youtube_dl/extractor/subtitles.py

   1 import socket
   2
   3 from .common import InfoExtractor
   4
   5 from ..utils import (
   6     compat_http_client,
   7     compat_urllib_error,
   8     compat_urllib_request,
   9     compat_str,
  10 )
  11
  12
  13 class SubtitlesIE(InfoExtractor):
  14
  15     def _list_available_subtitles(self, video_id):
  16         """ outputs the available subtitles for the video """
  17         sub_lang_list = self._get_available_subtitles(video_id)
  18         sub_lang = ",".join(list(sub_lang_list.keys()))
  19         self.to_screen(u'%s: Available subtitles for video: %s' %
  20                        (video_id, sub_lang))
  21
  22     def _extract_subtitles(self, video_id):
  23         """ returns {sub_lang: sub} or {} if subtitles not found """
  24         available_subs_list = self._get_available_subtitles(video_id)
  25         if not available_subs_list:  # error, it didn't get the available subtitles
  26             return {}
  27         if self._downloader.params.get('allsubtitles', False):
  28             sub_lang_list = available_subs_list
  29         else:
  30             if self._downloader.params.get('writesubtitles', False):
  31                 if self._downloader.params.get('subtitleslangs', False):
  32                     requested_langs = self._downloader.params.get('subtitleslangs')
  33                 elif 'en' in available_subs_list:
  34                     requested_langs = ['en']
  35                 else:
  36                     requested_langs = [list(available_subs_list.keys())[0]]
  37
  38                 sub_lang_list = {}
  39                 for sub_lang in requested_langs:
  40                     if not sub_lang in available_subs_list:
  41                         self._downloader.report_warning(u'no closed captions found in the specified language "%s"' % sub_lang)
  42                         continue
  43                     sub_lang_list[sub_lang] = available_subs_list[sub_lang]
  44
  45         subtitles = {}
  46         for sub_lang, url in sub_lang_list.items():
  47             subtitle = self._request_subtitle_url(sub_lang, url)
  48             if subtitle:
  49                 subtitles[sub_lang] = subtitle
  50         return subtitles
  51
  52     def _request_subtitle_url(self, sub_lang, url):
  53         """ makes the http request for the subtitle """
  54         try:
  55             sub = compat_urllib_request.urlopen(url).read().decode('utf-8')
  56         except (compat_urllib_error.URLError, compat_http_client.HTTPException, socket.error) as err:
  57             self._downloader.report_warning(u'unable to download video subtitles for %s: %s' % (sub_lang, compat_str(err)))
  58             return
  59         if not sub:
  60             self._downloader.report_warning(u'Did not fetch video subtitles')
  61             return
  62         return sub
  63
  64     def _get_available_subtitles(self, video_id):
  65         """ returns {sub_lang: url} or {} if not available """
  66         """ Must be redefined by the subclasses """
  67         pass
  68
  69     def _request_automatic_caption(self, video_id, webpage):
  70         """ returns {sub_lang: sub} or {} if not available """
  71         """ Must be redefined by the subclasses """
  72         pass
  73
  74
  75 class NoAutoSubtitlesIE(SubtitlesIE):
  76     """ A subtitle class for the servers that don't support auto-captions"""
  77
  78     def _request_automatic_caption(self, video_id, webpage):
  79         self._downloader.report_warning(u'Automatic Captions not supported by this server')
  80         return {}