yt_dlp/extractor/stageplus.py

   1 import json
   2 import uuid
   3
   4 from .common import InfoExtractor
   5 from ..utils import (
   6     float_or_none,
   7     traverse_obj,
   8     try_call,
   9     unified_timestamp,
  10     url_or_none,
  11 )
  12
  13
  14 class StagePlusVODConcertIE(InfoExtractor):
  15     _NETRC_MACHINE = 'stageplus'
  16     _VALID_URL = r'https?://(?:www\.)?stage-plus\.com/video/(?P<id>vod_concert_\w+)'
  17     _TESTS = [{
  18         'url': 'https://www.stage-plus.com/video/vod_concert_APNM8GRFDPHMASJKBSPJACG',
  19         'playlist_count': 6,
  20         'info_dict': {
  21             'id': 'vod_concert_APNM8GRFDPHMASJKBSPJACG',
  22             'title': 'Yuja Wang plays Rachmaninoff\'s Piano Concerto No. 2 – from Odeonsplatz',
  23             'description': 'md5:50f78ec180518c9bdb876bac550996fc',
  24             'artists': ['Yuja Wang', 'Lorenzo Viotti'],
  25             'upload_date': '20230331',
  26             'timestamp': 1680249600,
  27             'release_date': '20210709',
  28             'release_timestamp': 1625788800,
  29             'thumbnails': 'count:3',
  30         },
  31         'playlist': [{
  32             'info_dict': {
  33                 'id': 'performance_work_A1IN4PJFE9MM2RJ3CLBMUSJBBSOJAD9O',
  34                 'ext': 'mp4',
  35                 'title': 'Piano Concerto No. 2 in C Minor, Op. 18',
  36                 'description': 'md5:50f78ec180518c9bdb876bac550996fc',
  37                 'upload_date': '20230331',
  38                 'timestamp': 1680249600,
  39                 'release_date': '20210709',
  40                 'release_timestamp': 1625788800,
  41                 'duration': 2207,
  42                 'chapters': 'count:5',
  43                 'artists': ['Yuja Wang'],
  44                 'composers': ['Sergei Rachmaninoff'],
  45                 'album': 'Yuja Wang plays Rachmaninoff\'s Piano Concerto No. 2 – from Odeonsplatz',
  46                 'album_artists': ['Yuja Wang', 'Lorenzo Viotti'],
  47                 'track': 'Piano Concerto No. 2 in C Minor, Op. 18',
  48                 'track_number': 1,
  49                 'genre': 'Instrumental Concerto',
  50             },
  51         }],
  52         'params': {'skip_download': 'm3u8'},
  53     }]
  54
  55     # TODO: Prune this after livestream and/or album extractors are added
  56     _GRAPHQL_QUERY = '''query videoDetailPage($videoId: ID!, $sliderItemsFirst: Int = 24) {
  57   node(id: $videoId) {
  58     __typename
  59     ...LiveConcertFields
  60     ... on LiveConcert {
  61       artists {
  62         edges {
  63           role {
  64             ...RoleFields
  65           }
  66           node {
  67             id
  68             name
  69             sortName
  70           }
  71         }
  72       }
  73       isAtmos
  74       maxResolution
  75       groups {
  76         id
  77         name
  78         typeDisplayName
  79       }
  80       shortDescription
  81       performanceWorks {
  82         ...livePerformanceWorkFields
  83       }
  84       totalDuration
  85       sliders {
  86         ...contentContainerFields
  87       }
  88       vodConcert {
  89         __typename
  90         id
  91       }
  92     }
  93     ...VideoFields
  94     ... on Video {
  95       artists {
  96         edges {
  97           role {
  98             ...RoleFields
  99           }
 100           node {
 101             id
 102             name
 103             sortName
 104           }
 105         }
 106       }
 107       isAtmos
 108       maxResolution
 109       isLossless
 110       description
 111       productionDate
 112       takedownDate
 113       sliders {
 114         ...contentContainerFields
 115       }
 116     }
 117     ...VodConcertFields
 118     ... on VodConcert {
 119       artists {
 120         edges {
 121           role {
 122             ...RoleFields
 123           }
 124           node {
 125             id
 126             name
 127             sortName
 128           }
 129         }
 130       }
 131       isAtmos
 132       maxResolution
 133       groups {
 134         id
 135         name
 136         typeDisplayName
 137       }
 138       performanceWorks {
 139         ...PerformanceWorkFields
 140       }
 141       shortDescription
 142       productionDate
 143       takedownDate
 144       sliders {
 145         ...contentContainerFields
 146       }
 147     }
 148   }
 149 }
 150
 151 fragment LiveConcertFields on LiveConcert {
 152   endTime
 153   id
 154   pictures {
 155     ...PictureFields
 156   }
 157   reruns {
 158     ...liveConcertRerunFields
 159   }
 160   publicationLevel
 161   startTime
 162   streamStartTime
 163   subtitle
 164   title
 165   typeDisplayName
 166   stream {
 167     ...liveStreamFields
 168   }
 169   trailerStream {
 170     ...streamFields
 171   }
 172   geoAccessCountries
 173   geoAccessMode
 174 }
 175
 176 fragment PictureFields on Picture {
 177   id
 178   url
 179   type
 180 }
 181
 182 fragment liveConcertRerunFields on LiveConcertRerun {
 183   streamStartTime
 184   endTime
 185   startTime
 186   stream {
 187     ...rerunStreamFields
 188   }
 189 }
 190
 191 fragment rerunStreamFields on RerunStream {
 192   publicationLevel
 193   streamType
 194   url
 195 }
 196
 197 fragment liveStreamFields on LiveStream {
 198   publicationLevel
 199   streamType
 200   url
 201 }
 202
 203 fragment streamFields on Stream {
 204   publicationLevel
 205   streamType
 206   url
 207 }
 208
 209 fragment RoleFields on Role {
 210   __typename
 211   id
 212   type
 213   displayName
 214 }
 215
 216 fragment livePerformanceWorkFields on LivePerformanceWork {
 217   __typename
 218   id
 219   artists {
 220     ...artistWithRoleFields
 221   }
 222   groups {
 223     edges {
 224       node {
 225         id
 226         name
 227         typeDisplayName
 228       }
 229     }
 230   }
 231   work {
 232     ...workFields
 233   }
 234 }
 235
 236 fragment artistWithRoleFields on ArtistWithRoleConnection {
 237   edges {
 238     role {
 239       ...RoleFields
 240     }
 241     node {
 242       id
 243       name
 244       sortName
 245     }
 246   }
 247 }
 248
 249 fragment workFields on Work {
 250   id
 251   title
 252   movements {
 253     id
 254     title
 255   }
 256   composers {
 257     id
 258     name
 259   }
 260   genre {
 261     id
 262     title
 263   }
 264 }
 265
 266 fragment contentContainerFields on CuratedContentContainer {
 267   __typename
 268   ...SliderFields
 269   ...BannerFields
 270 }
 271
 272 fragment SliderFields on Slider {
 273   id
 274   headline
 275   items(first: $sliderItemsFirst) {
 276     edges {
 277       node {
 278         id
 279         __typename
 280         ...AlbumFields
 281         ...ArtistFields
 282         ...EpochFields
 283         ...GenreFields
 284         ...GroupFields
 285         ...LiveConcertFields
 286         ...PartnerFields
 287         ...PerformanceWorkFields
 288         ...VideoFields
 289         ...VodConcertFields
 290       }
 291     }
 292   }
 293 }
 294
 295 fragment AlbumFields on Album {
 296   artistAndGroupDisplayInfo
 297   id
 298   pictures {
 299     ...PictureFields
 300   }
 301   title
 302 }
 303
 304 fragment ArtistFields on Artist {
 305   id
 306   name
 307   roles {
 308     ...RoleFields
 309   }
 310   pictures {
 311     ...PictureFields
 312   }
 313 }
 314
 315 fragment EpochFields on Epoch {
 316   id
 317   endYear
 318   pictures {
 319     ...PictureFields
 320   }
 321   startYear
 322   title
 323 }
 324
 325 fragment GenreFields on Genre {
 326   id
 327   pictures {
 328     ...PictureFields
 329   }
 330   title
 331 }
 332
 333 fragment GroupFields on Group {
 334   id
 335   name
 336   typeDisplayName
 337   pictures {
 338     ...PictureFields
 339   }
 340 }
 341
 342 fragment PartnerFields on Partner {
 343   id
 344   name
 345   typeDisplayName
 346   subtypeDisplayName
 347   pictures {
 348     ...PictureFields
 349   }
 350 }
 351
 352 fragment PerformanceWorkFields on PerformanceWork {
 353   __typename
 354   id
 355   artists {
 356     ...artistWithRoleFields
 357   }
 358   groups {
 359     edges {
 360       node {
 361         id
 362         name
 363         typeDisplayName
 364       }
 365     }
 366   }
 367   work {
 368     ...workFields
 369   }
 370   stream {
 371     ...streamFields
 372   }
 373   vodConcert {
 374     __typename
 375     id
 376   }
 377   duration
 378   cuePoints {
 379     mark
 380     title
 381   }
 382 }
 383
 384 fragment VideoFields on Video {
 385   id
 386   archiveReleaseDate
 387   title
 388   subtitle
 389   pictures {
 390     ...PictureFields
 391   }
 392   stream {
 393     ...streamFields
 394   }
 395   trailerStream {
 396     ...streamFields
 397   }
 398   duration
 399   typeDisplayName
 400   duration
 401   geoAccessCountries
 402   geoAccessMode
 403   publicationLevel
 404   takedownDate
 405 }
 406
 407 fragment VodConcertFields on VodConcert {
 408   id
 409   archiveReleaseDate
 410   pictures {
 411     ...PictureFields
 412   }
 413   subtitle
 414   title
 415   typeDisplayName
 416   totalDuration
 417   geoAccessCountries
 418   geoAccessMode
 419   trailerStream {
 420    ...streamFields
 421   }
 422   publicationLevel
 423   takedownDate
 424 }
 425
 426 fragment BannerFields on Banner {
 427   description
 428   link
 429   pictures {
 430     ...PictureFields
 431   }
 432   title
 433 }'''
 434
 435     _TOKEN = None
 436
 437     def _perform_login(self, username, password):
 438         auth = self._download_json('https://audience.api.stageplus.io/oauth/token', None, headers={
 439             'Content-Type': 'application/json',
 440             'Origin': 'https://www.stage-plus.com',
 441         }, data=json.dumps({
 442             'grant_type': 'password',
 443             'username': username,
 444             'password': password,
 445             'device_info': 'Chrome (Windows)',
 446             'client_device_id': str(uuid.uuid4()),
 447         }, separators=(',', ':')).encode(), note='Logging in')
 448
 449         if auth.get('access_token'):
 450             self._TOKEN = auth['access_token']
 451
 452     def _real_initialize(self):
 453         if self._TOKEN:
 454             return
 455
 456         self._TOKEN = try_call(
 457             lambda: self._get_cookies('https://www.stage-plus.com/')['dgplus_access_token'].value)
 458         if not self._TOKEN:
 459             self.raise_login_required()
 460
 461     def _real_extract(self, url):
 462         concert_id = self._match_id(url)
 463
 464         data = self._download_json('https://audience.api.stageplus.io/graphql', concert_id, headers={
 465             'authorization': f'Bearer {self._TOKEN}',
 466             'content-type': 'application/json',
 467             'Origin': 'https://www.stage-plus.com',
 468         }, data=json.dumps({
 469             'query': self._GRAPHQL_QUERY,
 470             'variables': {'videoId': concert_id},
 471             'operationName': 'videoDetailPage',
 472         }, separators=(',', ':')).encode())['data']['node']
 473
 474         metadata = traverse_obj(data, {
 475             'title': 'title',
 476             'description': ('shortDescription', {str}),
 477             'artists': ('artists', 'edges', ..., 'node', 'name'),
 478             'timestamp': ('archiveReleaseDate', {unified_timestamp}),
 479             'release_timestamp': ('productionDate', {unified_timestamp}),
 480         })
 481
 482         thumbnails = traverse_obj(data, ('pictures', lambda _, v: url_or_none(v['url']), {
 483             'id': 'name',
 484             'url': 'url',
 485         })) or None
 486
 487         entries = []
 488         for idx, video in enumerate(traverse_obj(data, (
 489                 'performanceWorks', lambda _, v: v['id'] and url_or_none(v['stream']['url']))), 1):
 490             formats, subtitles = self._extract_m3u8_formats_and_subtitles(
 491                 video['stream']['url'], video['id'], 'mp4', m3u8_id='hls', query={'token': self._TOKEN})
 492             entries.append({
 493                 'id': video['id'],
 494                 'formats': formats,
 495                 'subtitles': subtitles,
 496                 'album': metadata.get('title'),
 497                 'album_artists': metadata.get('artist'),
 498                 'track_number': idx,
 499                 **metadata,
 500                 **traverse_obj(video, {
 501                     'title': ('work', 'title'),
 502                     'track': ('work', 'title'),
 503                     'duration': ('duration', {float_or_none}),
 504                     'chapters': (
 505                         'cuePoints', lambda _, v: float_or_none(v['mark']) is not None, {
 506                             'title': 'title',
 507                             'start_time': ('mark', {float_or_none}),
 508                         }),
 509                     'artists': ('artists', 'edges', ..., 'node', 'name'),
 510                     'composers': ('work', 'composers', ..., 'name'),
 511                     'genre': ('work', 'genre', 'title'),
 512                 }),
 513             })
 514
 515         return self.playlist_result(entries, concert_id, thumbnails=thumbnails, **metadata)