Raphaël G. Git Repositories - youtubedl/blob - youtube_dl/extractor/subtitles.py

   1 from __future__ import unicode_literals
   2 from .common import InfoExtractor
   3
   4 from ..compat import compat_str
   5 from ..utils import (
   6     ExtractorError,
   7 )
   8
   9
  10 class SubtitlesInfoExtractor(InfoExtractor):
  11     @property
  12     def _have_to_download_any_subtitles(self):
  13         return any([self._downloader.params.get('writesubtitles', False),
  14                     self._downloader.params.get('writeautomaticsub')])
  15
  16     def _list_available_subtitles(self, video_id, webpage):
  17         """ outputs the available subtitles for the video """
  18         sub_lang_list = self._get_available_subtitles(video_id, webpage)
  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('%s: Available subtitles for video: %s' %
  22                        (video_id, sub_lang))
  23         auto_lang = ",".join(auto_captions_list.keys())
  24         self.to_screen('%s: Available automatic captions for video: %s' %
  25                        (video_id, auto_lang))
  26
  27     def extract_subtitles(self, video_id, webpage):
  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, webpage))
  37         if self._downloader.params.get('writesubtitles', False):
  38             available_subs_list.update(self._get_available_subtitles(video_id, webpage))
  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 sub_lang not in available_subs_list:
  55                     self._downloader.report_warning('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 _download_subtitle_url(self, sub_lang, url):
  67         return self._download_webpage(url, None, note=False)
  68
  69     def _request_subtitle_url(self, sub_lang, url):
  70         """ makes the http request for the subtitle """
  71         try:
  72             sub = self._download_subtitle_url(sub_lang, url)
  73         except ExtractorError as err:
  74             self._downloader.report_warning('unable to download video subtitles for %s: %s' % (sub_lang, compat_str(err)))
  75             return
  76         if not sub:
  77             self._downloader.report_warning('Did not fetch video subtitles')
  78             return
  79         return sub
  80
  81     def _get_available_subtitles(self, video_id, webpage):
  82         """
  83         returns {sub_lang: url} or {} if not available
  84         Must be redefined by the subclasses
  85         """
  86
  87         # By default, allow implementations to simply pass in the result
  88         assert isinstance(webpage, dict), \
  89             '_get_available_subtitles not implemented'
  90         return webpage
  91
  92     def _get_available_automatic_caption(self, video_id, webpage):
  93         """
  94         returns {sub_lang: url} or {} if not available
  95         Must be redefined by the subclasses that support automatic captions,
  96         otherwise it will return {}
  97         """
  98         self._downloader.report_warning('Automatic Captions not supported by this server')
  99         return {}