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

   1 # coding: utf-8
   2 from __future__ import unicode_literals
   3
   4 import re
   5
   6 from .common import InfoExtractor
   7
   8 class TenPlayIE(InfoExtractor):
   9     _VALID_URL = r'https?://(?:www\.)?ten(play)?\.com\.au/.+'
  10     _TEST = {
  11         'url': 'http://tenplay.com.au/ten-insider/extra/season-2013/tenplay-tv-your-way',
  12         'md5': 'c9dda6aac8f814352ad2aee8899b1612',
  13         'info_dict': {
  14             'id': '2695695426001',
  15             'ext': 'flv',
  16             'title': 'TENplay: TV your way',
  17             'description': 'Welcome to a new TV experience. Enjoy a taste of the TENplay benefits.',
  18             'timestamp': 1380150606.889,
  19             'upload_date': '20130925',
  20             'uploader': 'TENplay'
  21         }
  22     }
  23
  24     _video_fields = ["id","name","shortDescription","longDescription","creationDate","publishedDate","lastModifiedDate","customFields","videoStillURL","thumbnailURL","referenceId","length","playsTotal","playsTrailingWeek","renditions","captioning","startDate","endDate"]
  25
  26     def _real_extract(self, url):
  27         webpage = self._download_webpage(url, url)
  28         video_id = self._html_search_regex(r'videoID: "(\d+?)"', webpage, 'video_id')
  29         api_token = self._html_search_regex(r'apiToken: "([a-zA-Z0-9-_\.]+?)"', webpage, 'api_token')
  30         title = self._html_search_regex(r'<meta property="og:title" content="\s*(.*?)\s*"\s*/?\s*>', webpage, 'title')
  31
  32         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)
  33
  34         formats = []
  35         for rendition in json['renditions']:
  36             url = rendition['remoteUrl'] or rendition['url']
  37             protocol = 'rtmp' if url.startswith('rtmp') else 'http'
  38             ext = 'flv' if protocol == 'rtmp' else rendition['videoContainer'].lower()
  39
  40             if protocol == 'rtmp':
  41                 url = url.replace('&mp4:', '')
  42
  43             formats.append({
  44                 'format_id': '_'.join(['rtmp', rendition['videoContainer'].lower(), rendition['videoCodec'].lower()]),
  45                 'width': rendition['frameWidth'],
  46                 'height': rendition['frameHeight'],
  47                 'tbr': rendition['encodingRate'] / 1024,
  48                 'filesize': rendition['size'],
  49                 'protocol': protocol,
  50                 'ext': ext,
  51                 'vcodec': rendition['videoCodec'].lower(),
  52                 'container': rendition['videoContainer'].lower(),
  53                 'url': url
  54                 })
  55
  56         return {
  57             'id': video_id,
  58             'display_id': json['referenceId'],
  59             'title': json['name'],
  60             'description': json['shortDescription'] or json['longDescription'],
  61             'formats': formats,
  62             'thumbnails': [{
  63                 'url': json['videoStillURL']
  64             }, {
  65                 'url': json['thumbnailURL']
  66             }],
  67             'thumbnail': json['videoStillURL'],
  68             'duration': json['length'] / 1000,
  69             'timestamp': float(json['creationDate']) / 1000,
  70             'uploader': json['customFields']['production_company_distributor'] if 'production_company_distributor' in json['customFields'] else 'TENplay',
  71             'view_count': json['playsTotal']
  72         }