X-Git-Url: https://git.rapsys.eu/youtubedl/blobdiff_plain/e76f531201cd41dfc0ce00be28bcc5c575c7acc5..50ac49cc58db6cb928094be89415983aca6bbd34:/youtube_dl/extractor/periscope.py diff --git a/youtube_dl/extractor/periscope.py b/youtube_dl/extractor/periscope.py index 887c802..75f5884 100644 --- a/youtube_dl/extractor/periscope.py +++ b/youtube_dl/extractor/periscope.py @@ -2,16 +2,16 @@ from __future__ import unicode_literals from .common import InfoExtractor -from ..compat import ( - compat_urllib_parse, - compat_urllib_request, +from ..utils import ( + parse_iso8601, + unescapeHTML, ) -from ..utils import parse_iso8601 class PeriscopeIE(InfoExtractor): IE_DESC = 'Periscope' - _VALID_URL = r'https?://(?:www\.)?periscope\.tv/w/(?P[^/?#]+)' + IE_NAME = 'periscope' + _VALID_URL = r'https?://(?:www\.)?periscope\.tv/[^/]+/(?P[^/?#]+)' # Alive example URLs can be found here http://onperiscope.com/ _TESTS = [{ 'url': 'https://www.periscope.tv/w/aJUQnjY3MjA3ODF8NTYxMDIyMDl2zCg2pECBgwTqRpQuQD352EMPTKQjT4uqlM3cgWFA-g==', @@ -29,12 +29,14 @@ class PeriscopeIE(InfoExtractor): }, { 'url': 'https://www.periscope.tv/w/1ZkKzPbMVggJv', 'only_matching': True, + }, { + 'url': 'https://www.periscope.tv/bastaakanoggano/1OdKrlkZZjOJX', + 'only_matching': True, }] def _call_api(self, method, value): - attribute = 'token' if len(value) > 13 else 'broadcast_id' return self._download_json( - 'https://api.periscope.tv/api/v2/%s?%s=%s' % (method, attribute, value), value) + 'https://api.periscope.tv/api/v2/%s?broadcast_id=%s' % (method, value), value) def _real_extract(self, url): token = self._match_id(url) @@ -43,8 +45,11 @@ class PeriscopeIE(InfoExtractor): broadcast = broadcast_data['broadcast'] status = broadcast['status'] - uploader = broadcast.get('user_display_name') or broadcast_data.get('user', {}).get('display_name') - uploader_id = broadcast.get('user_id') or broadcast_data.get('user', {}).get('id') + user = broadcast_data.get('user', {}) + + uploader = broadcast.get('user_display_name') or user.get('display_name') + uploader_id = (broadcast.get('username') or user.get('username') or + broadcast.get('user_id') or user.get('id')) title = '%s - %s' % (uploader, status) if uploader else status state = broadcast.get('state').lower() @@ -83,22 +88,44 @@ class PeriscopeIE(InfoExtractor): } -class QuickscopeIE(InfoExtractor): - IE_DESC = 'Quick Scope' - _VALID_URL = r'https?://watchonperiscope\.com/broadcast/(?P\d+)' +class PeriscopeUserIE(InfoExtractor): + _VALID_URL = r'https?://www\.periscope\.tv/(?P[^/]+)/?$' + IE_DESC = 'Periscope user videos' + IE_NAME = 'periscope:user' + _TEST = { - 'url': 'https://watchonperiscope.com/broadcast/56180087', - 'only_matching': True, + 'url': 'https://www.periscope.tv/LularoeHusbandMike/', + 'info_dict': { + 'id': 'LularoeHusbandMike', + 'title': 'LULAROE HUSBAND MIKE', + 'description': 'md5:6cf4ec8047768098da58e446e82c82f0', + }, + # Periscope only shows videos in the last 24 hours, so it's possible to + # get 0 videos + 'playlist_mincount': 0, } def _real_extract(self, url): - broadcast_id = self._match_id(url) - request = compat_urllib_request.Request( - 'https://watchonperiscope.com/api/accessChannel', compat_urllib_parse.urlencode({ - 'broadcast_id': broadcast_id, - 'entry_ticket': '', - 'from_push': 'false', - 'uses_sessions': 'true', - }).encode('utf-8')) - return self.url_result( - self._download_json(request, broadcast_id)['share_url'], 'Periscope') + user_id = self._match_id(url) + + webpage = self._download_webpage(url, user_id) + + data_store = self._parse_json( + unescapeHTML(self._search_regex( + r'data-store=(["\'])(?P.+?)\1', + webpage, 'data store', default='{}', group='data')), + user_id) + + user = data_store.get('User', {}).get('user', {}) + title = user.get('display_name') or user.get('username') + description = user.get('description') + + broadcast_ids = (data_store.get('UserBroadcastHistory', {}).get('broadcastIds') or + data_store.get('BroadcastCache', {}).get('broadcastIds', [])) + + entries = [ + self.url_result( + 'https://www.periscope.tv/%s/%s' % (user_id, broadcast_id)) + for broadcast_id in broadcast_ids] + + return self.playlist_result(entries, user_id, title, description)