X-Git-Url: https://git.rapsys.eu/youtubedl/blobdiff_plain/f46044c66663049e286c20ee015db99d47d9dd8a..4e090bc3ceacc4e3cd464d12ea97700e3acad37d:/youtube_dl/extractor/twitch.py diff --git a/youtube_dl/extractor/twitch.py b/youtube_dl/extractor/twitch.py index 4b0ce54..bbba394 100644 --- a/youtube_dl/extractor/twitch.py +++ b/youtube_dl/extractor/twitch.py @@ -7,13 +7,23 @@ import random from .common import InfoExtractor from ..compat import ( + compat_HTTPError, + compat_parse_qs, compat_str, - compat_urllib_parse, - compat_urllib_request, + compat_urllib_parse_urlencode, + compat_urllib_parse_urlparse, + compat_urlparse, ) from ..utils import ( + clean_html, ExtractorError, + int_or_none, + js_to_json, + orderedSet, + parse_duration, parse_iso8601, + update_url_query, + urlencode_postdata, ) @@ -21,8 +31,10 @@ class TwitchBaseIE(InfoExtractor): _VALID_URL_BASE = r'https?://(?:www\.)?twitch\.tv' _API_BASE = 'https://api.twitch.tv' - _USHER_BASE = 'http://usher.twitch.tv' - _LOGIN_URL = 'https://secure.twitch.tv/user/login' + _USHER_BASE = 'https://usher.ttvnw.net' + _LOGIN_URL = 'http://www.twitch.tv/login' + _CLIENT_ID = 'jzkbprff40iqj646a697cyrvl0zt2m6' + _NETRC_MACHINE = 'twitch' def _handle_error(self, response): if not isinstance(response, dict): @@ -33,16 +45,10 @@ class TwitchBaseIE(InfoExtractor): '%s returned error: %s - %s' % (self.IE_NAME, error, response.get('message')), expected=True) - def _download_json(self, url, video_id, note='Downloading JSON metadata'): - 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) + def _call_api(self, path, item_id, note): + response = self._download_json( + '%s/%s' % (self._API_BASE, path), item_id, note, + headers={'Client-ID': self._CLIENT_ID}) self._handle_error(response) return response @@ -54,47 +60,71 @@ class TwitchBaseIE(InfoExtractor): if username is None: return - login_page = self._download_webpage( + def fail(message): + raise ExtractorError( + 'Unable to login. Twitch said: %s' % message, expected=True) + + login_page, handle = self._download_webpage_handle( self._LOGIN_URL, None, 'Downloading login page') - authenticity_token = self._search_regex( - r']*>(?P[^<]+)", response) - if m: - raise ExtractorError( - 'Unable to login: %s' % m.group('msg').strip(), expected=True) + login_form.update({ + 'username': username, + 'password': password, + }) + + redirect_url = handle.geturl() + + post_url = self._search_regex( + r']+action=(["\'])(?P.+?)\1', login_page, + 'post url', default=redirect_url, group='url') + + if not post_url.startswith('http'): + post_url = compat_urlparse.urljoin(redirect_url, post_url) + + headers = {'Referer': redirect_url} + + try: + response = self._download_json( + post_url, None, 'Logging in as %s' % username, + data=urlencode_postdata(login_form), + headers=headers) + except ExtractorError as e: + if isinstance(e.cause, compat_HTTPError) and e.cause.code == 400: + response = self._parse_json( + e.cause.read().decode('utf-8'), None) + fail(response['message']) + raise + + if response.get('redirect'): + self._download_webpage( + response['redirect'], None, 'Downloading login redirect page', + headers=headers) + + 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): - return self._extract_info(self._download_json( - '%s/kraken/videos/%s%s' % (self._API_BASE, item, item_id), item_id, + return self._extract_info(self._call_api( + 'kraken/videos/%s%s' % (item, item_id), item_id, 'Downloading %s info JSON' % self._ITEM_TYPE)) def _extract_media(self, item_id): info = self._download_info(self._ITEM_SHORTCUT, item_id) - response = self._download_json( - '%s/api/videos/%s%s' % (self._API_BASE, self._ITEM_SHORTCUT, item_id), item_id, + response = self._call_api( + 'api/videos/%s%s' % (self._ITEM_SHORTCUT, item_id), item_id, 'Downloading %s playlist JSON' % self._ITEM_TYPE) entries = [] chunks = response['chunks'] @@ -123,14 +153,14 @@ class TwitchItemBaseIE(TwitchBaseIE): def _extract_info(self, info): return { 'id': info['_id'], - 'title': info['title'], - 'description': info['description'], - 'duration': info['length'], - 'thumbnail': info['preview'], - 'uploader': info['channel']['display_name'], - 'uploader_id': info['channel']['name'], - 'timestamp': parse_iso8601(info['recorded_at']), - 'view_count': info['views'], + 'title': info.get('title') or 'Untitled Broadcast', + 'description': info.get('description'), + 'duration': int_or_none(info.get('length')), + 'thumbnail': info.get('preview'), + 'uploader': info.get('channel', {}).get('display_name'), + 'uploader_id': info.get('channel', {}).get('name'), + 'timestamp': parse_iso8601(info.get('recorded_at')), + 'view_count': int_or_none(info.get('views')), } def _real_extract(self, url): @@ -139,7 +169,7 @@ 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' @@ -150,12 +180,13 @@ class TwitchVideoIE(TwitchItemBaseIE): 'title': 'Worlds Semifinals - Star Horn Royal Club vs. OMG', }, 'playlist_mincount': 12, + 'skip': 'HTTP Error 404: Not Found', } 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' @@ -166,6 +197,7 @@ class TwitchChapterIE(TwitchItemBaseIE): 'title': 'ACRL Off Season - Sports Cars @ Nordschleife', }, 'playlist_mincount': 3, + 'skip': 'HTTP Error 404: Not Found', }, { 'url': 'http://www.twitch.tv/tsm_theoddone/c/2349361', 'only_matching': True, @@ -174,67 +206,151 @@ class TwitchChapterIE(TwitchItemBaseIE): class TwitchVodIE(TwitchItemBaseIE): IE_NAME = 'twitch:vod' - _VALID_URL = r'%s/[^/]+/v/(?P[^/]+)' % TwitchBaseIE._VALID_URL_BASE + _VALID_URL = r'''(?x) + https?:// + (?: + (?:www\.)?twitch\.tv/(?:[^/]+/v|videos)/| + player\.twitch\.tv/\?.*?\bvideo=v + ) + (?P\d+) + ''' _ITEM_TYPE = 'vod' _ITEM_SHORTCUT = 'v' - _TEST = { - 'url': 'http://www.twitch.tv/ksptv/v/3622000', + _TESTS = [{ + 'url': 'http://www.twitch.tv/riotgames/v/6528877?t=5m10s', 'info_dict': { - 'id': 'v3622000', + 'id': 'v6528877', 'ext': 'mp4', - 'title': '''KSPTV: Squadcast: "Everyone's on vacation so here's Dahud" Edition!''', - 'thumbnail': 're:^https?://.*\.jpg$', - 'duration': 6951, - 'timestamp': 1419028564, - 'upload_date': '20141219', - 'uploader': 'KSPTV', - 'uploader_id': 'ksptv', + 'title': 'LCK Summer Split - Week 6 Day 1', + 'thumbnail': r're:^https?://.*\.jpg$', + 'duration': 17208, + 'timestamp': 1435131709, + 'upload_date': '20150624', + 'uploader': 'Riot Games', + 'uploader_id': 'riotgames', 'view_count': int, + 'start_time': 310, }, 'params': { # m3u8 download 'skip_download': True, }, - } + }, { + # Untitled broadcast (title is None) + 'url': 'http://www.twitch.tv/belkao_o/v/11230755', + 'info_dict': { + 'id': 'v11230755', + 'ext': 'mp4', + 'title': 'Untitled Broadcast', + 'thumbnail': r're:^https?://.*\.jpg$', + 'duration': 1638, + 'timestamp': 1439746708, + 'upload_date': '20150816', + 'uploader': 'BelkAO_o', + 'uploader_id': 'belkao_o', + 'view_count': int, + }, + 'params': { + # m3u8 download + 'skip_download': True, + }, + 'skip': 'HTTP Error 404: Not Found', + }, { + 'url': 'http://player.twitch.tv/?t=5m10s&video=v6528877', + 'only_matching': True, + }, { + 'url': 'https://www.twitch.tv/videos/6528877', + 'only_matching': True, + }] def _real_extract(self, url): item_id = self._match_id(url) + info = self._download_info(self._ITEM_SHORTCUT, item_id) - access_token = self._download_json( - '%s/api/vods/%s/access_token' % (self._API_BASE, item_id), item_id, + access_token = self._call_api( + 'api/vods/%s/access_token' % item_id, item_id, 'Downloading %s access token' % self._ITEM_TYPE) + formats = self._extract_m3u8_formats( - '%s/vod/%s?nauth=%s&nauthsig=%s' - % (self._USHER_BASE, item_id, access_token['token'], access_token['sig']), - item_id, 'mp4') + '%s/vod/%s?%s' % ( + self._USHER_BASE, item_id, + compat_urllib_parse_urlencode({ + 'allow_source': 'true', + 'allow_audio_only': 'true', + 'allow_spectre': 'true', + 'player': 'twitchweb', + 'nauth': access_token['token'], + 'nauthsig': access_token['sig'], + })), + item_id, 'mp4', entry_protocol='m3u8_native') + + self._prefer_source(formats) info['formats'] = formats + + parsed_url = compat_urllib_parse_urlparse(url) + query = compat_parse_qs(parsed_url.query) + if 't' in query: + info['start_time'] = parse_duration(query['t'][0]) + + if info.get('timestamp') is not None: + info['subtitles'] = { + 'rechat': [{ + 'url': update_url_query( + 'https://rechat.twitch.tv/rechat-messages', { + 'video_id': 'v%s' % item_id, + 'start': info['timestamp'], + }), + 'ext': 'json', + }], + } + return info class TwitchPlaylistBaseIE(TwitchBaseIE): - _PLAYLIST_URL = '%s/kraken/channels/%%s/videos/?offset=%%d&limit=%%d' % TwitchBaseIE._API_BASE + _PLAYLIST_PATH = 'kraken/channels/%s/videos/?offset=%d&limit=%d' _PAGE_LIMIT = 100 def _extract_playlist(self, channel_id): - info = self._download_json( - '%s/kraken/channels/%s' % (self._API_BASE, channel_id), + info = self._call_api( + 'kraken/channels/%s' % channel_id, channel_id, 'Downloading channel info JSON') channel_name = info.get('display_name') or info.get('name') entries = [] offset = 0 limit = self._PAGE_LIMIT + broken_paging_detected = False + counter_override = None for counter in itertools.count(1): - response = self._download_json( - self._PLAYLIST_URL % (channel_id, offset, limit), - channel_id, 'Downloading %s videos JSON page %d' % (self._PLAYLIST_TYPE, counter)) + response = self._call_api( + self._PLAYLIST_PATH % (channel_id, offset, limit), + channel_id, + 'Downloading %s JSON page %s' + % (self._PLAYLIST_TYPE, counter_override or counter)) page_entries = self._extract_playlist_page(response) if not page_entries: break + total = int_or_none(response.get('_total')) + # Since the beginning of March 2016 twitch's paging mechanism + # is completely broken on the twitch side. It simply ignores + # a limit and returns the whole offset number of videos. + # Working around by just requesting all videos at once. + # Upd: pagination bug was fixed by twitch on 15.03.2016. + if not broken_paging_detected and total and len(page_entries) > limit: + self.report_warning( + 'Twitch pagination is broken on twitch side, requesting all videos at once', + channel_id) + broken_paging_detected = True + offset = total + counter_override = '(all at once)' + continue entries.extend(page_entries) + if broken_paging_detected or total and len(page_entries) >= total: + break offset += limit return self.playlist_result( - [self.url_result(entry) for entry in set(entries)], + [self.url_result(entry) for entry in orderedSet(entries)], channel_id, channel_name) def _extract_playlist_page(self, response): @@ -260,52 +376,87 @@ class TwitchProfileIE(TwitchPlaylistBaseIE): } -class TwitchPastBroadcastsIE(TwitchPlaylistBaseIE): - IE_NAME = 'twitch:past_broadcasts' - _VALID_URL = r'%s/(?P[^/]+)/profile/past_broadcasts/?(?:\#.*)?$' % TwitchBaseIE._VALID_URL_BASE - _PLAYLIST_URL = TwitchPlaylistBaseIE._PLAYLIST_URL + '&broadcasts=true' - _PLAYLIST_TYPE = 'past broadcasts' +class TwitchVideosBaseIE(TwitchPlaylistBaseIE): + _VALID_URL_VIDEOS_BASE = r'%s/(?P[^/]+)/videos' % TwitchBaseIE._VALID_URL_BASE + _PLAYLIST_PATH = TwitchPlaylistBaseIE._PLAYLIST_PATH + '&broadcast_type=' + + +class TwitchAllVideosIE(TwitchVideosBaseIE): + IE_NAME = 'twitch:videos:all' + _VALID_URL = r'%s/all' % TwitchVideosBaseIE._VALID_URL_VIDEOS_BASE + _PLAYLIST_PATH = TwitchVideosBaseIE._PLAYLIST_PATH + 'archive,upload,highlight' + _PLAYLIST_TYPE = 'all videos' _TEST = { - 'url': 'http://www.twitch.tv/spamfish/profile/past_broadcasts', + 'url': 'https://www.twitch.tv/spamfish/videos/all', 'info_dict': { 'id': 'spamfish', 'title': 'Spamfish', }, - 'playlist_mincount': 54, + 'playlist_mincount': 869, } -class TwitchBookmarksIE(TwitchPlaylistBaseIE): - IE_NAME = 'twitch:bookmarks' - _VALID_URL = r'%s/(?P[^/]+)/profile/bookmarks/?(?:\#.*)?$' % TwitchBaseIE._VALID_URL_BASE - _PLAYLIST_URL = '%s/api/bookmark/?user=%%s&offset=%%d&limit=%%d' % TwitchBaseIE._API_BASE - _PLAYLIST_TYPE = 'bookmarks' +class TwitchUploadsIE(TwitchVideosBaseIE): + IE_NAME = 'twitch:videos:uploads' + _VALID_URL = r'%s/uploads' % TwitchVideosBaseIE._VALID_URL_VIDEOS_BASE + _PLAYLIST_PATH = TwitchVideosBaseIE._PLAYLIST_PATH + 'upload' + _PLAYLIST_TYPE = 'uploads' _TEST = { - 'url': 'http://www.twitch.tv/ognos/profile/bookmarks', + 'url': 'https://www.twitch.tv/spamfish/videos/uploads', 'info_dict': { - 'id': 'ognos', - 'title': 'Ognos', + 'id': 'spamfish', + 'title': 'Spamfish', }, - 'playlist_mincount': 3, + 'playlist_mincount': 0, } - def _extract_playlist_page(self, response): - entries = [] - for bookmark in response.get('bookmarks', []): - video = bookmark.get('video') - if not video: - continue - entries.append(video['url']) - return entries + +class TwitchPastBroadcastsIE(TwitchVideosBaseIE): + IE_NAME = 'twitch:videos:past-broadcasts' + _VALID_URL = r'%s/past-broadcasts' % TwitchVideosBaseIE._VALID_URL_VIDEOS_BASE + _PLAYLIST_PATH = TwitchVideosBaseIE._PLAYLIST_PATH + 'archive' + _PLAYLIST_TYPE = 'past broadcasts' + + _TEST = { + 'url': 'https://www.twitch.tv/spamfish/videos/past-broadcasts', + 'info_dict': { + 'id': 'spamfish', + 'title': 'Spamfish', + }, + 'playlist_mincount': 0, + } + + +class TwitchHighlightsIE(TwitchVideosBaseIE): + IE_NAME = 'twitch:videos:highlights' + _VALID_URL = r'%s/highlights' % TwitchVideosBaseIE._VALID_URL_VIDEOS_BASE + _PLAYLIST_PATH = TwitchVideosBaseIE._PLAYLIST_PATH + 'highlight' + _PLAYLIST_TYPE = 'highlights' + + _TEST = { + 'url': 'https://www.twitch.tv/spamfish/videos/highlights', + 'info_dict': { + 'id': 'spamfish', + 'title': 'Spamfish', + }, + 'playlist_mincount': 805, + } class TwitchStreamIE(TwitchBaseIE): IE_NAME = 'twitch:stream' - _VALID_URL = r'%s/(?P[^/]+)/?(?:\#.*)?$' % TwitchBaseIE._VALID_URL_BASE + _VALID_URL = r'''(?x) + https?:// + (?: + (?:www\.)?twitch\.tv/| + player\.twitch\.tv/\?.*?\bchannel= + ) + (?P[^/#?]+) + ''' - _TEST = { + _TESTS = [{ 'url': 'http://www.twitch.tv/shroomztv', 'info_dict': { 'id': '12772022048', @@ -324,45 +475,63 @@ class TwitchStreamIE(TwitchBaseIE): # m3u8 download 'skip_download': True, }, - } + }, { + 'url': 'http://www.twitch.tv/miracle_doto#profile-0', + 'only_matching': True, + }, { + 'url': 'https://player.twitch.tv/?channel=lotsofs', + 'only_matching': True, + }] + + @classmethod + def suitable(cls, url): + return (False + if any(ie.suitable(url) for ie in ( + TwitchVideoIE, + TwitchChapterIE, + TwitchVodIE, + TwitchProfileIE, + TwitchAllVideosIE, + TwitchUploadsIE, + TwitchPastBroadcastsIE, + TwitchHighlightsIE)) + else super(TwitchStreamIE, cls).suitable(url)) def _real_extract(self, url): channel_id = self._match_id(url) - stream = self._download_json( - '%s/kraken/streams/%s' % (self._API_BASE, channel_id), channel_id, + stream = self._call_api( + 'kraken/streams/%s?stream_type=all' % channel_id, channel_id, 'Downloading stream JSON').get('stream') - # Fallback on profile extraction if stream is offline if not stream: - return self.url_result( - 'http://www.twitch.tv/%s/profile' % channel_id, - 'TwitchProfile', channel_id) + raise ExtractorError('%s is offline' % channel_id, expected=True) + + # Channel name may be typed if different case than the original channel name + # (e.g. http://www.twitch.tv/TWITCHPLAYSPOKEMON) that will lead to constructing + # an invalid m3u8 URL. Working around by use of original channel name from stream + # JSON and fallback to lowercase if it's not available. + channel_id = stream.get('channel', {}).get('name') or channel_id.lower() - access_token = self._download_json( - '%s/api/channels/%s/access_token' % (self._API_BASE, channel_id), channel_id, + access_token = self._call_api( + 'api/channels/%s/access_token' % channel_id, channel_id, 'Downloading channel access token') query = { 'allow_source': 'true', + 'allow_audio_only': 'true', + 'allow_spectre': 'true', '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') - - # prefer the 'source' stream, the others are limited to 30 fps - def _sort_source(f): - if f.get('m3u8_media') is not None and f['m3u8_media'].get('NAME') == 'Source': - return 1 - return 0 - formats = sorted(formats, key=_sort_source) + self._prefer_source(formats) view_count = stream.get('viewers') timestamp = parse_iso8601(stream.get('created_at')) @@ -395,3 +564,61 @@ class TwitchStreamIE(TwitchBaseIE): 'formats': formats, 'is_live': True, } + + +class TwitchClipsIE(InfoExtractor): + IE_NAME = 'twitch:clips' + _VALID_URL = r'https?://clips\.twitch\.tv/(?:[^/]+/)*(?P[^/?#&]+)' + + _TESTS = [{ + 'url': 'https://clips.twitch.tv/ea/AggressiveCobraPoooound', + 'md5': '761769e1eafce0ffebfb4089cb3847cd', + 'info_dict': { + 'id': 'AggressiveCobraPoooound', + 'ext': 'mp4', + 'title': 'EA Play 2016 Live from the Novo Theatre', + 'thumbnail': r're:^https?://.*\.jpg', + 'creator': 'EA', + 'uploader': 'stereotype_', + 'uploader_id': 'stereotype_', + }, + }, { + # multiple formats + 'url': 'https://clips.twitch.tv/rflegendary/UninterestedBeeDAESuppy', + 'only_matching': True, + }] + + def _real_extract(self, url): + video_id = self._match_id(url) + + webpage = self._download_webpage(url, video_id) + + clip = self._parse_json( + self._search_regex( + r'(?s)clipInfo\s*=\s*({.+?});', webpage, 'clip info'), + video_id, transform_source=js_to_json) + + title = clip.get('channel_title') or self._og_search_title(webpage) + + formats = [{ + 'url': option['source'], + 'format_id': option.get('quality'), + 'height': int_or_none(option.get('quality')), + } for option in clip.get('quality_options', []) if option.get('source')] + + if not formats: + formats = [{ + 'url': clip['clip_video_url'], + }] + + self._sort_formats(formats) + + return { + 'id': video_id, + 'title': title, + 'thumbnail': self._og_search_thumbnail(webpage), + 'creator': clip.get('broadcaster_display_name') or clip.get('broadcaster_login'), + 'uploader': clip.get('curator_login'), + 'uploader_id': clip.get('curator_display_name'), + 'formats': formats, + }