X-Git-Url: https://git.cielonegro.org/gitweb.cgi?a=blobdiff_plain;f=youtube_dl%2Fextractor%2Fyoutube.py;h=f49665925fdeadeb9f3f64f9eafb7c7a12080fee;hb=74ac9bdd82b8a625ea9782251258ab7da1463877;hp=e71cd62ec3d43e2b541a7e024c6a02adbe16a355;hpb=cf1dd0c59e8ff4c66130bc47954c415be5db8cc3;p=youtube-dl.git diff --git a/youtube_dl/extractor/youtube.py b/youtube_dl/extractor/youtube.py index e71cd62ec..f49665925 100644 --- a/youtube_dl/extractor/youtube.py +++ b/youtube_dl/extractor/youtube.py @@ -5,9 +5,10 @@ import netrc import re import socket import itertools +import xml.etree.ElementTree from .common import InfoExtractor, SearchInfoExtractor -from .subtitles import SubtitlesIE +from .subtitles import SubtitlesInfoExtractor from ..utils import ( compat_http_client, compat_parse_qs, @@ -131,65 +132,8 @@ class YoutubeBaseInfoExtractor(InfoExtractor): return self._confirm_age() -class YoutubeSubtitlesIE(SubtitlesIE): - def _get_available_subtitles(self, video_id): - request = compat_urllib_request.Request('http://video.google.com/timedtext?hl=en&type=list&v=%s' % video_id) - try: - sub_list = compat_urllib_request.urlopen(request).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: %s' % compat_str(err)) - return {} - lang_list = re.findall(r'name="([^"]*)"[^>]+lang_code="([\w\-]+)"', sub_list) - - sub_lang_list = {} - for l in lang_list: - lang = l[1] - params = compat_urllib_parse.urlencode({ - 'lang': lang, - 'v': video_id, - 'fmt': self._downloader.params.get('subtitlesformat'), - }) - url = u'http://www.youtube.com/api/timedtext?' + params - sub_lang_list[lang] = url - if not sub_lang_list: - self._downloader.report_warning(u'video doesn\'t have subtitles') - return {} - return sub_lang_list - - def _request_automatic_caption(self, video_id, webpage): - """We need the webpage for getting the captions url, pass it as an - argument to speed up the process.""" - sub_lang = (self._downloader.params.get('subtitleslangs') or ['en'])[0] - sub_format = self._downloader.params.get('subtitlesformat') - self.to_screen(u'%s: Looking for automatic captions' % video_id) - mobj = re.search(r';ytplayer.config = ({.*?});', webpage) - err_msg = u'Couldn\'t find automatic captions for "%s"' % sub_lang - if mobj is None: - self._downloader.report_warning(err_msg) - return {} - player_config = json.loads(mobj.group(1)) - try: - args = player_config[u'args'] - caption_url = args[u'ttsurl'] - timestamp = args[u'timestamp'] - params = compat_urllib_parse.urlencode({ - 'lang': 'en', - 'tlang': sub_lang, - 'fmt': sub_format, - 'ts': timestamp, - 'kind': 'asr', - }) - subtitles_url = caption_url + '&' + params - sub = self._download_webpage(subtitles_url, video_id, u'Downloading automatic captions') - return {sub_lang: sub} - # An extractor error can be raise by the download process if there are - # no automatic captions but there are subtitles - except (KeyError, ExtractorError): - self._downloader.report_warning(err_msg) - return {} - -class YoutubeIE(YoutubeSubtitlesIE, YoutubeBaseInfoExtractor): +class YoutubeIE(YoutubeBaseInfoExtractor, SubtitlesInfoExtractor): IE_DESC = u'YouTube.com' _VALID_URL = r"""^ ( @@ -490,7 +434,7 @@ class YoutubeIE(YoutubeSubtitlesIE, YoutubeBaseInfoExtractor): elif len(s) == 83: return s[81:64:-1] + s[82] + s[63:52:-1] + s[45] + s[51:45:-1] + s[1] + s[44:1:-1] + s[0] elif len(s) == 82: - return s[1:19] + s[0] + s[20:68] + s[19] + s[69:82] + return s[80:73:-1] + s[81] + s[72:54:-1] + s[2] + s[53:43:-1] + s[0] + s[42:2:-1] + s[43] + s[1] + s[54] elif len(s) == 81: return s[56] + s[79:56:-1] + s[41] + s[55:41:-1] + s[80] + s[40:34:-1] + s[0] + s[33:29:-1] + s[34] + s[28:9:-1] + s[29] + s[8:0:-1] + s[9] elif len(s) == 80: @@ -510,6 +454,79 @@ class YoutubeIE(YoutubeSubtitlesIE, YoutubeBaseInfoExtractor): # Fallback to the other algortihms return self._decrypt_signature(s) + def _get_available_subtitles(self, video_id): + try: + sub_list = self._download_webpage( + 'http://video.google.com/timedtext?hl=en&type=list&v=%s' % video_id, + video_id, note=False) + except ExtractorError as err: + self._downloader.report_warning(u'unable to download video subtitles: %s' % compat_str(err)) + return {} + lang_list = re.findall(r'name="([^"]*)"[^>]+lang_code="([\w\-]+)"', sub_list) + + sub_lang_list = {} + for l in lang_list: + lang = l[1] + params = compat_urllib_parse.urlencode({ + 'lang': lang, + 'v': video_id, + 'fmt': self._downloader.params.get('subtitlesformat'), + }) + url = u'http://www.youtube.com/api/timedtext?' + params + sub_lang_list[lang] = url + if not sub_lang_list: + self._downloader.report_warning(u'video doesn\'t have subtitles') + return {} + return sub_lang_list + + def _get_available_automatic_caption(self, video_id, webpage): + """We need the webpage for getting the captions url, pass it as an + argument to speed up the process.""" + sub_format = self._downloader.params.get('subtitlesformat') + self.to_screen(u'%s: Looking for automatic captions' % video_id) + mobj = re.search(r';ytplayer.config = ({.*?});', webpage) + err_msg = u'Couldn\'t find automatic captions for %s' % video_id + if mobj is None: + self._downloader.report_warning(err_msg) + return {} + player_config = json.loads(mobj.group(1)) + try: + args = player_config[u'args'] + caption_url = args[u'ttsurl'] + timestamp = args[u'timestamp'] + # We get the available subtitles + list_params = compat_urllib_parse.urlencode({ + 'type': 'list', + 'tlangs': 1, + 'asrs': 1, + }) + list_url = caption_url + '&' + list_params + list_page = self._download_webpage(list_url, video_id) + caption_list = xml.etree.ElementTree.fromstring(list_page.encode('utf-8')) + original_lang_node = caption_list.find('track') + if original_lang_node.attrib.get('kind') != 'asr' : + self._downloader.report_warning(u'Video doesn\'t have automatic captions') + return {} + original_lang = original_lang_node.attrib['lang_code'] + + sub_lang_list = {} + for lang_node in caption_list.findall('target'): + sub_lang = lang_node.attrib['lang_code'] + params = compat_urllib_parse.urlencode({ + 'lang': original_lang, + 'tlang': sub_lang, + 'fmt': sub_format, + 'ts': timestamp, + 'kind': 'asr', + }) + sub_lang_list[sub_lang] = caption_url + '&' + params + return sub_lang_list + # An extractor error can be raise by the download process if there are + # no automatic captions but there are subtitles + except (KeyError, ExtractorError): + self._downloader.report_warning(err_msg) + return {} + def _print_formats(self, formats): print('Available formats:') for x in formats: @@ -708,15 +725,10 @@ class YoutubeIE(YoutubeSubtitlesIE, YoutubeBaseInfoExtractor): video_description = u'' # subtitles - video_subtitles = None - - if self._downloader.params.get('writesubtitles', False) or self._downloader.params.get('allsubtitles', False): - video_subtitles = self._extract_subtitles(video_id) - elif self._downloader.params.get('writeautomaticsub', False): - video_subtitles = self._request_automatic_caption(video_id, video_webpage) + video_subtitles = self.extract_subtitles(video_id, video_webpage) if self._downloader.params.get('listsubtitles', False): - self._list_available_subtitles(video_id) + self._list_available_subtitles(video_id, video_webpage) return if 'length_seconds' not in video_info: