yt_dlp/extractor/abematv.py

   1 import base64
   2 import binascii
   3 import hashlib
   4 import hmac
   5 import io
   6 import json
   7 import re
   8 import struct
   9 import time
  10 import urllib.parse
  11 import urllib.request
  12 import urllib.response
  13 import uuid
  14
  15 from .common import InfoExtractor
  16 from ..aes import aes_ecb_decrypt
  17 from ..utils import (
  18     ExtractorError,
  19     bytes_to_intlist,
  20     decode_base_n,
  21     int_or_none,
  22     intlist_to_bytes,
  23     request_to_url,
  24     time_seconds,
  25     traverse_obj,
  26     update_url_query,
  27     urljoin,
  28 )
  29
  30 # NOTE: network handler related code is temporary thing until network stack overhaul PRs are merged (#2861/#2862)
  31
  32
  33 def add_opener(ydl, handler):
  34     ''' Add a handler for opening URLs, like _download_webpage '''
  35     # https://github.com/python/cpython/blob/main/Lib/urllib/request.py#L426
  36     # https://github.com/python/cpython/blob/main/Lib/urllib/request.py#L605
  37     assert isinstance(ydl._opener, urllib.request.OpenerDirector)
  38     ydl._opener.add_handler(handler)
  39
  40
  41 def remove_opener(ydl, handler):
  42     '''
  43     Remove handler(s) for opening URLs
  44     @param handler Either handler object itself or handler type.
  45     Specifying handler type will remove all handler which isinstance returns True.
  46     '''
  47     # https://github.com/python/cpython/blob/main/Lib/urllib/request.py#L426
  48     # https://github.com/python/cpython/blob/main/Lib/urllib/request.py#L605
  49     opener = ydl._opener
  50     assert isinstance(ydl._opener, urllib.request.OpenerDirector)
  51     if isinstance(handler, (type, tuple)):
  52         find_cp = lambda x: isinstance(x, handler)
  53     else:
  54         find_cp = lambda x: x is handler
  55
  56     removed = []
  57     for meth in dir(handler):
  58         if meth in ["redirect_request", "do_open", "proxy_open"]:
  59             # oops, coincidental match
  60             continue
  61
  62         i = meth.find("_")
  63         protocol = meth[:i]
  64         condition = meth[i + 1:]
  65
  66         if condition.startswith("error"):
  67             j = condition.find("_") + i + 1
  68             kind = meth[j + 1:]
  69             try:
  70                 kind = int(kind)
  71             except ValueError:
  72                 pass
  73             lookup = opener.handle_error.get(protocol, {})
  74             opener.handle_error[protocol] = lookup
  75         elif condition == "open":
  76             kind = protocol
  77             lookup = opener.handle_open
  78         elif condition == "response":
  79             kind = protocol
  80             lookup = opener.process_response
  81         elif condition == "request":
  82             kind = protocol
  83             lookup = opener.process_request
  84         else:
  85             continue
  86
  87         handlers = lookup.setdefault(kind, [])
  88         if handlers:
  89             handlers[:] = [x for x in handlers if not find_cp(x)]
  90
  91         removed.append(x for x in handlers if find_cp(x))
  92
  93     if removed:
  94         for x in opener.handlers:
  95             if find_cp(x):
  96                 x.add_parent(None)
  97         opener.handlers[:] = [x for x in opener.handlers if not find_cp(x)]
  98
  99
 100 class AbemaLicenseHandler(urllib.request.BaseHandler):
 101     handler_order = 499
 102     STRTABLE = '123456789ABCDEFGHJKLMNPQRSTUVWXYZabcdefghijkmnopqrstuvwxyz'
 103     HKEY = b'3AF0298C219469522A313570E8583005A642E73EDD58E3EA2FB7339D3DF1597E'
 104
 105     def __init__(self, ie: 'AbemaTVIE'):
 106         # the protcol that this should really handle is 'abematv-license://'
 107         # abematv_license_open is just a placeholder for development purposes
 108         # ref. https://github.com/python/cpython/blob/f4c03484da59049eb62a9bf7777b963e2267d187/Lib/urllib/request.py#L510
 109         setattr(self, 'abematv-license_open', getattr(self, 'abematv_license_open'))
 110         self.ie = ie
 111
 112     def _get_videokey_from_ticket(self, ticket):
 113         to_show = self.ie.get_param('verbose', False)
 114         media_token = self.ie._get_media_token(to_show=to_show)
 115
 116         license_response = self.ie._download_json(
 117             'https://license.abema.io/abematv-hls', None, note='Requesting playback license' if to_show else False,
 118             query={'t': media_token},
 119             data=json.dumps({
 120                 'kv': 'a',
 121                 'lt': ticket
 122             }).encode('utf-8'),
 123             headers={
 124                 'Content-Type': 'application/json',
 125             })
 126
 127         res = decode_base_n(license_response['k'], table=self.STRTABLE)
 128         encvideokey = bytes_to_intlist(struct.pack('>QQ', res >> 64, res & 0xffffffffffffffff))
 129
 130         h = hmac.new(
 131             binascii.unhexlify(self.HKEY),
 132             (license_response['cid'] + self.ie._DEVICE_ID).encode('utf-8'),
 133             digestmod=hashlib.sha256)
 134         enckey = bytes_to_intlist(h.digest())
 135
 136         return intlist_to_bytes(aes_ecb_decrypt(encvideokey, enckey))
 137
 138     def abematv_license_open(self, url):
 139         url = request_to_url(url)
 140         ticket = urllib.parse.urlparse(url).netloc
 141         response_data = self._get_videokey_from_ticket(ticket)
 142         return urllib.response.addinfourl(io.BytesIO(response_data), headers={
 143             'Content-Length': len(response_data),
 144         }, url=url, code=200)
 145
 146
 147 class AbemaTVBaseIE(InfoExtractor):
 148     def _extract_breadcrumb_list(self, webpage, video_id):
 149         for jld in re.finditer(
 150                 r'(?is)</span></li></ul><script[^>]+type=(["\']?)application/ld\+json\1[^>]*>(?P<json_ld>.+?)</script>',
 151                 webpage):
 152             jsonld = self._parse_json(jld.group('json_ld'), video_id, fatal=False)
 153             if jsonld:
 154                 if jsonld.get('@type') != 'BreadcrumbList':
 155                     continue
 156                 trav = traverse_obj(jsonld, ('itemListElement', ..., 'name'))
 157                 if trav:
 158                     return trav
 159         return []
 160
 161
 162 class AbemaTVIE(AbemaTVBaseIE):
 163     _VALID_URL = r'https?://abema\.tv/(?P<type>now-on-air|video/episode|channels/.+?/slots)/(?P<id>[^?/]+)'
 164     _NETRC_MACHINE = 'abematv'
 165     _TESTS = [{
 166         'url': 'https://abema.tv/video/episode/194-25_s2_p1',
 167         'info_dict': {
 168             'id': '194-25_s2_p1',
 169             'title': '第1話 「チーズケーキ」　「モーニング再び」',
 170             'series': '異世界食堂２',
 171             'series_number': 2,
 172             'episode': '第1話 「チーズケーキ」　「モーニング再び」',
 173             'episode_number': 1,
 174         },
 175         'skip': 'expired',
 176     }, {
 177         'url': 'https://abema.tv/channels/anime-live2/slots/E8tvAnMJ7a9a5d',
 178         'info_dict': {
 179             'id': 'E8tvAnMJ7a9a5d',
 180             'title': 'ゆるキャン△ SEASON２ 全話一挙【無料ビデオ72時間】',
 181             'series': 'ゆるキャン△ SEASON２',
 182             'episode': 'ゆるキャン△ SEASON２ 全話一挙【無料ビデオ72時間】',
 183             'series_number': 2,
 184             'episode_number': 1,
 185             'description': 'md5:9c5a3172ae763278f9303922f0ea5b17',
 186         },
 187         'skip': 'expired',
 188     }, {
 189         'url': 'https://abema.tv/video/episode/87-877_s1282_p31047',
 190         'info_dict': {
 191             'id': 'E8tvAnMJ7a9a5d',
 192             'title': '第5話『光射す』',
 193             'description': 'md5:56d4fc1b4f7769ded5f923c55bb4695d',
 194             'thumbnail': r're:https://hayabusa\.io/.+',
 195             'series': '相棒',
 196             'episode': '第5話『光射す』',
 197         },
 198         'skip': 'expired',
 199     }, {
 200         'url': 'https://abema.tv/now-on-air/abema-anime',
 201         'info_dict': {
 202             'id': 'abema-anime',
 203             # this varies
 204             # 'title': '女子高生の無駄づかい 全話一挙【無料ビデオ72時間】',
 205             'description': 'md5:55f2e61f46a17e9230802d7bcc913d5f',
 206             'is_live': True,
 207         },
 208         'skip': 'Not supported until yt-dlp implements native live downloader OR AbemaTV can start a local HTTP server',
 209     }]
 210     _USERTOKEN = None
 211     _DEVICE_ID = None
 212     _TIMETABLE = None
 213     _MEDIATOKEN = None
 214
 215     _SECRETKEY = b'v+Gjs=25Aw5erR!J8ZuvRrCx*rGswhB&qdHd_SYerEWdU&a?3DzN9BRbp5KwY4hEmcj5#fykMjJ=AuWz5GSMY-d@H7DMEh3M@9n2G552Us$$k9cD=3TxwWe86!x#Zyhe'
 216
 217     def _generate_aks(self, deviceid):
 218         deviceid = deviceid.encode('utf-8')
 219         # add 1 hour and then drop minute and secs
 220         ts_1hour = int((time_seconds(hours=9) // 3600 + 1) * 3600)
 221         time_struct = time.gmtime(ts_1hour)
 222         ts_1hour_str = str(ts_1hour).encode('utf-8')
 223
 224         tmp = None
 225
 226         def mix_once(nonce):
 227             nonlocal tmp
 228             h = hmac.new(self._SECRETKEY, digestmod=hashlib.sha256)
 229             h.update(nonce)
 230             tmp = h.digest()
 231
 232         def mix_tmp(count):
 233             nonlocal tmp
 234             for i in range(count):
 235                 mix_once(tmp)
 236
 237         def mix_twist(nonce):
 238             nonlocal tmp
 239             mix_once(base64.urlsafe_b64encode(tmp).rstrip(b'=') + nonce)
 240
 241         mix_once(self._SECRETKEY)
 242         mix_tmp(time_struct.tm_mon)
 243         mix_twist(deviceid)
 244         mix_tmp(time_struct.tm_mday % 5)
 245         mix_twist(ts_1hour_str)
 246         mix_tmp(time_struct.tm_hour % 5)
 247
 248         return base64.urlsafe_b64encode(tmp).rstrip(b'=').decode('utf-8')
 249
 250     def _get_device_token(self):
 251         if self._USERTOKEN:
 252             return self._USERTOKEN
 253
 254         self._DEVICE_ID = str(uuid.uuid4())
 255         aks = self._generate_aks(self._DEVICE_ID)
 256         user_data = self._download_json(
 257             'https://api.abema.io/v1/users', None, note='Authorizing',
 258             data=json.dumps({
 259                 'deviceId': self._DEVICE_ID,
 260                 'applicationKeySecret': aks,
 261             }).encode('utf-8'),
 262             headers={
 263                 'Content-Type': 'application/json',
 264             })
 265         self._USERTOKEN = user_data['token']
 266
 267         # don't allow adding it 2 times or more, though it's guarded
 268         remove_opener(self._downloader, AbemaLicenseHandler)
 269         add_opener(self._downloader, AbemaLicenseHandler(self))
 270
 271         return self._USERTOKEN
 272
 273     def _get_media_token(self, invalidate=False, to_show=True):
 274         if not invalidate and self._MEDIATOKEN:
 275             return self._MEDIATOKEN
 276
 277         self._MEDIATOKEN = self._download_json(
 278             'https://api.abema.io/v1/media/token', None, note='Fetching media token' if to_show else False,
 279             query={
 280                 'osName': 'android',
 281                 'osVersion': '6.0.1',
 282                 'osLang': 'ja_JP',
 283                 'osTimezone': 'Asia/Tokyo',
 284                 'appId': 'tv.abema',
 285                 'appVersion': '3.27.1'
 286             }, headers={
 287                 'Authorization': 'bearer ' + self._get_device_token()
 288             })['token']
 289
 290         return self._MEDIATOKEN
 291
 292     def _perform_login(self, username, password):
 293         if '@' in username:  # don't strictly check if it's email address or not
 294             ep, method = 'user/email', 'email'
 295         else:
 296             ep, method = 'oneTimePassword', 'userId'
 297
 298         login_response = self._download_json(
 299             f'https://api.abema.io/v1/auth/{ep}', None, note='Logging in',
 300             data=json.dumps({
 301                 method: username,
 302                 'password': password
 303             }).encode('utf-8'), headers={
 304                 'Authorization': 'bearer ' + self._get_device_token(),
 305                 'Origin': 'https://abema.tv',
 306                 'Referer': 'https://abema.tv/',
 307                 'Content-Type': 'application/json',
 308             })
 309
 310         self._USERTOKEN = login_response['token']
 311         self._get_media_token(True)
 312
 313     def _real_extract(self, url):
 314         # starting download using infojson from this extractor is undefined behavior,
 315         # and never be fixed in the future; you must trigger downloads by directly specifing URL.
 316         # (unless there's a way to hook before downloading by extractor)
 317         video_id, video_type = self._match_valid_url(url).group('id', 'type')
 318         headers = {
 319             'Authorization': 'Bearer ' + self._get_device_token(),
 320         }
 321         video_type = video_type.split('/')[-1]
 322
 323         webpage = self._download_webpage(url, video_id)
 324         canonical_url = self._search_regex(
 325             r'<link\s+rel="canonical"\s*href="(.+?)"', webpage, 'canonical URL',
 326             default=url)
 327         info = self._search_json_ld(webpage, video_id, default={})
 328
 329         title = self._search_regex(
 330             r'<span\s*class=".+?EpisodeTitleBlock__title">(.+?)</span>', webpage, 'title', default=None)
 331         if not title:
 332             jsonld = None
 333             for jld in re.finditer(
 334                     r'(?is)<span\s*class="com-m-Thumbnail__image">(?:</span>)?<script[^>]+type=(["\']?)application/ld\+json\1[^>]*>(?P<json_ld>.+?)</script>',
 335                     webpage):
 336                 jsonld = self._parse_json(jld.group('json_ld'), video_id, fatal=False)
 337                 if jsonld:
 338                     break
 339             if jsonld:
 340                 title = jsonld.get('caption')
 341         if not title and video_type == 'now-on-air':
 342             if not self._TIMETABLE:
 343                 # cache the timetable because it goes to 5MiB in size (!!)
 344                 self._TIMETABLE = self._download_json(
 345                     'https://api.abema.io/v1/timetable/dataSet?debug=false', video_id,
 346                     headers=headers)
 347             now = time_seconds(hours=9)
 348             for slot in self._TIMETABLE.get('slots', []):
 349                 if slot.get('channelId') != video_id:
 350                     continue
 351                 if slot['startAt'] <= now and now < slot['endAt']:
 352                     title = slot['title']
 353                     break
 354
 355         # read breadcrumb on top of page
 356         breadcrumb = self._extract_breadcrumb_list(webpage, video_id)
 357         if breadcrumb:
 358             # breadcrumb list translates to: (example is 1st test for this IE)
 359             # Home > Anime (genre) > Isekai Shokudo 2 (series name) > Episode 1 "Cheese cakes" "Morning again" (episode title)
 360             # hence this works
 361             info['series'] = breadcrumb[-2]
 362             info['episode'] = breadcrumb[-1]
 363             if not title:
 364                 title = info['episode']
 365
 366         description = self._html_search_regex(
 367             (r'<p\s+class="com-video-EpisodeDetailsBlock__content"><span\s+class=".+?">(.+?)</span></p><div',
 368              r'<span\s+class=".+?SlotSummary.+?">(.+?)</span></div><div',),
 369             webpage, 'description', default=None, group=1)
 370         if not description:
 371             og_desc = self._html_search_meta(
 372                 ('description', 'og:description', 'twitter:description'), webpage)
 373             if og_desc:
 374                 description = re.sub(r'''(?sx)
 375                     ^(.+?)(?:
 376                         アニメの動画を無料で見るならABEMA！| # anime
 377                         等、.+ # applies for most of categories
 378                     )?
 379                 ''', r'\1', og_desc)
 380
 381         # canonical URL may contain series and episode number
 382         mobj = re.search(r's(\d+)_p(\d+)$', canonical_url)
 383         if mobj:
 384             seri = int_or_none(mobj.group(1), default=float('inf'))
 385             epis = int_or_none(mobj.group(2), default=float('inf'))
 386             info['series_number'] = seri if seri < 100 else None
 387             # some anime like Detective Conan (though not available in AbemaTV)
 388             # has more than 1000 episodes (1026 as of 2021/11/15)
 389             info['episode_number'] = epis if epis < 2000 else None
 390
 391         is_live, m3u8_url = False, None
 392         if video_type == 'now-on-air':
 393             is_live = True
 394             channel_url = 'https://api.abema.io/v1/channels'
 395             if video_id == 'news-global':
 396                 channel_url = update_url_query(channel_url, {'division': '1'})
 397             onair_channels = self._download_json(channel_url, video_id)
 398             for ch in onair_channels['channels']:
 399                 if video_id == ch['id']:
 400                     m3u8_url = ch['playback']['hls']
 401                     break
 402             else:
 403                 raise ExtractorError(f'Cannot find on-air {video_id} channel.', expected=True)
 404         elif video_type == 'episode':
 405             api_response = self._download_json(
 406                 f'https://api.abema.io/v1/video/programs/{video_id}', video_id,
 407                 note='Checking playability',
 408                 headers=headers)
 409             ondemand_types = traverse_obj(api_response, ('terms', ..., 'onDemandType'), default=[])
 410             if 3 not in ondemand_types:
 411                 # cannot acquire decryption key for these streams
 412                 self.report_warning('This is a premium-only stream')
 413
 414             m3u8_url = f'https://vod-abematv.akamaized.net/program/{video_id}/playlist.m3u8'
 415         elif video_type == 'slots':
 416             api_response = self._download_json(
 417                 f'https://api.abema.io/v1/media/slots/{video_id}', video_id,
 418                 note='Checking playability',
 419                 headers=headers)
 420             if not traverse_obj(api_response, ('slot', 'flags', 'timeshiftFree'), default=False):
 421                 self.report_warning('This is a premium-only stream')
 422
 423             m3u8_url = f'https://vod-abematv.akamaized.net/slot/{video_id}/playlist.m3u8'
 424         else:
 425             raise ExtractorError('Unreachable')
 426
 427         if is_live:
 428             self.report_warning("This is a livestream; yt-dlp doesn't support downloading natively, but FFmpeg cannot handle m3u8 manifests from AbemaTV")
 429             self.report_warning('Please consider using Streamlink to download these streams (https://github.com/streamlink/streamlink)')
 430         formats = self._extract_m3u8_formats(
 431             m3u8_url, video_id, ext='mp4', live=is_live)
 432
 433         info.update({
 434             'id': video_id,
 435             'title': title,
 436             'description': description,
 437             'formats': formats,
 438             'is_live': is_live,
 439         })
 440         return info
 441
 442
 443 class AbemaTVTitleIE(AbemaTVBaseIE):
 444     _VALID_URL = r'https?://abema\.tv/video/title/(?P<id>[^?/]+)'
 445
 446     _TESTS = [{
 447         'url': 'https://abema.tv/video/title/90-1597',
 448         'info_dict': {
 449             'id': '90-1597',
 450             'title': 'シャッフルアイランド',
 451         },
 452         'playlist_mincount': 2,
 453     }, {
 454         'url': 'https://abema.tv/video/title/193-132',
 455         'info_dict': {
 456             'id': '193-132',
 457             'title': '真心が届く~僕とスターのオフィス・ラブ!?~',
 458         },
 459         'playlist_mincount': 16,
 460     }]
 461
 462     def _real_extract(self, url):
 463         video_id = self._match_id(url)
 464         webpage = self._download_webpage(url, video_id)
 465
 466         playlist_title, breadcrumb = None, self._extract_breadcrumb_list(webpage, video_id)
 467         if breadcrumb:
 468             playlist_title = breadcrumb[-1]
 469
 470         playlist = [
 471             self.url_result(urljoin('https://abema.tv/', mobj.group(1)))
 472             for mobj in re.finditer(r'<li\s*class=".+?EpisodeList.+?"><a\s*href="(/[^"]+?)"', webpage)]
 473
 474         return self.playlist_result(playlist, playlist_title=playlist_title, playlist_id=video_id)