youtube_dl/extractor/ringtv.py

   1 from __future__ import unicode_literals
   2
   3 import re
   4
   5 from .common import InfoExtractor
   6
   7
   8 class RingTVIE(InfoExtractor):
   9     _VALID_URL = r'(?:http://)?(?:www\.)?ringtv\.craveonline\.com/(?P<type>news|videos/video)/(?P<id>[^/?#]+)'
  10     _TEST = {
  11         "url": "http://ringtv.craveonline.com/news/310833-luis-collazo-says-victor-ortiz-better-not-quit-on-jan-30",
  12         "file": "857645.mp4",
  13         "md5": "d25945f5df41cdca2d2587165ac28720",
  14         "info_dict": {
  15             "title": 'Video: Luis Collazo says Victor Ortiz "better not quit on Jan. 30" - Ring TV',
  16             "description": 'Luis Collazo is excited about his Jan. 30 showdown with fellow former welterweight titleholder Victor Ortiz at Barclays Center in his hometown of Brooklyn. The SuperBowl week fight headlines a Golden Boy Live! card on Fox Sports 1.',
  17         }
  18     }
  19
  20     def _real_extract(self, url):
  21         mobj = re.match(self._VALID_URL, url)
  22         video_id = mobj.group('id').split('-')[0]
  23         webpage = self._download_webpage(url, video_id)
  24
  25         if mobj.group('type') == 'news':
  26             video_id = self._search_regex(
  27                 r'''(?x)<iframe[^>]+src="http://cms\.springboardplatform\.com/
  28                         embed_iframe/[0-9]+/video/([0-9]+)/''',
  29                 webpage, 'real video ID')
  30         title = self._og_search_title(webpage)
  31         description = self._html_search_regex(
  32             r'addthis:description="([^"]+)"',
  33             webpage, 'description', fatal=False)
  34         final_url = "http://ringtv.craveonline.springboardplatform.com/storage/ringtv.craveonline.com/conversion/%s.mp4" % video_id
  35         thumbnail_url = "http://ringtv.craveonline.springboardplatform.com/storage/ringtv.craveonline.com/snapshots/%s.jpg" % video_id
  36
  37         return {
  38             'id': video_id,
  39             'url': final_url,
  40             'title': title,
  41             'thumbnail': thumbnail_url,
  42             'description': description,
  43         }
  44