-import socket
-
from .common import InfoExtractor
from ..utils import (
- compat_http_client,
- compat_urllib_error,
- compat_urllib_request,
compat_str,
+ ExtractorError,
)
-class SubtitlesIE(InfoExtractor):
-
- def report_video_subtitles_available(self, video_id, sub_lang_list):
- """Report available subtitles."""
- sub_lang = ",".join(list(sub_lang_list.keys()))
- self.to_screen(u'%s: Available subtitles for video: %s' % (video_id, sub_lang))
+class SubtitlesInfoExtractor(InfoExtractor):
+ @property
+ def _have_to_download_any_subtitles(self):
+ return any([self._downloader.params.get('writesubtitles', False),
+ self._downloader.params.get('writeautomaticsub')])
- def _list_available_subtitles(self, video_id):
+ def _list_available_subtitles(self, video_id, webpage=None):
+ """ outputs the available subtitles for the video """
sub_lang_list = self._get_available_subtitles(video_id)
- self.report_video_subtitles_available(video_id, sub_lang_list)
+ auto_captions_list = self._get_available_automatic_caption(video_id, webpage)
+ sub_lang = ",".join(list(sub_lang_list.keys()))
+ self.to_screen(u'%s: Available subtitles for video: %s' %
+ (video_id, sub_lang))
+ auto_lang = ",".join(auto_captions_list.keys())
+ self.to_screen(u'%s: Available automatic captions for video: %s' %
+ (video_id, auto_lang))
- def _extract_subtitles(self, video_id):
+ def extract_subtitles(self, video_id, video_webpage=None):
"""
- Return a dictionary: {language: subtitles} or {} if the subtitles
- couldn't be found
+ returns {sub_lang: sub} ,{} if subtitles not found or None if the
+ subtitles aren't requested.
"""
- sub_lang_list = self._get_available_subtitles(video_id)
- sub_format = self._downloader.params.get('subtitlesformat')
- if not sub_lang_list: #There was some error, it didn't get the available subtitles
- return {}
+ if not self._have_to_download_any_subtitles:
+ return None
+ available_subs_list = {}
+ if self._downloader.params.get('writeautomaticsub', False):
+ available_subs_list.update(self._get_available_automatic_caption(video_id, video_webpage))
if self._downloader.params.get('writesubtitles', False):
- if self._downloader.params.get('subtitleslang', False):
- sub_lang = self._downloader.params.get('subtitleslang')
- elif 'en' in sub_lang_list:
- sub_lang = 'en'
+ available_subs_list.update(self._get_available_subtitles(video_id))
+
+ if not available_subs_list: # error, it didn't get the available subtitles
+ return {}
+ if self._downloader.params.get('allsubtitles', False):
+ sub_lang_list = available_subs_list
+ else:
+ if self._downloader.params.get('subtitleslangs', False):
+ requested_langs = self._downloader.params.get('subtitleslangs')
+ elif 'en' in available_subs_list:
+ requested_langs = ['en']
else:
- sub_lang = list(sub_lang_list.keys())[0]
- if not sub_lang in sub_lang_list:
- self._downloader.report_warning(u'no closed captions found in the specified language "%s"' % sub_lang)
- return {}
- sub_lang_list = {sub_lang: sub_lang_list[sub_lang]}
+ requested_langs = [list(available_subs_list.keys())[0]]
+
+ sub_lang_list = {}
+ for sub_lang in requested_langs:
+ if not sub_lang in available_subs_list:
+ self._downloader.report_warning(u'no closed captions found in the specified language "%s"' % sub_lang)
+ continue
+ sub_lang_list[sub_lang] = available_subs_list[sub_lang]
+
subtitles = {}
- for sub_lang in sub_lang_list:
- subtitle = self._request_subtitle(sub_lang, sub_lang_list[sub_lang].encode('utf-8'), video_id, sub_format)
+ for sub_lang, url in sub_lang_list.items():
+ subtitle = self._request_subtitle_url(sub_lang, url)
if subtitle:
subtitles[sub_lang] = subtitle
return subtitles
- def _request_subtitle(self, sub_lang, sub_name, video_id, format):
- """ Return the subtitle as a string or None if they are not found """
- # return (u'Did not fetch video subtitles for %s' % sub_lang, None, None)
- self.to_screen(u'%s: Downloading video subtitles for %s.%s' % (video_id, sub_lang, format))
- url = self._get_subtitle_url(sub_lang, sub_name, video_id, format)
+ def _request_subtitle_url(self, sub_lang, url):
+ """ makes the http request for the subtitle """
try:
- sub = compat_urllib_request.urlopen(url).read().decode('utf-8')
- except (compat_urllib_error.URLError, compat_http_client.HTTPException, socket.error) as err:
+ sub = self._download_webpage(url, None, note=False)
+ except ExtractorError as err:
self._downloader.report_warning(u'unable to download video subtitles for %s: %s' % (sub_lang, compat_str(err)))
return
if not sub:
return sub
def _get_available_subtitles(self, video_id):
- """Get available subtitles. Redefine in subclasses."""
- """returns {(lang, url)} """
- # return {}
- pass
-
- def _get_subtitle_url(self, sub_lang, sub_name, video_id, format):
- """returns the url for the given subtitle. Redefine in subclasses."""
+ """
+ returns {sub_lang: url} or {} if not available
+ Must be redefined by the subclasses
+ """
pass
- def _request_automatic_caption(self, video_id, webpage):
- """Request automatic caption. Redefine in subclasses."""
- """returns a tuple of ... """
- # return [(err_msg, None, None)]
- pass
+ def _get_available_automatic_caption(self, video_id, webpage):
+ """
+ returns {sub_lang: url} or {} if not available
+ Must be redefined by the subclasses that support automatic captions,
+ otherwise it will return {}
+ """
+ self._downloader.report_warning(u'Automatic Captions not supported by this server')
+ return {}