youtube_dl/extractor/subtitles.py

   1 from .common import InfoExtractor
   2
   3 from ..utils import (
   4     compat_str,
   5     ExtractorError,
   6 )
   7
   8
   9 class SubtitlesInfoExtractor(InfoExtractor):
  10
  11     def _list_available_subtitles(self, video_id):
  12         """ outputs the available subtitles for the video """
  13         sub_lang_list = self._get_available_subtitles(video_id)
  14         sub_lang = ",".join(list(sub_lang_list.keys()))
  15         self.to_screen(u'%s: Available subtitles for video: %s' %
  16                        (video_id, sub_lang))
  17
  18     def extract_subtitles(self, video_id, video_webpage=None):
  19         """ returns {sub_lang: sub} or {} if subtitles not found """
  20         if self._downloader.params.get('writesubtitles', False) or self._downloader.params.get('allsubtitles', False):
  21             available_subs_list = self._get_available_subtitles(video_id)
  22         elif self._downloader.params.get('writeautomaticsub', False):
  23             available_subs_list = self._get_available_automatic_caption(video_id, video_webpage)
  24         else:
  25             return None
  26
  27         if not available_subs_list:  # error, it didn't get the available subtitles
  28             return {}
  29         if self._downloader.params.get('allsubtitles', False):
  30             sub_lang_list = available_subs_list
  31         else:
  32             if self._downloader.params.get('subtitleslangs', False):
  33                 requested_langs = self._downloader.params.get('subtitleslangs')
  34             elif 'en' in available_subs_list:
  35                 requested_langs = ['en']
  36             else:
  37                 requested_langs = [list(available_subs_list.keys())[0]]
  38
  39             sub_lang_list = {}
  40             for sub_lang in requested_langs:
  41                 if not sub_lang in available_subs_list:
  42                     self._downloader.report_warning(u'no closed captions found in the specified language "%s"' % sub_lang)
  43                     continue
  44                 sub_lang_list[sub_lang] = available_subs_list[sub_lang]
  45
  46         subtitles = {}
  47         for sub_lang, url in sub_lang_list.items():
  48             subtitle = self._request_subtitle_url(sub_lang, url)
  49             if subtitle:
  50                 subtitles[sub_lang] = subtitle
  51         return subtitles
  52
  53     def _request_subtitle_url(self, sub_lang, url):
  54         """ makes the http request for the subtitle """
  55         try:
  56             sub = self._download_webpage(url, None, note=False)
  57         except ExtractorError as err:
  58             self._downloader.report_warning(u'unable to download video subtitles for %s: %s' % (sub_lang, compat_str(err)))
  59             return
  60         if not sub:
  61             self._downloader.report_warning(u'Did not fetch video subtitles')
  62             return
  63         return sub
  64
  65     def _get_available_subtitles(self, video_id):
  66         """
  67         returns {sub_lang: url} or {} if not available
  68         Must be redefined by the subclasses
  69         """
  70         pass
  71
  72     def _get_available_automatic_caption(self, video_id, webpage):
  73         """
  74         returns {sub_lang: url} or {} if not available
  75         Must be redefined by the subclasses that support automatic captions,
  76         otherwise it will return {}
  77         """
  78         self._downloader.report_warning(u'Automatic Captions not supported by this server')
  79         return {}