X-Git-Url: https://git.rapsys.eu/youtubedl/blobdiff_plain/47d80ec0b18245caeb97018d4c1af18d0b5b972b..e070b5f283f0eca1a445b8382dd7292803ba2c3f:/youtube_dl/extractor/periscope.py?ds=inline diff --git a/youtube_dl/extractor/periscope.py b/youtube_dl/extractor/periscope.py index 0e36230..b337a56 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, ) @@ -20,7 +21,7 @@ class PeriscopeBaseIE(InfoExtractor): class PeriscopeIE(PeriscopeBaseIE): IE_DESC = 'Periscope' IE_NAME = 'periscope' - _VALID_URL = r'https?://(?:www\.)?periscope\.tv/[^/]+/(?P[^/?#]+)' + _VALID_URL = r'https?://(?:www\.)?(?:periscope|pscp)\.tv/[^/]+/(?P[^/?#]+)' # Alive example URLs can be found here http://onperiscope.com/ _TESTS = [{ 'url': 'https://www.periscope.tv/w/aJUQnjY3MjA3ODF8NTYxMDIyMDl2zCg2pECBgwTqRpQuQD352EMPTKQjT4uqlM3cgWFA-g==', @@ -41,30 +42,31 @@ 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'] + stream = self._call_api( + 'accessVideoPublic', {'broadcast_id': token}, token) - user = broadcast_data.get('user', {}) + broadcast = stream['broadcast'] + title = broadcast['status'] - 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')) + uploader = broadcast.get('user_display_name') or broadcast.get('username') + uploader_id = (broadcast.get('user_id') or broadcast.get('username')) - title = '%s - %s' % (uploader, status) if uploader else status + title = '%s - %s' % (uploader, title) if uploader else title state = broadcast.get('state').lower() if state == 'running': title = self._live_title(title) @@ -74,21 +76,37 @@ class PeriscopeIE(PeriscopeBaseIE): 'url': broadcast[image], } for image in ('image_url', 'image_url_small') if broadcast.get(image)] - stream = self._call_api( - 'getAccessPublic', {'broadcast_id': token}, token) + width = int_or_none(broadcast.get('width')) + height = int_or_none(broadcast.get('height')) + 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 + video_urls.add(video_url) + if format_id != 'rtmp': + m3u8_formats = self._extract_m3u8_formats( + video_url, token, 'mp4', + entry_protocol='m3u8_native' + if state in ('ended', 'timed_out') else 'm3u8', + m3u8_id=format_id, fatal=False) + if len(m3u8_formats) == 1: + add_width_and_height(m3u8_formats[0]) + formats.extend(m3u8_formats) continue - f = { + rtmp_format = { 'url': video_url, 'ext': 'flv' if format_id == 'rtmp' else 'mp4', } - if format_id != 'rtmp': - f['protocol'] = 'm3u8_native' if state in ('ended', 'timed_out') else 'm3u8' - formats.append(f) + add_width_and_height(rtmp_format) + formats.append(rtmp_format) self._sort_formats(formats) return { @@ -103,7 +121,7 @@ class PeriscopeIE(PeriscopeBaseIE): 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'