X-Git-Url: http://git.bitcoin.ninja/index.cgi?a=blobdiff_plain;f=youtube_dl%2Fextractor%2Fyoutube.py;h=cfe9eed551088dfac9d2690cac0de03c375c214e;hb=5c2266df4b9aeb7881ed8c026a038e2a25e43734;hp=8e2da46e3ad68d902a2757dcc9b4e4613c7adb45;hpb=cbaed4bb5e5e90103a1164d9326043a3abd0bf83;p=youtube-dl diff --git a/youtube_dl/extractor/youtube.py b/youtube_dl/extractor/youtube.py index 8e2da46e3..cfe9eed55 100644 --- a/youtube_dl/extractor/youtube.py +++ b/youtube_dl/extractor/youtube.py @@ -20,12 +20,12 @@ from ..compat import ( compat_urllib_parse_unquote, compat_urllib_parse_unquote_plus, compat_urllib_parse_urlparse, - compat_urllib_request, compat_urlparse, compat_str, ) from ..utils import ( clean_html, + encode_dict, ExtractorError, float_or_none, get_element_by_attribute, @@ -34,6 +34,7 @@ from ..utils import ( orderedSet, parse_duration, remove_start, + sanitized_Request, smuggle_url, str_to_int, unescapeHTML, @@ -111,12 +112,9 @@ class YoutubeBaseInfoExtractor(InfoExtractor): 'hl': 'en_US', } - # Convert to UTF-8 *before* urlencode because Python 2.x's urlencode - # chokes on unicode - login_form = dict((k.encode('utf-8'), v.encode('utf-8')) for k, v in login_form_strs.items()) - login_data = compat_urllib_parse.urlencode(login_form).encode('ascii') + login_data = compat_urllib_parse.urlencode(encode_dict(login_form_strs)).encode('ascii') - req = compat_urllib_request.Request(self._LOGIN_URL, login_data) + req = sanitized_Request(self._LOGIN_URL, login_data) login_results = self._download_webpage( req, None, note='Logging in', errnote='unable to log in', fatal=False) @@ -147,10 +145,9 @@ class YoutubeBaseInfoExtractor(InfoExtractor): 'TrustDevice': 'on', }) - tfa_form = dict((k.encode('utf-8'), v.encode('utf-8')) for k, v in tfa_form_strs.items()) - tfa_data = compat_urllib_parse.urlencode(tfa_form).encode('ascii') + tfa_data = compat_urllib_parse.urlencode(encode_dict(tfa_form_strs)).encode('ascii') - tfa_req = compat_urllib_request.Request(self._TWOFACTOR_URL, tfa_data) + tfa_req = sanitized_Request(self._TWOFACTOR_URL, tfa_data) tfa_results = self._download_webpage( tfa_req, None, note='Submitting TFA code', errnote='unable to submit tfa', fatal=False) @@ -181,6 +178,69 @@ class YoutubeBaseInfoExtractor(InfoExtractor): return +class YoutubeEntryListBaseInfoExtractor(InfoExtractor): + # Extract entries from page with "Load more" button + def _entries(self, page, playlist_id): + more_widget_html = content_html = page + for page_num in itertools.count(1): + for entry in self._process_page(content_html): + yield entry + + mobj = re.search(r'data-uix-load-more-href="/?(?P[^"]+)"', more_widget_html) + if not mobj: + break + + more = self._download_json( + 'https://youtube.com/%s' % mobj.group('more'), playlist_id, + 'Downloading page #%s' % page_num, + transform_source=uppercase_escape) + content_html = more['content_html'] + if not content_html.strip(): + # Some webpages show a "Load more" button but they don't + # have more videos + break + more_widget_html = more['load_more_widget_html'] + + +class YoutubePlaylistBaseInfoExtractor(YoutubeEntryListBaseInfoExtractor): + def _process_page(self, content): + for video_id, video_title in self.extract_videos_from_page(content): + yield self.url_result(video_id, 'Youtube', video_id, video_title) + + def extract_videos_from_page(self, page): + ids_in_page = [] + titles_in_page = [] + for mobj in re.finditer(self._VIDEO_RE, page): + # The link with index 0 is not the first video of the playlist (not sure if still actual) + if 'index' in mobj.groupdict() and mobj.group('id') == '0': + continue + video_id = mobj.group('id') + video_title = unescapeHTML(mobj.group('title')) + if video_title: + video_title = video_title.strip() + try: + idx = ids_in_page.index(video_id) + if video_title and not titles_in_page[idx]: + titles_in_page[idx] = video_title + except ValueError: + ids_in_page.append(video_id) + titles_in_page.append(video_title) + return zip(ids_in_page, titles_in_page) + + +class YoutubePlaylistsBaseInfoExtractor(YoutubeEntryListBaseInfoExtractor): + def _process_page(self, content): + for playlist_id in re.findall(r'href="/?playlist\?list=(.+?)"', content): + yield self.url_result( + 'https://www.youtube.com/playlist?list=%s' % playlist_id, 'YoutubePlaylist') + + def _real_extract(self, url): + playlist_id = self._match_id(url) + webpage = self._download_webpage(url, playlist_id) + title = self._og_search_title(webpage, fatal=False) + return self.playlist_result(self._entries(webpage, playlist_id), playlist_id, title) + + class YoutubeIE(YoutubeBaseInfoExtractor): IE_DESC = 'YouTube.com' _VALID_URL = r"""(?x)^ @@ -366,7 +426,8 @@ class YoutubeIE(YoutubeBaseInfoExtractor): 'title': 'Principal Sexually Assaults A Teacher - Episode 117 - 8th June 2012', 'description': 'md5:09b78bd971f1e3e289601dfba15ca4f7', 'uploader': 'SET India', - 'uploader_id': 'setindia' + 'uploader_id': 'setindia', + 'age_limit': 18, } }, { @@ -503,7 +564,7 @@ class YoutubeIE(YoutubeBaseInfoExtractor): 'info_dict': { 'id': 'lqQg6PlCWgI', 'ext': 'mp4', - 'upload_date': '20120724', + 'upload_date': '20150827', 'uploader_id': 'olympic', 'description': 'HO09 - Women - GER-AUS - Hockey - 31 July 2012 - London 2012 Olympic Games', 'uploader': 'Olympics', @@ -631,7 +692,28 @@ class YoutubeIE(YoutubeBaseInfoExtractor): { 'url': 'http://vid.plus/FlRa-iH7PGw', 'only_matching': True, - } + }, + { + # Title with JS-like syntax "};" (see https://github.com/rg3/youtube-dl/issues/7468) + 'url': 'https://www.youtube.com/watch?v=lsguqyKfVQg', + 'info_dict': { + 'id': 'lsguqyKfVQg', + 'ext': 'mp4', + 'title': '{dark walk}; Loki/AC/Dishonored; collab w/Elflover21', + 'description': 'md5:8085699c11dc3f597ce0410b0dcbb34a', + 'upload_date': '20151119', + 'uploader_id': 'IronSoulElf', + 'uploader': 'IronSoulElf', + }, + 'params': { + 'skip_download': True, + }, + }, + { + # Tags with '};' (see https://github.com/rg3/youtube-dl/issues/7468) + 'url': 'https://www.youtube.com/watch?v=Ms7iBXnlUO8', + 'only_matching': True, + }, ] def __init__(self, *args, **kwargs): @@ -660,7 +742,7 @@ class YoutubeIE(YoutubeBaseInfoExtractor): def _extract_signature_function(self, video_id, player_url, example_sig): id_m = re.match( - r'.*?-(?P[a-zA-Z0-9_-]+)(?:/watch_as3|/html5player)?\.(?P[a-z]+)$', + r'.*?-(?P[a-zA-Z0-9_-]+)(?:/watch_as3|/html5player(?:-new)?|/base)?\.(?P[a-z]+)$', player_url) if not id_m: raise ExtractorError('Cannot identify player %r' % player_url) @@ -815,16 +897,33 @@ class YoutubeIE(YoutubeBaseInfoExtractor): return {} return sub_lang_list + def _get_ytplayer_config(self, video_id, webpage): + patterns = ( + # User data may contain arbitrary character sequences that may affect + # JSON extraction with regex, e.g. when '};' is contained the second + # regex won't capture the whole JSON. Yet working around by trying more + # concrete regex first keeping in mind proper quoted string handling + # to be implemented in future that will replace this workaround (see + # https://github.com/rg3/youtube-dl/issues/7468, + # https://github.com/rg3/youtube-dl/pull/7599) + r';ytplayer\.config\s*=\s*({.+?});ytplayer', + r';ytplayer\.config\s*=\s*({.+?});', + ) + config = self._search_regex( + patterns, webpage, 'ytplayer.config', default=None) + if config: + return self._parse_json( + uppercase_escape(config), video_id, fatal=False) + def _get_automatic_captions(self, video_id, webpage): """We need the webpage for getting the captions url, pass it as an argument to speed up the process.""" self.to_screen('%s: Looking for automatic captions' % video_id) - mobj = re.search(r';ytplayer.config = ({.*?});', webpage) + player_config = self._get_ytplayer_config(video_id, webpage) err_msg = 'Couldn\'t find automatic captions for %s' % video_id - if mobj is None: + if not player_config: self._downloader.report_warning(err_msg) return {} - player_config = json.loads(mobj.group(1)) try: args = player_config['args'] caption_url = args['ttsurl'] @@ -1031,10 +1130,8 @@ class YoutubeIE(YoutubeBaseInfoExtractor): age_gate = False video_info = None # Try looking directly into the video webpage - mobj = re.search(r';ytplayer\.config\s*=\s*({.*?});', video_webpage) - if mobj: - json_code = uppercase_escape(mobj.group(1)) - ytplayer_config = json.loads(json_code) + ytplayer_config = self._get_ytplayer_config(video_id, video_webpage) + if ytplayer_config: args = ytplayer_config['args'] if args.get('url_encoded_fmt_stream_map'): # Convert to the same format returned by compat_parse_qs @@ -1064,6 +1161,17 @@ class YoutubeIE(YoutubeBaseInfoExtractor): if not video_info: video_info = get_video_info if 'token' in get_video_info: + # Different get_video_info requests may report different results, e.g. + # some may report video unavailability, but some may serve it without + # any complaint (see https://github.com/rg3/youtube-dl/issues/7362, + # the original webpage as well as el=info and el=embedded get_video_info + # requests report video unavailability due to geo restriction while + # el=detailpage succeeds and returns valid data). This is probably + # due to YouTube measures against IP ranges of hosting providers. + # Working around by preferring the first succeeded video_info containing + # the token if no such video_info yet was found. + if 'token' not in video_info: + video_info = get_video_info break if 'token' not in video_info: if 'reason' in video_info: @@ -1243,7 +1351,7 @@ class YoutubeIE(YoutubeBaseInfoExtractor): encoded_url_map = video_info.get('url_encoded_fmt_stream_map', [''])[0] + ',' + video_info.get('adaptive_fmts', [''])[0] if 'rtmpe%3Dyes' in encoded_url_map: raise ExtractorError('rtmpe downloads are not supported, see https://github.com/rg3/youtube-dl/issues/343 for more information.', expected=True) - url_map = {} + formats = [] for url_data_str in encoded_url_map.split(','): url_data = compat_parse_qs(url_data_str) if 'itag' not in url_data or 'url' not in url_data: @@ -1289,7 +1397,7 @@ class YoutubeIE(YoutubeBaseInfoExtractor): player_desc = 'flash player %s' % player_version else: player_version = self._search_regex( - r'html5player-([^/]+?)(?:/html5player)?\.js', + [r'html5player-([^/]+?)(?:/html5player(?:-new)?)?\.js', r'(?:www|player)-([^/]+)/base\.js'], player_url, 'html5 player', fatal=False) player_desc = 'html5 player %s' % player_version @@ -1303,8 +1411,50 @@ class YoutubeIE(YoutubeBaseInfoExtractor): url += '&signature=' + signature if 'ratebypass' not in url: url += '&ratebypass=yes' - url_map[format_id] = url - formats = _map_to_format_list(url_map) + + # Some itags are not included in DASH manifest thus corresponding formats will + # lack metadata (see https://github.com/rg3/youtube-dl/pull/5993). + # Trying to extract metadata from url_encoded_fmt_stream_map entry. + mobj = re.search(r'^(?P\d+)[xX](?P\d+)$', url_data.get('size', [''])[0]) + width, height = (int(mobj.group('width')), int(mobj.group('height'))) if mobj else (None, None) + dct = { + 'format_id': format_id, + 'url': url, + 'player_url': player_url, + 'filesize': int_or_none(url_data.get('clen', [None])[0]), + 'tbr': float_or_none(url_data.get('bitrate', [None])[0], 1000), + 'width': width, + 'height': height, + 'fps': int_or_none(url_data.get('fps', [None])[0]), + 'format_note': url_data.get('quality_label', [None])[0] or url_data.get('quality', [None])[0], + } + type_ = url_data.get('type', [None])[0] + if type_: + type_split = type_.split(';') + kind_ext = type_split[0].split('/') + if len(kind_ext) == 2: + kind, ext = kind_ext + dct['ext'] = ext + if kind in ('audio', 'video'): + codecs = None + for mobj in re.finditer( + r'(?P[a-zA-Z_-]+)=(?P["\']?)(?P.+?)(?P=quote)(?:;|$)', type_): + if mobj.group('key') == 'codecs': + codecs = mobj.group('val') + break + if codecs: + codecs = codecs.split(',') + if len(codecs) == 2: + acodec, vcodec = codecs[0], codecs[1] + else: + acodec, vcodec = (codecs[0], 'none') if kind == 'audio' else ('none', codecs[0]) + dct.update({ + 'acodec': acodec, + 'vcodec': vcodec, + }) + if format_id in self._formats: + dct.update(self._formats[format_id]) + formats.append(dct) elif video_info.get('hlsvp'): manifest_url = video_info['hlsvp'][0] url_map = self._extract_from_m3u8(manifest_url, video_id) @@ -1380,7 +1530,7 @@ class YoutubeIE(YoutubeBaseInfoExtractor): } -class YoutubePlaylistIE(YoutubeBaseInfoExtractor): +class YoutubePlaylistIE(YoutubeBaseInfoExtractor, YoutubePlaylistBaseInfoExtractor): IE_DESC = 'YouTube.com playlists' _VALID_URL = r"""(?x)(?: (?:https?://)? @@ -1401,7 +1551,7 @@ class YoutubePlaylistIE(YoutubeBaseInfoExtractor): ((?:PL|LL|EC|UU|FL|RD|UL)[0-9A-Za-z-_]{10,}) )""" _TEMPLATE_URL = 'https://www.youtube.com/playlist?list=%s' - _VIDEO_RE = r'href="\s*/watch\?v=(?P[0-9A-Za-z_-]{11})&[^"]*?index=(?P\d+)' + _VIDEO_RE = r'href="\s*/watch\?v=(?P[0-9A-Za-z_-]{11})&[^"]*?index=(?P\d+)(?:[^>]+>(?P[^<]+))?' IE_NAME = 'youtube:playlist' _TESTS = [{ 'url': 'https://www.youtube.com/playlist?list=PLwiyx1dc3P2JR9N8gQaQN_BCvlSlap7re', @@ -1518,37 +1668,11 @@ class YoutubePlaylistIE(YoutubeBaseInfoExtractor): else: self.report_warning('Youtube gives an alert message: ' + match) - # Extract the video ids from the playlist pages - def _entries(): - more_widget_html = content_html = page - for page_num in itertools.count(1): - matches = re.finditer(self._VIDEO_RE, content_html) - # We remove the duplicates and the link with index 0 - # (it's not the first video of the playlist) - new_ids = orderedSet(m.group('id') for m in matches if m.group('index') != '0') - for vid_id in new_ids: - yield self.url_result(vid_id, 'Youtube', video_id=vid_id) - - mobj = re.search(r'data-uix-load-more-href="/?(?P<more>[^"]+)"', more_widget_html) - if not mobj: - break - - more = self._download_json( - 'https://youtube.com/%s' % mobj.group('more'), playlist_id, - 'Downloading page #%s' % page_num, - transform_source=uppercase_escape) - content_html = more['content_html'] - if not content_html.strip(): - # Some webpages show a "Load more" button but they don't - # have more videos - break - more_widget_html = more['load_more_widget_html'] - playlist_title = self._html_search_regex( - r'(?s)<h1 class="pl-header-title[^"]*">\s*(.*?)\s*</h1>', + r'(?s)<h1 class="pl-header-title[^"]*"[^>]*>\s*(.*?)\s*</h1>', page, 'title') - return self.playlist_result(_entries(), playlist_id, playlist_title) + return self.playlist_result(self._entries(page, playlist_id), playlist_id, playlist_title) def _real_extract(self, url): # Extract playlist id @@ -1574,36 +1698,31 @@ class YoutubePlaylistIE(YoutubeBaseInfoExtractor): return self._extract_playlist(playlist_id) -class YoutubeChannelIE(InfoExtractor): +class YoutubeChannelIE(YoutubePlaylistBaseInfoExtractor): IE_DESC = 'YouTube.com channels' _VALID_URL = r'https?://(?:youtu\.be|(?:\w+\.)?youtube(?:-nocookie)?\.com)/channel/(?P<id>[0-9A-Za-z_-]+)' _TEMPLATE_URL = 'https://www.youtube.com/channel/%s/videos' + _VIDEO_RE = r'(?:title="(?P<title>[^"]+)"[^>]+)?href="/watch\?v=(?P<id>[0-9A-Za-z_-]+)&?' IE_NAME = 'youtube:channel' _TESTS = [{ 'note': 'paginated channel', 'url': 'https://www.youtube.com/channel/UCKfVa3S1e4PHvxWcwyMMg8w', 'playlist_mincount': 91, 'info_dict': { - 'id': 'UCKfVa3S1e4PHvxWcwyMMg8w', + 'id': 'UUKfVa3S1e4PHvxWcwyMMg8w', + 'title': 'Uploads from lex will', } + }, { + 'note': 'Age restricted channel', + # from https://www.youtube.com/user/DeusExOfficial + 'url': 'https://www.youtube.com/channel/UCs0ifCMCm1icqRbqhUINa0w', + 'playlist_mincount': 64, + 'info_dict': { + 'id': 'UUs0ifCMCm1icqRbqhUINa0w', + 'title': 'Uploads from Deus Ex', + }, }] - @staticmethod - def extract_videos_from_page(page): - ids_in_page = [] - titles_in_page = [] - for mobj in re.finditer(r'(?:title="(?P<title>[^"]+)"[^>]+)?href="/watch\?v=(?P<id>[0-9A-Za-z_-]+)&?', page): - video_id = mobj.group('id') - video_title = unescapeHTML(mobj.group('title')) - try: - idx = ids_in_page.index(video_id) - if video_title and not titles_in_page[idx]: - titles_in_page[idx] = video_title - except ValueError: - ids_in_page.append(video_id) - titles_in_page.append(video_title) - return zip(ids_in_page, titles_in_page) - def _real_extract(self, url): channel_id = self._match_id(url) @@ -1615,12 +1734,15 @@ class YoutubeChannelIE(InfoExtractor): channel_page = self._download_webpage( url + '?view=57', channel_id, 'Downloading channel page', fatal=False) - channel_playlist_id = self._html_search_meta( - 'channelId', channel_page, 'channel id', default=None) - if not channel_playlist_id: - channel_playlist_id = self._search_regex( - r'data-channel-external-id="([^"]+)"', - channel_page, 'channel id', default=None) + if channel_page is False: + channel_playlist_id = False + else: + channel_playlist_id = self._html_search_meta( + 'channelId', channel_page, 'channel id', default=None) + if not channel_playlist_id: + channel_playlist_id = self._search_regex( + r'data-(?:channel-external-|yt)id="([^"]+)"', + channel_page, 'channel id', default=None) if channel_playlist_id and channel_playlist_id.startswith('UC'): playlist_id = 'UU' + channel_playlist_id[2:] return self.url_result( @@ -1643,29 +1765,7 @@ class YoutubeChannelIE(InfoExtractor): for video_id, video_title in self.extract_videos_from_page(channel_page)] return self.playlist_result(entries, channel_id) - def _entries(): - more_widget_html = content_html = channel_page - for pagenum in itertools.count(1): - - for video_id, video_title in self.extract_videos_from_page(content_html): - yield self.url_result( - video_id, 'Youtube', video_id=video_id, - video_title=video_title) - - mobj = re.search( - r'data-uix-load-more-href="/?(?P<more>[^"]+)"', - more_widget_html) - if not mobj: - break - - more = self._download_json( - 'https://youtube.com/%s' % mobj.group('more'), channel_id, - 'Downloading page #%s' % (pagenum + 1), - transform_source=uppercase_escape) - content_html = more['content_html'] - more_widget_html = more['load_more_widget_html'] - - return self.playlist_result(_entries(), channel_id) + return self.playlist_result(self._entries(channel_page, channel_id), channel_id) class YoutubeUserIE(YoutubeChannelIE): @@ -1696,6 +1796,29 @@ class YoutubeUserIE(YoutubeChannelIE): return super(YoutubeUserIE, cls).suitable(url) +class YoutubeUserPlaylistsIE(YoutubePlaylistsBaseInfoExtractor): + IE_DESC = 'YouTube.com user playlists' + _VALID_URL = r'https?://(?:\w+\.)?youtube\.com/user/(?P<id>[^/]+)/playlists' + IE_NAME = 'youtube:user:playlists' + + _TESTS = [{ + 'url': 'http://www.youtube.com/user/ThirstForScience/playlists', + 'playlist_mincount': 4, + 'info_dict': { + 'id': 'ThirstForScience', + 'title': 'Thirst for Science', + }, + }, { + # with "Load more" button + 'url': 'http://www.youtube.com/user/igorkle1/playlists?view=1&sort=dd', + 'playlist_mincount': 70, + 'info_dict': { + 'id': 'igorkle1', + 'title': 'Игорь Клейнер', + }, + }] + + class YoutubeSearchIE(SearchInfoExtractor, YoutubePlaylistIE): IE_DESC = 'YouTube.com searches' # there doesn't appear to be a real limit, for example if you search for @@ -1791,13 +1914,13 @@ class YoutubeSearchURLIE(InfoExtractor): } -class YoutubeShowIE(InfoExtractor): +class YoutubeShowIE(YoutubePlaylistsBaseInfoExtractor): IE_DESC = 'YouTube.com (multi-season) shows' _VALID_URL = r'https?://www\.youtube\.com/show/(?P<id>[^?#]*)' IE_NAME = 'youtube:show' _TESTS = [{ - 'url': 'http://www.youtube.com/show/airdisasters', - 'playlist_mincount': 3, + 'url': 'https://www.youtube.com/show/airdisasters', + 'playlist_mincount': 5, 'info_dict': { 'id': 'airdisasters', 'title': 'Air Disasters', @@ -1805,26 +1928,9 @@ class YoutubeShowIE(InfoExtractor): }] def _real_extract(self, url): - mobj = re.match(self._VALID_URL, url) - playlist_id = mobj.group('id') - webpage = self._download_webpage( - url, playlist_id, 'Downloading show webpage') - # There's one playlist for each season of the show - m_seasons = list(re.finditer(r'href="(/playlist\?list=.*?)"', webpage)) - self.to_screen('%s: Found %s seasons' % (playlist_id, len(m_seasons))) - entries = [ - self.url_result( - 'https://www.youtube.com' + season.group(1), 'YoutubePlaylist') - for season in m_seasons - ] - title = self._og_search_title(webpage, fatal=False) - - return { - '_type': 'playlist', - 'id': playlist_id, - 'title': title, - 'entries': entries, - } + playlist_id = self._match_id(url) + return super(YoutubeShowIE, self)._real_extract( + 'https://www.youtube.com/show/%s/playlists' % playlist_id) class YoutubeFeedsInfoExtractor(YoutubeBaseInfoExtractor): @@ -1931,6 +2037,7 @@ class YoutubeTruncatedURLIE(InfoExtractor): annotation_id=annotation_[^&]+| x-yt-cl=[0-9]+| hl=[^&]*| + t=[0-9]+ )? | attribution_link\?a=[^&]+ @@ -1953,6 +2060,9 @@ class YoutubeTruncatedURLIE(InfoExtractor): }, { 'url': 'https://www.youtube.com/watch?hl=en-GB', 'only_matching': True, + }, { + 'url': 'https://www.youtube.com/watch?t=2372', + 'only_matching': True, }] def _real_extract(self, url):