953e32b2c1
generic subtitle download. The idea is that all subtitle downloaders must descend from SubtitlesIE and implement only three basic methods to achieve the complete subtitle download functionality. This will allow to reduce the code in YoutubeIE once it is rewritten.
81 lines
3.3 KiB
Python
81 lines
3.3 KiB
Python
import socket
|
|
|
|
from .common import InfoExtractor
|
|
|
|
from ..utils import (
|
|
compat_http_client,
|
|
compat_urllib_error,
|
|
compat_urllib_request,
|
|
compat_str,
|
|
)
|
|
|
|
|
|
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))
|
|
|
|
def _list_available_subtitles(self, video_id):
|
|
sub_lang_list = self._get_available_subtitles(video_id)
|
|
self.report_video_subtitles_available(video_id, sub_lang_list)
|
|
|
|
def _extract_subtitles(self, video_id):
|
|
"""
|
|
Return a dictionary: {language: subtitles} or {} if the subtitles
|
|
couldn't be found
|
|
"""
|
|
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 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'
|
|
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]}
|
|
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)
|
|
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)
|
|
try:
|
|
sub = compat_urllib_request.urlopen(url).read().decode('utf-8')
|
|
except (compat_urllib_error.URLError, compat_http_client.HTTPException, socket.error) as err:
|
|
self._downloader.report_warning(u'unable to download video subtitles for %s: %s' % (sub_lang, compat_str(err)))
|
|
return
|
|
if not sub:
|
|
self._downloader.report_warning(u'Did not fetch video subtitles')
|
|
return
|
|
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."""
|
|
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
|