Raphaël G. Git Repositories - youtubedl/blob - youtube_dl/extractor/thisoldhouse.py

   1 # coding: utf-8
   2 from __future__ import unicode_literals
   3
   4 from .common import InfoExtractor
   5
   6
   7 class ThisOldHouseIE(InfoExtractor):
   8     _VALID_URL = r'https?://(?:www\.)?thisoldhouse\.com/(?:watch|how-to|tv-episode|(?:[^/]+/)?\d+)/(?P<id>[^/?#]+)'
   9     _TESTS = [{
  10         'url': 'https://www.thisoldhouse.com/how-to/how-to-build-storage-bench',
  11         'info_dict': {
  12             'id': '5dcdddf673c3f956ef5db202',
  13             'ext': 'mp4',
  14             'title': 'How to Build a Storage Bench',
  15             'description': 'In the workshop, Tom Silva and Kevin O\'Connor build a storage bench for an entryway.',
  16             'timestamp': 1442548800,
  17             'upload_date': '20150918',
  18         },
  19         'params': {
  20             'skip_download': True,
  21         },
  22     }, {
  23         'url': 'https://www.thisoldhouse.com/watch/arlington-arts-crafts-arts-and-crafts-class-begins',
  24         'only_matching': True,
  25     }, {
  26         'url': 'https://www.thisoldhouse.com/tv-episode/ask-toh-shelf-rough-electric',
  27         'only_matching': True,
  28     }, {
  29         'url': 'https://www.thisoldhouse.com/furniture/21017078/how-to-build-a-storage-bench',
  30         'only_matching': True,
  31     }, {
  32         'url': 'https://www.thisoldhouse.com/21113884/s41-e13-paradise-lost',
  33         'only_matching': True,
  34     }, {
  35         # iframe www.thisoldhouse.com
  36         'url': 'https://www.thisoldhouse.com/21083431/seaside-transformation-the-westerly-project',
  37         'only_matching': True,
  38     }]
  39     _ZYPE_TMPL = 'https://player.zype.com/embed/%s.html?api_key=hsOk_yMSPYNrT22e9pu8hihLXjaZf0JW5jsOWv4ZqyHJFvkJn6rtToHl09tbbsbe'
  40
  41     def _real_extract(self, url):
  42         display_id = self._match_id(url)
  43         webpage = self._download_webpage(url, display_id)
  44         video_id = self._search_regex(
  45             r'<iframe[^>]+src=[\'"](?:https?:)?//(?:www\.)?thisoldhouse\.(?:chorus\.build|com)/videos/zype/([0-9a-f]{24})',
  46             webpage, 'video id')
  47         return self.url_result(self._ZYPE_TMPL % video_id, 'Zype', video_id)