yt_dlp/extractor/abematv.py

   1 import io
   2 import json
   3 import time
   4 import hashlib
   5 import hmac
   6 import re
   7 import struct
   8 from base64 import urlsafe_b64encode
   9 from binascii import unhexlify
  10
  11 from .common import InfoExtractor
  12 from ..aes import aes_ecb_decrypt
  13 from ..compat import (
  14     compat_urllib_response,
  15     compat_urllib_parse_urlparse,
  16     compat_urllib_request,
  17 )
  18 from ..utils import (
  19     ExtractorError,
  20     decode_base,
  21     int_or_none,
  22     random_uuidv4,
  23     request_to_url,
  24     time_seconds,
  25     update_url_query,
  26     traverse_obj,
  27     intlist_to_bytes,
  28     bytes_to_intlist,
  29     urljoin,
  30 )
  31
  32
  33 # NOTE: network handler related code is temporary thing until network stack overhaul PRs are merged (#2861/#2862)
  34
  35 def add_opener(ydl, handler):
  36     ''' Add a handler for opening URLs, like _download_webpage '''
  37     # https://github.com/python/cpython/blob/main/Lib/urllib/request.py#L426
  38     # https://github.com/python/cpython/blob/main/Lib/urllib/request.py#L605
  39     assert isinstance(ydl._opener, compat_urllib_request.OpenerDirector)
  40     ydl._opener.add_handler(handler)
  41
  42
  43 def remove_opener(ydl, handler):
  44     '''
  45     Remove handler(s) for opening URLs
  46     @param handler Either handler object itself or handler type.
  47     Specifying handler type will remove all handler which isinstance returns True.
  48     '''
  49     # https://github.com/python/cpython/blob/main/Lib/urllib/request.py#L426
  50     # https://github.com/python/cpython/blob/main/Lib/urllib/request.py#L605
  51     opener = ydl._opener
  52     assert isinstance(ydl._opener, compat_urllib_request.OpenerDirector)
  53     if isinstance(handler, (type, tuple)):
  54         find_cp = lambda x: isinstance(x, handler)
  55     else:
  56         find_cp = lambda x: x is handler
  57
  58     removed = []
  59     for meth in dir(handler):
  60         if meth in ["redirect_request", "do_open", "proxy_open"]:
  61             # oops, coincidental match
  62             continue
  63
  64         i = meth.find("_")
  65         protocol = meth[:i]
  66         condition = meth[i + 1:]
  67
  68         if condition.startswith("error"):
  69             j = condition.find("_") + i + 1
  70             kind = meth[j + 1:]
  71             try:
  72                 kind = int(kind)
  73             except ValueError:
  74                 pass
  75             lookup = opener.handle_error.get(protocol, {})
  76             opener.handle_error[protocol] = lookup
  77         elif condition == "open":
  78             kind = protocol
  79             lookup = opener.handle_open
  80         elif condition == "response":
  81             kind = protocol
  82             lookup = opener.process_response
  83         elif condition == "request":
  84             kind = protocol
  85             lookup = opener.process_request
  86         else:
  87             continue
  88
  89         handlers = lookup.setdefault(kind, [])
  90         if handlers:
  91             handlers[:] = [x for x in handlers if not find_cp(x)]
  92
  93         removed.append(x for x in handlers if find_cp(x))
  94
  95     if removed:
  96         for x in opener.handlers:
  97             if find_cp(x):
  98                 x.add_parent(None)
  99         opener.handlers[:] = [x for x in opener.handlers if not find_cp(x)]
 100
 101
 102 class AbemaLicenseHandler(compat_urllib_request.BaseHandler):
 103     handler_order = 499
 104     STRTABLE = '123456789ABCDEFGHJKLMNPQRSTUVWXYZabcdefghijkmnopqrstuvwxyz'
 105     HKEY = b'3AF0298C219469522A313570E8583005A642E73EDD58E3EA2FB7339D3DF1597E'
 106
 107     def __init__(self, ie: 'AbemaTVIE'):
 108         # the protcol that this should really handle is 'abematv-license://'
 109         # abematv_license_open is just a placeholder for development purposes
 110         # ref. https://github.com/python/cpython/blob/f4c03484da59049eb62a9bf7777b963e2267d187/Lib/urllib/request.py#L510
 111         setattr(self, 'abematv-license_open', getattr(self, 'abematv_license_open'))
 112         self.ie = ie
 113
 114     def _get_videokey_from_ticket(self, ticket):
 115         to_show = self.ie._downloader.params.get('verbose', False)
 116         media_token = self.ie._get_media_token(to_show=to_show)
 117
 118         license_response = self.ie._download_json(
 119             'https://license.abema.io/abematv-hls', None, note='Requesting playback license' if to_show else False,
 120             query={'t': media_token},
 121             data=json.dumps({
 122                 'kv': 'a',
 123                 'lt': ticket
 124             }).encode('utf-8'),
 125             headers={
 126                 'Content-Type': 'application/json',
 127             })
 128
 129         res = decode_base(license_response['k'], self.STRTABLE)
 130         encvideokey = bytes_to_intlist(struct.pack('>QQ', res >> 64, res & 0xffffffffffffffff))
 131
 132         h = hmac.new(
 133             unhexlify(self.HKEY),
 134             (license_response['cid'] + self.ie._DEVICE_ID).encode('utf-8'),
 135             digestmod=hashlib.sha256)
 136         enckey = bytes_to_intlist(h.digest())
 137
 138         return intlist_to_bytes(aes_ecb_decrypt(encvideokey, enckey))
 139
 140     def abematv_license_open(self, url):
 141         url = request_to_url(url)
 142         ticket = compat_urllib_parse_urlparse(url).netloc
 143         response_data = self._get_videokey_from_ticket(ticket)
 144         return compat_urllib_response.addinfourl(io.BytesIO(response_data), headers={
 145             'Content-Length': len(response_data),
 146         }, url=url, code=200)
 147
 148
 149 class AbemaTVBaseIE(InfoExtractor):
 150     def _extract_breadcrumb_list(self, webpage, video_id):
 151         for jld in re.finditer(
 152                 r'(?is)</span></li></ul><script[^>]+type=(["\']?)application/ld\+json\1[^>]*>(?P<json_ld>.+?)</script>',
 153                 webpage):
 154             jsonld = self._parse_json(jld.group('json_ld'), video_id, fatal=False)
 155             if jsonld:
 156                 if jsonld.get('@type') != 'BreadcrumbList':
 157                     continue
 158                 trav = traverse_obj(jsonld, ('itemListElement', ..., 'name'))
 159                 if trav:
 160                     return trav
 161         return []
 162
 163
 164 class AbemaTVIE(AbemaTVBaseIE):
 165     _VALID_URL = r'https?://abema\.tv/(?P<type>now-on-air|video/episode|channels/.+?/slots)/(?P<id>[^?/]+)'
 166     _NETRC_MACHINE = 'abematv'
 167     _TESTS = [{
 168         'url': 'https://abema.tv/video/episode/194-25_s2_p1',
 169         'info_dict': {
 170             'id': '194-25_s2_p1',
 171             'title': '第1話 「チーズケーキ」　「モーニング再び」',
 172             'series': '異世界食堂２',
 173             'series_number': 2,
 174             'episode': '第1話 「チーズケーキ」　「モーニング再び」',
 175             'episode_number': 1,
 176         },
 177         'skip': 'expired',
 178     }, {
 179         'url': 'https://abema.tv/channels/anime-live2/slots/E8tvAnMJ7a9a5d',
 180         'info_dict': {
 181             'id': 'E8tvAnMJ7a9a5d',
 182             'title': 'ゆるキャン△ SEASON２ 全話一挙【無料ビデオ72時間】',
 183             'series': 'ゆるキャン△ SEASON２',
 184             'episode': 'ゆるキャン△ SEASON２ 全話一挙【無料ビデオ72時間】',
 185             'series_number': 2,
 186             'episode_number': 1,
 187             'description': 'md5:9c5a3172ae763278f9303922f0ea5b17',
 188         },
 189         'skip': 'expired',
 190     }, {
 191         'url': 'https://abema.tv/video/episode/87-877_s1282_p31047',
 192         'info_dict': {
 193             'id': 'E8tvAnMJ7a9a5d',
 194             'title': '第5話『光射す』',
 195             'description': 'md5:56d4fc1b4f7769ded5f923c55bb4695d',
 196             'thumbnail': r're:https://hayabusa\.io/.+',
 197             'series': '相棒',
 198             'episode': '第5話『光射す』',
 199         },
 200         'skip': 'expired',
 201     }, {
 202         'url': 'https://abema.tv/now-on-air/abema-anime',
 203         'info_dict': {
 204             'id': 'abema-anime',
 205             # this varies
 206             # 'title': '女子高生の無駄づかい 全話一挙【無料ビデオ72時間】',
 207             'description': 'md5:55f2e61f46a17e9230802d7bcc913d5f',
 208             'is_live': True,
 209         },
 210         'skip': 'Not supported until yt-dlp implements native live downloader OR AbemaTV can start a local HTTP server',
 211     }]
 212     _USERTOKEN = None
 213     _DEVICE_ID = None
 214     _TIMETABLE = None
 215     _MEDIATOKEN = None
 216
 217     _SECRETKEY = b'v+Gjs=25Aw5erR!J8ZuvRrCx*rGswhB&qdHd_SYerEWdU&a?3DzN9BRbp5KwY4hEmcj5#fykMjJ=AuWz5GSMY-d@H7DMEh3M@9n2G552Us$$k9cD=3TxwWe86!x#Zyhe'
 218
 219     def _generate_aks(self, deviceid):
 220         deviceid = deviceid.encode('utf-8')
 221         # add 1 hour and then drop minute and secs
 222         ts_1hour = int((time_seconds(hours=9) // 3600 + 1) * 3600)
 223         time_struct = time.gmtime(ts_1hour)
 224         ts_1hour_str = str(ts_1hour).encode('utf-8')
 225
 226         tmp = None
 227
 228         def mix_once(nonce):
 229             nonlocal tmp
 230             h = hmac.new(self._SECRETKEY, digestmod=hashlib.sha256)
 231             h.update(nonce)
 232             tmp = h.digest()
 233
 234         def mix_tmp(count):
 235             nonlocal tmp
 236             for i in range(count):
 237                 mix_once(tmp)
 238
 239         def mix_twist(nonce):
 240             nonlocal tmp
 241             mix_once(urlsafe_b64encode(tmp).rstrip(b'=') + nonce)
 242
 243         mix_once(self._SECRETKEY)
 244         mix_tmp(time_struct.tm_mon)
 245         mix_twist(deviceid)
 246         mix_tmp(time_struct.tm_mday % 5)
 247         mix_twist(ts_1hour_str)
 248         mix_tmp(time_struct.tm_hour % 5)
 249
 250         return urlsafe_b64encode(tmp).rstrip(b'=').decode('utf-8')
 251
 252     def _get_device_token(self):
 253         if self._USERTOKEN:
 254             return self._USERTOKEN
 255
 256         self._DEVICE_ID = random_uuidv4()
 257         aks = self._generate_aks(self._DEVICE_ID)
 258         user_data = self._download_json(
 259             'https://api.abema.io/v1/users', None, note='Authorizing',
 260             data=json.dumps({
 261                 'deviceId': self._DEVICE_ID,
 262                 'applicationKeySecret': aks,
 263             }).encode('utf-8'),
 264             headers={
 265                 'Content-Type': 'application/json',
 266             })
 267         self._USERTOKEN = user_data['token']
 268
 269         # don't allow adding it 2 times or more, though it's guarded
 270         remove_opener(self._downloader, AbemaLicenseHandler)
 271         add_opener(self._downloader, AbemaLicenseHandler(self))
 272
 273         return self._USERTOKEN
 274
 275     def _get_media_token(self, invalidate=False, to_show=True):
 276         if not invalidate and self._MEDIATOKEN:
 277             return self._MEDIATOKEN
 278
 279         self._MEDIATOKEN = self._download_json(
 280             'https://api.abema.io/v1/media/token', None, note='Fetching media token' if to_show else False,
 281             query={
 282                 'osName': 'android',
 283                 'osVersion': '6.0.1',
 284                 'osLang': 'ja_JP',
 285                 'osTimezone': 'Asia/Tokyo',
 286                 'appId': 'tv.abema',
 287                 'appVersion': '3.27.1'
 288             }, headers={
 289                 'Authorization': 'bearer ' + self._get_device_token()
 290             })['token']
 291
 292         return self._MEDIATOKEN
 293
 294     def _perform_login(self, username, password):
 295         if '@' in username:  # don't strictly check if it's email address or not
 296             ep, method = 'user/email', 'email'
 297         else:
 298             ep, method = 'oneTimePassword', 'userId'
 299
 300         login_response = self._download_json(
 301             f'https://api.abema.io/v1/auth/{ep}', None, note='Logging in',
 302             data=json.dumps({
 303                 method: username,
 304                 'password': password
 305             }).encode('utf-8'), headers={
 306                 'Authorization': 'bearer ' + self._get_device_token(),
 307                 'Origin': 'https://abema.tv',
 308                 'Referer': 'https://abema.tv/',
 309                 'Content-Type': 'application/json',
 310             })
 311
 312         self._USERTOKEN = login_response['token']
 313         self._get_media_token(True)
 314
 315     def _real_extract(self, url):
 316         # starting download using infojson from this extractor is undefined behavior,
 317         # and never be fixed in the future; you must trigger downloads by directly specifing URL.
 318         # (unless there's a way to hook before downloading by extractor)
 319         video_id, video_type = self._match_valid_url(url).group('id', 'type')
 320         headers = {
 321             'Authorization': 'Bearer ' + self._get_device_token(),
 322         }
 323         video_type = video_type.split('/')[-1]
 324
 325         webpage = self._download_webpage(url, video_id)
 326         canonical_url = self._search_regex(
 327             r'<link\s+rel="canonical"\s*href="(.+?)"', webpage, 'canonical URL',
 328             default=url)
 329         info = self._search_json_ld(webpage, video_id, default={})
 330
 331         title = self._search_regex(
 332             r'<span\s*class=".+?EpisodeTitleBlock__title">(.+?)</span>', webpage, 'title', default=None)
 333         if not title:
 334             jsonld = None
 335             for jld in re.finditer(
 336                     r'(?is)<span\s*class="com-m-Thumbnail__image">(?:</span>)?<script[^>]+type=(["\']?)application/ld\+json\1[^>]*>(?P<json_ld>.+?)</script>',
 337                     webpage):
 338                 jsonld = self._parse_json(jld.group('json_ld'), video_id, fatal=False)
 339                 if jsonld:
 340                     break
 341             if jsonld:
 342                 title = jsonld.get('caption')
 343         if not title and video_type == 'now-on-air':
 344             if not self._TIMETABLE:
 345                 # cache the timetable because it goes to 5MiB in size (!!)
 346                 self._TIMETABLE = self._download_json(
 347                     'https://api.abema.io/v1/timetable/dataSet?debug=false', video_id,
 348                     headers=headers)
 349             now = time_seconds(hours=9)
 350             for slot in self._TIMETABLE.get('slots', []):
 351                 if slot.get('channelId') != video_id:
 352                     continue
 353                 if slot['startAt'] <= now and now < slot['endAt']:
 354                     title = slot['title']
 355                     break
 356
 357         # read breadcrumb on top of page
 358         breadcrumb = self._extract_breadcrumb_list(webpage, video_id)
 359         if breadcrumb:
 360             # breadcrumb list translates to: (example is 1st test for this IE)
 361             # Home > Anime (genre) > Isekai Shokudo 2 (series name) > Episode 1 "Cheese cakes" "Morning again" (episode title)
 362             # hence this works
 363             info['series'] = breadcrumb[-2]
 364             info['episode'] = breadcrumb[-1]
 365             if not title:
 366                 title = info['episode']
 367
 368         description = self._html_search_regex(
 369             (r'<p\s+class="com-video-EpisodeDetailsBlock__content"><span\s+class=".+?">(.+?)</span></p><div',
 370              r'<span\s+class=".+?SlotSummary.+?">(.+?)</span></div><div',),
 371             webpage, 'description', default=None, group=1)
 372         if not description:
 373             og_desc = self._html_search_meta(
 374                 ('description', 'og:description', 'twitter:description'), webpage)
 375             if og_desc:
 376                 description = re.sub(r'''(?sx)
 377                     ^(.+?)(?:
 378                         アニメの動画を無料で見るならABEMA！| # anime
 379                         等、.+ # applies for most of categories
 380                     )?
 381                 ''', r'\1', og_desc)
 382
 383         # canonical URL may contain series and episode number
 384         mobj = re.search(r's(\d+)_p(\d+)$', canonical_url)
 385         if mobj:
 386             seri = int_or_none(mobj.group(1), default=float('inf'))
 387             epis = int_or_none(mobj.group(2), default=float('inf'))
 388             info['series_number'] = seri if seri < 100 else None
 389             # some anime like Detective Conan (though not available in AbemaTV)
 390             # has more than 1000 episodes (1026 as of 2021/11/15)
 391             info['episode_number'] = epis if epis < 2000 else None
 392
 393         is_live, m3u8_url = False, None
 394         if video_type == 'now-on-air':
 395             is_live = True
 396             channel_url = 'https://api.abema.io/v1/channels'
 397             if video_id == 'news-global':
 398                 channel_url = update_url_query(channel_url, {'division': '1'})
 399             onair_channels = self._download_json(channel_url, video_id)
 400             for ch in onair_channels['channels']:
 401                 if video_id == ch['id']:
 402                     m3u8_url = ch['playback']['hls']
 403                     break
 404             else:
 405                 raise ExtractorError(f'Cannot find on-air {video_id} channel.', expected=True)
 406         elif video_type == 'episode':
 407             api_response = self._download_json(
 408                 f'https://api.abema.io/v1/video/programs/{video_id}', video_id,
 409                 note='Checking playability',
 410                 headers=headers)
 411             ondemand_types = traverse_obj(api_response, ('terms', ..., 'onDemandType'), default=[])
 412             if 3 not in ondemand_types:
 413                 # cannot acquire decryption key for these streams
 414                 self.report_warning('This is a premium-only stream')
 415
 416             m3u8_url = f'https://vod-abematv.akamaized.net/program/{video_id}/playlist.m3u8'
 417         elif video_type == 'slots':
 418             api_response = self._download_json(
 419                 f'https://api.abema.io/v1/media/slots/{video_id}', video_id,
 420                 note='Checking playability',
 421                 headers=headers)
 422             if not traverse_obj(api_response, ('slot', 'flags', 'timeshiftFree'), default=False):
 423                 self.report_warning('This is a premium-only stream')
 424
 425             m3u8_url = f'https://vod-abematv.akamaized.net/slot/{video_id}/playlist.m3u8'
 426         else:
 427             raise ExtractorError('Unreachable')
 428
 429         if is_live:
 430             self.report_warning("This is a livestream; yt-dlp doesn't support downloading natively, but FFmpeg cannot handle m3u8 manifests from AbemaTV")
 431             self.report_warning('Please consider using Streamlink to download these streams (https://github.com/streamlink/streamlink)')
 432         formats = self._extract_m3u8_formats(
 433             m3u8_url, video_id, ext='mp4', live=is_live)
 434
 435         info.update({
 436             'id': video_id,
 437             'title': title,
 438             'description': description,
 439             'formats': formats,
 440             'is_live': is_live,
 441         })
 442         return info
 443
 444
 445 class AbemaTVTitleIE(AbemaTVBaseIE):
 446     _VALID_URL = r'https?://abema\.tv/video/title/(?P<id>[^?/]+)'
 447
 448     _TESTS = [{
 449         'url': 'https://abema.tv/video/title/90-1597',
 450         'info_dict': {
 451             'id': '90-1597',
 452             'title': 'シャッフルアイランド',
 453         },
 454         'playlist_mincount': 2,
 455     }, {
 456         'url': 'https://abema.tv/video/title/193-132',
 457         'info_dict': {
 458             'id': '193-132',
 459             'title': '真心が届く~僕とスターのオフィス・ラブ!?~',
 460         },
 461         'playlist_mincount': 16,
 462     }]
 463
 464     def _real_extract(self, url):
 465         video_id = self._match_id(url)
 466         webpage = self._download_webpage(url, video_id)
 467
 468         playlist_title, breadcrumb = None, self._extract_breadcrumb_list(webpage, video_id)
 469         if breadcrumb:
 470             playlist_title = breadcrumb[-1]
 471
 472         playlist = [
 473             self.url_result(urljoin('https://abema.tv/', mobj.group(1)))
 474             for mobj in re.finditer(r'<li\s*class=".+?EpisodeList.+?"><a\s*href="(/[^"]+?)"', webpage)]
 475
 476         return self.playlist_result(playlist, playlist_title=playlist_title, playlist_id=video_id)