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,
(
(?:https?://)? # http(s):// (optional)
(?:(?:(?:(?:\w+\.)?youtube(?:-nocookie)?\.com/|
- tube\.majestyc\.net/) # the various hostnames, with wildcard subdomains
+ tube\.majestyc\.net/|
+ youtube\.googleapis\.com/) # the various hostnames, with wildcard subdomains
(?:.*?\#/)? # handle anchor (#/) redirect urls
(?: # the various things that can precede the ID:
(?:(?:v|embed|e)/) # v/ or embed/ or e/
u"info_dict": {
u"upload_date": u"20120506",
u"title": u"Icona Pop - I Love It (feat. Charli XCX) [OFFICIAL VIDEO]",
- u"description": u"md5:3e2666e0a55044490499ea45fe9037b7",
+ u"description": u"md5:bdac09887d209a4ed54b8f76b2bdaa8b",
u"uploader": u"Icona Pop",
u"uploader_id": u"IconaPop"
}
def _decrypt_signature(self, s):
"""Turn the encrypted s field into a working signature"""
- if len(s) == 92:
+ if len(s) == 93:
+ return s[86:29:-1] + s[88] + s[28:5:-1]
+ elif len(s) == 92:
return s[25] + s[3:25] + s[0] + s[26:42] + s[79] + s[43:79] + s[91] + s[80:83]
+ elif len(s) == 91:
+ return s[84:27:-1] + s[86] + s[26:5:-1]
elif len(s) == 90:
return s[25] + s[3:25] + s[2] + s[26:40] + s[77] + s[41:77] + s[89] + s[78:81]
elif len(s) == 89:
elif len(s) == 87:
return s[6:27] + s[4] + s[28:39] + s[27] + s[40:59] + s[2] + s[60:]
elif len(s) == 86:
- return s[5:34] + s[0] + s[35:38] + s[3] + s[39:45] + s[38] + s[46:53] + s[73] + s[54:73] + s[85] + s[74:85] + s[53]
+ return s[80:72:-1] + s[16] + s[71:39:-1] + s[72] + s[38:16:-1] + s[82] + s[:16][::-1]
elif len(s) == 85:
- return s[40] + s[82:43:-1] + s[22] + s[42:40:-1] + s[83] + s[39:22:-1] + s[0] + s[21:2:-1]
+ return s[3:11] + s[0] + s[12:55] + s[84] + s[56:84]
elif len(s) == 84:
- return s[81:36:-1] + s[0] + s[35:2:-1]
+ return s[78:70:-1] + s[14] + s[69:37:-1] + s[70] + s[36:14:-1] + s[80] + s[:14][::-1]
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:
return self._decrypt_signature(s)
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:
+ 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)
return {}
return sub_lang_list
- def _request_automatic_caption(self, video_id, webpage):
+ 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_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
+ err_msg = u'Couldn\'t find automatic captions for %s' % video_id
if mobj is None:
self._downloader.report_warning(err_msg)
return {}
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',
+ # We get the available subtitles
+ list_params = compat_urllib_parse.urlencode({
+ 'type': 'list',
+ 'tlangs': 1,
+ 'asrs': 1,
})
- subtitles_url = caption_url + '&' + params
- sub = self._download_webpage(subtitles_url, video_id, u'Downloading automatic captions')
- return {sub_lang: sub}
+ 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):
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:
if self._downloader.params.get('verbose'):
s = url_data['s'][0]
if age_gate:
- player_version = self._search_regex(r'ad3-(.+?)\.swf',
- video_info['ad3_module'][0] if 'ad3_module' in video_info else 'NOT FOUND',
- 'flash player', fatal=False)
- player = 'flash player %s' % player_version
+ player = 'flash player'
else:
player = u'html5 player %s' % self._search_regex(r'html5player-(.+?)\.js', video_webpage,
'html5 player', fatal=False)
response = json.loads(page)
except ValueError as err:
raise ExtractorError(u'Invalid JSON in API response: ' + compat_str(err))
+ if 'entry' not in response['feed']:
+ # Number of videos is a multiple of self._MAX_RESULTS
+ break
# Extract video identifiers
ids_in_page = []