_ Git - youtube-dl/blob - 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     @property
  11     def _have_to_download_any_subtitles(self):
  12         return any([self._downloader.params.get('writesubtitles', False),
  13                     self._downloader.params.get('writeautomaticsub'),
  14                     self._downloader.params.get('allsubtitles', False)])
  15
  16     def _list_available_subtitles(self, video_id, webpage=None):
  17         """ outputs the available subtitles for the video """
  18         sub_lang_list = self._get_available_subtitles(video_id)
  19         auto_captions_list = self._get_available_automatic_caption(video_id, webpage)
  20         sub_lang = ",".join(list(sub_lang_list.keys()))
  21         self.to_screen(u'%s: Available subtitles for video: %s' %
  22                        (video_id, sub_lang))
  23         auto_lang = ",".join(auto_captions_list.keys())
  24         self.to_screen(u'%s: Available automatic captions for video: %s' %
  25                        (video_id, auto_lang))
  26
  27     def extract_subtitles(self, video_id, video_webpage=None):
  28         """
  29         returns {sub_lang: sub} ,{} if subtitles not found or None if the
  30         subtitles aren't requested.
  31         """
  32         if not self._have_to_download_any_subtitles:
  33             return None
  34         available_subs_list = {}
  35         if self._downloader.params.get('writeautomaticsub', False):
  36             available_subs_list.update(self._get_available_automatic_caption(video_id, video_webpage))
  37         if self._downloader.params.get('writesubtitles', False) or self._downloader.params.get('allsubtitles', False):
  38             available_subs_list.update(self._get_available_subtitles(video_id))
  39
  40         if not available_subs_list:  # error, it didn't get the available subtitles
  41             return {}
  42         if self._downloader.params.get('allsubtitles', False):
  43             sub_lang_list = available_subs_list
  44         else:
  45             if self._downloader.params.get('subtitleslangs', False):
  46                 requested_langs = self._downloader.params.get('subtitleslangs')
  47             elif 'en' in available_subs_list:
  48                 requested_langs = ['en']
  49             else:
  50                 requested_langs = [list(available_subs_list.keys())[0]]
  51
  52             sub_lang_list = {}
  53             for sub_lang in requested_langs:
  54                 if not sub_lang in available_subs_list:
  55                     self._downloader.report_warning(u'no closed captions found in the specified language "%s"' % sub_lang)
  56                     continue
  57                 sub_lang_list[sub_lang] = available_subs_list[sub_lang]
  58
  59         subtitles = {}
  60         for sub_lang, url in sub_lang_list.items():
  61             subtitle = self._request_subtitle_url(sub_lang, url)
  62             if subtitle:
  63                 subtitles[sub_lang] = subtitle
  64         return subtitles
  65
  66     def _request_subtitle_url(self, sub_lang, url):
  67         """ makes the http request for the subtitle """
  68         try:
  69             sub = self._download_webpage(url, None, note=False)
  70         except ExtractorError as err:
  71             self._downloader.report_warning(u'unable to download video subtitles for %s: %s' % (sub_lang, compat_str(err)))
  72             return
  73         if not sub:
  74             self._downloader.report_warning(u'Did not fetch video subtitles')
  75             return
  76         return sub
  77
  78     def _get_available_subtitles(self, video_id):
  79         """
  80         returns {sub_lang: url} or {} if not available
  81         Must be redefined by the subclasses
  82         """
  83         pass
  84
  85     def _get_available_automatic_caption(self, video_id, webpage):
  86         """
  87         returns {sub_lang: url} or {} if not available
  88         Must be redefined by the subclasses that support automatic captions,
  89         otherwise it will return {}
  90         """
  91         self._downloader.report_warning(u'Automatic Captions not supported by this server')
  92         return {}