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
  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     request_to_url,
  26     time_seconds,
  27     traverse_obj,
  28     update_url_query,
  29 )
  30
  31 # NOTE: network handler related code is temporary thing until network stack overhaul PRs are merged (#2861/#2862)
  32
  33
  34 def add_opener(ydl, handler):
  35     ''' Add a handler for opening URLs, like _download_webpage '''
  36     # https://github.com/python/cpython/blob/main/Lib/urllib/request.py#L426
  37     # https://github.com/python/cpython/blob/main/Lib/urllib/request.py#L605
  38     assert isinstance(ydl._opener, urllib.request.OpenerDirector)
  39     ydl._opener.add_handler(handler)
  40
  41
  42 def remove_opener(ydl, handler):
  43     '''
  44     Remove handler(s) for opening URLs
  45     @param handler Either handler object itself or handler type.
  46     Specifying handler type will remove all handler which isinstance returns True.
  47     '''
  48     # https://github.com/python/cpython/blob/main/Lib/urllib/request.py#L426
  49     # https://github.com/python/cpython/blob/main/Lib/urllib/request.py#L605
  50     opener = ydl._opener
  51     assert isinstance(ydl._opener, urllib.request.OpenerDirector)
  52     if isinstance(handler, (type, tuple)):
  53         find_cp = lambda x: isinstance(x, handler)
  54     else:
  55         find_cp = lambda x: x is handler
  56
  57     removed = []
  58     for meth in dir(handler):
  59         if meth in ["redirect_request", "do_open", "proxy_open"]:
  60             # oops, coincidental match
  61             continue
  62
  63         i = meth.find("_")
  64         protocol = meth[:i]
  65         condition = meth[i + 1:]
  66
  67         if condition.startswith("error"):
  68             j = condition.find("_") + i + 1
  69             kind = meth[j + 1:]
  70             try:
  71                 kind = int(kind)
  72             except ValueError:
  73                 pass
  74             lookup = opener.handle_error.get(protocol, {})
  75             opener.handle_error[protocol] = lookup
  76         elif condition == "open":
  77             kind = protocol
  78             lookup = opener.handle_open
  79         elif condition == "response":
  80             kind = protocol
  81             lookup = opener.process_response
  82         elif condition == "request":
  83             kind = protocol
  84             lookup = opener.process_request
  85         else:
  86             continue
  87
  88         handlers = lookup.setdefault(kind, [])
  89         if handlers:
  90             handlers[:] = [x for x in handlers if not find_cp(x)]
  91
  92         removed.append(x for x in handlers if find_cp(x))
  93
  94     if removed:
  95         for x in opener.handlers:
  96             if find_cp(x):
  97                 x.add_parent(None)
  98         opener.handlers[:] = [x for x in opener.handlers if not find_cp(x)]
  99
 100
 101 class AbemaLicenseHandler(urllib.request.BaseHandler):
 102     handler_order = 499
 103     STRTABLE = '123456789ABCDEFGHJKLMNPQRSTUVWXYZabcdefghijkmnopqrstuvwxyz'
 104     HKEY = b'3AF0298C219469522A313570E8583005A642E73EDD58E3EA2FB7339D3DF1597E'
 105
 106     def __init__(self, ie: 'AbemaTVIE'):
 107         # the protocol that this should really handle is 'abematv-license://'
 108         # abematv_license_open is just a placeholder for development purposes
 109         # ref. https://github.com/python/cpython/blob/f4c03484da59049eb62a9bf7777b963e2267d187/Lib/urllib/request.py#L510
 110         setattr(self, 'abematv-license_open', getattr(self, 'abematv_license_open'))
 111         self.ie = ie
 112
 113     def _get_videokey_from_ticket(self, ticket):
 114         to_show = self.ie.get_param('verbose', False)
 115         media_token = self.ie._get_media_token(to_show=to_show)
 116
 117         license_response = self.ie._download_json(
 118             'https://license.abema.io/abematv-hls', None, note='Requesting playback license' if to_show else False,
 119             query={'t': media_token},
 120             data=json.dumps({
 121                 'kv': 'a',
 122                 'lt': ticket
 123             }).encode('utf-8'),
 124             headers={
 125                 'Content-Type': 'application/json',
 126             })
 127
 128         res = decode_base_n(license_response['k'], table=self.STRTABLE)
 129         encvideokey = bytes_to_intlist(struct.pack('>QQ', res >> 64, res & 0xffffffffffffffff))
 130
 131         h = hmac.new(
 132             binascii.unhexlify(self.HKEY),
 133             (license_response['cid'] + self.ie._DEVICE_ID).encode('utf-8'),
 134             digestmod=hashlib.sha256)
 135         enckey = bytes_to_intlist(h.digest())
 136
 137         return intlist_to_bytes(aes_ecb_decrypt(encvideokey, enckey))
 138
 139     def abematv_license_open(self, url):
 140         url = request_to_url(url)
 141         ticket = urllib.parse.urlparse(url).netloc
 142         response_data = self._get_videokey_from_ticket(ticket)
 143         return urllib.response.addinfourl(io.BytesIO(response_data), headers={
 144             'Content-Length': len(response_data),
 145         }, url=url, code=200)
 146
 147
 148 class AbemaTVBaseIE(InfoExtractor):
 149     _USERTOKEN = None
 150     _DEVICE_ID = None
 151     _MEDIATOKEN = None
 152
 153     _SECRETKEY = b'v+Gjs=25Aw5erR!J8ZuvRrCx*rGswhB&qdHd_SYerEWdU&a?3DzN9BRbp5KwY4hEmcj5#fykMjJ=AuWz5GSMY-d@H7DMEh3M@9n2G552Us$$k9cD=3TxwWe86!x#Zyhe'
 154
 155     @classmethod
 156     def _generate_aks(cls, deviceid):
 157         deviceid = deviceid.encode('utf-8')
 158         # add 1 hour and then drop minute and secs
 159         ts_1hour = int((time_seconds(hours=9) // 3600 + 1) * 3600)
 160         time_struct = time.gmtime(ts_1hour)
 161         ts_1hour_str = str(ts_1hour).encode('utf-8')
 162
 163         tmp = None
 164
 165         def mix_once(nonce):
 166             nonlocal tmp
 167             h = hmac.new(cls._SECRETKEY, digestmod=hashlib.sha256)
 168             h.update(nonce)
 169             tmp = h.digest()
 170
 171         def mix_tmp(count):
 172             nonlocal tmp
 173             for i in range(count):
 174                 mix_once(tmp)
 175
 176         def mix_twist(nonce):
 177             nonlocal tmp
 178             mix_once(base64.urlsafe_b64encode(tmp).rstrip(b'=') + nonce)
 179
 180         mix_once(cls._SECRETKEY)
 181         mix_tmp(time_struct.tm_mon)
 182         mix_twist(deviceid)
 183         mix_tmp(time_struct.tm_mday % 5)
 184         mix_twist(ts_1hour_str)
 185         mix_tmp(time_struct.tm_hour % 5)
 186
 187         return base64.urlsafe_b64encode(tmp).rstrip(b'=').decode('utf-8')
 188
 189     def _get_device_token(self):
 190         if self._USERTOKEN:
 191             return self._USERTOKEN
 192
 193         AbemaTVBaseIE._DEVICE_ID = str(uuid.uuid4())
 194         aks = self._generate_aks(self._DEVICE_ID)
 195         user_data = self._download_json(
 196             'https://api.abema.io/v1/users', None, note='Authorizing',
 197             data=json.dumps({
 198                 'deviceId': self._DEVICE_ID,
 199                 'applicationKeySecret': aks,
 200             }).encode('utf-8'),
 201             headers={
 202                 'Content-Type': 'application/json',
 203             })
 204         AbemaTVBaseIE._USERTOKEN = user_data['token']
 205
 206         # don't allow adding it 2 times or more, though it's guarded
 207         remove_opener(self._downloader, AbemaLicenseHandler)
 208         add_opener(self._downloader, AbemaLicenseHandler(self))
 209
 210         return self._USERTOKEN
 211
 212     def _get_media_token(self, invalidate=False, to_show=True):
 213         if not invalidate and self._MEDIATOKEN:
 214             return self._MEDIATOKEN
 215
 216         AbemaTVBaseIE._MEDIATOKEN = self._download_json(
 217             'https://api.abema.io/v1/media/token', None, note='Fetching media token' if to_show else False,
 218             query={
 219                 'osName': 'android',
 220                 'osVersion': '6.0.1',
 221                 'osLang': 'ja_JP',
 222                 'osTimezone': 'Asia/Tokyo',
 223                 'appId': 'tv.abema',
 224                 'appVersion': '3.27.1'
 225             }, headers={
 226                 'Authorization': f'bearer {self._get_device_token()}',
 227             })['token']
 228
 229         return self._MEDIATOKEN
 230
 231     def _call_api(self, endpoint, video_id, query=None, note='Downloading JSON metadata'):
 232         return self._download_json(
 233             f'https://api.abema.io/{endpoint}', video_id, query=query or {},
 234             note=note,
 235             headers={
 236                 'Authorization': f'bearer {self._get_device_token()}',
 237             })
 238
 239     def _extract_breadcrumb_list(self, webpage, video_id):
 240         for jld in re.finditer(
 241                 r'(?is)</span></li></ul><script[^>]+type=(["\']?)application/ld\+json\1[^>]*>(?P<json_ld>.+?)</script>',
 242                 webpage):
 243             jsonld = self._parse_json(jld.group('json_ld'), video_id, fatal=False)
 244             if traverse_obj(jsonld, '@type') != 'BreadcrumbList':
 245                 continue
 246             items = traverse_obj(jsonld, ('itemListElement', ..., 'name'))
 247             if items:
 248                 return items
 249         return []
 250
 251
 252 class AbemaTVIE(AbemaTVBaseIE):
 253     _VALID_URL = r'https?://abema\.tv/(?P<type>now-on-air|video/episode|channels/.+?/slots)/(?P<id>[^?/]+)'
 254     _NETRC_MACHINE = 'abematv'
 255     _TESTS = [{
 256         'url': 'https://abema.tv/video/episode/194-25_s2_p1',
 257         'info_dict': {
 258             'id': '194-25_s2_p1',
 259             'title': '第1話 「チーズケーキ」　「モーニング再び」',
 260             'series': '異世界食堂２',
 261             'series_number': 2,
 262             'episode': '第1話 「チーズケーキ」　「モーニング再び」',
 263             'episode_number': 1,
 264         },
 265         'skip': 'expired',
 266     }, {
 267         'url': 'https://abema.tv/channels/anime-live2/slots/E8tvAnMJ7a9a5d',
 268         'info_dict': {
 269             'id': 'E8tvAnMJ7a9a5d',
 270             'title': 'ゆるキャン△ SEASON２ 全話一挙【無料ビデオ72時間】',
 271             'series': 'ゆるキャン△ SEASON２',
 272             'episode': 'ゆるキャン△ SEASON２ 全話一挙【無料ビデオ72時間】',
 273             'series_number': 2,
 274             'episode_number': 1,
 275             'description': 'md5:9c5a3172ae763278f9303922f0ea5b17',
 276         },
 277         'skip': 'expired',
 278     }, {
 279         'url': 'https://abema.tv/video/episode/87-877_s1282_p31047',
 280         'info_dict': {
 281             'id': 'E8tvAnMJ7a9a5d',
 282             'title': '第5話『光射す』',
 283             'description': 'md5:56d4fc1b4f7769ded5f923c55bb4695d',
 284             'thumbnail': r're:https://hayabusa\.io/.+',
 285             'series': '相棒',
 286             'episode': '第5話『光射す』',
 287         },
 288         'skip': 'expired',
 289     }, {
 290         'url': 'https://abema.tv/now-on-air/abema-anime',
 291         'info_dict': {
 292             'id': 'abema-anime',
 293             # this varies
 294             # 'title': '女子高生の無駄づかい 全話一挙【無料ビデオ72時間】',
 295             'description': 'md5:55f2e61f46a17e9230802d7bcc913d5f',
 296             'is_live': True,
 297         },
 298         'skip': 'Not supported until yt-dlp implements native live downloader OR AbemaTV can start a local HTTP server',
 299     }]
 300     _TIMETABLE = None
 301
 302     def _perform_login(self, username, password):
 303         if '@' in username:  # don't strictly check if it's email address or not
 304             ep, method = 'user/email', 'email'
 305         else:
 306             ep, method = 'oneTimePassword', 'userId'
 307
 308         login_response = self._download_json(
 309             f'https://api.abema.io/v1/auth/{ep}', None, note='Logging in',
 310             data=json.dumps({
 311                 method: username,
 312                 'password': password
 313             }).encode('utf-8'), headers={
 314                 'Authorization': f'bearer {self._get_device_token()}',
 315                 'Origin': 'https://abema.tv',
 316                 'Referer': 'https://abema.tv/',
 317                 'Content-Type': 'application/json',
 318             })
 319
 320         AbemaTVBaseIE._USERTOKEN = login_response['token']
 321         self._get_media_token(True)
 322
 323     def _real_extract(self, url):
 324         # starting download using infojson from this extractor is undefined behavior,
 325         # and never be fixed in the future; you must trigger downloads by directly specifying URL.
 326         # (unless there's a way to hook before downloading by extractor)
 327         video_id, video_type = self._match_valid_url(url).group('id', 'type')
 328         headers = {
 329             'Authorization': 'Bearer ' + self._get_device_token(),
 330         }
 331         video_type = video_type.split('/')[-1]
 332
 333         webpage = self._download_webpage(url, video_id)
 334         canonical_url = self._search_regex(
 335             r'<link\s+rel="canonical"\s*href="(.+?)"', webpage, 'canonical URL',
 336             default=url)
 337         info = self._search_json_ld(webpage, video_id, default={})
 338
 339         title = self._search_regex(
 340             r'<span\s*class=".+?EpisodeTitleBlock__title">(.+?)</span>', webpage, 'title', default=None)
 341         if not title:
 342             jsonld = None
 343             for jld in re.finditer(
 344                     r'(?is)<span\s*class="com-m-Thumbnail__image">(?:</span>)?<script[^>]+type=(["\']?)application/ld\+json\1[^>]*>(?P<json_ld>.+?)</script>',
 345                     webpage):
 346                 jsonld = self._parse_json(jld.group('json_ld'), video_id, fatal=False)
 347                 if jsonld:
 348                     break
 349             if jsonld:
 350                 title = jsonld.get('caption')
 351         if not title and video_type == 'now-on-air':
 352             if not self._TIMETABLE:
 353                 # cache the timetable because it goes to 5MiB in size (!!)
 354                 self._TIMETABLE = self._download_json(
 355                     'https://api.abema.io/v1/timetable/dataSet?debug=false', video_id,
 356                     headers=headers)
 357             now = time_seconds(hours=9)
 358             for slot in self._TIMETABLE.get('slots', []):
 359                 if slot.get('channelId') != video_id:
 360                     continue
 361                 if slot['startAt'] <= now and now < slot['endAt']:
 362                     title = slot['title']
 363                     break
 364
 365         # read breadcrumb on top of page
 366         breadcrumb = self._extract_breadcrumb_list(webpage, video_id)
 367         if breadcrumb:
 368             # breadcrumb list translates to: (example is 1st test for this IE)
 369             # Home > Anime (genre) > Isekai Shokudo 2 (series name) > Episode 1 "Cheese cakes" "Morning again" (episode title)
 370             # hence this works
 371             info['series'] = breadcrumb[-2]
 372             info['episode'] = breadcrumb[-1]
 373             if not title:
 374                 title = info['episode']
 375
 376         description = self._html_search_regex(
 377             (r'<p\s+class="com-video-EpisodeDetailsBlock__content"><span\s+class=".+?">(.+?)</span></p><div',
 378              r'<span\s+class=".+?SlotSummary.+?">(.+?)</span></div><div',),
 379             webpage, 'description', default=None, group=1)
 380         if not description:
 381             og_desc = self._html_search_meta(
 382                 ('description', 'og:description', 'twitter:description'), webpage)
 383             if og_desc:
 384                 description = re.sub(r'''(?sx)
 385                     ^(.+?)(?:
 386                         アニメの動画を無料で見るならABEMA！| # anime
 387                         等、.+ # applies for most of categories
 388                     )?
 389                 ''', r'\1', og_desc)
 390
 391         # canonical URL may contain series and episode number
 392         mobj = re.search(r's(\d+)_p(\d+)$', canonical_url)
 393         if mobj:
 394             seri = int_or_none(mobj.group(1), default=float('inf'))
 395             epis = int_or_none(mobj.group(2), default=float('inf'))
 396             info['series_number'] = seri if seri < 100 else None
 397             # some anime like Detective Conan (though not available in AbemaTV)
 398             # has more than 1000 episodes (1026 as of 2021/11/15)
 399             info['episode_number'] = epis if epis < 2000 else None
 400
 401         is_live, m3u8_url = False, None
 402         if video_type == 'now-on-air':
 403             is_live = True
 404             channel_url = 'https://api.abema.io/v1/channels'
 405             if video_id == 'news-global':
 406                 channel_url = update_url_query(channel_url, {'division': '1'})
 407             onair_channels = self._download_json(channel_url, video_id)
 408             for ch in onair_channels['channels']:
 409                 if video_id == ch['id']:
 410                     m3u8_url = ch['playback']['hls']
 411                     break
 412             else:
 413                 raise ExtractorError(f'Cannot find on-air {video_id} channel.', expected=True)
 414         elif video_type == 'episode':
 415             api_response = self._download_json(
 416                 f'https://api.abema.io/v1/video/programs/{video_id}', video_id,
 417                 note='Checking playability',
 418                 headers=headers)
 419             ondemand_types = traverse_obj(api_response, ('terms', ..., 'onDemandType'), default=[])
 420             if 3 not in ondemand_types:
 421                 # cannot acquire decryption key for these streams
 422                 self.report_warning('This is a premium-only stream')
 423
 424             m3u8_url = f'https://vod-abematv.akamaized.net/program/{video_id}/playlist.m3u8'
 425         elif video_type == 'slots':
 426             api_response = self._download_json(
 427                 f'https://api.abema.io/v1/media/slots/{video_id}', video_id,
 428                 note='Checking playability',
 429                 headers=headers)
 430             if not traverse_obj(api_response, ('slot', 'flags', 'timeshiftFree'), default=False):
 431                 self.report_warning('This is a premium-only stream')
 432
 433             m3u8_url = f'https://vod-abematv.akamaized.net/slot/{video_id}/playlist.m3u8'
 434         else:
 435             raise ExtractorError('Unreachable')
 436
 437         if is_live:
 438             self.report_warning("This is a livestream; yt-dlp doesn't support downloading natively, but FFmpeg cannot handle m3u8 manifests from AbemaTV")
 439             self.report_warning('Please consider using Streamlink to download these streams (https://github.com/streamlink/streamlink)')
 440         formats = self._extract_m3u8_formats(
 441             m3u8_url, video_id, ext='mp4', live=is_live)
 442
 443         info.update({
 444             'id': video_id,
 445             'title': title,
 446             'description': description,
 447             'formats': formats,
 448             'is_live': is_live,
 449         })
 450         return info
 451
 452
 453 class AbemaTVTitleIE(AbemaTVBaseIE):
 454     _VALID_URL = r'https?://abema\.tv/video/title/(?P<id>[^?/]+)'
 455     _PAGE_SIZE = 25
 456
 457     _TESTS = [{
 458         'url': 'https://abema.tv/video/title/90-1597',
 459         'info_dict': {
 460             'id': '90-1597',
 461             'title': 'シャッフルアイランド',
 462         },
 463         'playlist_mincount': 2,
 464     }, {
 465         'url': 'https://abema.tv/video/title/193-132',
 466         'info_dict': {
 467             'id': '193-132',
 468             'title': '真心が届く~僕とスターのオフィス・ラブ!?~',
 469         },
 470         'playlist_mincount': 16,
 471     }, {
 472         'url': 'https://abema.tv/video/title/25-102',
 473         'info_dict': {
 474             'id': '25-102',
 475             'title': 'ソードアート・オンライン アリシゼーション',
 476         },
 477         'playlist_mincount': 24,
 478     }]
 479
 480     def _fetch_page(self, playlist_id, series_version, page):
 481         programs = self._call_api(
 482             f'v1/video/series/{playlist_id}/programs', playlist_id,
 483             note=f'Downloading page {page + 1}',
 484             query={
 485                 'seriesVersion': series_version,
 486                 'offset': str(page * self._PAGE_SIZE),
 487                 'order': 'seq',
 488                 'limit': str(self._PAGE_SIZE),
 489             })
 490         yield from (
 491             self.url_result(f'https://abema.tv/video/episode/{x}')
 492             for x in traverse_obj(programs, ('programs', ..., 'id'), default=[]))
 493
 494     def _entries(self, playlist_id, series_version):
 495         return OnDemandPagedList(
 496             functools.partial(self._fetch_page, playlist_id, series_version),
 497             self._PAGE_SIZE)
 498
 499     def _real_extract(self, url):
 500         playlist_id = self._match_id(url)
 501         series_info = self._call_api(f'v1/video/series/{playlist_id}', playlist_id)
 502
 503         return self.playlist_result(
 504             self._entries(playlist_id, series_info['version']), playlist_id=playlist_id,
 505             playlist_title=series_info.get('title'),
 506             playlist_description=series_info.get('content'))