudelft.nl/Mediasite/Play/585a43626e544bdd97aeb71a0ec907a01d',
  47             'md5': '481fda1c11f67588c0d9d8fbdced4e39',
  48             'info_dict': {
  49                 'id': '585a43626e544bdd97aeb71a0ec907a01d',
  50                 'ext': 'mp4',
  51                 'title': 'Een nieuwe wereld: waarden, bewustzijn en techniek van de mensheid 2.0.',
  52                 'description': '',
  53                 'thumbnail': r're:^https?://.*\.jpg(?:\?.*)?$',
  54                 'duration': 7713.088,
  55                 'timestamp': 1413309600,
  56                 'upload_date': '20141014',
  57             },
  58         },
  59         {
  60             'url': 'https://collegerama.tudelft.nl/Mediasite/Play/86a9ea9f53e149079fbdb4202b521ed21d?catalog=fd32fd35-6c99-466c-89d4-cd3c431bc8a4',
  61             'md5': 'ef1fdded95bdf19b12c5999949419c92',
  62             'info_dict': {
  63                 'id': '86a9ea9f53e149079fbdb4202b521ed21d',
  64                 'ext': 'wmv',
  65                 'title': '64ste Vakantiecursus: Afvalwater',
  66                 'description': 'md5:7fd774865cc69d972f542b157c328305',
  67                 'thumbnail': r're:^https?://.*\.jpg(?:\?.*?)?$',
  68                 'duration': 10853,
  69                 'timestamp': 1326446400,
  70                 'upload_date': '20120113',
  71             },
  72         },
  73         {
  74             'url': 'http://digitalops.sandia.gov/Mediasite/Play/24aace4429fc450fb5b38cdbf424a66e1d',
  75             'md5': '9422edc9b9a60151727e4b6d8bef393d',
  76             'info_dict': {
  77                 'id': '24aace4429fc450fb5b38cdbf424a66e1d',
  78                 'ext': 'mp4',
  79                 'title': 'Xyce Software Training - Section 1',
  80                 'description': r're:(?s)SAND Number: SAND 2013-7800.{200,}',
  81                 'upload_date': '20120409',
  82                 'timestamp': 1333983600,
  83                 'duration': 7794,
  84             }
  85         }
  86     ]
  87
  88     # look in Mediasite.Core.js (Mediasite.ContentStreamType[*])
  89     _STREAM_TYPES = {
  90         0: 'video1', # the main video
  91         2: 'slide',
  92         3: 'presentation',
  93         4: 'video2', # screencast?
  94         5: 'video3',
  95     }
  96
  97     def _real_extract(self, url):
  98         url, data = unsmuggle_url(url, {})
  99         mobj = re.match(self._VALID_URL, url)
 100         ResourceId = mobj.group('id')
 101         QueryString = mobj.group('QueryString')
 102
 103         webpage = self._download_webpage(url, ResourceId) # XXX: add UrlReferrer?
 104
 105         # XXX: might have also extracted UrlReferrer and QueryString from the html
 106         ServicePath = compat_urlparse.urljoin(url, self._html_search_regex(
 107             r'<div id="ServicePath">(.+?)</div>', webpage, ResourceId,
 108             default='/Mediasite/PlayerService/PlayerService.svc/json'))
 109
 110         PlayerOptions = self._download_json(
 111             '%s/GetPlayerOptions' % (ServicePath), ResourceId,
 112             headers={
 113                 'Content-type': 'application/json; charset=utf-8',
 114                 'X-Requested-With': 'XMLHttpRequest',
 115             },
 116             data=json.dumps({
 117                 'getPlayerOptionsRequest': {
 118                     'ResourceId': ResourceId,
 119                     'QueryString': QueryString,
 120                     'UrlReferrer': data.get('UrlReferrer', ''),
 121                     'UseScreenReader': False,
 122                 }
 123             }).encode('utf-8'))
 124         Presentation = PlayerOptions['d']['Presentation']
 125         if Presentation is None:
 126             raise ExtractorError('Mediasite says: %s' %
 127                 (PlayerOptions['d']['PlayerPresentationStatusMessage'],),
 128                 expected=True)
 129
 130         thumbnails = []
 131         formats = []
 132         for snum, Stream in enumerate(Presentation['Streams']):
 133             stream_type = self._STREAM_TYPES.get(
 134                 Stream['StreamType'], 'type%u' % Stream['StreamType'])
 135
 136             stream_formats = []
 137             for unum, VideoUrl in enumerate(Stream['VideoUrls']):
 138                 url = VideoUrl['Location']
 139                 # XXX: if Stream.get('CanChangeScheme', False), switch scheme to HTTP/HTTPS
 140
 141                 if VideoUrl['MediaType'] == 'SS':
 142                     stream_formats.extend(self._extract_ism_formats(
 143                         url, ResourceId, ism_id='%s-%u.%u' % (stream_type, snum, unum)))
 144                     continue
 145
 146                 stream_formats.append({
 147                     'format_id': '%s-%u.%u' % (stream_type, snum, unum),
 148                     'url': url,
 149                     'ext': mimetype2ext(VideoUrl['MimeType']),
 150                 })
 151
 152             # TODO: if Stream['HasSlideContent']:
 153             # synthesise an MJPEG video stream '%s-%u.slides' % (stream_type, snum)
 154             # from Stream['Slides']
 155             # this will require writing a custom downloader...
 156
 157             # disprefer 'secondary' streams
 158             if Stream['StreamType'] != 0:
 159                 for fmt in stream_formats:
 160                     fmt['preference'] = -1
 161
 162             ThumbnailUrl = Stream.get('ThumbnailUrl')
 163             if ThumbnailUrl:
 164                 thumbnails.append({
 165                     'id': '%s-%u' % (stream_type, snum),
 166                     'url': compat_urlparse.urljoin(url, ThumbnailUrl),
 167                     'preference': -1 if Stream['StreamType'] != 0 else 0,
 168                 })
 169             formats.extend(stream_formats)
 170
 171         self._sort_formats(formats)
 172
 173         # XXX: Presentation['Presenters']
 174         # XXX: Presentation['Transcript']
 175
 176         return {
 177             'id': ResourceId,
 178             'title': Presentation['Title'],
 179             'description': Presentation.get('Description'),
 180             'duration': float_or_none(Presentation.get('Duration'), 1000),
 181             'timestamp': float_or_none(Presentation.get('UnixTime'), 1000),
 182             'formats': formats,
 183             'thumbnails': thumbnails,
 184         }