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 urllib.request
  13 import urllib.response
  14 import uuid
  15 from ..utils.networking import clean_proxies
  16 from .common import InfoExtractor
  17 from ..aes import aes_ecb_decrypt
  18 from ..utils import (
  19     ExtractorError,
  20     bytes_to_intlist,
  21     decode_base_n,
  22     int_or_none,
  23     intlist_to_bytes,
  24     OnDemandPagedList,
  25     time_seconds,
  26     traverse_obj,
  27     update_url_query,
  28 )
  29
  30
  31 def add_opener(ydl, handler):  # FIXME: Create proper API in .networking
  32     """Add a handler for opening URLs, like _download_webpage"""
  33     # https://github.com/python/cpython/blob/main/Lib/urllib/request.py#L426
  34     # https://github.com/python/cpython/blob/main/Lib/urllib/request.py#L605
  35     rh = ydl._request_director.handlers['Urllib']
  36     if 'abematv-license' in rh._SUPPORTED_URL_SCHEMES:
  37         return
  38     headers = ydl.params['http_headers'].copy()
  39     proxies = ydl.proxies.copy()
  40     clean_proxies(proxies, headers)
  41     opener = rh._get_instance(cookiejar=ydl.cookiejar, proxies=proxies)
  42     assert isinstance(opener, urllib.request.OpenerDirector)
  43     opener.add_handler(handler)
  44     rh._SUPPORTED_URL_SCHEMES = (*rh._SUPPORTED_URL_SCHEMES, 'abematv-license')
  45
  46
  47 class AbemaLicenseHandler(urllib.request.BaseHandler):
  48     handler_order = 499
  49     STRTABLE = '123456789ABCDEFGHJKLMNPQRSTUVWXYZabcdefghijkmnopqrstuvwxyz'
  50     HKEY = b'3AF0298C219469522A313570E8583005A642E73EDD58E3EA2FB7339D3DF1597E'
  51
  52     def __init__(self, ie: 'AbemaTVIE'):
  53         # the protocol that this should really handle is 'abematv-license://'
  54         # abematv_license_open is just a placeholder for development purposes
  55         # ref. https://github.com/python/cpython/blob/f4c03484da59049eb62a9bf7777b963e2267d187/Lib/urllib/request.py#L510
  56         setattr(self, 'abematv-license_open', getattr(self, 'abematv_license_open'))
  57         self.ie = ie
  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('utf-8'),
  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('utf-8'),
  80             digestmod=hashlib.sha256)
  81         enckey = bytes_to_intlist(h.digest())
  82
  83         return intlist_to_bytes(aes_ecb_decrypt(encvideokey, enckey))
  84
  85     def abematv_license_open(self, url):
  86         url = url.get_full_url() if isinstance(url, urllib.request.Request) else url
  87         ticket = urllib.parse.urlparse(url).netloc
  88         response_data = self._get_videokey_from_ticket(ticket)
  89         return urllib.response.addinfourl(io.BytesIO(response_data), headers={
  90             'Content-Length': str(len(response_data)),
  91         }, url=url, code=200)
  92
  93
  94 class AbemaTVBaseIE(InfoExtractor):
  95     _USERTOKEN = None
  96     _DEVICE_ID = None
  97     _MEDIATOKEN = None
  98
  99     _SECRETKEY = b'v+Gjs=25Aw5erR!J8ZuvRrCx*rGswhB&qdHd_SYerEWdU&a?3DzN9BRbp5KwY4hEmcj5#fykMjJ=AuWz5GSMY-d@H7DMEh3M@9n2G552Us$$k9cD=3TxwWe86!x#Zyhe'
 100
 101     @classmethod
 102     def _generate_aks(cls, deviceid):
 103         deviceid = deviceid.encode('utf-8')
 104         # add 1 hour and then drop minute and secs
 105         ts_1hour = int((time_seconds() // 3600 + 1) * 3600)
 106         time_struct = time.gmtime(ts_1hour)
 107         ts_1hour_str = str(ts_1hour).encode('utf-8')
 108
 109         tmp = None
 110
 111         def mix_once(nonce):
 112             nonlocal tmp
 113             h = hmac.new(cls._SECRETKEY, digestmod=hashlib.sha256)
 114             h.update(nonce)
 115             tmp = h.digest()
 116
 117         def mix_tmp(count):
 118             nonlocal tmp
 119             for i in range(count):
 120                 mix_once(tmp)
 121
 122         def mix_twist(nonce):
 123             nonlocal tmp
 124             mix_once(base64.urlsafe_b64encode(tmp).rstrip(b'=') + nonce)
 125
 126         mix_once(cls._SECRETKEY)
 127         mix_tmp(time_struct.tm_mon)
 128         mix_twist(deviceid)
 129         mix_tmp(time_struct.tm_mday % 5)
 130         mix_twist(ts_1hour_str)
 131         mix_tmp(time_struct.tm_hour % 5)
 132
 133         return base64.urlsafe_b64encode(tmp).rstrip(b'=').decode('utf-8')
 134
 135     def _get_device_token(self):
 136         if self._USERTOKEN:
 137             return self._USERTOKEN
 138
 139         username, _ = self._get_login_info()
 140         AbemaTVBaseIE._USERTOKEN = username and self.cache.load(self._NETRC_MACHINE, username)
 141         if AbemaTVBaseIE._USERTOKEN:
 142             # try authentication with locally stored token
 143             try:
 144                 self._get_media_token(True)
 145                 return
 146             except ExtractorError as e:
 147                 self.report_warning(f'Failed to login with cached user token; obtaining a fresh one ({e})')
 148
 149         AbemaTVBaseIE._DEVICE_ID = str(uuid.uuid4())
 150         aks = self._generate_aks(self._DEVICE_ID)
 151         user_data = self._download_json(
 152             'https://api.abema.io/v1/users', None, note='Authorizing',
 153             data=json.dumps({
 154                 'deviceId': self._DEVICE_ID,
 155                 'applicationKeySecret': aks,
 156             }).encode('utf-8'),
 157             headers={
 158                 'Content-Type': 'application/json',
 159             })
 160         AbemaTVBaseIE._USERTOKEN = user_data['token']
 161
 162         add_opener(self._downloader, AbemaLicenseHandler(self))
 163         return self._USERTOKEN
 164
 165     def _get_media_token(self, invalidate=False, to_show=True):
 166         if not invalidate and self._MEDIATOKEN:
 167             return self._MEDIATOKEN
 168
 169         AbemaTVBaseIE._MEDIATOKEN = self._download_json(
 170             'https://api.abema.io/v1/media/token', None, note='Fetching media token' if to_show else False,
 171             query={
 172                 'osName': 'android',
 173                 'osVersion': '6.0.1',
 174                 'osLang': 'ja_JP',
 175                 'osTimezone': 'Asia/Tokyo',
 176                 'appId': 'tv.abema',
 177                 'appVersion': '3.27.1'
 178             }, headers={
 179                 'Authorization': f'bearer {self._get_device_token()}',
 180             })['token']
 181
 182         return self._MEDIATOKEN
 183
 184     def _call_api(self, endpoint, video_id, query=None, note='Downloading JSON metadata'):
 185         return self._download_json(
 186             f'https://api.abema.io/{endpoint}', video_id, query=query or {},
 187             note=note,
 188             headers={
 189                 'Authorization': f'bearer {self._get_device_token()}',
 190             })
 191
 192     def _extract_breadcrumb_list(self, webpage, video_id):
 193         for jld in re.finditer(
 194                 r'(?is)</span></li></ul><script[^>]+type=(["\']?)application/ld\+json\1[^>]*>(?P<json_ld>.+?)</script>',
 195                 webpage):
 196             jsonld = self._parse_json(jld.group('json_ld'), video_id, fatal=False)
 197             if traverse_obj(jsonld, '@type') != 'BreadcrumbList':
 198                 continue
 199             items = traverse_obj(jsonld, ('itemListElement', ..., 'name'))
 200             if items:
 201                 return items
 202         return []
 203
 204
 205 class AbemaTVIE(AbemaTVBaseIE):
 206     _VALID_URL = r'https?://abema\.tv/(?P<type>now-on-air|video/episode|channels/.+?/slots)/(?P<id>[^?/]+)'
 207     _NETRC_MACHINE = 'abematv'
 208     _TESTS = [{
 209         'url': 'https://abema.tv/video/episode/194-25_s2_p1',
 210         'info_dict': {
 211             'id': '194-25_s2_p1',
 212             'title': '第1話 「チーズケーキ」　「モーニング再び」',
 213             'series': '異世界食堂２',
 214             'season': 'シーズン2',
 215             'season_number': 2,
 216             'episode': '第1話 「チーズケーキ」　「モーニング再び」',
 217             'episode_number': 1,
 218         },
 219         'skip': 'expired',
 220     }, {
 221         'url': 'https://abema.tv/channels/anime-live2/slots/E8tvAnMJ7a9a5d',
 222         'info_dict': {
 223             'id': 'E8tvAnMJ7a9a5d',
 224             'title': 'ゆるキャン△ SEASON２ 全話一挙【無料ビデオ72時間】',
 225             'series': 'ゆるキャン△ SEASON２',
 226             'episode': 'ゆるキャン△ SEASON２ 全話一挙【無料ビデオ72時間】',
 227             'series_number': 2,
 228             'episode_number': 1,
 229             'description': 'md5:9c5a3172ae763278f9303922f0ea5b17',
 230         },
 231         'skip': 'expired',
 232     }, {
 233         'url': 'https://abema.tv/video/episode/87-877_s1282_p31047',
 234         'info_dict': {
 235             'id': 'E8tvAnMJ7a9a5d',
 236             'title': '第5話『光射す』',
 237             'description': 'md5:56d4fc1b4f7769ded5f923c55bb4695d',
 238             'thumbnail': r're:https://hayabusa\.io/.+',
 239             'series': '相棒',
 240             'episode': '第5話『光射す』',
 241         },
 242         'skip': 'expired',
 243     }, {
 244         'url': 'https://abema.tv/now-on-air/abema-anime',
 245         'info_dict': {
 246             'id': 'abema-anime',
 247             # this varies
 248             # 'title': '女子高生の無駄づかい 全話一挙【無料ビデオ72時間】',
 249             'description': 'md5:55f2e61f46a17e9230802d7bcc913d5f',
 250             'is_live': True,
 251         },
 252         'skip': 'Not supported until yt-dlp implements native live downloader OR AbemaTV can start a local HTTP server',
 253     }]
 254     _TIMETABLE = None
 255
 256     def _perform_login(self, username, password):
 257         self._get_device_token()
 258         if self.cache.load(self._NETRC_MACHINE, username) and self._get_media_token():
 259             self.write_debug('Skipping logging in')
 260             return
 261
 262         if '@' in username:  # don't strictly check if it's email address or not
 263             ep, method = 'user/email', 'email'
 264         else:
 265             ep, method = 'oneTimePassword', 'userId'
 266
 267         login_response = self._download_json(
 268             f'https://api.abema.io/v1/auth/{ep}', None, note='Logging in',
 269             data=json.dumps({
 270                 method: username,
 271                 'password': password
 272             }).encode('utf-8'), headers={
 273                 'Authorization': f'bearer {self._get_device_token()}',
 274                 'Origin': 'https://abema.tv',
 275                 'Referer': 'https://abema.tv/',
 276                 'Content-Type': 'application/json',
 277             })
 278
 279         AbemaTVBaseIE._USERTOKEN = login_response['token']
 280         self._get_media_token(True)
 281         self.cache.store(self._NETRC_MACHINE, username, AbemaTVBaseIE._USERTOKEN)
 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         if video_type == 'now-on-air':
 363             is_live = True
 364             channel_url = 'https://api.abema.io/v1/channels'
 365             if video_id == 'news-global':
 366                 channel_url = update_url_query(channel_url, {'division': '1'})
 367             onair_channels = self._download_json(channel_url, video_id)
 368             for ch in onair_channels['channels']:
 369                 if video_id == ch['id']:
 370                     m3u8_url = ch['playback']['hls']
 371                     break
 372             else:
 373                 raise ExtractorError(f'Cannot find on-air {video_id} channel.', expected=True)
 374         elif video_type == 'episode':
 375             api_response = self._download_json(
 376                 f'https://api.abema.io/v1/video/programs/{video_id}', video_id,
 377                 note='Checking playability',
 378                 headers=headers)
 379             ondemand_types = traverse_obj(api_response, ('terms', ..., 'onDemandType'))
 380             if 3 not in ondemand_types:
 381                 # cannot acquire decryption key for these streams
 382                 self.report_warning('This is a premium-only stream')
 383             info.update(traverse_obj(api_response, {
 384                 'series': ('series', 'title'),
 385                 'season': ('season', 'name'),
 386                 'season_number': ('season', 'sequence'),
 387                 'episode_number': ('episode', 'number'),
 388             }))
 389             if not title:
 390                 title = traverse_obj(api_response, ('episode', 'title'))
 391             if not description:
 392                 description = traverse_obj(api_response, ('episode', 'content'))
 393
 394             m3u8_url = f'https://vod-abematv.akamaized.net/program/{video_id}/playlist.m3u8'
 395         elif video_type == 'slots':
 396             api_response = self._download_json(
 397                 f'https://api.abema.io/v1/media/slots/{video_id}', video_id,
 398                 note='Checking playability',
 399                 headers=headers)
 400             if not traverse_obj(api_response, ('slot', 'flags', 'timeshiftFree'), default=False):
 401                 self.report_warning('This is a premium-only stream')
 402
 403             m3u8_url = f'https://vod-abematv.akamaized.net/slot/{video_id}/playlist.m3u8'
 404         else:
 405             raise ExtractorError('Unreachable')
 406
 407         if is_live:
 408             self.report_warning("This is a livestream; yt-dlp doesn't support downloading natively, but FFmpeg cannot handle m3u8 manifests from AbemaTV")
 409             self.report_warning('Please consider using Streamlink to download these streams (https://github.com/streamlink/streamlink)')
 410         formats = self._extract_m3u8_formats(
 411             m3u8_url, video_id, ext='mp4', live=is_live)
 412
 413         info.update({
 414             'id': video_id,
 415             'title': title,
 416             'description': description,
 417             'formats': formats,
 418             'is_live': is_live,
 419         })
 420         return info
 421
 422
 423 class AbemaTVTitleIE(AbemaTVBaseIE):
 424     _VALID_URL = r'https?://abema\.tv/video/title/(?P<id>[^?/]+)'
 425     _PAGE_SIZE = 25
 426
 427     _TESTS = [{
 428         'url': 'https://abema.tv/video/title/90-1597',
 429         'info_dict': {
 430             'id': '90-1597',
 431             'title': 'シャッフルアイランド',
 432         },
 433         'playlist_mincount': 2,
 434     }, {
 435         'url': 'https://abema.tv/video/title/193-132',
 436         'info_dict': {
 437             'id': '193-132',
 438             'title': '真心が届く~僕とスターのオフィス・ラブ!?~',
 439         },
 440         'playlist_mincount': 16,
 441     }, {
 442         'url': 'https://abema.tv/video/title/25-102',
 443         'info_dict': {
 444             'id': '25-102',
 445             'title': 'ソードアート・オンライン アリシゼーション',
 446         },
 447         'playlist_mincount': 24,
 448     }]
 449
 450     def _fetch_page(self, playlist_id, series_version, page):
 451         programs = self._call_api(
 452             f'v1/video/series/{playlist_id}/programs', playlist_id,
 453             note=f'Downloading page {page + 1}',
 454             query={
 455                 'seriesVersion': series_version,
 456                 'offset': str(page * self._PAGE_SIZE),
 457                 'order': 'seq',
 458                 'limit': str(self._PAGE_SIZE),
 459             })
 460         yield from (
 461             self.url_result(f'https://abema.tv/video/episode/{x}')
 462             for x in traverse_obj(programs, ('programs', ..., 'id')))
 463
 464     def _entries(self, playlist_id, series_version):
 465         return OnDemandPagedList(
 466             functools.partial(self._fetch_page, playlist_id, series_version),
 467             self._PAGE_SIZE)
 468
 469     def _real_extract(self, url):
 470         playlist_id = self._match_id(url)
 471         series_info = self._call_api(f'v1/video/series/{playlist_id}', playlist_id)
 472
 473         return self.playlist_result(
 474             self._entries(playlist_id, series_info['version']), playlist_id=playlist_id,
 475             playlist_title=series_info.get('title'),
 476             playlist_description=series_info.get('content'))