inc.py 2.2 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657
  1. from .common import InfoExtractor
  2. from .kaltura import KalturaIE
  3. class IncIE(InfoExtractor):
  4. _VALID_URL = r'https?://(?:www\.)?inc\.com/(?:[^/]+/)+(?P<id>[^.]+).html'
  5. _TESTS = [{
  6. 'url': 'http://www.inc.com/tip-sheet/bill-gates-says-these-5-books-will-make-you-smarter.html',
  7. 'md5': '7416739c9c16438c09fa35619d6ba5cb',
  8. 'info_dict': {
  9. 'id': '1_wqig47aq',
  10. 'ext': 'mov',
  11. 'title': 'Bill Gates Says These 5 Books Will Make You Smarter',
  12. 'description': 'md5:bea7ff6cce100886fc1995acb743237e',
  13. 'timestamp': 1474414430,
  14. 'upload_date': '20160920',
  15. 'uploader_id': 'video@inc.com',
  16. },
  17. 'params': {
  18. 'skip_download': True,
  19. },
  20. }, {
  21. # div with id=kaltura_player_1_kqs38cgm
  22. 'url': 'https://www.inc.com/oscar-raymundo/richard-branson-young-entrepeneurs.html',
  23. 'info_dict': {
  24. 'id': '1_kqs38cgm',
  25. 'ext': 'mp4',
  26. 'title': 'Branson: "In the end, you have to say, Screw it. Just do it."',
  27. 'description': 'md5:21b832d034f9af5191ca5959da5e9cb6',
  28. 'timestamp': 1364403232,
  29. 'upload_date': '20130327',
  30. 'uploader_id': 'incdigital@inc.com',
  31. },
  32. 'params': {
  33. 'skip_download': True,
  34. },
  35. }, {
  36. 'url': 'http://www.inc.com/video/david-whitford/founders-forum-tripadvisor-steve-kaufer-most-enjoyable-moment-for-entrepreneur.html',
  37. 'only_matching': True,
  38. }]
  39. def _real_extract(self, url):
  40. display_id = self._match_id(url)
  41. webpage = self._download_webpage(url, display_id)
  42. partner_id = self._search_regex(
  43. r'var\s+_?bizo_data_partner_id\s*=\s*["\'](\d+)', webpage,
  44. 'partner id', default='1034971')
  45. kaltura_id = self._search_regex(
  46. r'id=(["\'])kaltura_player_(?P<id>.+?)\1', webpage, 'kaltura id',
  47. default=None, group='id') or self._parse_json(self._search_regex(
  48. r'pageInfo\.videos\s*=\s*\[(.+)\];', webpage, 'kaltura id'),
  49. display_id)['vid_kaltura_id']
  50. return self.url_result(
  51. f'kaltura:{partner_id}:{kaltura_id}', KalturaIE.ie_key())