franceinter.py 2.0 KB

1234567891011121314151617181920212223242526272829303132333435363738394041424344454647484950515253545556
  1. # coding: utf-8
  2. from __future__ import unicode_literals
  3. from .common import InfoExtractor
  4. from ..compat import compat_str
  5. from ..utils import month_by_name
  6. class FranceInterIE(InfoExtractor):
  7. _VALID_URL = r'https?://(?:www\.)?franceinter\.fr/emissions/(?P<id>[^?#]+)'
  8. _TEST = {
  9. 'url': 'https://www.franceinter.fr/emissions/la-tete-au-carre/la-tete-au-carre-14-septembre-2016',
  10. 'md5': '4e3aeb58fe0e83d7b0581fa213c409d0',
  11. 'info_dict': {
  12. 'id': 'la-tete-au-carre/la-tete-au-carre-14-septembre-2016',
  13. 'ext': 'mp3',
  14. 'title': 'Et si les rêves pouvaient nous aider à agir dans notre vie quotidienne ?',
  15. 'description': 'md5:a245dd62cf5bf51de915f8d9956d180a',
  16. 'upload_date': '20160914',
  17. },
  18. }
  19. def _real_extract(self, url):
  20. video_id = self._match_id(url)
  21. webpage = self._download_webpage(url, video_id)
  22. video_url = self._search_regex(
  23. r'(?s)<div[^>]+class=["\']page-diffusion["\'][^>]*>.*?<button[^>]+data-url=(["\'])(?P<url>(?:(?!\1).)+)\1',
  24. webpage, 'video url', group='url')
  25. title = self._og_search_title(webpage)
  26. description = self._og_search_description(webpage)
  27. upload_date_str = self._search_regex(
  28. r'class=["\']cover-emission-period["\'][^>]*>[^<]+\s+(\d{1,2}\s+[^\s]+\s+\d{4})<',
  29. webpage, 'upload date', fatal=False)
  30. if upload_date_str:
  31. upload_date_list = upload_date_str.split()
  32. upload_date_list.reverse()
  33. upload_date_list[1] = '%02d' % (month_by_name(upload_date_list[1], lang='fr') or 0)
  34. upload_date = ''.join(upload_date_list)
  35. else:
  36. upload_date = None
  37. return {
  38. 'id': video_id,
  39. 'title': title,
  40. 'description': description,
  41. 'upload_date': upload_date,
  42. 'formats': [{
  43. 'url': video_url,
  44. 'vcodec': 'none',
  45. }],
  46. }