diff options
Diffstat (limited to 'youtube_dl/extractor/youtube.py')
| -rw-r--r-- | youtube_dl/extractor/youtube.py | 322 |
1 files changed, 262 insertions, 60 deletions
diff --git a/youtube_dl/extractor/youtube.py b/youtube_dl/extractor/youtube.py index 56957a661..ce97fd75b 100644 --- a/youtube_dl/extractor/youtube.py +++ b/youtube_dl/extractor/youtube.py @@ -27,11 +27,14 @@ from ..compat import ( ) from ..jsinterp import JSInterpreter from ..utils import ( + bug_reports_message, clean_html, dict_get, error_to_compat_str, ExtractorError, + filter_dict, float_or_none, + get_first, extract_attributes, get_element_by_attribute, int_or_none, @@ -63,6 +66,7 @@ from ..utils import ( url_or_none, urlencode_postdata, urljoin, + variadic, ) @@ -82,9 +86,66 @@ class YoutubeBaseInfoExtractor(InfoExtractor): _PLAYLIST_ID_RE = r'(?:(?:PL|LL|EC|UU|FL|RD|UL|TL|PU|OLAK5uy_)[0-9A-Za-z-_]{10,}|RDMM)' + _INNERTUBE_CLIENTS = { + 'ios': { + 'INNERTUBE_CONTEXT': { + 'client': { + 'clientName': 'IOS', + 'clientVersion': '20.10.4', + 'deviceMake': 'Apple', + 'deviceModel': 'iPhone16,2', + 'userAgent': 'com.google.ios.youtube/20.10.4 (iPhone16,2; U; CPU iOS 18_3_2 like Mac OS X;)', + 'osName': 'iPhone', + 'osVersion': '18.3.2.22D82', + }, + }, + 'INNERTUBE_CONTEXT_CLIENT_NAME': 5, + 'REQUIRE_JS_PLAYER': False, + 'REQUIRE_PO_TOKEN': True, + }, + # mweb has 'ultralow' formats + # See: https://github.com/yt-dlp/yt-dlp/pull/557 + 'mweb': { + 'INNERTUBE_CONTEXT': { + 'client': { + 'clientName': 'MWEB', + 'clientVersion': '2.20250311.03.00', + # mweb previously did not require PO Token with this UA + 'userAgent': 'Mozilla/5.0 (iPad; CPU OS 16_7_10 like Mac OS X) AppleWebKit/605.1.15 (KHTML, like Gecko) Version/16.6 Mobile/15E148 Safari/604.1,gzip(gfe)', + }, + }, + 'INNERTUBE_CONTEXT_CLIENT_NAME': 2, + 'REQUIRE_PO_TOKEN': True, + 'SUPPORTS_COOKIES': True, + }, + 'tv': { + 'INNERTUBE_CONTEXT': { + 'client': { + 'clientName': 'TVHTML5', + 'clientVersion': '7.20250312.16.00', + 'userAgent': 'Mozilla/5.0 (ChromiumStylePlatform) Cobalt/Version', + }, + }, + 'INNERTUBE_CONTEXT_CLIENT_NAME': 7, + 'SUPPORTS_COOKIES': True, + }, + 'web': { + 'INNERTUBE_CONTEXT': { + 'client': { + 'clientName': 'WEB', + 'clientVersion': '2.20250312.04.00', + }, + }, + 'INNERTUBE_CONTEXT_CLIENT_NAME': 1, + 'REQUIRE_PO_TOKEN': True, + 'SUPPORTS_COOKIES': True, + }, + } + def _login(self): """ Attempt to log in to YouTube. + True is returned if successful or skipped. False is returned if login failed. @@ -321,19 +382,24 @@ class YoutubeBaseInfoExtractor(InfoExtractor): '{0} {1} {2}'.format(time_now, self._SAPISID, origin).encode('utf-8')).hexdigest() return 'SAPISIDHASH {0}_{1}'.format(time_now, sapisidhash) - def _call_api(self, ep, query, video_id, fatal=True, headers=None): + def _call_api(self, ep, query, video_id, fatal=True, headers=None, + note='Downloading API JSON'): data = self._DEFAULT_API_DATA.copy() data.update(query) real_headers = {'content-type': 'application/json'} if headers: real_headers.update(headers) + # was: 'AIzaSyAO_FJ2SlqU8Q4STEHLGCilw_Y9_11qcW8' + api_key = self.get_param('youtube_innertube_key') return self._download_json( 'https://www.youtube.com/youtubei/v1/%s' % ep, video_id=video_id, - note='Downloading API JSON', errnote='Unable to download API page', + note=note, errnote='Unable to download API page', data=json.dumps(data).encode('utf8'), fatal=fatal, - headers=real_headers, - query={'key': 'AIzaSyAO_FJ2SlqU8Q4STEHLGCilw_Y9_11qcW8'}) + headers=real_headers, query=filter_dict({ + 'key': api_key, + 'prettyPrint': 'false', + })) def _extract_yt_initial_data(self, video_id, webpage): return self._parse_json( @@ -342,6 +408,22 @@ class YoutubeBaseInfoExtractor(InfoExtractor): self._YT_INITIAL_DATA_RE), webpage, 'yt initial data'), video_id) + def _extract_visitor_data(self, *args): + """ + Extract visitorData from an API response or ytcfg + + Appears to be used to track session state + """ + visitor_data = self.get_param('youtube_visitor_data') + if visitor_data: + return visitor_data + + return get_first( + args, (('VISITOR_DATA', + ('INNERTUBE_CONTEXT', 'client', 'visitorData'), + ('responseContext', 'visitorData')), + T(compat_str))) + def _extract_ytcfg(self, video_id, webpage): return self._parse_json( self._search_regex( @@ -381,6 +463,26 @@ class YoutubeBaseInfoExtractor(InfoExtractor): 'uploader': uploader, } + @staticmethod + def _extract_thumbnails(data, *path_list, **kw_final_key): + """ + Extract thumbnails from thumbnails dict + @param path_list: path list to level that contains 'thumbnails' key + """ + final_key = kw_final_key.get('final_key', 'thumbnails') + + return traverse_obj(data, (( + tuple(variadic(path) + (final_key, Ellipsis) + for path in path_list or [()])), { + 'url': ('url', T(url_or_none), + # Sometimes youtube gives a wrong thumbnail URL. See: + # https://github.com/yt-dlp/yt-dlp/issues/233 + # https://github.com/ytdl-org/youtube-dl/issues/28023 + T(lambda u: update_url(u, query=None) if u and 'maxresdefault' in u else u)), + 'height': ('height', T(int_or_none)), + 'width': ('width', T(int_or_none)), + }, T(lambda t: t if t.get('url') else None))) + def _search_results(self, query, params): data = { 'context': { @@ -590,9 +692,9 @@ class YoutubeIE(YoutubeBaseInfoExtractor): 'invidious': '|'.join(_INVIDIOUS_SITES), } _PLAYER_INFO_RE = ( - r'/s/player/(?P<id>[a-zA-Z0-9_-]{8,})/player', - r'/(?P<id>[a-zA-Z0-9_-]{8,})/player(?:_ias\.vflset(?:/[a-zA-Z]{2,3}_[a-zA-Z]{2,3})?|-plasma-ias-(?:phone|tablet)-[a-z]{2}_[A-Z]{2}\.vflset)/base\.js$', - r'\b(?P<id>vfl[a-zA-Z0-9_-]+)\b.*?\.js$', + r'/s/player/(?P<id>[a-zA-Z0-9_-]{8,})/(?:tv-)?player', + r'/(?P<id>[a-zA-Z0-9_-]{8,})/player(?:_ias(?:_tce)?\.vflset(?:/[a-zA-Z]{2,3}_[a-zA-Z]{2,3})?|-plasma-ias-(?:phone|tablet)-[a-z]{2}_[A-Z]{2}\.vflset)/base\.js$', + r'\b(?P<id>vfl[a-zA-Z0-9_-]{6,})\b.*?\.js$', ) _SUBTITLE_FORMATS = ('json3', 'srv1', 'srv2', 'srv3', 'ttml', 'vtt') @@ -1524,15 +1626,13 @@ class YoutubeIE(YoutubeBaseInfoExtractor): """ Return a string representation of a signature """ return '.'.join(compat_str(len(part)) for part in example_sig.split('.')) - @classmethod - def _extract_player_info(cls, player_url): - for player_re in cls._PLAYER_INFO_RE: - id_m = re.search(player_re, player_url) - if id_m: - break - else: - raise ExtractorError('Cannot identify player %r' % player_url) - return id_m.group('id') + def _extract_player_info(self, player_url): + try: + return self._search_regex( + self._PLAYER_INFO_RE, player_url, 'player info', group='id') + except ExtractorError as e: + raise ExtractorError( + 'Cannot identify player %r' % (player_url,), cause=e) def _load_player(self, video_id, player_url, fatal=True, player_id=None): if not player_id: @@ -1609,6 +1709,23 @@ class YoutubeIE(YoutubeBaseInfoExtractor): ' return %s\n') % (signature_id_tuple, expr_code) self.to_screen('Extracted signature function:\n' + code) + def _extract_sig_fn(self, jsi, funcname): + var_ay = self._search_regex( + r'''(?x) + (?:\*/|\{|\n|^)\s*(?:'[^']+'\s*;\s*) + (var\s*[\w$]+\s*=\s*(?: + ('|")(?:\\\2|(?!\2).)+\2\s*\.\s*split\(\s*('|")\W+\3\s*\)| + \[\s*(?:('|")(?:\\\4|(?!\4).)*\4\s*(?:(?=\])|,\s*))+\] + ))(?=\s*[,;]) + ''', jsi.code, 'useful values', default='') + + sig_fn = jsi.extract_function_code(funcname) + + if var_ay: + sig_fn = (sig_fn[0], ';\n'.join((var_ay, sig_fn[1]))) + + return sig_fn + def _parse_sig_js(self, jscode): # Examples where `sig` is funcname: # sig=function(a){a=a.split(""); ... ;return a.join("")}; @@ -1634,8 +1751,12 @@ class YoutubeIE(YoutubeBaseInfoExtractor): jscode, 'Initial JS player signature function name', group='sig') jsi = JSInterpreter(jscode) - initial_function = jsi.extract_function(funcname) - return lambda s: initial_function([s]) + + initial_function = self._extract_sig_fn(jsi, funcname) + + func = jsi.extract_function_from_code(*initial_function) + + return lambda s: func([s]) def _cached(self, func, *cache_id): def inner(*args, **kwargs): @@ -1750,12 +1871,16 @@ class YoutubeIE(YoutubeBaseInfoExtractor): if func_code: return jsi, player_id, func_code + return self._extract_n_function_code_jsi(video_id, jsi, player_id) + + def _extract_n_function_code_jsi(self, video_id, jsi, player_id=None): - func_name = self._extract_n_function_name(jscode) + func_name = self._extract_n_function_name(jsi.code) - func_code = jsi.extract_function_code(func_name) + func_code = self._extract_sig_fn(jsi, func_name) - self.cache.store('youtube-nsig', player_id, func_code) + if player_id: + self.cache.store('youtube-nsig', player_id, func_code) return jsi, player_id, func_code def _extract_n_function_from_code(self, jsi, func_code): @@ -1944,6 +2069,8 @@ class YoutubeIE(YoutubeBaseInfoExtractor): player_response = self._extract_yt_initial_variable( webpage, self._YT_INITIAL_PLAYER_RESPONSE_RE, video_id, 'initial player response') + is_live = traverse_obj(player_response, ('videoDetails', 'isLive')) + if False and not player_response: player_response = self._call_api( 'player', {'videoId': video_id}, video_id) @@ -1957,37 +2084,74 @@ class YoutubeIE(YoutubeBaseInfoExtractor): if sts: pb_context['signatureTimestamp'] = sts - query = { - 'playbackContext': { - 'contentPlaybackContext': pb_context, - 'contentCheckOk': True, - 'racyCheckOk': True, - }, - 'context': { - 'client': { - 'clientName': 'MWEB', - 'clientVersion': '2.20241202.07.00', - 'hl': 'en', - 'userAgent': 'Mozilla/5.0 (iPad; CPU OS 16_7_10 like Mac OS X) AppleWebKit/605.1.15 (KHTML, like Gecko) Version/16.6 Mobile/15E148 Safari/604.1,gzip(gfe)', - 'timeZone': 'UTC', - 'utcOffsetMinutes': 0, - }, - }, - 'videoId': video_id, - } + client_names = traverse_obj(self._INNERTUBE_CLIENTS, ( + T(dict.items), lambda _, k_v: not k_v[1].get('REQUIRE_PO_TOKEN'), + 0))[:1] + if 'web' not in client_names: + # webpage links won't download: ignore links and playability + player_response = filter_dict( + player_response or {}, + lambda k, _: k not in ('streamingData', 'playabilityStatus')) + + if is_live and 'ios' not in client_names: + client_names.append('ios') + headers = { - 'X-YouTube-Client-Name': '2', - 'X-YouTube-Client-Version': '2.20241202.07.00', - 'Origin': origin, 'Sec-Fetch-Mode': 'navigate', - 'User-Agent': query['context']['client']['userAgent'], + 'Origin': origin, + 'X-Goog-Visitor-Id': self._extract_visitor_data(ytcfg) or '', } auth = self._generate_sapisidhash_header(origin) if auth is not None: headers['Authorization'] = auth headers['X-Origin'] = origin - player_response = self._call_api('player', query, video_id, fatal=False, headers=headers) + for client in traverse_obj(self._INNERTUBE_CLIENTS, (client_names, T(dict))): + + query = { + 'playbackContext': { + 'contentPlaybackContext': pb_context, + }, + 'contentCheckOk': True, + 'racyCheckOk': True, + 'context': { + 'client': merge_dicts( + traverse_obj(client, ('INNERTUBE_CONTEXT', 'client')), { + 'hl': 'en', + 'timeZone': 'UTC', + 'utcOffsetMinutes': 0, + }), + }, + 'videoId': video_id, + } + + api_headers = merge_dicts(headers, traverse_obj(client, { + 'X-YouTube-Client-Name': 'INNERTUBE_CONTEXT_CLIENT_NAME', + 'X-YouTube-Client-Version': ( + 'INNERTUBE_CONTEXT', 'client', 'clientVersion'), + 'User-Agent': ( + 'INNERTUBE_CONTEXT', 'client', 'userAgent'), + })) + + api_player_response = self._call_api( + 'player', query, video_id, fatal=False, headers=api_headers, + note=join_nonempty( + 'Downloading', traverse_obj(query, ( + 'context', 'client', 'clientName')), + 'API JSON', delim=' ')) + + hls = traverse_obj( + (player_response, api_player_response), + (Ellipsis, 'streamingData', 'hlsManifestUrl', T(url_or_none))) + if len(hls) == 2 and not hls[0] and hls[1]: + player_response['streamingData']['hlsManifestUrl'] = hls[1] + else: + video_details = merge_dicts(*traverse_obj( + (player_response, api_player_response), + (Ellipsis, 'videoDetails', T(dict)))) + player_response.update(filter_dict( + api_player_response or {}, cndn=lambda k, _: k != 'captions')) + player_response['videoDetails'] = video_details def is_agegated(playability): if not isinstance(playability, dict): @@ -2130,6 +2294,7 @@ class YoutubeIE(YoutubeBaseInfoExtractor): itag_qualities = {} q = qualities(['tiny', 'small', 'medium', 'large', 'hd720', 'hd1080', 'hd1440', 'hd2160', 'hd2880', 'highres']) CHUNK_SIZE = 10 << 20 + is_live = video_details.get('isLive') streaming_data = player_response.get('streamingData') or {} streaming_formats = streaming_data.get('formats') or [] @@ -2274,7 +2439,8 @@ class YoutubeIE(YoutubeBaseInfoExtractor): hls_manifest_url = streaming_data.get('hlsManifestUrl') if hls_manifest_url: for f in self._extract_m3u8_formats( - hls_manifest_url, video_id, 'mp4', fatal=False): + hls_manifest_url, video_id, 'mp4', + entry_protocol='m3u8_native', live=is_live, fatal=False): if process_manifest_format( f, 'hls', None, self._search_regex( r'/itag/(\d+)', f['url'], 'itag', default=None)): @@ -2380,8 +2546,6 @@ class YoutubeIE(YoutubeBaseInfoExtractor): # Strictly de-prioritize damaged formats f['preference'] = -10 - is_live = video_details.get('isLive') - owner_profile_url = self._yt_urljoin(self._extract_author_var( webpage, 'url', videodetails=video_details, metadata=microformat)) @@ -2416,8 +2580,8 @@ class YoutubeIE(YoutubeBaseInfoExtractor): } pctr = traverse_obj( - player_response, - ('captions', 'playerCaptionsTracklistRenderer', T(dict))) + (player_response, api_player_response), + (Ellipsis, 'captions', 'playerCaptionsTracklistRenderer', T(dict))) if pctr: def process_language(container, base_url, lang_code, query): lang_subs = [] @@ -2434,19 +2598,21 @@ class YoutubeIE(YoutubeBaseInfoExtractor): def process_subtitles(): subtitles = {} for caption_track in traverse_obj(pctr, ( - 'captionTracks', lambda _, v: v.get('baseUrl'))): + Ellipsis, 'captionTracks', lambda _, v: ( + v.get('baseUrl') and v.get('languageCode')))): + base_url = self._yt_urljoin(caption_track['baseUrl']) if not base_url: continue + lang_code = caption_track['languageCode'] if caption_track.get('kind') != 'asr': - lang_code = caption_track.get('languageCode') - if not lang_code: - continue process_language( subtitles, base_url, lang_code, {}) continue automatic_captions = {} + process_language( + automatic_captions, base_url, lang_code, {}) for translation_language in traverse_obj(pctr, ( - 'translationLanguages', lambda _, v: v.get('languageCode'))): + Ellipsis, 'translationLanguages', lambda _, v: v.get('languageCode'))): translation_language_code = translation_language['languageCode'] process_language( automatic_captions, base_url, translation_language_code, @@ -3065,8 +3231,12 @@ class YoutubeTabIE(YoutubeBaseInfoExtractor): expected_type=txt_or_none) def _grid_entries(self, grid_renderer): - for item in grid_renderer['items']: - if not isinstance(item, dict): + for item in traverse_obj(grid_renderer, ('items', Ellipsis, T(dict))): + lockup_view_model = traverse_obj(item, ('lockupViewModel', T(dict))) + if lockup_view_model: + entry = self._extract_lockup_view_model(lockup_view_model) + if entry: + yield entry continue renderer = self._extract_grid_item_renderer(item) if not isinstance(renderer, dict): @@ -3150,6 +3320,25 @@ class YoutubeTabIE(YoutubeBaseInfoExtractor): continue yield self._extract_video(renderer) + def _extract_lockup_view_model(self, view_model): + content_id = view_model.get('contentId') + if not content_id: + return + content_type = view_model.get('contentType') + if content_type not in ('LOCKUP_CONTENT_TYPE_PLAYLIST', 'LOCKUP_CONTENT_TYPE_PODCAST'): + self.report_warning( + 'Unsupported lockup view model content type "{0}"{1}'.format(content_type, bug_reports_message()), only_once=True) + return + return merge_dicts(self.url_result( + update_url_query('https://www.youtube.com/playlist', {'list': content_id}), + ie=YoutubeTabIE.ie_key(), video_id=content_id), { + 'title': traverse_obj(view_model, ( + 'metadata', 'lockupMetadataViewModel', 'title', 'content', T(compat_str))), + 'thumbnails': self._extract_thumbnails(view_model, ( + 'contentImage', 'collectionThumbnailViewModel', 'primaryThumbnail', + 'thumbnailViewModel', 'image'), final_key='sources'), + }) + def _video_entry(self, video_renderer): video_id = video_renderer.get('videoId') if video_id: @@ -3354,7 +3543,7 @@ class YoutubeTabIE(YoutubeBaseInfoExtractor): if not continuation: break if visitor_data: - headers['x-goog-visitor-id'] = visitor_data + headers['X-Goog-Visitor-Id'] = visitor_data data['continuation'] = continuation['continuation'] data['clickTracking'] = { 'clickTrackingParams': continuation['itct'], @@ -3536,10 +3725,23 @@ class YoutubeTabIE(YoutubeBaseInfoExtractor): def _real_extract(self, url): item_id = self._match_id(url) url = update_url(url, netloc='www.youtube.com') - # Handle both video/playlist URLs qs = parse_qs(url) - video_id = qs.get('v', [None])[0] - playlist_id = qs.get('list', [None])[0] + + def qs_get(key, default=None): + return qs.get(key, [default])[-1] + + # Go around for /feeds/videos.xml?playlist_id={pl_id} + if item_id == 'feeds' and '/feeds/videos.xml?' in url: + playlist_id = qs_get('playlist_id') + if playlist_id: + return self.url_result( + update_url_query('https://www.youtube.com/playlist', { + 'list': playlist_id, + }), ie=self.ie_key(), video_id=playlist_id) + + # Handle both video/playlist URLs + video_id = qs_get('v') + playlist_id = qs_get('list') if video_id and playlist_id: if self._downloader.params.get('noplaylist'): self.to_screen('Downloading just video %s because of --no-playlist' % video_id) |