subtitles.py 2.7 KB

12345678910111213141516171819202122232425262728293031323334353637383940414243444546474849505152535455565758596061626364656667686970717273
  1. import socket
  2. from .common import InfoExtractor
  3. from ..utils import (
  4. compat_http_client,
  5. compat_urllib_error,
  6. compat_urllib_request,
  7. compat_str,
  8. )
  9. class SubtitlesIE(InfoExtractor):
  10. def report_video_subtitles_available(self, video_id, sub_lang_list):
  11. """Report available subtitles."""
  12. sub_lang = ",".join(list(sub_lang_list.keys()))
  13. self.to_screen(u'%s: Available subtitles for video: %s' %
  14. (video_id, sub_lang))
  15. def _list_available_subtitles(self, video_id):
  16. sub_lang_list = self._get_available_subtitles(video_id)
  17. self.report_video_subtitles_available(video_id, sub_lang_list)
  18. def _extract_subtitles(self, video_id):
  19. """
  20. Return a dictionary: {language: subtitles} or {} if the subtitles
  21. couldn't be found
  22. """
  23. sub_lang_list = self._get_available_subtitles(video_id)
  24. if not sub_lang_list: # error, it didn't get the available subtitles
  25. return {}
  26. if self._downloader.params.get('writesubtitles', False):
  27. if self._downloader.params.get('subtitleslang', False):
  28. sub_lang = self._downloader.params.get('subtitleslang')
  29. elif 'en' in sub_lang_list:
  30. sub_lang = 'en'
  31. else:
  32. sub_lang = list(sub_lang_list.keys())[0]
  33. if not sub_lang in sub_lang_list:
  34. self._downloader.report_warning(u'no closed captions found in the specified language "%s"' % sub_lang)
  35. return {}
  36. sub_lang_list = {sub_lang: sub_lang_list[sub_lang]}
  37. subtitles = {}
  38. for sub_lang, url in sub_lang_list.iteritems():
  39. subtitle = self._request_subtitle_url(sub_lang, url)
  40. if subtitle:
  41. subtitles[sub_lang] = subtitle
  42. return subtitles
  43. def _request_subtitle_url(self, sub_lang, url):
  44. try:
  45. sub = compat_urllib_request.urlopen(url).read().decode('utf-8')
  46. except (compat_urllib_error.URLError, compat_http_client.HTTPException, socket.error) as err:
  47. self._downloader.report_warning(u'unable to download video subtitles for %s: %s' % (sub_lang, compat_str(err)))
  48. return
  49. if not sub:
  50. self._downloader.report_warning(u'Did not fetch video subtitles')
  51. return
  52. return sub
  53. def _get_available_subtitles(self, video_id):
  54. """returns the list of available subtitles like this {lang: url} """
  55. """or {} if not available. Must be redefined by the subclasses."""
  56. pass
  57. def _request_automatic_caption(self, video_id, webpage):
  58. """Request automatic caption. Redefine in subclasses."""
  59. """returns a tuple of ... """
  60. # return [(err_msg, None, None)]
  61. pass