import base64 import binascii import json import os import random import time from .common import InfoExtractor from ..aes import aes_cbc_decrypt_bytes, unpad_pkcs7 from ..compat import compat_b64decode from ..networking.exceptions import HTTPError from ..utils import ( ExtractorError, ass_subtitles_timecode, bytes_to_intlist, bytes_to_long, float_or_none, int_or_none, intlist_to_bytes, long_to_bytes, parse_iso8601, pkcs1pad, str_or_none, strip_or_none, try_get, unified_strdate, urlencode_postdata, ) from ..utils.traversal import traverse_obj class ADNBaseIE(InfoExtractor): IE_DESC = 'Animation Digital Network' _NETRC_MACHINE = 'animationdigitalnetwork' _BASE = 'animationdigitalnetwork.fr' _API_BASE_URL = f'https://gw.api.{_BASE}/' _PLAYER_BASE_URL = f'{_API_BASE_URL}player/' _HEADERS = {} _LOGIN_ERR_MESSAGE = 'Unable to log in' _RSA_KEY = (0x9B42B08905199A5CCE2026274399CA560ECB209EE9878A708B1C0812E1BB8CB5D1FB7441861147C1A1F2F3A0476DD63A9CAC20D3E983613346850AA6CB38F16DC7D720FD7D86FC6E5B3D5BBC72E14CD0BF9E869F2CEA2CCAD648F1DCE38F1FF916CEFB2D339B64AA0264372344BC775E265E8A852F88144AB0BD9AA06C1A4ABB, 65537) _POS_ALIGN_MAP = { 'start': 1, 'end': 3, } _LINE_ALIGN_MAP = { 'middle': 8, 'end': 4, } class ADNIE(ADNBaseIE): _VALID_URL = r'https?://(?:www\.)?(?:animation|anime)digitalnetwork\.(?Pfr|de)/video/[^/?#]+/(?P\d+)' _TESTS = [{ 'url': 'https://animationdigitalnetwork.fr/video/fruits-basket/9841-episode-1-a-ce-soir', 'md5': '1c9ef066ceb302c86f80c2b371615261', 'info_dict': { 'id': '9841', 'ext': 'mp4', 'title': 'Fruits Basket - Episode 1', 'description': 'md5:14be2f72c3c96809b0ca424b0097d336', 'series': 'Fruits Basket', 'duration': 1437, 'release_date': '20190405', 'comment_count': int, 'average_rating': float, 'season_number': 1, 'episode': 'À ce soir !', 'episode_number': 1, 'thumbnail': str, 'season': 'Season 1', }, 'skip': 'Only available in French and German speaking Europe', }, { 'url': 'http://animedigitalnetwork.fr/video/blue-exorcist-kyoto-saga/7778-episode-1-debut-des-hostilites', 'only_matching': True, }, { 'url': 'https://animationdigitalnetwork.de/video/the-eminence-in-shadow/23550-folge-1', 'md5': '5c5651bf5791fa6fcd7906012b9d94e8', 'info_dict': { 'id': '23550', 'ext': 'mp4', 'episode_number': 1, 'duration': 1417, 'release_date': '20231004', 'series': 'The Eminence in Shadow', 'season_number': 2, 'episode': str, 'title': str, 'thumbnail': str, 'season': 'Season 2', 'comment_count': int, 'average_rating': float, 'description': str, }, # 'skip': 'Only available in French and German speaking Europe', }] def _get_subtitles(self, sub_url, video_id): if not sub_url: return None enc_subtitles = self._download_webpage( sub_url, video_id, 'Downloading subtitles location', fatal=False) or '{}' subtitle_location = (self._parse_json(enc_subtitles, video_id, fatal=False) or {}).get('location') if subtitle_location: enc_subtitles = self._download_webpage( subtitle_location, video_id, 'Downloading subtitles data', fatal=False, headers={'Origin': 'https://' + self._BASE}) if not enc_subtitles: return None # http://animationdigitalnetwork.fr/components/com_vodvideo/videojs/adn-vjs.min.js dec_subtitles = unpad_pkcs7(aes_cbc_decrypt_bytes( compat_b64decode(enc_subtitles[24:]), binascii.unhexlify(self._K + '7fac1178830cfe0c'), compat_b64decode(enc_subtitles[:24]))) subtitles_json = self._parse_json(dec_subtitles.decode(), None, fatal=False) if not subtitles_json: return None subtitles = {} for sub_lang, sub in subtitles_json.items(): ssa = '''[Script Info] ScriptType:V4.00 [V4 Styles] Format: Name,Fontname,Fontsize,PrimaryColour,SecondaryColour,TertiaryColour,BackColour,Bold,Italic,BorderStyle,Outline,Shadow,Alignment,MarginL,MarginR,MarginV,AlphaLevel,Encoding Style: Default,Arial,18,16777215,16777215,16777215,0,-1,0,1,1,0,2,20,20,20,0,0 [Events] Format: Marked,Start,End,Style,Name,MarginL,MarginR,MarginV,Effect,Text''' for current in sub: start, end, text, line_align, position_align = ( float_or_none(current.get('startTime')), float_or_none(current.get('endTime')), current.get('text'), current.get('lineAlign'), current.get('positionAlign')) if start is None or end is None or text is None: continue alignment = self._POS_ALIGN_MAP.get(position_align, 2) + self._LINE_ALIGN_MAP.get(line_align, 0) ssa += os.linesep + 'Dialogue: Marked=0,%s,%s,Default,,0,0,0,,%s%s' % ( ass_subtitles_timecode(start), ass_subtitles_timecode(end), '{\\a%d}' % alignment if alignment != 2 else '', text.replace('\n', '\\N').replace('', '{\\i1}').replace('', '{\\i0}')) if sub_lang == 'vostf': sub_lang = 'fr' elif sub_lang == 'vostde': sub_lang = 'de' subtitles.setdefault(sub_lang, []).extend([{ 'ext': 'json', 'data': json.dumps(sub), }, { 'ext': 'ssa', 'data': ssa, }]) return subtitles def _perform_login(self, username, password): try: access_token = (self._download_json( self._API_BASE_URL + 'authentication/login', None, 'Logging in', self._LOGIN_ERR_MESSAGE, fatal=False, data=urlencode_postdata({ 'password': password, 'rememberMe': False, 'source': 'Web', 'username': username, })) or {}).get('accessToken') if access_token: self._HEADERS = {'authorization': 'Bearer ' + access_token} except ExtractorError as e: message = None if isinstance(e.cause, HTTPError) and e.cause.status == 401: resp = self._parse_json( e.cause.response.read().decode(), None, fatal=False) or {} message = resp.get('message') or resp.get('code') self.report_warning(message or self._LOGIN_ERR_MESSAGE) def _real_extract(self, url): lang, video_id = self._match_valid_url(url).group('lang', 'id') video_base_url = self._PLAYER_BASE_URL + 'video/%s/' % video_id player = self._download_json( video_base_url + 'configuration', video_id, 'Downloading player config JSON metadata', headers=self._HEADERS)['player'] options = player['options'] user = options['user'] if not user.get('hasAccess'): start_date = traverse_obj(options, ('video', 'startDate', {str})) if (parse_iso8601(start_date) or 0) > time.time(): raise ExtractorError(f'This video is not available yet. Release date: {start_date}', expected=True) self.raise_login_required('This video requires a subscription', method='password') token = self._download_json( user.get('refreshTokenUrl') or (self._PLAYER_BASE_URL + 'refresh/token'), video_id, 'Downloading access token', headers={ 'X-Player-Refresh-Token': user['refreshToken'], }, data=b'')['token'] links_url = try_get(options, lambda x: x['video']['url']) or (video_base_url + 'link') self._K = ''.join(random.choices('0123456789abcdef', k=16)) message = bytes_to_intlist(json.dumps({ 'k': self._K, 't': token, })) # Sometimes authentication fails for no good reason, retry with # a different random padding links_data = None for _ in range(3): padded_message = intlist_to_bytes(pkcs1pad(message, 128)) n, e = self._RSA_KEY encrypted_message = long_to_bytes(pow(bytes_to_long(padded_message), e, n)) authorization = base64.b64encode(encrypted_message).decode() try: links_data = self._download_json( links_url, video_id, 'Downloading links JSON metadata', headers={ 'X-Player-Token': authorization, 'X-Target-Distribution': lang, **self._HEADERS }, query={ 'freeWithAds': 'true', 'adaptive': 'false', 'withMetadata': 'true', 'source': 'Web' }) break except ExtractorError as e: if not isinstance(e.cause, HTTPError): raise e if e.cause.status == 401: # This usually goes away with a different random pkcs1pad, so retry continue error = self._parse_json(e.cause.response.read(), video_id) message = error.get('message') if e.cause.code == 403 and error.get('code') == 'player-bad-geolocation-country': self.raise_geo_restricted(msg=message) raise ExtractorError(message) else: raise ExtractorError('Giving up retrying') links = links_data.get('links') or {} metas = links_data.get('metadata') or {} sub_url = (links.get('subtitles') or {}).get('all') video_info = links_data.get('video') or {} title = metas['title'] formats = [] for format_id, qualities in (links.get('streaming') or {}).items(): if not isinstance(qualities, dict): continue for quality, load_balancer_url in qualities.items(): load_balancer_data = self._download_json( load_balancer_url, video_id, 'Downloading %s %s JSON metadata' % (format_id, quality), fatal=False) or {} m3u8_url = load_balancer_data.get('location') if not m3u8_url: continue m3u8_formats = self._extract_m3u8_formats( m3u8_url, video_id, 'mp4', 'm3u8_native', m3u8_id=format_id, fatal=False) if format_id == 'vf': for f in m3u8_formats: f['language'] = 'fr' elif format_id == 'vde': for f in m3u8_formats: f['language'] = 'de' formats.extend(m3u8_formats) if not formats: self.raise_login_required('This video requires a subscription', method='password') video = (self._download_json( self._API_BASE_URL + 'video/%s' % video_id, video_id, 'Downloading additional video metadata', fatal=False) or {}).get('video') or {} show = video.get('show') or {} return { 'id': video_id, 'title': title, 'description': strip_or_none(metas.get('summary') or video.get('summary')), 'thumbnail': video_info.get('image') or player.get('image'), 'formats': formats, 'subtitles': self.extract_subtitles(sub_url, video_id), 'episode': metas.get('subtitle') or video.get('name'), 'episode_number': int_or_none(video.get('shortNumber')), 'series': show.get('title'), 'season_number': int_or_none(video.get('season')), 'duration': int_or_none(video_info.get('duration') or video.get('duration')), 'release_date': unified_strdate(video.get('releaseDate')), 'average_rating': float_or_none(video.get('rating') or metas.get('rating')), 'comment_count': int_or_none(video.get('commentsCount')), } class ADNSeasonIE(ADNBaseIE): _VALID_URL = r'https?://(?:www\.)?(?:animation|anime)digitalnetwork\.(?Pfr|de)/video/(?P[^/?#]+)/?(?:$|[#?])' _TESTS = [{ 'url': 'https://animationdigitalnetwork.fr/video/tokyo-mew-mew-new', 'playlist_count': 12, 'info_dict': { 'id': '911', 'title': 'Tokyo Mew Mew New', }, # 'skip': 'Only available in French end German speaking Europe', }] def _real_extract(self, url): lang, video_show_slug = self._match_valid_url(url).group('lang', 'id') show = self._download_json( f'{self._API_BASE_URL}show/{video_show_slug}/', video_show_slug, 'Downloading show JSON metadata', headers=self._HEADERS)['show'] show_id = str(show['id']) episodes = self._download_json( f'{self._API_BASE_URL}video/show/{show_id}', video_show_slug, 'Downloading episode list', headers={ 'X-Target-Distribution': lang, **self._HEADERS }, query={ 'order': 'asc', 'limit': '-1', }) def entries(): for episode_id in traverse_obj(episodes, ('videos', ..., 'id', {str_or_none})): yield self.url_result( f'https://animationdigitalnetwork.{lang}/video/{video_show_slug}/{episode_id}', ADNIE, episode_id) return self.playlist_result(entries(), show_id, show.get('title'))