yt_dlp/extractor/abematv.py

   1 import base64
   2 import binascii
   3 import functools
   4 import hashlib
   5 import hmac
   6 import io
   7 import json
   8 import re
   9 import struct
  10 import time
  11 import urllib.parse
  12 import uuid
  13
  14 from .common import InfoExtractor
  15 from ..aes import aes_ecb_decrypt
  16 from ..networking import RequestHandler, Response
  17 from ..networking.exceptions import TransportError
  18 from ..utils import (
  19     ExtractorError,
  20     OnDemandPagedList,
  21     bytes_to_intlist,
  22     decode_base_n,
  23     int_or_none,
  24     intlist_to_bytes,
  25     time_seconds,
  26     traverse_obj,
  27     update_url_query,
  28 )
  29
  30
  31 class AbemaLicenseRH(RequestHandler):
  32     _SUPPORTED_URL_SCHEMES = ('abematv-license',)
  33     _SUPPORTED_PROXY_SCHEMES = None
  34     _SUPPORTED_FEATURES = None
  35     RH_NAME = 'abematv_license'
  36
  37     _STRTABLE = '123456789ABCDEFGHJKLMNPQRSTUVWXYZabcdefghijkmnopqrstuvwxyz'
  38     _HKEY = b'3AF0298C219469522A313570E8583005A642E73EDD58E3EA2FB7339D3DF1597E'
  39
  40     def __init__(self, *, ie: 'AbemaTVIE', **kwargs):
  41         super().__init__(**kwargs)
  42         self.ie = ie
  43
  44     def _send(self, request):
  45         url = request.url
  46         ticket = urllib.parse.urlparse(url).netloc
  47
  48         try:
  49             response_data = self._get_videokey_from_ticket(ticket)
  50         except ExtractorError as e:
  51             raise TransportError(cause=e.cause) from e
  52         except (IndexError, KeyError, TypeError) as e:
  53             raise TransportError(cause=repr(e)) from e
  54
  55         return Response(
  56             io.BytesIO(response_data), url,
  57             headers={'Content-Length': str(len(response_data))})
  58
  59     def _get_videokey_from_ticket(self, ticket):
  60         to_show = self.ie.get_param('verbose', False)
  61         media_token = self.ie._get_media_token(to_show=to_show)
  62
  63         license_response = self.ie._download_json(
  64             'https://license.abema.io/abematv-hls', None, note='Requesting playback license' if to_show else False,
  65             query={'t': media_token},
  66             data=json.dumps({
  67                 'kv': 'a',
  68                 'lt': ticket,
  69             }).encode(),
  70             headers={
  71                 'Content-Type': 'application/json',
  72             })
  73
  74         res = decode_base_n(license_response['k'], table=self._STRTABLE)
  75         encvideokey = bytes_to_intlist(struct.pack('>QQ', res >> 64, res & 0xffffffffffffffff))
  76
  77         h = hmac.new(
  78             binascii.unhexlify(self._HKEY),
  79             (license_response['cid'] + self.ie._DEVICE_ID).encode(),
  80             digestmod=hashlib.sha256)
  81         enckey = bytes_to_intlist(h.digest())
  82
  83         return intlist_to_bytes(aes_ecb_decrypt(encvideokey, enckey))
  84
  85
  86 class AbemaTVBaseIE(InfoExtractor):
  87     _NETRC_MACHINE = 'abematv'
  88
  89     _USERTOKEN = None
  90     _DEVICE_ID = None
  91     _MEDIATOKEN = None
  92
  93     _SECRETKEY = b'v+Gjs=25Aw5erR!J8ZuvRrCx*rGswhB&qdHd_SYerEWdU&a?3DzN9BRbp5KwY4hEmcj5#fykMjJ=AuWz5GSMY-d@H7DMEh3M@9n2G552Us$$k9cD=3TxwWe86!x#Zyhe'
  94
  95     @classmethod
  96     def _generate_aks(cls, deviceid):
  97         deviceid = deviceid.encode()
  98         # add 1 hour and then drop minute and secs
  99         ts_1hour = int((time_seconds() // 3600 + 1) * 3600)
 100         time_struct = time.gmtime(ts_1hour)
 101         ts_1hour_str = str(ts_1hour).encode()
 102
 103         tmp = None
 104
 105         def mix_once(nonce):
 106             nonlocal tmp
 107             h = hmac.new(cls._SECRETKEY, digestmod=hashlib.sha256)
 108             h.update(nonce)
 109             tmp = h.digest()
 110
 111         def mix_tmp(count):
 112             nonlocal tmp
 113             for _ in range(count):
 114                 mix_once(tmp)
 115
 116         def mix_twist(nonce):
 117             nonlocal tmp
 118             mix_once(base64.urlsafe_b64encode(tmp).rstrip(b'=') + nonce)
 119
 120         mix_once(cls._SECRETKEY)
 121         mix_tmp(time_struct.tm_mon)
 122         mix_twist(deviceid)
 123         mix_tmp(time_struct.tm_mday % 5)
 124         mix_twist(ts_1hour_str)
 125         mix_tmp(time_struct.tm_hour % 5)
 126
 127         return base64.urlsafe_b64encode(tmp).rstrip(b'=').decode('utf-8')
 128
 129     def _get_device_token(self):
 130         if self._USERTOKEN:
 131             return self._USERTOKEN
 132
 133         self._downloader._request_director.add_handler(AbemaLicenseRH(ie=self, logger=None))
 134
 135         username, _ = self._get_login_info()
 136         auth_cache = username and self.cache.load(self._NETRC_MACHINE, username, min_ver='2024.01.19')
 137         AbemaTVBaseIE._USERTOKEN = auth_cache and auth_cache.get('usertoken')
 138         if AbemaTVBaseIE._USERTOKEN:
 139             # try authentication with locally stored token
 140             try:
 141                 AbemaTVBaseIE._DEVICE_ID = auth_cache.get('device_id')
 142                 self._get_media_token(True)
 143                 return
 144             except ExtractorError as e:
 145                 self.report_warning(f'Failed to login with cached user token; obtaining a fresh one ({e})')
 146
 147         AbemaTVBaseIE._DEVICE_ID = str(uuid.uuid4())
 148         aks = self._generate_aks(self._DEVICE_ID)
 149         user_data = self._download_json(
 150             'https://api.abema.io/v1/users', None, note='Authorizing',
 151             data=json.dumps({
 152                 'deviceId': self._DEVICE_ID,
 153                 'applicationKeySecret': aks,
 154             }).encode(),
 155             headers={
 156                 'Content-Type': 'application/json',
 157             })
 158         AbemaTVBaseIE._USERTOKEN = user_data['token']
 159
 160         return self._USERTOKEN
 161
 162     def _get_media_token(self, invalidate=False, to_show=True):
 163         if not invalidate and self._MEDIATOKEN:
 164             return self._MEDIATOKEN
 165
 166         AbemaTVBaseIE._MEDIATOKEN = self._download_json(
 167             'https://api.abema.io/v1/media/token', None, note='Fetching media token' if to_show else False,
 168             query={
 169                 'osName': 'android',
 170                 'osVersion': '6.0.1',
 171                 'osLang': 'ja_JP',
 172                 'osTimezone': 'Asia/Tokyo',
 173                 'appId': 'tv.abema',
 174                 'appVersion': '3.27.1',
 175             }, headers={
 176                 'Authorization': f'bearer {self._get_device_token()}',
 177             })['token']
 178
 179         return self._MEDIATOKEN
 180
 181     def _perform_login(self, username, password):
 182         self._get_device_token()
 183         if self.cache.load(self._NETRC_MACHINE, username, min_ver='2024.01.19') and self._get_media_token():
 184             self.write_debug('Skipping logging in')
 185             return
 186
 187         if '@' in username:  # don't strictly check if it's email address or not
 188             ep, method = 'user/email', 'email'
 189         else:
 190             ep, method = 'oneTimePassword', 'userId'
 191
 192         login_response = self._download_json(
 193             f'https://api.abema.io/v1/auth/{ep}', None, note='Logging in',
 194             data=json.dumps({
 195                 method: username,
 196                 'password': password,
 197             }).encode(), headers={
 198                 'Authorization': f'bearer {self._get_device_token()}',
 199                 'Origin': 'https://abema.tv',
 200                 'Referer': 'https://abema.tv/',
 201                 'Content-Type': 'application/json',
 202             })
 203
 204         AbemaTVBaseIE._USERTOKEN = login_response['token']
 205         self._get_media_token(True)
 206         auth_cache = {
 207             'device_id': AbemaTVBaseIE._DEVICE_ID,
 208             'usertoken': AbemaTVBaseIE._USERTOKEN,
 209         }
 210         self.cache.store(self._NETRC_MACHINE, username, auth_cache)
 211
 212     def _call_api(self, endpoint, video_id, query=None, note='Downloading JSON metadata'):
 213         return self._download_json(
 214             f'https://api.abema.io/{endpoint}', video_id, query=query or {},
 215             note=note,
 216             headers={
 217                 'Authorization': f'bearer {self._get_device_token()}',
 218             })
 219
 220     def _extract_breadcrumb_list(self, webpage, video_id):
 221         for jld in re.finditer(
 222                 r'(?is)</span></li></ul><script[^>]+type=(["\']?)application/ld\+json\1[^>]*>(?P<json_ld>.+?)</script>',
 223                 webpage):
 224             jsonld = self._parse_json(jld.group('json_ld'), video_id, fatal=False)
 225             if traverse_obj(jsonld, '@type') != 'BreadcrumbList':
 226                 continue
 227             items = traverse_obj(jsonld, ('itemListElement', ..., 'name'))
 228             if items:
 229                 return items
 230         return []
 231
 232
 233 class AbemaTVIE(AbemaTVBaseIE):
 234     _VALID_URL = r'https?://abema\.tv/(?P<type>now-on-air|video/episode|channels/.+?/slots)/(?P<id>[^?/]+)'
 235     _TESTS = [{
 236         'url': 'https://abema.tv/video/episode/194-25_s2_p1',
 237         'info_dict': {
 238             'id': '194-25_s2_p1',
 239             'title': '第1話 「チーズケーキ」　「モーニング再び」',
 240             'series': '異世界食堂２',
 241             'season': 'シーズン2',
 242             'season_number': 2,
 243             'episode': '第1話 「チーズケーキ」　「モーニング再び」',
 244             'episode_number': 1,
 245         },
 246         'skip': 'expired',
 247     }, {
 248         'url': 'https://abema.tv/channels/anime-live2/slots/E8tvAnMJ7a9a5d',
 249         'info_dict': {
 250             'id': 'E8tvAnMJ7a9a5d',
 251             'title': 'ゆるキャン△ SEASON２ 全話一挙【無料ビデオ72時間】',
 252             'series': 'ゆるキャン△ SEASON２',
 253             'episode': 'ゆるキャン△ SEASON２ 全話一挙【無料ビデオ72時間】',
 254             'season_number': 2,
 255             'episode_number': 1,
 256             'description': 'md5:9c5a3172ae763278f9303922f0ea5b17',
 257         },
 258         'skip': 'expired',
 259     }, {
 260         'url': 'https://abema.tv/video/episode/87-877_s1282_p31047',
 261         'info_dict': {
 262             'id': 'E8tvAnMJ7a9a5d',
 263             'title': '第5話『光射す』',
 264             'description': 'md5:56d4fc1b4f7769ded5f923c55bb4695d',
 265             'thumbnail': r're:https://hayabusa\.io/.+',
 266             'series': '相棒',
 267             'episode': '第5話『光射す』',
 268         },
 269         'skip': 'expired',
 270     }, {
 271         'url': 'https://abema.tv/now-on-air/abema-anime',
 272         'info_dict': {
 273             'id': 'abema-anime',
 274             # this varies
 275             # 'title': '女子高生の無駄づかい 全話一挙【無料ビデオ72時間】',
 276             'description': 'md5:55f2e61f46a17e9230802d7bcc913d5f',
 277             'is_live': True,
 278         },
 279         'skip': 'Not supported until yt-dlp implements native live downloader OR AbemaTV can start a local HTTP server',
 280     }]
 281     _TIMETABLE = None
 282
 283     def _real_extract(self, url):
 284         # starting download using infojson from this extractor is undefined behavior,
 285         # and never be fixed in the future; you must trigger downloads by directly specifying URL.
 286         # (unless there's a way to hook before downloading by extractor)
 287         video_id, video_type = self._match_valid_url(url).group('id', 'type')
 288         headers = {
 289             'Authorization': 'Bearer ' + self._get_device_token(),
 290         }
 291         video_type = video_type.split('/')[-1]
 292
 293         webpage = self._download_webpage(url, video_id)
 294         canonical_url = self._search_regex(
 295             r'<link\s+rel="canonical"\s*href="(.+?)"', webpage, 'canonical URL',
 296             default=url)
 297         info = self._search_json_ld(webpage, video_id, default={})
 298
 299         title = self._search_regex(
 300             r'<span\s*class=".+?EpisodeTitleBlock__title">(.+?)</span>', webpage, 'title', default=None)
 301         if not title:
 302             jsonld = None
 303             for jld in re.finditer(
 304                     r'(?is)<span\s*class="com-m-Thumbnail__image">(?:</span>)?<script[^>]+type=(["\']?)application/ld\+json\1[^>]*>(?P<json_ld>.+?)</script>',
 305                     webpage):
 306                 jsonld = self._parse_json(jld.group('json_ld'), video_id, fatal=False)
 307                 if jsonld:
 308                     break
 309             if jsonld:
 310                 title = jsonld.get('caption')
 311         if not title and video_type == 'now-on-air':
 312             if not self._TIMETABLE:
 313                 # cache the timetable because it goes to 5MiB in size (!!)
 314                 self._TIMETABLE = self._download_json(
 315                     'https://api.abema.io/v1/timetable/dataSet?debug=false', video_id,
 316                     headers=headers)
 317             now = time_seconds(hours=9)
 318             for slot in self._TIMETABLE.get('slots', []):
 319                 if slot.get('channelId') != video_id:
 320                     continue
 321                 if slot['startAt'] <= now and now < slot['endAt']:
 322                     title = slot['title']
 323                     break
 324
 325         # read breadcrumb on top of page
 326         breadcrumb = self._extract_breadcrumb_list(webpage, video_id)
 327         if breadcrumb:
 328             # breadcrumb list translates to: (e.g. 1st test for this IE)
 329             # Home > Anime (genre) > Isekai Shokudo 2 (series name) > Episode 1 "Cheese cakes" "Morning again" (episode title)
 330             # hence this works
 331             info['series'] = breadcrumb[-2]
 332             info['episode'] = breadcrumb[-1]
 333             if not title:
 334                 title = info['episode']
 335
 336         description = self._html_search_regex(
 337             (r'<p\s+class="com-video-EpisodeDetailsBlock__content"><span\s+class=".+?">(.+?)</span></p><div',
 338              r'<span\s+class=".+?SlotSummary.+?">(.+?)</span></div><div'),
 339             webpage, 'description', default=None, group=1)
 340         if not description:
 341             og_desc = self._html_search_meta(
 342                 ('description', 'og:description', 'twitter:description'), webpage)
 343             if og_desc:
 344                 description = re.sub(r'''(?sx)
 345                     ^(.+?)(?:
 346                         アニメの動画を無料で見るならABEMA！| # anime
 347                         等、.+ # applies for most of categories
 348                     )?
 349                 ''', r'\1', og_desc)
 350
 351         # canonical URL may contain season and episode number
 352         mobj = re.search(r's(\d+)_p(\d+)$', canonical_url)
 353         if mobj:
 354             seri = int_or_none(mobj.group(1), default=float('inf'))
 355             epis = int_or_none(mobj.group(2), default=float('inf'))
 356             info['season_number'] = seri if seri < 100 else None
 357             # some anime like Detective Conan (though not available in AbemaTV)
 358             # has more than 1000 episodes (1026 as of 2021/11/15)
 359             info['episode_number'] = epis if epis < 2000 else None
 360
 361         is_live, m3u8_url = False, None
 362         availability = 'public'
 363         if video_type == 'now-on-air':
 364             is_live = True
 365             channel_url = 'https://api.abema.io/v1/channels'
 366             if video_id == 'news-global':
 367                 channel_url = update_url_query(channel_url, {'division': '1'})
 368             onair_channels = self._download_json(channel_url, video_id)
 369             for ch in onair_channels['channels']:
 370                 if video_id == ch['id']:
 371                     m3u8_url = ch['playback']['hls']
 372                     break
 373             else:
 374                 raise ExtractorError(f'Cannot find on-air {video_id} channel.', expected=True)
 375         elif video_type == 'episode':
 376             api_response = self._download_json(
 377                 f'https://api.abema.io/v1/video/programs/{video_id}', video_id,
 378                 note='Checking playability',
 379                 headers=headers)
 380             if not traverse_obj(api_response, ('label', 'free', {bool})):
 381                 # cannot acquire decryption key for these streams
 382                 self.report_warning('This is a premium-only stream')
 383                 availability = 'premium_only'
 384             info.update(traverse_obj(api_response, {
 385                 'series': ('series', 'title'),
 386                 'season': ('season', 'name'),
 387                 'season_number': ('season', 'sequence'),
 388                 'episode_number': ('episode', 'number'),
 389             }))
 390             if not title:
 391                 title = traverse_obj(api_response, ('episode', 'title'))
 392             if not description:
 393                 description = traverse_obj(api_response, ('episode', 'content'))
 394
 395             m3u8_url = f'https://vod-abematv.akamaized.net/program/{video_id}/playlist.m3u8'
 396         elif video_type == 'slots':
 397             api_response = self._download_json(
 398                 f'https://api.abema.io/v1/media/slots/{video_id}', video_id,
 399                 note='Checking playability',
 400                 headers=headers)
 401             if not traverse_obj(api_response, ('slot', 'flags', 'timeshiftFree'), default=False):
 402                 self.report_warning('This is a premium-only stream')
 403                 availability = 'premium_only'
 404
 405             m3u8_url = f'https://vod-abematv.akamaized.net/slot/{video_id}/playlist.m3u8'
 406         else:
 407             raise ExtractorError('Unreachable')
 408
 409         if is_live:
 410             self.report_warning("This is a livestream; yt-dlp doesn't support downloading natively, but FFmpeg cannot handle m3u8 manifests from AbemaTV")
 411             self.report_warning('Please consider using Streamlink to download these streams (https://github.com/streamlink/streamlink)')
 412         formats = self._extract_m3u8_formats(
 413             m3u8_url, video_id, ext='mp4', live=is_live)
 414
 415         info.update({
 416             'id': video_id,
 417             'title': title,
 418             'description': description,
 419             'formats': formats,
 420             'is_live': is_live,
 421             'availability': availability,
 422         })
 423         return info
 424
 425
 426 class AbemaTVTitleIE(AbemaTVBaseIE):
 427     _VALID_URL = r'https?://abema\.tv/video/title/(?P<id>[^?/]+)'
 428     _PAGE_SIZE = 25
 429
 430     _TESTS = [{
 431         'url': 'https://abema.tv/video/title/90-1597',
 432         'info_dict': {
 433             'id': '90-1597',
 434             'title': 'シャッフルアイランド',
 435         },
 436         'playlist_mincount': 2,
 437     }, {
 438         'url': 'https://abema.tv/video/title/193-132',
 439         'info_dict': {
 440             'id': '193-132',
 441             'title': '真心が届く~僕とスターのオフィス・ラブ!?~',
 442         },
 443         'playlist_mincount': 16,
 444     }, {
 445         'url': 'https://abema.tv/video/title/25-102',
 446         'info_dict': {
 447             'id': '25-102',
 448             'title': 'ソードアート・オンライン アリシゼーション',
 449         },
 450         'playlist_mincount': 24,
 451     }]
 452
 453     def _fetch_page(self, playlist_id, series_version, page):
 454         programs = self._call_api(
 455             f'v1/video/series/{playlist_id}/programs', playlist_id,
 456             note=f'Downloading page {page + 1}',
 457             query={
 458                 'seriesVersion': series_version,
 459                 'offset': str(page * self._PAGE_SIZE),
 460                 'order': 'seq',
 461                 'limit': str(self._PAGE_SIZE),
 462             })
 463         yield from (
 464             self.url_result(f'https://abema.tv/video/episode/{x}')
 465             for x in traverse_obj(programs, ('programs', ..., 'id')))
 466
 467     def _entries(self, playlist_id, series_version):
 468         return OnDemandPagedList(
 469             functools.partial(self._fetch_page, playlist_id, series_version),
 470             self._PAGE_SIZE)
 471
 472     def _real_extract(self, url):
 473         playlist_id = self._match_id(url)
 474         series_info = self._call_api(f'v1/video/series/{playlist_id}', playlist_id)
 475
 476         return self.playlist_result(
 477             self._entries(playlist_id, series_info['version']), playlist_id=playlist_id,
 478             playlist_title=series_info.get('title'),
 479             playlist_description=series_info.get('content'))