X-Git-Url: https://jfr.im/git/yt-dlp.git/blobdiff_plain/8ceb07e870424c219dced8f4348729553f05c5cc..61edf57f8f13f6dfd81154174e647eb5fdd26089:/yt_dlp/extractor/tiktok.py diff --git a/yt_dlp/extractor/tiktok.py b/yt_dlp/extractor/tiktok.py index fb838d529..c3505b14f 100644 --- a/yt_dlp/extractor/tiktok.py +++ b/yt_dlp/extractor/tiktok.py @@ -1,25 +1,27 @@ +import functools import itertools import json import random import re import string import time +import urllib.parse +import uuid from .common import InfoExtractor -from ..compat import compat_urllib_parse_unquote, compat_urllib_parse_urlparse +from ..networking import HEADRequest from ..utils import ( ExtractorError, - HEADRequest, - LazyList, UnsupportedError, UserNotLive, determine_ext, + filter_dict, format_field, - get_element_by_id, - get_first, int_or_none, join_nonempty, merge_dicts, + mimetype2ext, + parse_qs, qualities, remove_start, srt_subtitles_timecode, @@ -28,125 +30,229 @@ try_call, try_get, url_or_none, + urlencode_postdata, ) class TikTokBaseIE(InfoExtractor): - _APP_VERSIONS = [('26.1.3', '260103'), ('26.1.2', '260102'), ('26.1.1', '260101'), ('25.6.2', '250602')] - _WORKING_APP_VERSION = None - _APP_NAME = 'trill' - _AID = 1180 _UPLOADER_URL_FORMAT = 'https://www.tiktok.com/@%s' _WEBPAGE_HOST = 'https://www.tiktok.com/' QUALITIES = ('360p', '540p', '720p', '1080p') - @property + _APP_INFO_DEFAULTS = { + # unique "install id" + 'iid': None, + # TikTok (KR/PH/TW/TH/VN) = trill, TikTok (rest of world) = musical_ly, Douyin = aweme + 'app_name': 'musical_ly', + 'app_version': '35.1.3', + 'manifest_app_version': '2023501030', + # "app id": aweme = 1128, trill = 1180, musical_ly = 1233, universal = 0 + 'aid': '0', + } + _APP_INFO_POOL = None + _APP_INFO = None + _APP_USER_AGENT = None + + @functools.cached_property + def _KNOWN_APP_INFO(self): + # If we have a genuine device ID, we may not need any IID + default = [''] if self._KNOWN_DEVICE_ID else [] + return self._configuration_arg('app_info', default, ie_key=TikTokIE) + + @functools.cached_property + def _KNOWN_DEVICE_ID(self): + return self._configuration_arg('device_id', [None], ie_key=TikTokIE)[0] + + @functools.cached_property + def _DEVICE_ID(self): + return self._KNOWN_DEVICE_ID or str(random.randint(7250000000000000000, 7351147085025500000)) + + @functools.cached_property def _API_HOSTNAME(self): return self._configuration_arg( 'api_hostname', ['api16-normal-c-useast1a.tiktokv.com'], ie_key=TikTokIE)[0] + def _get_next_app_info(self): + if self._APP_INFO_POOL is None: + defaults = { + key: self._configuration_arg(key, [default], ie_key=TikTokIE)[0] + for key, default in self._APP_INFO_DEFAULTS.items() + if key != 'iid' + } + self._APP_INFO_POOL = [ + {**defaults, **dict( + (k, v) for k, v in zip(self._APP_INFO_DEFAULTS, app_info.split('/')) if v + )} for app_info in self._KNOWN_APP_INFO + ] + + if not self._APP_INFO_POOL: + return False + + self._APP_INFO = self._APP_INFO_POOL.pop(0) + + app_name = self._APP_INFO['app_name'] + version = self._APP_INFO['manifest_app_version'] + if app_name == 'musical_ly': + package = f'com.zhiliaoapp.musically/{version}' + else: # trill, aweme + package = f'com.ss.android.ugc.{app_name}/{version}' + self._APP_USER_AGENT = f'{package} (Linux; U; Android 13; en_US; Pixel 7; Build/TD1A.220804.031; Cronet/58.0.2991.0)' + + return True + @staticmethod def _create_url(user_id, video_id): return f'https://www.tiktok.com/@{user_id or "_"}/video/{video_id}' def _get_sigi_state(self, webpage, display_id): - return self._parse_json(get_element_by_id( - 'SIGI_STATE|sigi-persisted-data', webpage, escape_value=False), display_id) + return self._search_json( + r']+\bid="(?:SIGI_STATE|sigi-persisted-data)"[^>]*>', webpage, + 'sigi state', display_id, end_pattern=r'', default={}) - def _call_api_impl(self, ep, query, manifest_app_version, video_id, fatal=True, + def _get_universal_data(self, webpage, display_id): + return traverse_obj(self._search_json( + r']+\bid="__UNIVERSAL_DATA_FOR_REHYDRATION__"[^>]*>', webpage, + 'universal data', display_id, end_pattern=r'', default={}), + ('__DEFAULT_SCOPE__', {dict})) or {} + + def _call_api_impl(self, ep, video_id, query=None, data=None, headers=None, fatal=True, note='Downloading API JSON', errnote='Unable to download API page'): self._set_cookie(self._API_HOSTNAME, 'odin_tt', ''.join(random.choices('0123456789abcdef', k=160))) webpage_cookies = self._get_cookies(self._WEBPAGE_HOST) if webpage_cookies.get('sid_tt'): self._set_cookie(self._API_HOSTNAME, 'sid_tt', webpage_cookies['sid_tt'].value) return self._download_json( - 'https://%s/aweme/v1/%s/' % (self._API_HOSTNAME, ep), video_id=video_id, + f'https://{self._API_HOSTNAME}/aweme/v1/{ep}/', video_id=video_id, fatal=fatal, note=note, errnote=errnote, headers={ - 'User-Agent': f'com.ss.android.ugc.{self._APP_NAME}/{manifest_app_version} (Linux; U; Android 10; en_US; Pixel 4; Build/QQ3A.200805.001; Cronet/58.0.2991.0)', + 'User-Agent': self._APP_USER_AGENT, 'Accept': 'application/json', - }, query=query) + **(headers or {}), + }, query=query, data=data) - def _build_api_query(self, query, app_version, manifest_app_version): - return { + def _build_api_query(self, query): + return filter_dict({ **query, - 'version_name': app_version, - 'version_code': manifest_app_version, - 'build_number': app_version, - 'manifest_version_code': manifest_app_version, - 'update_version_code': manifest_app_version, - 'openudid': ''.join(random.choices('0123456789abcdef', k=16)), - 'uuid': ''.join(random.choices(string.digits, k=16)), - '_rticket': int(time.time() * 1000), - 'ts': int(time.time()), - 'device_brand': 'Google', - 'device_type': 'Pixel 4', 'device_platform': 'android', - 'resolution': '1080*1920', + 'os': 'android', + 'ssmix': 'a', + '_rticket': int(time.time() * 1000), + 'cdid': str(uuid.uuid4()), + 'channel': 'googleplay', + 'aid': self._APP_INFO['aid'], + 'app_name': self._APP_INFO['app_name'], + 'version_code': ''.join(f'{int(v):02d}' for v in self._APP_INFO['app_version'].split('.')), + 'version_name': self._APP_INFO['app_version'], + 'manifest_version_code': self._APP_INFO['manifest_app_version'], + 'update_version_code': self._APP_INFO['manifest_app_version'], + 'ab_version': self._APP_INFO['app_version'], + 'resolution': '1080*2400', 'dpi': 420, - 'os_version': '10', + 'device_type': 'Pixel 7', + 'device_brand': 'Google', + 'language': 'en', 'os_api': '29', - 'carrier_region': 'US', + 'os_version': '13', + 'ac': 'wifi', + 'is_pad': '0', + 'current_region': 'US', + 'app_type': 'normal', 'sys_region': 'US', - 'region': 'US', - 'app_name': self._APP_NAME, - 'app_language': 'en', - 'language': 'en', + 'last_install_time': int(time.time()) - random.randint(86400, 1123200), 'timezone_name': 'America/New_York', + 'residence': 'US', + 'app_language': 'en', 'timezone_offset': '-14400', - 'channel': 'googleplay', - 'ac': 'wifi', - 'mcc_mnc': '310260', - 'is_my_cn': 0, - 'aid': self._AID, - 'ssmix': 'a', - 'as': 'a1qwert123', - 'cp': 'cbfhckdckkde1', - } + 'host_abi': 'armeabi-v7a', + 'locale': 'en', + 'ac2': 'wifi5g', + 'uoo': '1', + 'carrier_region': 'US', + 'op_region': 'US', + 'build_number': self._APP_INFO['app_version'], + 'region': 'US', + 'ts': int(time.time()), + 'iid': self._APP_INFO.get('iid'), + 'device_id': self._DEVICE_ID, + 'openudid': ''.join(random.choices('0123456789abcdef', k=16)), + }) - def _call_api(self, ep, query, video_id, fatal=True, + def _call_api(self, ep, video_id, query=None, data=None, headers=None, fatal=True, note='Downloading API JSON', errnote='Unable to download API page'): - if not self._WORKING_APP_VERSION: - app_version = self._configuration_arg('app_version', [''], ie_key=TikTokIE.ie_key())[0] - manifest_app_version = self._configuration_arg('manifest_app_version', [''], ie_key=TikTokIE.ie_key())[0] - if app_version and manifest_app_version: - self._WORKING_APP_VERSION = (app_version, manifest_app_version) - self.write_debug('Imported app version combo from extractor arguments') - elif app_version or manifest_app_version: - self.report_warning('Only one of the two required version params are passed as extractor arguments', only_once=True) - - if self._WORKING_APP_VERSION: - app_version, manifest_app_version = self._WORKING_APP_VERSION - real_query = self._build_api_query(query, app_version, manifest_app_version) - return self._call_api_impl(ep, real_query, manifest_app_version, video_id, fatal, note, errnote) - - for count, (app_version, manifest_app_version) in enumerate(self._APP_VERSIONS, start=1): - real_query = self._build_api_query(query, app_version, manifest_app_version) + if not self._APP_INFO and not self._get_next_app_info(): + message = 'No working app info is available' + if fatal: + raise ExtractorError(message, expected=True) + else: + self.report_warning(message) + return + + max_tries = len(self._APP_INFO_POOL) + 1 # _APP_INFO_POOL + _APP_INFO + for count in itertools.count(1): + self.write_debug(str(self._APP_INFO)) + real_query = self._build_api_query(query or {}) try: - res = self._call_api_impl(ep, real_query, manifest_app_version, video_id, fatal, note, errnote) - self._WORKING_APP_VERSION = (app_version, manifest_app_version) - return res + return self._call_api_impl( + ep, video_id, query=real_query, data=data, headers=headers, + fatal=fatal, note=note, errnote=errnote) except ExtractorError as e: if isinstance(e.cause, json.JSONDecodeError) and e.cause.pos == 0: - if count == len(self._APP_VERSIONS): + message = str(e.cause or e.msg) + if not self._get_next_app_info(): if fatal: - raise e + raise else: - self.report_warning(str(e.cause or e.msg)) + self.report_warning(message) return - self.report_warning('%s. Retrying... (attempt %s of %s)' % (str(e.cause or e.msg), count, len(self._APP_VERSIONS))) + self.report_warning(f'{message}. Retrying... (attempt {count} of {max_tries})') continue - raise e + raise def _extract_aweme_app(self, aweme_id): - feed_list = self._call_api( - 'feed', {'aweme_id': aweme_id}, aweme_id, note='Downloading video feed', - errnote='Unable to download video feed').get('aweme_list') or [] - aweme_detail = next((aweme for aweme in feed_list if str(aweme.get('aweme_id')) == aweme_id), None) + aweme_detail = traverse_obj( + self._call_api('multi/aweme/detail', aweme_id, data=urlencode_postdata({ + 'aweme_ids': f'[{aweme_id}]', + 'request_source': '0', + }), headers={'X-Argus': ''}), ('aweme_details', 0, {dict})) if not aweme_detail: - raise ExtractorError('Unable to find video in feed', video_id=aweme_id) + raise ExtractorError('Unable to extract aweme detail info', video_id=aweme_id) return self._parse_aweme_video_app(aweme_detail) - def _get_subtitles(self, aweme_detail, aweme_id): + def _extract_web_data_and_status(self, url, video_id, fatal=True): + video_data, status = {}, -1 + + res = self._download_webpage_handle(url, video_id, fatal=fatal, headers={'User-Agent': 'Mozilla/5.0'}) + if res is False: + return video_data, status + + webpage, urlh = res + if urllib.parse.urlparse(urlh.url).path == '/login': + message = 'TikTok is requiring login for access to this content' + if fatal: + self.raise_login_required(message) + self.report_warning(f'{message}. {self._login_hint()}') + return video_data, status + + if universal_data := self._get_universal_data(webpage, video_id): + self.write_debug('Found universal data for rehydration') + status = traverse_obj(universal_data, ('webapp.video-detail', 'statusCode', {int})) or 0 + video_data = traverse_obj(universal_data, ('webapp.video-detail', 'itemInfo', 'itemStruct', {dict})) + + elif sigi_data := self._get_sigi_state(webpage, video_id): + self.write_debug('Found sigi state data') + status = traverse_obj(sigi_data, ('VideoPage', 'statusCode', {int})) or 0 + video_data = traverse_obj(sigi_data, ('ItemModule', video_id, {dict})) + + elif next_data := self._search_nextjs_data(webpage, video_id, default={}): + self.write_debug('Found next.js data') + status = traverse_obj(next_data, ('props', 'pageProps', 'statusCode', {int})) or 0 + video_data = traverse_obj(next_data, ('props', 'pageProps', 'itemInfo', 'itemStruct', {dict})) + + elif fatal: + raise ExtractorError('Unable to extract webpage video data') + + return video_data, status + + def _get_subtitles(self, aweme_detail, aweme_id, user_name): # TODO: Extract text positioning info subtitles = {} # aweme/detail endpoint subs @@ -164,7 +270,7 @@ def _get_subtitles(self, aweme_detail, aweme_id): 'ext': 'srt', 'data': '\n\n'.join( f'{i + 1}\n{srt_subtitles_timecode(line["start_time"] / 1000)} --> {srt_subtitles_timecode(line["end_time"] / 1000)}\n{line["text"]}' - for i, line in enumerate(caption_json['utterances']) if line.get('text')) + for i, line in enumerate(caption_json['utterances']) if line.get('text')), }) # feed endpoint subs if not subtitles: @@ -177,49 +283,51 @@ def _get_subtitles(self, aweme_detail, aweme_id): }) # webpage subs if not subtitles: - for caption in traverse_obj(aweme_detail, ('video', 'subtitleInfos', ...), expected_type=dict): - if not caption.get('Url'): - continue + if user_name: # only _parse_aweme_video_app needs to extract the webpage here + aweme_detail, _ = self._extract_web_data_and_status( + self._create_url(user_name, aweme_id), aweme_id, fatal=False) + for caption in traverse_obj(aweme_detail, ('video', 'subtitleInfos', lambda _, v: v['Url'])): subtitles.setdefault(caption.get('LanguageCodeName') or 'en', []).append({ 'ext': remove_start(caption.get('Format'), 'web'), 'url': caption['Url'], }) return subtitles + def _parse_url_key(self, url_key): + format_id, codec, res, bitrate = self._search_regex( + r'v[^_]+_(?P(?P[^_]+)_(?P\d+p)_(?P\d+))', url_key, + 'url key', default=(None, None, None, None), group=('id', 'codec', 'res', 'bitrate')) + if not format_id: + return {}, None + return { + 'format_id': format_id, + 'vcodec': 'h265' if codec == 'bytevc1' else codec, + 'tbr': int_or_none(bitrate, scale=1000) or None, + 'quality': qualities(self.QUALITIES)(res), + }, res + def _parse_aweme_video_app(self, aweme_detail): aweme_id = aweme_detail['aweme_id'] video_info = aweme_detail['video'] - - def parse_url_key(url_key): - format_id, codec, res, bitrate = self._search_regex( - r'v[^_]+_(?P(?P[^_]+)_(?P\d+p)_(?P\d+))', url_key, - 'url key', default=(None, None, None, None), group=('id', 'codec', 'res', 'bitrate')) - if not format_id: - return {}, None - return { - 'format_id': format_id, - 'vcodec': 'h265' if codec == 'bytevc1' else codec, - 'tbr': int_or_none(bitrate, scale=1000) or None, - 'quality': qualities(self.QUALITIES)(res), - }, res - known_resolutions = {} - def mp3_meta(url): + def audio_meta(url): + ext = determine_ext(url, default_ext='m4a') return { 'format_note': 'Music track', - 'ext': 'mp3', - 'acodec': 'mp3', + 'ext': ext, + 'acodec': 'aac' if ext == 'm4a' else ext, 'vcodec': 'none', 'width': None, 'height': None, - } if determine_ext(url) == 'mp3' else {} + } if ext == 'mp3' or '-music-' in url else {} def extract_addr(addr, add_meta={}): - parsed_meta, res = parse_url_key(addr.get('url_key', '')) + parsed_meta, res = self._parse_url_key(addr.get('url_key', '')) + is_bytevc2 = parsed_meta.get('vcodec') == 'bytevc2' if res: - known_resolutions.setdefault(res, {}).setdefault('height', add_meta.get('height')) - known_resolutions[res].setdefault('width', add_meta.get('width')) + known_resolutions.setdefault(res, {}).setdefault('height', int_or_none(addr.get('height'))) + known_resolutions[res].setdefault('width', int_or_none(addr.get('width'))) parsed_meta.update(known_resolutions.get(res, {})) add_meta.setdefault('height', int_or_none(res[:-1])) return [{ @@ -229,29 +337,37 @@ def extract_addr(addr, add_meta={}): 'acodec': 'aac', 'source_preference': -2 if 'aweme/v1' in url else -1, # Downloads from API might get blocked **add_meta, **parsed_meta, + # bytevc2 is bytedance's own custom h266/vvc codec, as-of-yet unplayable + 'preference': -100 if is_bytevc2 else -1, 'format_note': join_nonempty( - add_meta.get('format_note'), '(API)' if 'aweme/v1' in url else None, delim=' '), - **mp3_meta(url), + add_meta.get('format_note'), '(API)' if 'aweme/v1' in url else None, + '(UNPLAYABLE)' if is_bytevc2 else None, delim=' '), + **audio_meta(url), } for url in addr.get('url_list') or []] # Hack: Add direct video links first to prioritize them when removing duplicate formats formats = [] + width = int_or_none(video_info.get('width')) + height = int_or_none(video_info.get('height')) + ratio = try_call(lambda: width / height) or 0.5625 if video_info.get('play_addr'): formats.extend(extract_addr(video_info['play_addr'], { 'format_id': 'play_addr', 'format_note': 'Direct video', 'vcodec': 'h265' if traverse_obj( video_info, 'is_bytevc1', 'is_h265') else 'h264', # TODO: Check for "direct iOS" videos, like https://www.tiktok.com/@cookierun_dev/video/7039716639834656002 - 'width': video_info.get('width'), - 'height': video_info.get('height'), + 'width': width, + 'height': height, })) if video_info.get('download_addr'): - formats.extend(extract_addr(video_info['download_addr'], { + download_addr = video_info['download_addr'] + dl_width = int_or_none(download_addr.get('width')) + formats.extend(extract_addr(download_addr, { 'format_id': 'download_addr', 'format_note': 'Download video%s' % (', watermarked' if video_info.get('has_watermark') else ''), 'vcodec': 'h264', - 'width': video_info.get('width'), - 'height': video_info.get('height'), + 'width': dl_width, + 'height': try_call(lambda: int(dl_width / ratio)), # download_addr['height'] is wrong 'preference': -2 if video_info.get('has_watermark') else -1, })) if video_info.get('play_addr_h264'): @@ -282,25 +398,19 @@ def extract_addr(addr, add_meta={}): auth_cookie = self._get_cookies(self._WEBPAGE_HOST).get('sid_tt') if auth_cookie: for f in formats: - self._set_cookie(compat_urllib_parse_urlparse(f['url']).hostname, 'sid_tt', auth_cookie.value) + self._set_cookie(urllib.parse.urlparse(f['url']).hostname, 'sid_tt', auth_cookie.value) thumbnails = [] for cover_id in ('cover', 'ai_dynamic_cover', 'animated_cover', 'ai_dynamic_cover_bak', 'origin_cover', 'dynamic_cover'): - cover = video_info.get(cover_id) - if cover: - for cover_url in cover['url_list']: - thumbnails.append({ - 'id': cover_id, - 'url': cover_url, - }) + for cover_url in traverse_obj(video_info, (cover_id, 'url_list', ...)): + thumbnails.append({ + 'id': cover_id, + 'url': cover_url, + }) - stats_info = aweme_detail.get('statistics', {}) - author_info = aweme_detail.get('author', {}) - music_info = aweme_detail.get('music', {}) - user_url = self._UPLOADER_URL_FORMAT % (traverse_obj(author_info, - 'sec_uid', 'id', 'uid', 'unique_id', - expected_type=str_or_none, get_all=False)) + stats_info = aweme_detail.get('statistics') or {} + music_info = aweme_detail.get('music') or {} labels = traverse_obj(aweme_detail, ('hybrid_label', ..., 'text'), expected_type=str) contained_music_track = traverse_obj( @@ -308,35 +418,46 @@ def extract_addr(addr, add_meta={}): contained_music_author = traverse_obj( music_info, ('matched_song', 'author'), ('matched_pgc_sound', 'author'), 'author', expected_type=str) - is_generic_og_trackname = music_info.get('is_original_sound') and music_info.get('title') == 'original sound - %s' % music_info.get('owner_handle') + is_generic_og_trackname = music_info.get('is_original_sound') and music_info.get('title') == 'original sound - {}'.format(music_info.get('owner_handle')) if is_generic_og_trackname: music_track, music_author = contained_music_track or 'original sound', contained_music_author else: - music_track, music_author = music_info.get('title'), music_info.get('author') + music_track, music_author = music_info.get('title'), traverse_obj(music_info, ('author', {str})) + + author_info = traverse_obj(aweme_detail, ('author', { + 'uploader': ('unique_id', {str}), + 'uploader_id': ('uid', {str_or_none}), + 'channel': ('nickname', {str}), + 'channel_id': ('sec_uid', {str}), + })) return { 'id': aweme_id, - 'extractor_key': TikTokIE.ie_key(), - 'extractor': TikTokIE.IE_NAME, - 'webpage_url': self._create_url(author_info.get('uid'), aweme_id), - 'title': aweme_detail.get('desc'), - 'description': aweme_detail.get('desc'), - 'view_count': int_or_none(stats_info.get('play_count')), - 'like_count': int_or_none(stats_info.get('digg_count')), - 'repost_count': int_or_none(stats_info.get('share_count')), - 'comment_count': int_or_none(stats_info.get('comment_count')), - 'uploader': str_or_none(author_info.get('unique_id')), - 'creator': str_or_none(author_info.get('nickname')), - 'uploader_id': str_or_none(author_info.get('uid')), - 'uploader_url': user_url, + **traverse_obj(aweme_detail, { + 'title': ('desc', {str}), + 'description': ('desc', {str}), + 'timestamp': ('create_time', {int_or_none}), + }), + **traverse_obj(stats_info, { + 'view_count': 'play_count', + 'like_count': 'digg_count', + 'repost_count': 'share_count', + 'comment_count': 'comment_count', + }, expected_type=int_or_none), + **author_info, + 'channel_url': format_field(author_info, 'channel_id', self._UPLOADER_URL_FORMAT, default=None), + 'uploader_url': format_field( + author_info, ['uploader', 'uploader_id'], self._UPLOADER_URL_FORMAT, default=None), 'track': music_track, 'album': str_or_none(music_info.get('album')) or None, - 'artist': music_author or None, - 'timestamp': int_or_none(aweme_detail.get('create_time')), + 'artists': re.split(r'(?:, | & )', music_author) if music_author else None, 'formats': formats, - 'subtitles': self.extract_subtitles(aweme_detail, aweme_id), + 'subtitles': self.extract_subtitles( + aweme_detail, aweme_id, traverse_obj(author_info, 'uploader', 'uploader_id', 'channel_id')), 'thumbnails': thumbnails, - 'duration': int_or_none(traverse_obj(video_info, 'duration', ('download_addr', 'duration')), scale=1000), + 'duration': (traverse_obj(video_info, ( + (None, 'download_addr'), 'duration', {functools.partial(int_or_none, scale=1000)}, any)) + or traverse_obj(music_info, ('duration', {int_or_none}))), 'availability': self._availability( is_private='Private' in labels, needs_subscription='Friends only' in labels, @@ -344,77 +465,136 @@ def extract_addr(addr, add_meta={}): '_format_sort_fields': ('quality', 'codec', 'size', 'br'), } - def _parse_aweme_video_web(self, aweme_detail, webpage_url): - video_info = aweme_detail['video'] - author_info = traverse_obj(aweme_detail, 'authorInfo', 'author', expected_type=dict, default={}) - music_info = aweme_detail.get('music') or {} - stats_info = aweme_detail.get('stats') or {} - user_url = self._UPLOADER_URL_FORMAT % (traverse_obj(author_info, - 'secUid', 'id', 'uid', 'uniqueId', - expected_type=str_or_none, get_all=False) - or aweme_detail.get('authorSecId')) - + def _extract_web_formats(self, aweme_detail): + COMMON_FORMAT_INFO = { + 'ext': 'mp4', + 'vcodec': 'h264', + 'acodec': 'aac', + } + video_info = traverse_obj(aweme_detail, ('video', {dict})) or {} + play_width = int_or_none(video_info.get('width')) + play_height = int_or_none(video_info.get('height')) + ratio = try_call(lambda: play_width / play_height) or 0.5625 formats = [] - play_url = video_info.get('playAddr') - width = video_info.get('width') - height = video_info.get('height') - if isinstance(play_url, str): - formats = [{ + + for bitrate_info in traverse_obj(video_info, ('bitrateInfo', lambda _, v: v['PlayAddr']['UrlList'])): + format_info, res = self._parse_url_key( + traverse_obj(bitrate_info, ('PlayAddr', 'UrlKey', {str})) or '') + # bytevc2 is bytedance's own custom h266/vvc codec, as-of-yet unplayable + is_bytevc2 = format_info.get('vcodec') == 'bytevc2' + format_info.update({ + 'format_note': 'UNPLAYABLE' if is_bytevc2 else None, + 'preference': -100 if is_bytevc2 else -1, + 'filesize': traverse_obj(bitrate_info, ('PlayAddr', 'DataSize', {int_or_none})), + }) + + if dimension := (res and int(res[:-1])): + if dimension == 540: # '540p' is actually 576p + dimension = 576 + if ratio < 1: # portrait: res/dimension is width + y = int(dimension / ratio) + format_info.update({ + 'width': dimension, + 'height': y - (y % 2), + }) + else: # landscape: res/dimension is height + x = int(dimension * ratio) + format_info.update({ + 'width': x + (x % 2), + 'height': dimension, + }) + + for video_url in traverse_obj(bitrate_info, ('PlayAddr', 'UrlList', ..., {url_or_none})): + formats.append({ + **COMMON_FORMAT_INFO, + **format_info, + 'url': self._proto_relative_url(video_url), + }) + + # We don't have res string for play formats, but need quality for sorting & de-duplication + play_quality = traverse_obj(formats, (lambda _, v: v['width'] == play_width, 'quality', any)) + + for play_url in traverse_obj(video_info, ('playAddr', ((..., 'src'), None), {url_or_none})): + formats.append({ + **COMMON_FORMAT_INFO, + 'format_id': 'play', 'url': self._proto_relative_url(play_url), - 'ext': 'mp4', - 'width': width, - 'height': height, - }] - elif isinstance(play_url, list): - formats = [{ - 'url': self._proto_relative_url(url), - 'ext': 'mp4', - 'width': width, - 'height': height, - } for url in traverse_obj(play_url, (..., 'src'), expected_type=url_or_none) if url] + 'width': play_width, + 'height': play_height, + 'quality': play_quality, + }) - download_url = url_or_none(video_info.get('downloadAddr')) or traverse_obj(video_info, ('download', 'url'), expected_type=url_or_none) - if download_url: + for download_url in traverse_obj(video_info, (('downloadAddr', ('download', 'url')), {url_or_none})): formats.append({ + **COMMON_FORMAT_INFO, 'format_id': 'download', 'url': self._proto_relative_url(download_url), - 'ext': 'mp4', - 'width': width, - 'height': height, }) + self._remove_duplicate_formats(formats) - thumbnails = [] - for thumbnail_name in ('thumbnail', 'cover', 'dynamicCover', 'originCover'): - if aweme_detail.get(thumbnail_name): - thumbnails = [{ - 'url': self._proto_relative_url(aweme_detail[thumbnail_name]), - 'width': width, - 'height': height - }] + for f in traverse_obj(formats, lambda _, v: 'unwatermarked' not in v['url']): + f.update({ + 'format_note': join_nonempty(f.get('format_note'), 'watermarked', delim=', '), + 'preference': f.get('preference') or -2, + }) + + # Is it a slideshow with only audio for download? + if not formats and traverse_obj(aweme_detail, ('music', 'playUrl', {url_or_none})): + audio_url = aweme_detail['music']['playUrl'] + ext = traverse_obj(parse_qs(audio_url), ( + 'mime_type', -1, {lambda x: x.replace('_', '/')}, {mimetype2ext})) or 'm4a' + formats.append({ + 'format_id': 'audio', + 'url': self._proto_relative_url(audio_url), + 'ext': ext, + 'acodec': 'aac' if ext == 'm4a' else ext, + 'vcodec': 'none', + }) + + return formats + + def _parse_aweme_video_web(self, aweme_detail, webpage_url, video_id, extract_flat=False): + author_info = traverse_obj(aweme_detail, (('authorInfo', 'author', None), { + 'channel': ('nickname', {str}), + 'channel_id': (('authorSecId', 'secUid'), {str}), + 'uploader': (('uniqueId', 'author'), {str}), + 'uploader_id': (('authorId', 'uid', 'id'), {str_or_none}), + }), get_all=False) return { - 'id': traverse_obj(aweme_detail, 'id', 'awemeId', expected_type=str_or_none), - 'title': aweme_detail.get('desc'), - 'duration': try_get(aweme_detail, lambda x: x['video']['duration'], int), - 'view_count': int_or_none(stats_info.get('playCount')), - 'like_count': int_or_none(stats_info.get('diggCount')), - 'repost_count': int_or_none(stats_info.get('shareCount')), - 'comment_count': int_or_none(stats_info.get('commentCount')), - 'timestamp': int_or_none(aweme_detail.get('createTime')), - 'creator': str_or_none(author_info.get('nickname')), - 'uploader': str_or_none(author_info.get('uniqueId') or aweme_detail.get('author')), - 'uploader_id': str_or_none(traverse_obj(author_info, 'id', 'uid', 'authorId')), - 'uploader_url': user_url, - 'track': str_or_none(music_info.get('title')), - 'album': str_or_none(music_info.get('album')) or None, - 'artist': str_or_none(music_info.get('authorName')), - 'formats': formats, - 'thumbnails': thumbnails, - 'description': str_or_none(aweme_detail.get('desc')), - 'http_headers': { - 'Referer': webpage_url - } + 'id': video_id, + 'formats': None if extract_flat else self._extract_web_formats(aweme_detail), + 'subtitles': None if extract_flat else self.extract_subtitles(aweme_detail, video_id, None), + 'http_headers': {'Referer': webpage_url}, + **author_info, + 'channel_url': format_field(author_info, 'channel_id', self._UPLOADER_URL_FORMAT, default=None), + 'uploader_url': format_field( + author_info, ['uploader', 'uploader_id'], self._UPLOADER_URL_FORMAT, default=None), + **traverse_obj(aweme_detail, ('music', { + 'track': ('title', {str}), + 'album': ('album', {str}, {lambda x: x or None}), + 'artists': ('authorName', {str}, {lambda x: re.split(r'(?:, | & )', x) if x else None}), + 'duration': ('duration', {int_or_none}), + })), + **traverse_obj(aweme_detail, { + 'title': ('desc', {str}), + 'description': ('desc', {str}), + # audio-only slideshows have a video duration of 0 and an actual audio duration + 'duration': ('video', 'duration', {int_or_none}, {lambda x: x or None}), + 'timestamp': ('createTime', {int_or_none}), + }), + **traverse_obj(aweme_detail, ('stats', { + 'view_count': 'playCount', + 'like_count': 'diggCount', + 'repost_count': 'shareCount', + 'comment_count': 'commentCount', + }), expected_type=int_or_none), + 'thumbnails': traverse_obj(aweme_detail, ( + (None, 'video'), ('thumbnail', 'cover', 'dynamicCover', 'originCover'), { + 'url': ({url_or_none}, {self._proto_relative_url}), + }, + )), } @@ -447,22 +627,25 @@ class TikTokIE(TikTokBaseIE): 'artist': 'Ysrbeats', 'album': 'Lehanga', 'track': 'Lehanga', - } + }, + 'skip': '404 Not Found', }, { 'url': 'https://www.tiktok.com/@patroxofficial/video/6742501081818877190?langCountry=en', - 'md5': '6f3cf8cdd9b28cb8363fe0a9a160695b', + 'md5': 'f21112672ee4ce05ca390fb6522e1b6f', 'info_dict': { 'id': '6742501081818877190', 'ext': 'mp4', 'title': 'md5:5e2a23877420bb85ce6521dbee39ba94', 'description': 'md5:5e2a23877420bb85ce6521dbee39ba94', 'duration': 27, - 'height': 960, - 'width': 540, + 'height': 1024, + 'width': 576, 'uploader': 'patrox', 'uploader_id': '18702747', - 'uploader_url': 'https://www.tiktok.com/@MS4wLjABAAAAiFnldaILebi5heDoVU6bn4jBWWycX6-9U3xuNPqZ8Ws', - 'creator': 'patroX', + 'uploader_url': 'https://www.tiktok.com/@patrox', + 'channel_url': 'https://www.tiktok.com/@MS4wLjABAAAAiFnldaILebi5heDoVU6bn4jBWWycX6-9U3xuNPqZ8Ws', + 'channel_id': 'MS4wLjABAAAAiFnldaILebi5heDoVU6bn4jBWWycX6-9U3xuNPqZ8Ws', + 'channel': 'patroX', 'thumbnail': r're:^https?://[\w\/\.\-]+(~[\w\-]+\.image)?', 'upload_date': '20190930', 'timestamp': 1569860870, @@ -470,11 +653,11 @@ class TikTokIE(TikTokBaseIE): 'like_count': int, 'repost_count': int, 'comment_count': int, - 'artist': 'Evan Todd, Jessica Keenan Wynn, Alice Lee, Barrett Wilbert Weed & Jon Eidson', + 'artists': ['Evan Todd', 'Jessica Keenan Wynn', 'Alice Lee', 'Barrett Wilbert Weed', 'Jon Eidson'], 'track': 'Big Fun', - } + }, }, { - # Banned audio, only available on the app + # Banned audio, was available on the app, now works with web too 'url': 'https://www.tiktok.com/@barudakhb_/video/6984138651336838402', 'info_dict': { 'id': '6984138651336838402', @@ -482,11 +665,13 @@ class TikTokIE(TikTokBaseIE): 'title': 'Balas @yolaaftwsr hayu yu ? #SquadRandom_ 🔥', 'description': 'Balas @yolaaftwsr hayu yu ? #SquadRandom_ 🔥', 'uploader': 'barudakhb_', - 'creator': 'md5:29f238c49bc0c176cb3cef1a9cea9fa6', + 'channel': 'md5:29f238c49bc0c176cb3cef1a9cea9fa6', 'uploader_id': '6974687867511718913', - 'uploader_url': 'https://www.tiktok.com/@MS4wLjABAAAAbhBwQC-R1iKoix6jDFsF-vBdfx2ABoDjaZrM9fX6arU3w71q3cOWgWuTXn1soZ7d', + 'uploader_url': 'https://www.tiktok.com/@barudakhb_', + 'channel_url': 'https://www.tiktok.com/@MS4wLjABAAAAbhBwQC-R1iKoix6jDFsF-vBdfx2ABoDjaZrM9fX6arU3w71q3cOWgWuTXn1soZ7d', + 'channel_id': 'MS4wLjABAAAAbhBwQC-R1iKoix6jDFsF-vBdfx2ABoDjaZrM9fX6arU3w71q3cOWgWuTXn1soZ7d', 'track': 'Boka Dance', - 'artist': 'md5:29f238c49bc0c176cb3cef1a9cea9fa6', + 'artists': ['md5:29f238c49bc0c176cb3cef1a9cea9fa6'], 'timestamp': 1626121503, 'duration': 18, 'thumbnail': r're:^https?://[\w\/\.\-]+(~[\w\-]+\.image)?', @@ -495,7 +680,7 @@ class TikTokIE(TikTokBaseIE): 'like_count': int, 'repost_count': int, 'comment_count': int, - } + }, }, { # Sponsored video, only available with feed workaround 'url': 'https://www.tiktok.com/@MS4wLjABAAAATh8Vewkn0LYM7Fo03iec3qKdeCUOcBIouRk1mkiag6h3o_pQu_dUXvZ2EZlGST7_/video/7042692929109986561', @@ -505,9 +690,10 @@ class TikTokIE(TikTokBaseIE): 'title': 'Slap and Run!', 'description': 'Slap and Run!', 'uploader': 'user440922249', - 'creator': 'Slap And Run', + 'channel': 'Slap And Run', 'uploader_id': '7036055384943690754', 'uploader_url': 'https://www.tiktok.com/@MS4wLjABAAAATh8Vewkn0LYM7Fo03iec3qKdeCUOcBIouRk1mkiag6h3o_pQu_dUXvZ2EZlGST7_', + 'channel_id': 'MS4wLjABAAAATh8Vewkn0LYM7Fo03iec3qKdeCUOcBIouRk1mkiag6h3o_pQu_dUXvZ2EZlGST7_', 'track': 'Promoted Music', 'timestamp': 1639754738, 'duration': 30, @@ -518,7 +704,7 @@ class TikTokIE(TikTokBaseIE): 'repost_count': int, 'comment_count': int, }, - 'expected_warnings': ['trying with webpage', 'Unable to find video in feed'] + 'skip': 'This video is unavailable', }, { # Video without title and description 'url': 'https://www.tiktok.com/@pokemonlife22/video/7059698374567611694', @@ -528,15 +714,17 @@ class TikTokIE(TikTokBaseIE): 'title': 'TikTok video #7059698374567611694', 'description': '', 'uploader': 'pokemonlife22', - 'creator': 'Pokemon', + 'channel': 'Pokemon', 'uploader_id': '6820838815978423302', - 'uploader_url': 'https://www.tiktok.com/@MS4wLjABAAAA0tF1nBwQVVMyrGu3CqttkNgM68Do1OXUFuCY0CRQk8fEtSVDj89HqoqvbSTmUP2W', + 'uploader_url': 'https://www.tiktok.com/@pokemonlife22', + 'channel_url': 'https://www.tiktok.com/@MS4wLjABAAAA0tF1nBwQVVMyrGu3CqttkNgM68Do1OXUFuCY0CRQk8fEtSVDj89HqoqvbSTmUP2W', + 'channel_id': 'MS4wLjABAAAA0tF1nBwQVVMyrGu3CqttkNgM68Do1OXUFuCY0CRQk8fEtSVDj89HqoqvbSTmUP2W', 'track': 'original sound', 'timestamp': 1643714123, 'duration': 6, 'thumbnail': r're:^https?://[\w\/\.\-]+(~[\w\-]+\.image)?', 'upload_date': '20220201', - 'artist': 'Pokemon', + 'artists': ['Pokemon'], 'view_count': int, 'like_count': int, 'repost_count': int, @@ -573,149 +761,271 @@ class TikTokIE(TikTokBaseIE): 'ext': 'mp3', 'title': 'TikTok video #7139980461132074283', 'description': '', - 'creator': 'Antaura', + 'channel': 'Antaura', 'uploader': '_le_cannibale_', 'uploader_id': '6604511138619654149', - 'uploader_url': 'https://www.tiktok.com/@MS4wLjABAAAAoShJqaw_5gvy48y3azFeFcT4jeyKWbB0VVYasOCt2tTLwjNFIaDcHAM4D-QGXFOP', - 'artist': 'nathan !', + 'uploader_url': 'https://www.tiktok.com/@_le_cannibale_', + 'channel_url': 'https://www.tiktok.com/@MS4wLjABAAAAoShJqaw_5gvy48y3azFeFcT4jeyKWbB0VVYasOCt2tTLwjNFIaDcHAM4D-QGXFOP', + 'channel_id': 'MS4wLjABAAAAoShJqaw_5gvy48y3azFeFcT4jeyKWbB0VVYasOCt2tTLwjNFIaDcHAM4D-QGXFOP', + 'artists': ['nathan !'], 'track': 'grahamscott canon', + 'duration': 10, 'upload_date': '20220905', 'timestamp': 1662406249, 'view_count': int, 'like_count': int, 'repost_count': int, 'comment_count': int, + 'thumbnail': r're:^https://.+\.(?:webp|jpe?g)', + }, + }, { + # only available via web + 'url': 'https://www.tiktok.com/@moxypatch/video/7206382937372134662', + 'md5': '4cdefa501ac8ac20bf04986e10916fea', + 'info_dict': { + 'id': '7206382937372134662', + 'ext': 'mp4', + 'title': 'md5:1d95c0b96560ca0e8a231af4172b2c0a', + 'description': 'md5:1d95c0b96560ca0e8a231af4172b2c0a', + 'channel': 'MoxyPatch', + 'uploader': 'moxypatch', + 'uploader_id': '7039142049363379205', + 'uploader_url': 'https://www.tiktok.com/@moxypatch', + 'channel_url': 'https://www.tiktok.com/@MS4wLjABAAAAFhqKnngMHJSsifL0w1vFOP5kn3Ndo1ODp0XuIBkNMBCkALTvwILdpu12g3pTtL4V', + 'channel_id': 'MS4wLjABAAAAFhqKnngMHJSsifL0w1vFOP5kn3Ndo1ODp0XuIBkNMBCkALTvwILdpu12g3pTtL4V', + 'artists': ['your worst nightmare'], + 'track': 'original sound', + 'upload_date': '20230303', + 'timestamp': 1677866781, + 'duration': 10, + 'view_count': int, + 'like_count': int, + 'repost_count': int, + 'comment_count': int, + 'thumbnail': r're:^https://.+', + 'thumbnails': 'count:3', + }, + 'expected_warnings': ['Unable to find video in feed'], + }, { + # 1080p format + 'url': 'https://www.tiktok.com/@tatemcrae/video/7107337212743830830', # FIXME: Web can only get audio + 'md5': '982512017a8a917124d5a08c8ae79621', + 'info_dict': { + 'id': '7107337212743830830', + 'ext': 'mp4', + 'title': 'new music video 4 don’t come backkkk🧸🖤 i hope u enjoy !! @musicontiktok', + 'description': 'new music video 4 don’t come backkkk🧸🖤 i hope u enjoy !! @musicontiktok', + 'uploader': 'tatemcrae', + 'uploader_id': '86328792343818240', + 'uploader_url': 'https://www.tiktok.com/@MS4wLjABAAAA-0bQT0CqebTRr6I4IkYvMDMKSRSJHLNPBo5HrSklJwyA2psXLSZG5FP-LMNpHnJd', + 'channel_id': 'MS4wLjABAAAA-0bQT0CqebTRr6I4IkYvMDMKSRSJHLNPBo5HrSklJwyA2psXLSZG5FP-LMNpHnJd', + 'channel': 'tate mcrae', + 'artists': ['tate mcrae'], + 'track': 'original sound', + 'upload_date': '20220609', + 'timestamp': 1654805899, + 'duration': 150, + 'view_count': int, + 'like_count': int, + 'repost_count': int, + 'comment_count': int, 'thumbnail': r're:^https://.+\.webp', }, + 'skip': 'Unavailable via feed API, only audio available via web', + }, { + # Slideshow, audio-only m4a format + 'url': 'https://www.tiktok.com/@hara_yoimiya/video/7253412088251534594', + 'md5': '2ff8fe0174db2dbf49c597a7bef4e47d', + 'info_dict': { + 'id': '7253412088251534594', + 'ext': 'm4a', + 'title': 'я ред флаг простите #переписка #щитпост #тревожныйтиппривязанности #рекомендации ', + 'description': 'я ред флаг простите #переписка #щитпост #тревожныйтиппривязанности #рекомендации ', + 'uploader': 'hara_yoimiya', + 'uploader_id': '6582536342634676230', + 'uploader_url': 'https://www.tiktok.com/@hara_yoimiya', + 'channel_url': 'https://www.tiktok.com/@MS4wLjABAAAAIAlDxriiPWLE-p8p1R_0Bx8qWKfi-7zwmGhzU8Mv25W8sNxjfIKrol31qTczzuLB', + 'channel_id': 'MS4wLjABAAAAIAlDxriiPWLE-p8p1R_0Bx8qWKfi-7zwmGhzU8Mv25W8sNxjfIKrol31qTczzuLB', + 'channel': 'лампочка(!)', + 'artists': ['Øneheart'], + 'album': 'watching the stars', + 'track': 'watching the stars', + 'duration': 60, + 'upload_date': '20230708', + 'timestamp': 1688816612, + 'view_count': int, + 'like_count': int, + 'comment_count': int, + 'repost_count': int, + 'thumbnail': r're:^https://.+\.(?:webp|jpe?g)', + }, }, { # Auto-captions available 'url': 'https://www.tiktok.com/@hankgreen1/video/7047596209028074758', - 'only_matching': True + 'only_matching': True, }] def _real_extract(self, url): video_id, user_id = self._match_valid_url(url).group('id', 'user_id') - try: - return self._extract_aweme_app(video_id) - except ExtractorError as e: - self.report_warning(f'{e}; trying with webpage') + + if self._KNOWN_APP_INFO: + try: + return self._extract_aweme_app(video_id) + except ExtractorError as e: + e.expected = True + self.report_warning(f'{e}; trying with webpage') url = self._create_url(user_id, video_id) - webpage = self._download_webpage(url, video_id, headers={'User-Agent': 'Mozilla/5.0'}) - next_data = self._search_nextjs_data(webpage, video_id, default='{}') - if next_data: - status = traverse_obj(next_data, ('props', 'pageProps', 'statusCode'), expected_type=int) or 0 - video_data = traverse_obj(next_data, ('props', 'pageProps', 'itemInfo', 'itemStruct'), expected_type=dict) - else: - sigi_data = self._get_sigi_state(webpage, video_id) - status = traverse_obj(sigi_data, ('VideoPage', 'statusCode'), expected_type=int) or 0 - video_data = traverse_obj(sigi_data, ('ItemModule', video_id), expected_type=dict) + video_data, status = self._extract_web_data_and_status(url, video_id) - if status == 0: - return self._parse_aweme_video_web(video_data, url) + if video_data and status == 0: + return self._parse_aweme_video_web(video_data, url, video_id) elif status == 10216: raise ExtractorError('This video is private', expected=True) - raise ExtractorError('Video not available', video_id=video_id) + raise ExtractorError(f'Video not available, status code {status}', video_id=video_id) class TikTokUserIE(TikTokBaseIE): IE_NAME = 'tiktok:user' - _VALID_URL = r'https?://(?:www\.)?tiktok\.com/@(?P[\w\.-]+)/?(?:$|[#?])' - _WORKING = False + _VALID_URL = r'(?:tiktokuser:|https?://(?:www\.)?tiktok\.com/@)(?P[\w.-]+)/?(?:$|[#?])' _TESTS = [{ 'url': 'https://tiktok.com/@corgibobaa?lang=en', 'playlist_mincount': 45, 'info_dict': { - 'id': '6935371178089399301', + 'id': 'MS4wLjABAAAAepiJKgwWhulvCpSuUVsp7sgVVsFJbbNaLeQ6OQ0oAJERGDUIXhb2yxxHZedsItgT', 'title': 'corgibobaa', - 'thumbnail': r're:https://.+_1080x1080\.webp' }, - 'expected_warnings': ['Retrying'] }, { 'url': 'https://www.tiktok.com/@6820838815978423302', 'playlist_mincount': 5, 'info_dict': { - 'id': '6820838815978423302', + 'id': 'MS4wLjABAAAA0tF1nBwQVVMyrGu3CqttkNgM68Do1OXUFuCY0CRQk8fEtSVDj89HqoqvbSTmUP2W', 'title': '6820838815978423302', - 'thumbnail': r're:https://.+_1080x1080\.webp' }, - 'expected_warnings': ['Retrying'] }, { 'url': 'https://www.tiktok.com/@meme', 'playlist_mincount': 593, 'info_dict': { - 'id': '79005827461758976', + 'id': 'MS4wLjABAAAAiKfaDWeCsT3IHwY77zqWGtVRIy9v4ws1HbVi7auP1Vx7dJysU_hc5yRiGywojRD6', 'title': 'meme', - 'thumbnail': r're:https://.+_1080x1080\.webp' }, - 'expected_warnings': ['Retrying'] + }, { + 'url': 'tiktokuser:MS4wLjABAAAAM3R2BtjzVT-uAtstkl2iugMzC6AtnpkojJbjiOdDDrdsTiTR75-8lyWJCY5VvDrZ', + 'playlist_mincount': 31, + 'info_dict': { + 'id': 'MS4wLjABAAAAM3R2BtjzVT-uAtstkl2iugMzC6AtnpkojJbjiOdDDrdsTiTR75-8lyWJCY5VvDrZ', + }, }] + _USER_AGENT = 'Mozilla/5.0 (Windows NT 10.0; Win64; x64; rv:115.0) Gecko/20100101 Firefox/115.0' + _API_BASE_URL = 'https://www.tiktok.com/api/creator/item_list/' - r''' # TODO: Fix by adding _signature to api_url - def _entries(self, webpage, user_id, username): - secuid = self._search_regex(r'\"secUid\":\"(?P[^\"]+)', webpage, username) - verifyfp_cookie = self._get_cookies('https://www.tiktok.com').get('s_v_web_id') - if not verifyfp_cookie: - raise ExtractorError('Improper cookies (missing s_v_web_id).', expected=True) - api_url = f'https://m.tiktok.com/api/post/item_list/?aid=1988&cookie_enabled=true&count=30&verifyFp={verifyfp_cookie.value}&secUid={secuid}&cursor=' - cursor = '0' - for page in itertools.count(): - data_json = self._download_json(api_url + cursor, username, note='Downloading Page %d' % page) - for video in data_json.get('itemList', []): - video_id = video['id'] - video_url = f'https://www.tiktok.com/@{user_id}/video/{video_id}' - yield self._url_result(video_url, 'TikTok', video_id, str_or_none(video.get('desc'))) - if not data_json.get('hasMore'): - break - cursor = data_json['cursor'] - ''' - - def _video_entries_api(self, webpage, user_id, username): - query = { - 'user_id': user_id, - 'count': 21, - 'max_cursor': 0, - 'min_cursor': 0, - 'retry_type': 'no_retry', - 'device_id': ''.join(random.choices(string.digits, k=19)), # Some endpoints don't like randomized device_id, so it isn't directly set in _call_api. + def _build_web_query(self, sec_uid, cursor): + return { + 'aid': '1988', + 'app_language': 'en', + 'app_name': 'tiktok_web', + 'browser_language': 'en-US', + 'browser_name': 'Mozilla', + 'browser_online': 'true', + 'browser_platform': 'Win32', + 'browser_version': '5.0 (Windows)', + 'channel': 'tiktok_web', + 'cookie_enabled': 'true', + 'count': '15', + 'cursor': cursor, + 'device_id': self._DEVICE_ID, + 'device_platform': 'web_pc', + 'focus_state': 'true', + 'from_page': 'user', + 'history_len': '2', + 'is_fullscreen': 'false', + 'is_page_visible': 'true', + 'language': 'en', + 'os': 'windows', + 'priority_region': '', + 'referer': '', + 'region': 'US', + 'screen_height': '1080', + 'screen_width': '1920', + 'secUid': sec_uid, + 'type': '1', # pagination type: 0 == oldest-to-newest, 1 == newest-to-oldest + 'tz_name': 'UTC', + 'verifyFp': f'verify_{"".join(random.choices(string.hexdigits, k=7))}', + 'webcast_language': 'en', } + def _entries(self, sec_uid, user_name): + display_id = user_name or sec_uid + seen_ids = set() + + cursor = int(time.time() * 1E3) for page in itertools.count(1): - for retry in self.RetryManager(): - try: - post_list = self._call_api( - 'aweme/post', query, username, note=f'Downloading user video list page {page}', - errnote='Unable to download user video list') - except ExtractorError as e: - if isinstance(e.cause, json.JSONDecodeError) and e.cause.pos == 0: - retry.error = e - continue - raise - yield from post_list.get('aweme_list', []) - if not post_list.get('has_more'): + response = self._download_json( + self._API_BASE_URL, display_id, f'Downloading page {page}', + query=self._build_web_query(sec_uid, cursor), headers={'User-Agent': self._USER_AGENT}) + + for video in traverse_obj(response, ('itemList', lambda _, v: v['id'])): + video_id = video['id'] + if video_id in seen_ids: + continue + seen_ids.add(video_id) + webpage_url = self._create_url(display_id, video_id) + yield self.url_result( + webpage_url, TikTokIE, + **self._parse_aweme_video_web(video, webpage_url, video_id, extract_flat=True)) + + old_cursor = cursor + cursor = traverse_obj( + response, ('itemList', -1, 'createTime', {lambda x: int(x * 1E3)})) + if not cursor or old_cursor == cursor: + # User may not have posted within this ~1 week lookback, so manually adjust cursor + cursor = old_cursor - 7 * 86_400_000 + # In case 'hasMorePrevious' is wrong, break if we have gone back before TikTok existed + if cursor < 1472706000000 or not traverse_obj(response, 'hasMorePrevious'): break - query['max_cursor'] = post_list['max_cursor'] - - def _entries_api(self, user_id, videos): - for video in videos: - yield { - **self._parse_aweme_video_app(video), - 'extractor_key': TikTokIE.ie_key(), - 'extractor': 'TikTok', - 'webpage_url': f'https://tiktok.com/@{user_id}/video/{video["aweme_id"]}', - } - def _real_extract(self, url): - user_name = self._match_id(url) - webpage = self._download_webpage(url, user_name, headers={ - 'User-Agent': 'facebookexternalhit/1.1 (+http://www.facebook.com/externalhit_uatext.php)' - }) - user_id = self._html_search_regex(r'snssdk\d*://user/profile/(\d+)', webpage, 'user ID', default=None) or user_name + def _get_sec_uid(self, user_url, user_name, msg): + webpage = self._download_webpage( + user_url, user_name, fatal=False, headers={'User-Agent': 'Mozilla/5.0'}, + note=f'Downloading {msg} webpage', errnote=f'Unable to download {msg} webpage') or '' + return (traverse_obj(self._get_universal_data(webpage, user_name), + ('webapp.user-detail', 'userInfo', 'user', 'secUid', {str})) + or traverse_obj(self._get_sigi_state(webpage, user_name), + ('LiveRoom', 'liveRoomUserInfo', 'user', 'secUid', {str}), + ('UserModule', 'users', ..., 'secUid', {str}, any))) - videos = LazyList(self._video_entries_api(webpage, user_id, user_name)) - thumbnail = traverse_obj(videos, (0, 'author', 'avatar_larger', 'url_list', 0)) + def _real_extract(self, url): + user_name, sec_uid = self._match_id(url), None + if mobj := re.fullmatch(r'MS4wLjABAAAA[\w-]{64}', user_name): + user_name, sec_uid = None, mobj.group(0) + else: + sec_uid = (self._get_sec_uid(self._UPLOADER_URL_FORMAT % user_name, user_name, 'user') + or self._get_sec_uid(self._UPLOADER_URL_FORMAT % f'{user_name}/live', user_name, 'live')) + + if not sec_uid: + webpage = self._download_webpage( + f'https://www.tiktok.com/embed/@{user_name}', user_name, + note='Downloading user embed page', fatal=False) or '' + data = traverse_obj(self._search_json( + r']+\bid=[\'"]__FRONTITY_CONNECT_STATE__[\'"][^>]*>', + webpage, 'data', user_name, default={}), + ('source', 'data', f'/embed/@{user_name}', {dict})) + + for aweme_id in traverse_obj(data, ('videoList', ..., 'id', {str})): + webpage_url = self._create_url(user_name, aweme_id) + video_data, _ = self._extract_web_data_and_status(webpage_url, aweme_id, fatal=False) + sec_uid = self._parse_aweme_video_web( + video_data, webpage_url, aweme_id, extract_flat=True).get('channel_id') + if sec_uid: + break + + if not sec_uid: + raise ExtractorError( + 'Unable to extract secondary user ID. If you are able to get the channel_id ' + 'from a video posted by this user, try using "tiktokuser:channel_id" as the ' + 'input URL (replacing `channel_id` with its actual value)', expected=True) - return self.playlist_result(self._entries_api(user_id, videos), user_id, user_name, thumbnail=thumbnail) + return self.playlist_result(self._entries(sec_uid, user_name), sec_uid, user_name) class TikTokBaseListIE(TikTokBaseIE): # XXX: Conventionally, base classes should end with BaseIE/InfoExtractor @@ -725,14 +1035,15 @@ def _entries(self, list_id, display_id): 'cursor': 0, 'count': 20, 'type': 5, - 'device_id': ''.join(random.choices(string.digits, k=19)) + 'device_id': self._DEVICE_ID, } for page in itertools.count(1): for retry in self.RetryManager(): try: post_list = self._call_api( - self._API_ENDPOINT, query, display_id, note=f'Downloading video list page {page}', + self._API_ENDPOINT, display_id, query=query, + note=f'Downloading video list page {page}', errnote='Unable to download video list') except ExtractorError as e: if isinstance(e.cause, json.JSONDecodeError) and e.cause.pos == 0: @@ -765,17 +1076,17 @@ class TikTokSoundIE(TikTokBaseListIE): 'url': 'https://www.tiktok.com/music/Build-a-Btch-6956990112127585029?lang=en', 'playlist_mincount': 100, 'info_dict': { - 'id': '6956990112127585029' + 'id': '6956990112127585029', }, - 'expected_warnings': ['Retrying'] + 'expected_warnings': ['Retrying'], }, { # Actual entries are less than listed video count 'url': 'https://www.tiktok.com/music/jiefei-soap-remix-7036843036118469381', 'playlist_mincount': 2182, 'info_dict': { - 'id': '7036843036118469381' + 'id': '7036843036118469381', }, - 'expected_warnings': ['Retrying'] + 'expected_warnings': ['Retrying'], }] @@ -791,11 +1102,11 @@ class TikTokEffectIE(TikTokBaseListIE): 'info_dict': { 'id': '1258156', }, - 'expected_warnings': ['Retrying'] + 'expected_warnings': ['Retrying'], }, { # Different entries between mobile and web, depending on region 'url': 'https://www.tiktok.com/sticker/Elf-Friend-479565', - 'only_matching': True + 'only_matching': True, }] @@ -812,82 +1123,152 @@ class TikTokTagIE(TikTokBaseListIE): 'id': '46294678', 'title': 'hello2018', }, - 'expected_warnings': ['Retrying'] + 'expected_warnings': ['Retrying'], }, { 'url': 'https://tiktok.com/tag/fypシ?is_copy_url=0&is_from_webapp=v1', - 'only_matching': True + 'only_matching': True, }] def _real_extract(self, url): display_id = self._match_id(url) webpage = self._download_webpage(url, display_id, headers={ - 'User-Agent': 'facebookexternalhit/1.1 (+http://www.facebook.com/externalhit_uatext.php)' + 'User-Agent': 'facebookexternalhit/1.1 (+http://www.facebook.com/externalhit_uatext.php)', }) tag_id = self._html_search_regex(r'snssdk\d*://challenge/detail/(\d+)', webpage, 'tag ID') return self.playlist_result(self._entries(tag_id, display_id), tag_id, display_id) +class TikTokCollectionIE(TikTokBaseIE): + IE_NAME = 'tiktok:collection' + _VALID_URL = r'https?://www\.tiktok\.com/@(?P[\w.-]+)/collection/(?P[^/?#]+)-(?P<id>\d+)/?(?:[?#]|$)' + _TESTS = [{ + # playlist should have exactly 9 videos + 'url': 'https://www.tiktok.com/@imanoreotwe/collection/count-test-7371330159376370462', + 'info_dict': { + 'id': '7371330159376370462', + 'title': 'imanoreotwe-count-test', + }, + 'playlist_count': 9, + }, { + # tests returning multiple pages of a large collection + 'url': 'https://www.tiktok.com/@imanoreotwe/collection/%F0%9F%98%82-7111887189571160875', + 'info_dict': { + 'id': '7111887189571160875', + 'title': 'imanoreotwe-%F0%9F%98%82', + }, + 'playlist_mincount': 100, + }] + _API_BASE_URL = 'https://www.tiktok.com/api/collection/item_list/' + _PAGE_COUNT = 30 + + def _build_web_query(self, collection_id, cursor): + return { + 'aid': '1988', + 'collectionId': collection_id, + 'count': self._PAGE_COUNT, + 'cursor': cursor, + 'sourceType': '113', + } + + def _entries(self, collection_id): + cursor = 0 + for page in itertools.count(1): + response = self._download_json( + self._API_BASE_URL, collection_id, f'Downloading page {page}', + query=self._build_web_query(collection_id, cursor)) + + for video in traverse_obj(response, ('itemList', lambda _, v: v['id'])): + video_id = video['id'] + author = traverse_obj(video, ('author', ('uniqueId', 'secUid', 'id'), {str}, any)) or '_' + webpage_url = self._create_url(author, video_id) + yield self.url_result( + webpage_url, TikTokIE, + **self._parse_aweme_video_web(video, webpage_url, video_id, extract_flat=True)) + + if not traverse_obj(response, 'hasMore'): + break + cursor += self._PAGE_COUNT + + def _real_extract(self, url): + collection_id, title, user_name = self._match_valid_url(url).group('id', 'title', 'user_id') + + return self.playlist_result( + self._entries(collection_id), collection_id, '-'.join((user_name, title))) + + class DouyinIE(TikTokBaseIE): _VALID_URL = r'https?://(?:www\.)?douyin\.com/video/(?P<id>[0-9]+)' _TESTS = [{ 'url': 'https://www.douyin.com/video/6961737553342991651', - 'md5': 'a97db7e3e67eb57bf40735c022ffa228', + 'md5': '9ecce7bc5b302601018ecb2871c63a75', 'info_dict': { 'id': '6961737553342991651', 'ext': 'mp4', 'title': '#杨超越 小小水手带你去远航❤️', 'description': '#杨超越 小小水手带你去远航❤️', + 'uploader': '6897520xka', 'uploader_id': '110403406559', 'uploader_url': 'https://www.douyin.com/user/MS4wLjABAAAAEKnfa654JAJ_N5lgZDQluwsxmY0lhfmEYNQBBkwGG98', - 'creator': '杨超越', - 'duration': 19782, + 'channel_id': 'MS4wLjABAAAAEKnfa654JAJ_N5lgZDQluwsxmY0lhfmEYNQBBkwGG98', + 'channel': '杨超越', + 'duration': 19, 'timestamp': 1620905839, 'upload_date': '20210513', 'track': '@杨超越创作的原声', + 'artists': ['杨超越'], 'view_count': int, 'like_count': int, 'repost_count': int, 'comment_count': int, + 'thumbnail': r're:https?://.+\.jpe?g', }, }, { 'url': 'https://www.douyin.com/video/6982497745948921092', - 'md5': '34a87ebff3833357733da3fe17e37c0e', + 'md5': '15c5e660b7048af3707304e3cc02bbb5', 'info_dict': { 'id': '6982497745948921092', 'ext': 'mp4', 'title': '这个夏日和小羊@杨超越 一起遇见白色幻想', 'description': '这个夏日和小羊@杨超越 一起遇见白色幻想', + 'uploader': '0731chaoyue', 'uploader_id': '408654318141572', 'uploader_url': 'https://www.douyin.com/user/MS4wLjABAAAAZJpnglcjW2f_CMVcnqA_6oVBXKWMpH0F8LIHuUu8-lA', - 'creator': '杨超越工作室', - 'duration': 42608, + 'channel_id': 'MS4wLjABAAAAZJpnglcjW2f_CMVcnqA_6oVBXKWMpH0F8LIHuUu8-lA', + 'channel': '杨超越工作室', + 'duration': 42, 'timestamp': 1625739481, 'upload_date': '20210708', 'track': '@杨超越工作室创作的原声', + 'artists': ['杨超越工作室'], 'view_count': int, 'like_count': int, 'repost_count': int, 'comment_count': int, + 'thumbnail': r're:https?://.+\.jpe?g', }, }, { 'url': 'https://www.douyin.com/video/6953975910773099811', - 'md5': 'dde3302460f19db59c47060ff013b902', + 'md5': '0e6443758b8355db9a3c34864a4276be', 'info_dict': { 'id': '6953975910773099811', 'ext': 'mp4', 'title': '#一起看海 出现在你的夏日里', 'description': '#一起看海 出现在你的夏日里', + 'uploader': '6897520xka', 'uploader_id': '110403406559', 'uploader_url': 'https://www.douyin.com/user/MS4wLjABAAAAEKnfa654JAJ_N5lgZDQluwsxmY0lhfmEYNQBBkwGG98', - 'creator': '杨超越', - 'duration': 17228, + 'channel_id': 'MS4wLjABAAAAEKnfa654JAJ_N5lgZDQluwsxmY0lhfmEYNQBBkwGG98', + 'channel': '杨超越', + 'duration': 17, 'timestamp': 1619098692, 'upload_date': '20210422', 'track': '@杨超越创作的原声', + 'artists': ['杨超越'], 'view_count': int, 'like_count': int, 'repost_count': int, 'comment_count': int, + 'thumbnail': r're:https?://.+\.jpe?g', }, }, { 'url': 'https://www.douyin.com/video/6950251282489675042', @@ -908,55 +1289,46 @@ class DouyinIE(TikTokBaseIE): 'skip': 'No longer available', }, { 'url': 'https://www.douyin.com/video/6963263655114722595', - 'md5': 'cf9f11f0ec45d131445ec2f06766e122', + 'md5': '1440bcf59d8700f8e014da073a4dfea8', 'info_dict': { 'id': '6963263655114722595', 'ext': 'mp4', 'title': '#哪个爱豆的105度最甜 换个角度看看我哈哈', 'description': '#哪个爱豆的105度最甜 换个角度看看我哈哈', + 'uploader': '6897520xka', 'uploader_id': '110403406559', 'uploader_url': 'https://www.douyin.com/user/MS4wLjABAAAAEKnfa654JAJ_N5lgZDQluwsxmY0lhfmEYNQBBkwGG98', - 'creator': '杨超越', - 'duration': 15115, + 'channel_id': 'MS4wLjABAAAAEKnfa654JAJ_N5lgZDQluwsxmY0lhfmEYNQBBkwGG98', + 'channel': '杨超越', + 'duration': 15, 'timestamp': 1621261163, 'upload_date': '20210517', 'track': '@杨超越创作的原声', + 'artists': ['杨超越'], 'view_count': int, 'like_count': int, 'repost_count': int, 'comment_count': int, + 'thumbnail': r're:https?://.+\.jpe?g', }, }] - _APP_VERSIONS = [('23.3.0', '230300')] - _APP_NAME = 'aweme' - _AID = 1128 - _API_HOSTNAME = 'aweme.snssdk.com' _UPLOADER_URL_FORMAT = 'https://www.douyin.com/user/%s' _WEBPAGE_HOST = 'https://www.douyin.com/' def _real_extract(self, url): video_id = self._match_id(url) - try: - return self._extract_aweme_app(video_id) - except ExtractorError as e: - e.expected = True - self.to_screen(f'{e}; trying with webpage') - - webpage = self._download_webpage(url, video_id) - render_data_json = self._search_regex( - r'<script [^>]*\bid=[\'"]RENDER_DATA[\'"][^>]*>(%7B.+%7D)</script>', - webpage, 'render data', default=None) - if not render_data_json: + detail = traverse_obj(self._download_json( + 'https://www.douyin.com/aweme/v1/web/aweme/detail/', video_id, + 'Downloading web detail JSON', 'Failed to download web detail JSON', + query={'aweme_id': video_id}, fatal=False), ('aweme_detail', {dict})) + if not detail: # TODO: Run verification challenge code to generate signature cookies - cookies = self._get_cookies(self._WEBPAGE_HOST) - expected = not cookies.get('s_v_web_id') or not cookies.get('ttwid') raise ExtractorError( - 'Fresh cookies (not necessarily logged in) are needed', expected=expected) + 'Fresh cookies (not necessarily logged in) are needed', + expected=not self._get_cookies(self._WEBPAGE_HOST).get('s_v_web_id')) - render_data = self._parse_json( - render_data_json, video_id, transform_source=compat_urllib_parse_unquote) - return self._parse_aweme_video_web(get_first(render_data, ('aweme', 'detail')), url) + return self._parse_aweme_video_app(detail) class TikTokVMIE(InfoExtractor): @@ -1015,7 +1387,7 @@ class TikTokVMIE(InfoExtractor): def _real_extract(self, url): new_url = self._request_webpage( - HEADRequest(url), self._match_id(url), headers={'User-Agent': 'facebookexternalhit/1.1'}).geturl() + HEADRequest(url), self._match_id(url), headers={'User-Agent': 'facebookexternalhit/1.1'}).url if self.suitable(new_url): # Prevent infinite loop in case redirect fails raise UnsupportedError(new_url) return self.url_result(new_url) @@ -1085,7 +1457,7 @@ def _real_extract(self, url): url, uploader or room_id, headers={'User-Agent': 'Mozilla/5.0'}, fatal=not room_id) if webpage: - data = try_call(lambda: self._get_sigi_state(webpage, uploader or room_id)) + data = self._get_sigi_state(webpage, uploader or room_id) room_id = (traverse_obj(data, ('UserModule', 'users', ..., 'roomId', {str_or_none}), get_all=False) or self._search_regex(r'snssdk\d*://live\?room_id=(\d+)', webpage, 'room ID', default=None) or room_id)