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
  15     def _list_available_subtitles(self, video_id, webpage):
  16         """ outputs the available subtitles for the video """
  17         sub_lang_list = self._get_available_subtitles(video_id, webpage)
  18         auto_captions_list = self._get_available_automatic_caption(video_id, webpage)
  19         sub_lang = ",".join(list(sub_lang_list.keys()))
  20         self.to_screen(u'%s: Available subtitles for video: %s' %
  21                        (video_id, sub_lang))
  22         auto_lang = ",".join(auto_captions_list.keys())
  23         self.to_screen(u'%s: Available automatic captions for video: %s' %
  24                        (video_id, auto_lang))
  25
  26     def extract_subtitles(self, video_id, webpage):
  27         """
  28         returns {sub_lang: sub} ,{} if subtitles not found or None if the
  29         subtitles aren't requested.
  30         """
  31         if not self._have_to_download_any_subtitles:
  32             return None
  33         available_subs_list = {}
  34         if self._downloader.params.get('writeautomaticsub', False):
  35             available_subs_list.update(self._get_available_automatic_caption(video_id, webpage))
  36         if self._downloader.params.get('writesubtitles', False):
  37             available_subs_list.update(self._get_available_subtitles(video_id, webpage))
  38
  39         if not available_subs_list:  # error, it didn't get the available subtitles
  40             return {}
  41         if self._downloader.params.get('allsubtitles', False):
  42             sub_lang_list = available_subs_list
  43         else:
  44             if self._downloader.params.get('subtitleslangs', False):
  45                 requested_langs = self._downloader.params.get('subtitleslangs')
  46             elif 'en' in available_subs_list:
  47                 requested_langs = ['en']
  48             else:
  49                 requested_langs = [list(available_subs_list.keys())[0]]
  50
  51             sub_lang_list = {}
  52             for sub_lang in requested_langs:
  53                 if not sub_lang in available_subs_list:
  54                     self._downloader.report_warning(u'no closed captions found in the specified language "%s"' % sub_lang)
  55                     continue
  56                 sub_lang_list[sub_lang] = available_subs_list[sub_lang]
  57
  58         subtitles = {}
  59         for sub_lang, url in sub_lang_list.items():
  60             subtitle = self._request_subtitle_url(sub_lang, url)
  61             if subtitle:
  62                 subtitles[sub_lang] = subtitle
  63         return subtitles
  64
  65     def _download_subtitle_url(self, sub_lang, url):
  66         return self._download_webpage(url, None, note=False)
  67
  68     def _request_subtitle_url(self, sub_lang, url):
  69         """ makes the http request for the subtitle """
  70         try:
  71             sub = self._download_subtitle_url(sub_lang, url)
  72         except ExtractorError as err:
  73             self._downloader.report_warning(u'unable to download video subtitles for %s: %s' % (sub_lang, compat_str(err)))
  74             return
  75         if not sub:
  76             self._downloader.report_warning(u'Did not fetch video subtitles')
  77             return
  78         return sub
  79
  80     def _get_available_subtitles(self, video_id, webpage):
  81         """
  82         returns {sub_lang: url} or {} if not available
  83         Must be redefined by the subclasses
  84         """
  85
  86         # By default, allow implementations to simply pass in the result
  87         assert isinstance(webpage, dict), \
  88             '_get_available_subtitles not implemented'
  89         return webpage
  90
  91     def _get_available_automatic_caption(self, video_id, webpage):
  92         """
  93         returns {sub_lang: url} or {} if not available
  94         Must be redefined by the subclasses that support automatic captions,
  95         otherwise it will return {}
  96         """
  97         self._downloader.report_warning(u'Automatic Captions not supported by this server')
  98         return {}