X-Git-Url: http://git.bitcoin.ninja/index.cgi?a=blobdiff_plain;f=youtube_dl%2Fextractor%2Fperiscope.py;h=b15906390d07715494b5653dce5499ca0ad72141;hb=7016e24ebea09c717f413273cda16721c4098325;hp=e8b2f11c6b2e0bd7842ef4c026e2dbb3cd76bcbf;hpb=f0bc5a8609786633d8b51ab4255c1f0fdb941f73;p=youtube-dl diff --git a/youtube_dl/extractor/periscope.py b/youtube_dl/extractor/periscope.py index e8b2f11c6..b15906390 100644 --- a/youtube_dl/extractor/periscope.py +++ b/youtube_dl/extractor/periscope.py @@ -5,6 +5,7 @@ import re from .common import InfoExtractor from ..utils import ( + int_or_none, parse_iso8601, unescapeHTML, ) @@ -16,12 +17,54 @@ class PeriscopeBaseIE(InfoExtractor): 'https://api.periscope.tv/api/v2/%s' % method, item_id, query=query) + def _parse_broadcast_data(self, broadcast, video_id): + title = broadcast.get('status') or 'Periscope Broadcast' + uploader = broadcast.get('user_display_name') or broadcast.get('username') + title = '%s - %s' % (uploader, title) if uploader else title + is_live = broadcast.get('state').lower() == 'running' + + thumbnails = [{ + 'url': broadcast[image], + } for image in ('image_url', 'image_url_small') if broadcast.get(image)] + + return { + 'id': broadcast.get('id') or video_id, + 'title': self._live_title(title) if is_live else title, + 'timestamp': parse_iso8601(broadcast.get('created_at')), + 'uploader': uploader, + 'uploader_id': broadcast.get('user_id') or broadcast.get('username'), + 'thumbnails': thumbnails, + 'view_count': int_or_none(broadcast.get('total_watched')), + 'tags': broadcast.get('tags'), + 'is_live': is_live, + } + + @staticmethod + def _extract_common_format_info(broadcast): + return broadcast.get('state').lower(), int_or_none(broadcast.get('width')), int_or_none(broadcast.get('height')) + + @staticmethod + def _add_width_and_height(f, width, height): + for key, val in (('width', width), ('height', height)): + if not f.get(key): + f[key] = val + + def _extract_pscp_m3u8_formats(self, m3u8_url, video_id, format_id, state, width, height, fatal=True): + m3u8_formats = self._extract_m3u8_formats( + m3u8_url, video_id, 'mp4', + entry_protocol='m3u8_native' + if state in ('ended', 'timed_out') else 'm3u8', + m3u8_id=format_id, fatal=fatal) + if len(m3u8_formats) == 1: + self._add_width_and_height(m3u8_formats[0], width, height) + return m3u8_formats + class PeriscopeIE(PeriscopeBaseIE): IE_DESC = 'Periscope' IE_NAME = 'periscope' - _VALID_URL = r'https?://(?:www\.)?periscope\.tv/[^/]+/(?P[^/?#]+)' - # Alive example URLs can be found here http://onperiscope.com/ + _VALID_URL = r'https?://(?:www\.)?(?:periscope|pscp)\.tv/[^/]+/(?P[^/?#]+)' + # Alive example URLs can be found here https://www.periscope.tv/ _TESTS = [{ 'url': 'https://www.periscope.tv/w/aJUQnjY3MjA3ODF8NTYxMDIyMDl2zCg2pECBgwTqRpQuQD352EMPTKQjT4uqlM3cgWFA-g==', 'md5': '65b57957972e503fcbbaeed8f4fa04ca', @@ -41,69 +84,62 @@ class PeriscopeIE(PeriscopeBaseIE): }, { 'url': 'https://www.periscope.tv/bastaakanoggano/1OdKrlkZZjOJX', 'only_matching': True, + }, { + 'url': 'https://www.periscope.tv/w/1ZkKzPbMVggJv', + 'only_matching': True, }] @staticmethod def _extract_url(webpage): mobj = re.search( - r']+src=([\'"])(?P(?:https?:)?//(?:www\.)?periscope\.tv/(?:(?!\1).)+)\1', webpage) + r']+src=([\'"])(?P(?:https?:)?//(?:www\.)?(?:periscope|pscp)\.tv/(?:(?!\1).)+)\1', webpage) if mobj: return mobj.group('url') def _real_extract(self, url): token = self._match_id(url) - broadcast_data = self._call_api( - 'getBroadcastPublic', {'broadcast_id': token}, token) - broadcast = broadcast_data['broadcast'] - status = broadcast['status'] - - user = broadcast_data.get('user', {}) + stream = self._call_api( + 'accessVideoPublic', {'broadcast_id': token}, token) - 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')) + broadcast = stream['broadcast'] + info = self._parse_broadcast_data(broadcast, token) - title = '%s - %s' % (uploader, status) if uploader else status state = broadcast.get('state').lower() - if state == 'running': - title = self._live_title(title) - timestamp = parse_iso8601(broadcast.get('created_at')) - - thumbnails = [{ - 'url': broadcast[image], - } for image in ('image_url', 'image_url_small') if broadcast.get(image)] + width = int_or_none(broadcast.get('width')) + height = int_or_none(broadcast.get('height')) - stream = self._call_api( - 'getAccessPublic', {'broadcast_id': token}, token) + def add_width_and_height(f): + for key, val in (('width', width), ('height', height)): + if not f.get(key): + f[key] = val + video_urls = set() formats = [] - for format_id in ('replay', 'rtmp', 'hls', 'https_hls'): + for format_id in ('replay', 'rtmp', 'hls', 'https_hls', 'lhls', 'lhlsweb'): video_url = stream.get(format_id + '_url') - if not video_url: + if not video_url or video_url in video_urls: continue - f = { + video_urls.add(video_url) + if format_id != 'rtmp': + m3u8_formats = self._extract_pscp_m3u8_formats( + video_url, token, format_id, state, width, height, False) + formats.extend(m3u8_formats) + continue + rtmp_format = { 'url': video_url, 'ext': 'flv' if format_id == 'rtmp' else 'mp4', } - if format_id != 'rtmp': - f['protocol'] = 'm3u8_native' if state == 'ended' else 'm3u8' - formats.append(f) + self._add_width_and_height(rtmp_format) + formats.append(rtmp_format) self._sort_formats(formats) - return { - 'id': broadcast.get('id') or token, - 'title': title, - 'timestamp': timestamp, - 'uploader': uploader, - 'uploader_id': uploader_id, - 'thumbnails': thumbnails, - 'formats': formats, - } + info['formats'] = formats + return info class PeriscopeUserIE(PeriscopeBaseIE): - _VALID_URL = r'https?://(?:www\.)?periscope\.tv/(?P[^/]+)/?$' + _VALID_URL = r'https?://(?:www\.)?(?:periscope|pscp)\.tv/(?P[^/]+)/?$' IE_DESC = 'Periscope user videos' IE_NAME = 'periscope:user' @@ -132,7 +168,7 @@ class PeriscopeUserIE(PeriscopeBaseIE): user = list(data_store['UserCache']['users'].values())[0]['user'] user_id = user['id'] - session_id = data_store['SessionToken']['broadcastHistory']['token']['session_id'] + session_id = data_store['SessionToken']['public']['broadcastHistory']['token']['session_id'] broadcasts = self._call_api( 'getUserBroadcastsPublic',