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

   1 import json
   2 import re
   3
   4 from .common import InfoExtractor
   5
   6 from ..utils import (
   7     compat_urllib_parse,
   8 )
   9
  10
  11 class WatIE(InfoExtractor):
  12     _VALID_URL=r'http://www.wat.tv/.*-(?P<shortID>.*?)_.*?.html'
  13     IE_NAME = 'wat.tv'
  14     _TEST = {
  15         u'url': u'http://www.wat.tv/video/world-war-philadelphia-vost-6bv55_2fjr7_.html',
  16         u'file': u'6bv55.mp4',
  17         u'md5': u'0a4fe7870f31eaeabb5e25fd8da8414a',
  18         u'info_dict': {
  19             u"title": u"World War Z - Philadelphia VOST"
  20         }
  21     }
  22
  23     def _real_extract(self, url):
  24         mobj = re.match(self._VALID_URL, url)
  25         short_id = mobj.group('shortID')
  26
  27         player_data = compat_urllib_parse.urlencode({'shortVideoId': short_id,
  28                                                      'html5': '1'})
  29         player_info = self._download_webpage('http://www.wat.tv/player?' + player_data,
  30                                              short_id, u'Downloading player info')
  31         player = json.loads(player_info)['player']
  32         html5_player = self._html_search_regex(r'iframe src="(.*?)"', player,
  33                                                'html5 player')
  34         player_webpage = self._download_webpage(html5_player, short_id,
  35                                                 u'Downloading player webpage')
  36
  37         video_url = self._search_regex(r'urlhtml5 : "(.*?)"', player_webpage,
  38                                        'video url')
  39         title = self._search_regex(r'contentTitle : "(.*?)"', player_webpage,
  40                                    'title')
  41         thumbnail = self._search_regex(r'previewMedia : "(.*?)"', player_webpage,
  42                                        'thumbnail')
  43         return {'id': short_id,
  44                 'url': video_url,
  45                 'ext': 'mp4',
  46                 'title': title,
  47                 'thumbnail': thumbnail,
  48                 }