X-Git-Url: http://git.bitcoin.ninja/index.cgi?a=blobdiff_plain;f=youtube_dl%2Fextractor%2Ftwitch.py;h=94bd6345da18815a50b72502a8b91ae4e30ae2b5;hb=233c1c0e76d64c9e13dc8968bfd8a014c49e66a8;hp=340cadcf5f2a00f855d4cbb47c1b46ba09ec502a;hpb=5f0d813d9395848e92a1c6d83335360652d654c1;p=youtube-dl diff --git a/youtube_dl/extractor/twitch.py b/youtube_dl/extractor/twitch.py index 340cadcf5..94bd6345d 100644 --- a/youtube_dl/extractor/twitch.py +++ b/youtube_dl/extractor/twitch.py @@ -23,6 +23,8 @@ class TwitchBaseIE(InfoExtractor): _API_BASE = 'https://api.twitch.tv' _USHER_BASE = 'http://usher.twitch.tv' _LOGIN_URL = 'https://secure.twitch.tv/user/login' + _LOGIN_POST_URL = 'https://secure-login.twitch.tv/login' + _NETRC_MACHINE = 'twitch' def _handle_error(self, response): if not isinstance(response, dict): @@ -34,7 +36,15 @@ class TwitchBaseIE(InfoExtractor): expected=True) def _download_json(self, url, video_id, note='Downloading JSON metadata'): - response = super(TwitchBaseIE, self)._download_json(url, video_id, note) + headers = { + 'Referer': 'http://api.twitch.tv/crossdomain/receiver.html?v=2', + 'X-Requested-With': 'XMLHttpRequest', + } + for cookie in self._downloader.cookiejar: + if cookie.name == 'api_token': + headers['Twitch-Api-Token'] = cookie.value + request = compat_urllib_request.Request(url, headers=headers) + response = super(TwitchBaseIE, self)._download_json(request, video_id, note) self._handle_error(response) return response @@ -58,14 +68,14 @@ class TwitchBaseIE(InfoExtractor): 'authenticity_token': authenticity_token, 'redirect_on_login': '', 'embed_form': 'false', - 'mp_source_action': '', + 'mp_source_action': 'login-button', 'follow': '', - 'user[login]': username, - 'user[password]': password, + 'login': username, + 'password': password, } request = compat_urllib_request.Request( - self._LOGIN_URL, compat_urllib_parse.urlencode(login_form).encode('utf-8')) + self._LOGIN_POST_URL, compat_urllib_parse.urlencode(login_form).encode('utf-8')) request.add_header('Referer', self._LOGIN_URL) response = self._download_webpage( request, None, 'Logging in as %s' % username) @@ -76,6 +86,14 @@ class TwitchBaseIE(InfoExtractor): raise ExtractorError( 'Unable to login: %s' % m.group('msg').strip(), expected=True) + def _prefer_source(self, formats): + try: + source = next(f for f in formats if f['format_id'] == 'Source') + source['preference'] = 10 + except StopIteration: + pass # No Source stream present + self._sort_formats(formats) + class TwitchItemBaseIE(TwitchBaseIE): def _download_info(self, item, item_id): @@ -131,7 +149,7 @@ class TwitchItemBaseIE(TwitchBaseIE): class TwitchVideoIE(TwitchItemBaseIE): IE_NAME = 'twitch:video' - _VALID_URL = r'%s/[^/]+/b/(?P[^/]+)' % TwitchBaseIE._VALID_URL_BASE + _VALID_URL = r'%s/[^/]+/b/(?P\d+)' % TwitchBaseIE._VALID_URL_BASE _ITEM_TYPE = 'video' _ITEM_SHORTCUT = 'a' @@ -147,7 +165,7 @@ class TwitchVideoIE(TwitchItemBaseIE): class TwitchChapterIE(TwitchItemBaseIE): IE_NAME = 'twitch:chapter' - _VALID_URL = r'%s/[^/]+/c/(?P[^/]+)' % TwitchBaseIE._VALID_URL_BASE + _VALID_URL = r'%s/[^/]+/c/(?P\d+)' % TwitchBaseIE._VALID_URL_BASE _ITEM_TYPE = 'chapter' _ITEM_SHORTCUT = 'c' @@ -166,7 +184,7 @@ class TwitchChapterIE(TwitchItemBaseIE): class TwitchVodIE(TwitchItemBaseIE): IE_NAME = 'twitch:vod' - _VALID_URL = r'%s/[^/]+/v/(?P[^/]+)' % TwitchBaseIE._VALID_URL_BASE + _VALID_URL = r'%s/[^/]+/v/(?P\d+)' % TwitchBaseIE._VALID_URL_BASE _ITEM_TYPE = 'vod' _ITEM_SHORTCUT = 'v' @@ -200,6 +218,7 @@ class TwitchVodIE(TwitchItemBaseIE): '%s/vod/%s?nauth=%s&nauthsig=%s' % (self._USHER_BASE, item_id, access_token['token'], access_token['sig']), item_id, 'mp4') + self._prefer_source(formats) info['formats'] = formats return info @@ -220,12 +239,18 @@ class TwitchPlaylistBaseIE(TwitchBaseIE): response = self._download_json( self._PLAYLIST_URL % (channel_id, offset, limit), channel_id, 'Downloading %s videos JSON page %d' % (self._PLAYLIST_TYPE, counter)) - videos = response['videos'] - if not videos: + page_entries = self._extract_playlist_page(response) + if not page_entries: break - entries.extend([self.url_result(video['url']) for video in videos]) + entries.extend(page_entries) offset += limit - return self.playlist_result(entries, channel_id, channel_name) + return self.playlist_result( + [self.url_result(entry) for entry in set(entries)], + channel_id, channel_name) + + def _extract_playlist_page(self, response): + videos = response.get('videos') + return [video['url'] for video in videos] if videos else [] def _real_extract(self, url): return self._extract_playlist(self._match_id(url)) @@ -262,6 +287,31 @@ class TwitchPastBroadcastsIE(TwitchPlaylistBaseIE): } +class TwitchBookmarksIE(TwitchPlaylistBaseIE): + IE_NAME = 'twitch:bookmarks' + _VALID_URL = r'%s/(?P[^/]+)/profile/bookmarks/?(?:\#.*)?$' % TwitchBaseIE._VALID_URL_BASE + _PLAYLIST_URL = '%s/api/bookmark/?user=%%s&offset=%%d&limit=%%d' % TwitchBaseIE._API_BASE + _PLAYLIST_TYPE = 'bookmarks' + + _TEST = { + 'url': 'http://www.twitch.tv/ognos/profile/bookmarks', + 'info_dict': { + 'id': 'ognos', + 'title': 'Ognos', + }, + 'playlist_mincount': 3, + } + + def _extract_playlist_page(self, response): + entries = [] + for bookmark in response.get('bookmarks', []): + video = bookmark.get('video') + if not video: + continue + entries.append(video['url']) + return entries + + class TwitchStreamIE(TwitchBaseIE): IE_NAME = 'twitch:stream' _VALID_URL = r'%s/(?P[^/]+)/?(?:\#.*)?$' % TwitchBaseIE._VALID_URL_BASE @@ -309,14 +359,14 @@ class TwitchStreamIE(TwitchBaseIE): 'p': random.randint(1000000, 10000000), 'player': 'twitchweb', 'segment_preference': '4', - 'sig': access_token['sig'], - 'token': access_token['token'], + 'sig': access_token['sig'].encode('utf-8'), + 'token': access_token['token'].encode('utf-8'), } - formats = self._extract_m3u8_formats( '%s/api/channel/hls/%s.m3u8?%s' - % (self._USHER_BASE, channel_id, compat_urllib_parse.urlencode(query).encode('utf-8')), + % (self._USHER_BASE, channel_id, compat_urllib_parse.urlencode(query)), channel_id, 'mp4') + self._prefer_source(formats) view_count = stream.get('viewers') timestamp = parse_iso8601(stream.get('created_at'))