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

   1 # coding: utf-8
   2 from __future__ import unicode_literals
   3
   4 from .common import InfoExtractor
   5
   6
   7 class TenPlayIE(InfoExtractor):
   8     _VALID_URL = r'https?://(?:www\.)?ten(play)?\.com\.au/.+'
   9     _TEST = {
  10         'url': 'http://tenplay.com.au/ten-insider/extra/season-2013/tenplay-tv-your-way',
  11         'info_dict': {
  12             'id': '2695695426001',
  13             'ext': 'flv',
  14             'title': 'TENplay: TV your way',
  15             'description': 'Welcome to a new TV experience. Enjoy a taste of the TENplay benefits.',
  16             'timestamp': 1380150606.889,
  17             'upload_date': '20130925',
  18             'uploader': 'TENplay',
  19         },
  20         'params': {
  21             'skip_download': True,  # Requires rtmpdump
  22         }
  23     }
  24
  25     _video_fields = [
  26         "id", "name", "shortDescription", "longDescription", "creationDate",
  27         "publishedDate", "lastModifiedDate", "customFields", "videoStillURL",
  28         "thumbnailURL", "referenceId", "length", "playsTotal",
  29         "playsTrailingWeek", "renditions", "captioning", "startDate", "endDate"]
  30
  31     def _real_extract(self, url):
  32         webpage = self._download_webpage(url, url)
  33         video_id = self._html_search_regex(
  34             r'videoID: "(\d+?)"', webpage, 'video_id')
  35         api_token = self._html_search_regex(
  36             r'apiToken: "([a-zA-Z0-9-_\.]+?)"', webpage, 'api_token')
  37         title = self._html_search_regex(
  38             r'<meta property="og:title" content="\s*(.*?)\s*"\s*/?\s*>',
  39             webpage, 'title')
  40
  41         json = self._download_json('https://api.brightcove.com/services/library?command=find_video_by_id&video_id=%s&token=%s&video_fields=%s' % (video_id, api_token, ','.join(self._video_fields)), title)
  42
  43         formats = []
  44         for rendition in json['renditions']:
  45             url = rendition['remoteUrl'] or rendition['url']
  46             protocol = 'rtmp' if url.startswith('rtmp') else 'http'
  47             ext = 'flv' if protocol == 'rtmp' else rendition['videoContainer'].lower()
  48
  49             if protocol == 'rtmp':
  50                 url = url.replace('&mp4:', '')
  51
  52             formats.append({
  53                 'format_id': '_'.join(['rtmp', rendition['videoContainer'].lower(), rendition['videoCodec'].lower()]),
  54                 'width': rendition['frameWidth'],
  55                 'height': rendition['frameHeight'],
  56                 'tbr': rendition['encodingRate'] / 1024,
  57                 'filesize': rendition['size'],
  58                 'protocol': protocol,
  59                 'ext': ext,
  60                 'vcodec': rendition['videoCodec'].lower(),
  61                 'container': rendition['videoContainer'].lower(),
  62                 'url': url,
  63             })
  64
  65         return {
  66             'id': video_id,
  67             'display_id': json['referenceId'],
  68             'title': json['name'],
  69             'description': json['shortDescription'] or json['longDescription'],
  70             'formats': formats,
  71             'thumbnails': [{
  72                 'url': json['videoStillURL']
  73             }, {
  74                 'url': json['thumbnailURL']
  75             }],
  76             'thumbnail': json['videoStillURL'],
  77             'duration': json['length'] / 1000,
  78             'timestamp': float(json['creationDate']) / 1000,
  79             'uploader': json['customFields']['production_company_distributor'] if 'production_company_distributor' in json['customFields'] else 'TENplay',
  80             'view_count': json['playsTotal']
  81         }