yt_dlp/extractor/bandlab.py

   1 from .common import InfoExtractor
   2 from ..utils import (
   3     ExtractorError,
   4     float_or_none,
   5     format_field,
   6     int_or_none,
   7     parse_iso8601,
   8     parse_qs,
   9     truncate_string,
  10     url_or_none,
  11 )
  12 from ..utils.traversal import traverse_obj, value
  13
  14
  15 class BandlabBaseIE(InfoExtractor):
  16     def _call_api(self, endpoint, asset_id, **kwargs):
  17         headers = kwargs.pop('headers', None) or {}
  18         return self._download_json(
  19             f'https://www.bandlab.com/api/v1.3/{endpoint}/{asset_id}',
  20             asset_id, headers={
  21                 'accept': 'application/json',
  22                 'referer': 'https://www.bandlab.com/',
  23                 'x-client-id': 'BandLab-Web',
  24                 'x-client-version': '10.1.124',
  25                 **headers,
  26             }, **kwargs)
  27
  28     def _parse_revision(self, revision_data, url=None):
  29         return {
  30             'vcodec': 'none',
  31             'media_type': 'revision',
  32             'extractor_key': BandlabIE.ie_key(),
  33             'extractor': BandlabIE.IE_NAME,
  34             **traverse_obj(revision_data, {
  35                 'webpage_url': (
  36                     'id', ({value(url)}, {format_field(template='https://www.bandlab.com/revision/%s')}), filter, any),
  37                 'id': (('revisionId', 'id'), {str}, any),
  38                 'title': ('song', 'name', {str}),
  39                 'track': ('song', 'name', {str}),
  40                 'url': ('mixdown', 'file', {url_or_none}),
  41                 'thumbnail': ('song', 'picture', 'url', {url_or_none}),
  42                 'description': ('description', {str}),
  43                 'uploader': ('creator', 'name', {str}),
  44                 'uploader_id': ('creator', 'username', {str}),
  45                 'timestamp': ('createdOn', {parse_iso8601}),
  46                 'duration': ('mixdown', 'duration', {float_or_none}),
  47                 'view_count': ('counters', 'plays', {int_or_none}),
  48                 'like_count': ('counters', 'likes', {int_or_none}),
  49                 'comment_count': ('counters', 'comments', {int_or_none}),
  50                 'genres': ('genres', ..., 'name', {str}),
  51             }),
  52         }
  53
  54     def _parse_track(self, track_data, url=None):
  55         return {
  56             'vcodec': 'none',
  57             'media_type': 'track',
  58             'extractor_key': BandlabIE.ie_key(),
  59             'extractor': BandlabIE.IE_NAME,
  60             **traverse_obj(track_data, {
  61                 'webpage_url': (
  62                     'id', ({value(url)}, {format_field(template='https://www.bandlab.com/post/%s')}), filter, any),
  63                 'id': (('revisionId', 'id'), {str}, any),
  64                 'url': ('track', 'sample', 'audioUrl', {url_or_none}),
  65                 'title': ('track', 'name', {str}),
  66                 'track': ('track', 'name', {str}),
  67                 'description': ('caption', {str}),
  68                 'thumbnail': ('track', 'picture', ('original', 'url'), {url_or_none}, any),
  69                 'view_count': ('counters', 'plays', {int_or_none}),
  70                 'like_count': ('counters', 'likes', {int_or_none}),
  71                 'comment_count': ('counters', 'comments', {int_or_none}),
  72                 'duration': ('track', 'sample', 'duration', {float_or_none}),
  73                 'uploader': ('creator', 'name', {str}),
  74                 'uploader_id': ('creator', 'username', {str}),
  75                 'timestamp': ('createdOn', {parse_iso8601}),
  76             }),
  77         }
  78
  79     def _parse_video(self, video_data, url=None):
  80         return {
  81             'media_type': 'video',
  82             'extractor_key': BandlabIE.ie_key(),
  83             'extractor': BandlabIE.IE_NAME,
  84             **traverse_obj(video_data, {
  85                 'id': ('id', {str}),
  86                 'webpage_url': (
  87                     'id', ({value(url)}, {format_field(template='https://www.bandlab.com/post/%s')}), filter, any),
  88                 'url': ('video', 'url', {url_or_none}),
  89                 'title': ('caption', {lambda x: x.replace('\n', ' ')}, {truncate_string(left=50)}),
  90                 'description': ('caption', {str}),
  91                 'thumbnail': ('video', 'picture', 'url', {url_or_none}),
  92                 'view_count': ('video', 'counters', 'plays', {int_or_none}),
  93                 'like_count': ('video', 'counters', 'likes', {int_or_none}),
  94                 'comment_count': ('counters', 'comments', {int_or_none}),
  95                 'duration': ('video', 'duration', {float_or_none}),
  96                 'uploader': ('creator', 'name', {str}),
  97                 'uploader_id': ('creator', 'username', {str}),
  98             }),
  99         }
 100
 101
 102 class BandlabIE(BandlabBaseIE):
 103     _VALID_URL = [
 104         r'https?://(?:www\.)?bandlab.com/(?P<url_type>track|post|revision)/(?P<id>[\da-f_-]+)',
 105         r'https?://(?:www\.)?bandlab.com/(?P<url_type>embed)/\?(?:[^#]*&)?id=(?P<id>[\da-f-]+)',
 106     ]
 107     _EMBED_REGEX = [rf'<iframe[^>]+src=[\'"](?P<url>{_VALID_URL[1]})[\'"]']
 108     _TESTS = [{
 109         'url': 'https://www.bandlab.com/track/04b37e88dba24967b9dac8eb8567ff39_07d7f906fc96ee11b75e000d3a428fff',
 110         'md5': '46f7b43367dd268bbcf0bbe466753b2c',
 111         'info_dict': {
 112             'id': '02d7f906-fc96-ee11-b75e-000d3a428fff',
 113             'ext': 'm4a',
 114             'uploader_id': 'ender_milze',
 115             'track': 'sweet black',
 116             'description': 'composed by juanjn3737',
 117             'timestamp': 1702171963,
 118             'view_count': int,
 119             'like_count': int,
 120             'duration': 54.629999999999995,
 121             'title': 'sweet black',
 122             'upload_date': '20231210',
 123             'thumbnail': 'https://bandlabimages.azureedge.net/v1.0/songs/fa082beb-b856-4730-9170-a57e4e32cc2c/',
 124             'genres': ['Lofi'],
 125             'uploader': 'ender milze',
 126             'comment_count': int,
 127             'media_type': 'revision',
 128         },
 129     }, {
 130         # Same track as above but post URL
 131         'url': 'https://www.bandlab.com/post/07d7f906-fc96-ee11-b75e-000d3a428fff',
 132         'md5': '46f7b43367dd268bbcf0bbe466753b2c',
 133         'info_dict': {
 134             'id': '02d7f906-fc96-ee11-b75e-000d3a428fff',
 135             'ext': 'm4a',
 136             'uploader_id': 'ender_milze',
 137             'track': 'sweet black',
 138             'description': 'composed by juanjn3737',
 139             'timestamp': 1702171973,
 140             'view_count': int,
 141             'like_count': int,
 142             'duration': 54.629999999999995,
 143             'title': 'sweet black',
 144             'upload_date': '20231210',
 145             'thumbnail': 'https://bandlabimages.azureedge.net/v1.0/songs/fa082beb-b856-4730-9170-a57e4e32cc2c/',
 146             'genres': ['Lofi'],
 147             'uploader': 'ender milze',
 148             'comment_count': int,
 149             'media_type': 'revision',
 150         },
 151     }, {
 152         # SharedKey Example
 153         'url': 'https://www.bandlab.com/track/048916c2-c6da-ee11-85f9-6045bd2e11f9?sharedKey=0NNWX8qYAEmI38lWAzCNDA',
 154         'md5': '15174b57c44440e2a2008be9cae00250',
 155         'info_dict': {
 156             'id': '038916c2-c6da-ee11-85f9-6045bd2e11f9',
 157             'ext': 'm4a',
 158             'comment_count': int,
 159             'genres': ['Other'],
 160             'uploader_id': 'user8353034818103753',
 161             'thumbnail': 'https://bandlabimages.azureedge.net/v1.0/songs/51b18363-da23-4b9b-a29c-2933a3e561ca/',
 162             'timestamp': 1709625771,
 163             'track': 'PodcastMaerchen4b',
 164             'duration': 468.14,
 165             'view_count': int,
 166             'description': 'Podcast: Neues aus der Märchenwelt',
 167             'like_count': int,
 168             'upload_date': '20240305',
 169             'uploader': 'Erna Wageneder',
 170             'title': 'PodcastMaerchen4b',
 171             'media_type': 'revision',
 172         },
 173     }, {
 174         # Different Revision selected
 175         'url': 'https://www.bandlab.com/track/130343fc-148b-ea11-96d2-0003ffd1fc09?revId=110343fc-148b-ea11-96d2-0003ffd1fc09',
 176         'md5': '74e055ef9325d63f37088772fbfe4454',
 177         'info_dict': {
 178             'id': '110343fc-148b-ea11-96d2-0003ffd1fc09',
 179             'ext': 'm4a',
 180             'timestamp': 1588273294,
 181             'thumbnail': 'https://bandlabimages.azureedge.net/v1.0/users/b612e533-e4f7-4542-9f50-3fcfd8dd822c/',
 182             'description': 'Final Revision.',
 183             'title': 'Replay ( Instrumental)',
 184             'uploader': 'David R Sparks',
 185             'uploader_id': 'davesnothome69',
 186             'view_count': int,
 187             'comment_count': int,
 188             'track': 'Replay ( Instrumental)',
 189             'genres': ['Rock'],
 190             'upload_date': '20200430',
 191             'like_count': int,
 192             'duration': 279.43,
 193             'media_type': 'revision',
 194         },
 195     }, {
 196         # Video
 197         'url': 'https://www.bandlab.com/post/5cdf9036-3857-ef11-991a-6045bd36e0d9',
 198         'md5': '8caa2ef28e86c1dacf167293cfdbeba9',
 199         'info_dict': {
 200             'id': '5cdf9036-3857-ef11-991a-6045bd36e0d9',
 201             'ext': 'mp4',
 202             'duration': 44.705,
 203             'thumbnail': 'https://bandlabimages.azureedge.net/v1.0/videos/67c6cef1-cef6-40d3-831e-a55bc1dcb972/',
 204             'comment_count': int,
 205             'title': 'backing vocals',
 206             'uploader_id': 'marliashya',
 207             'uploader': 'auraa',
 208             'like_count': int,
 209             'description': 'backing vocals',
 210             'media_type': 'video',
 211         },
 212     }, {
 213         # Embed Example
 214         'url': 'https://www.bandlab.com/embed/?blur=false&id=014de0a4-7d82-ea11-a94c-0003ffd19c0f',
 215         'md5': 'a4ad05cb68c54faaed9b0a8453a8cf4a',
 216         'info_dict': {
 217             'id': '014de0a4-7d82-ea11-a94c-0003ffd19c0f',
 218             'ext': 'm4a',
 219             'comment_count': int,
 220             'genres': ['Electronic'],
 221             'uploader': 'Charlie Henson',
 222             'timestamp': 1587328674,
 223             'upload_date': '20200419',
 224             'view_count': int,
 225             'track': 'Positronic Meltdown',
 226             'duration': 318.55,
 227             'thumbnail': 'https://bandlabimages.azureedge.net/v1.0/songs/87165bc3-5439-496e-b1f7-a9f13b541ff2/',
 228             'description': 'Checkout my tracks at AOMX http://aomxsounds.com/',
 229             'uploader_id': 'microfreaks',
 230             'title': 'Positronic Meltdown',
 231             'like_count': int,
 232             'media_type': 'revision',
 233         },
 234     }, {
 235         # Track without revisions available
 236         'url': 'https://www.bandlab.com/track/55767ac51789ea11a94c0003ffd1fc09_2f007b0a37b94ec7a69bc25ae15108a5',
 237         'md5': 'f05d68a3769952c2d9257c473e14c15f',
 238         'info_dict': {
 239             'id': '55767ac51789ea11a94c0003ffd1fc09_2f007b0a37b94ec7a69bc25ae15108a5',
 240             'ext': 'm4a',
 241             'track': 'insame',
 242             'like_count': int,
 243             'duration': 84.03,
 244             'title': 'insame',
 245             'view_count': int,
 246             'comment_count': int,
 247             'uploader': 'Sorakime',
 248             'uploader_id': 'sorakime',
 249             'thumbnail': 'https://bandlabimages.azureedge.net/v1.0/users/572a351a-0f3a-4c6a-ac39-1a5defdeeb1c/',
 250             'timestamp': 1691162128,
 251             'upload_date': '20230804',
 252             'media_type': 'track',
 253         },
 254     }, {
 255         'url': 'https://www.bandlab.com/revision/014de0a4-7d82-ea11-a94c-0003ffd19c0f',
 256         'only_matching': True,
 257     }]
 258     _WEBPAGE_TESTS = [{
 259         'url': 'https://phantomluigi.github.io/',
 260         'info_dict': {
 261             'id': 'e14223c3-7871-ef11-bdfd-000d3a980db3',
 262             'ext': 'm4a',
 263             'view_count': int,
 264             'upload_date': '20240913',
 265             'uploader_id': 'phantommusicofficial',
 266             'timestamp': 1726194897,
 267             'uploader': 'Phantom',
 268             'comment_count': int,
 269             'genres': ['Progresive Rock'],
 270             'description': 'md5:a38cd668f7a2843295ef284114f18429',
 271             'duration': 225.23,
 272             'like_count': int,
 273             'title': 'Vermilion Pt. 2 (Cover)',
 274             'track': 'Vermilion Pt. 2 (Cover)',
 275             'thumbnail': 'https://bandlabimages.azureedge.net/v1.0/songs/62b10750-7aef-4f42-ad08-1af52f577e97/',
 276             'media_type': 'revision',
 277         },
 278     }]
 279
 280     def _real_extract(self, url):
 281         display_id, url_type = self._match_valid_url(url).group('id', 'url_type')
 282
 283         qs = parse_qs(url)
 284         revision_id = traverse_obj(qs, (('revId', 'id'), 0, any))
 285         if url_type == 'revision':
 286             revision_id = display_id
 287
 288         revision_data = None
 289         if not revision_id:
 290             post_data = self._call_api(
 291                 'posts', display_id, note='Downloading post data',
 292                 query=traverse_obj(qs, {'sharedKey': ('sharedKey', 0)}))
 293
 294             revision_id = traverse_obj(post_data, (('revisionId', ('revision', 'id')), {str}, any))
 295             revision_data = traverse_obj(post_data, ('revision', {dict}))
 296
 297             if not revision_data and not revision_id:
 298                 post_type = post_data.get('type')
 299                 if post_type == 'Video':
 300                     return self._parse_video(post_data, url=url)
 301                 if post_type == 'Track':
 302                     return self._parse_track(post_data, url=url)
 303                 raise ExtractorError(f'Could not extract data for post type {post_type!r}')
 304
 305         if not revision_data:
 306             revision_data = self._call_api(
 307                 'revisions', revision_id, note='Downloading revision data', query={'edit': 'false'})
 308
 309         return self._parse_revision(revision_data, url=url)
 310
 311
 312 class BandlabPlaylistIE(BandlabBaseIE):
 313     _VALID_URL = [
 314         r'https?://(?:www\.)?bandlab.com/(?:[\w]+/)?(?P<type>albums|collections)/(?P<id>[\da-f-]+)',
 315         r'https?://(?:www\.)?bandlab.com/(?P<type>embed)/collection/\?(?:[^#]*&)?id=(?P<id>[\da-f-]+)',
 316     ]
 317     _EMBED_REGEX = [rf'<iframe[^>]+src=[\'"](?P<url>{_VALID_URL[1]})[\'"]']
 318     _TESTS = [{
 319         'url': 'https://www.bandlab.com/davesnothome69/albums/89b79ea6-de42-ed11-b495-00224845aac7',
 320         'info_dict': {
 321             'thumbnail': 'https://bl-prod-images.azureedge.net/v1.3/albums/69507ff3-579a-45be-afca-9e87eddec944/',
 322             'release_date': '20221003',
 323             'title': 'Remnants',
 324             'album': 'Remnants',
 325             'like_count': int,
 326             'album_type': 'LP',
 327             'description': 'A collection of some feel good, rock hits.',
 328             'comment_count': int,
 329             'view_count': int,
 330             'id': '89b79ea6-de42-ed11-b495-00224845aac7',
 331             'uploader': 'David R Sparks',
 332             'uploader_id': 'davesnothome69',
 333         },
 334         'playlist_count': 10,
 335     }, {
 336         'url': 'https://www.bandlab.com/slytheband/collections/955102d4-1040-ef11-86c3-000d3a42581b',
 337         'info_dict': {
 338             'id': '955102d4-1040-ef11-86c3-000d3a42581b',
 339             'timestamp': 1720762659,
 340             'view_count': int,
 341             'title': 'My Shit 🖤',
 342             'uploader_id': 'slytheband',
 343             'uploader': '𝓢𝓛𝓨',
 344             'upload_date': '20240712',
 345             'like_count': int,
 346             'thumbnail': 'https://bandlabimages.azureedge.net/v1.0/collections/2c64ca12-b180-4b76-8587-7a8da76bddc8/',
 347         },
 348         'playlist_count': 15,
 349     }, {
 350         # Embeds can contain both albums and collections with the same URL pattern. This is an album
 351         'url': 'https://www.bandlab.com/embed/collection/?id=12cc6f7f-951b-ee11-907c-00224844f303',
 352         'info_dict': {
 353             'id': '12cc6f7f-951b-ee11-907c-00224844f303',
 354             'release_date': '20230706',
 355             'description': 'This is a collection of songs I created when I had an Amiga computer.',
 356             'view_count': int,
 357             'title': 'Mark Salud The Amiga Collection',
 358             'uploader_id': 'mssirmooth1962',
 359             'comment_count': int,
 360             'thumbnail': 'https://bl-prod-images.azureedge.net/v1.3/albums/d618bd7b-0537-40d5-bdd8-61b066e77d59/',
 361             'like_count': int,
 362             'uploader': 'Mark Salud',
 363             'album': 'Mark Salud The Amiga Collection',
 364             'album_type': 'LP',
 365         },
 366         'playlist_count': 24,
 367     }, {
 368         # Tracks without revision id
 369         'url': 'https://www.bandlab.com/embed/collection/?id=e98aafb5-d932-ee11-b8f0-00224844c719',
 370         'info_dict': {
 371             'like_count': int,
 372             'uploader_id': 'sorakime',
 373             'comment_count': int,
 374             'uploader': 'Sorakime',
 375             'view_count': int,
 376             'description': 'md5:4ec31c568a5f5a5a2b17572ea64c3825',
 377             'release_date': '20230812',
 378             'title': 'Art',
 379             'album': 'Art',
 380             'album_type': 'Album',
 381             'id': 'e98aafb5-d932-ee11-b8f0-00224844c719',
 382             'thumbnail': 'https://bl-prod-images.azureedge.net/v1.3/albums/20c890de-e94a-4422-828a-2da6377a13c8/',
 383         },
 384         'playlist_count': 13,
 385     }, {
 386         'url': 'https://www.bandlab.com/albums/89b79ea6-de42-ed11-b495-00224845aac7',
 387         'only_matching': True,
 388     }]
 389
 390     def _entries(self, album_data):
 391         for post in traverse_obj(album_data, ('posts', lambda _, v: v['type'])):
 392             post_type = post['type']
 393             if post_type == 'Revision':
 394                 yield self._parse_revision(post.get('revision'))
 395             elif post_type == 'Track':
 396                 yield self._parse_track(post)
 397             elif post_type == 'Video':
 398                 yield self._parse_video(post)
 399             else:
 400                 self.report_warning(f'Skipping unknown post type: "{post_type}"')
 401
 402     def _real_extract(self, url):
 403         playlist_id, playlist_type = self._match_valid_url(url).group('id', 'type')
 404
 405         endpoints = {
 406             'albums': ['albums'],
 407             'collections': ['collections'],
 408             'embed': ['collections', 'albums'],
 409         }.get(playlist_type)
 410         for endpoint in endpoints:
 411             playlist_data = self._call_api(
 412                 endpoint, playlist_id, note=f'Downloading {endpoint[:-1]} data',
 413                 fatal=False, expected_status=404)
 414             if not playlist_data.get('errorCode'):
 415                 playlist_type = endpoint
 416                 break
 417         if error_code := playlist_data.get('errorCode'):
 418             raise ExtractorError(f'Could not find playlist data. Error code: "{error_code}"')
 419
 420         return self.playlist_result(
 421             self._entries(playlist_data), playlist_id,
 422             **traverse_obj(playlist_data, {
 423                 'title': ('name', {str}),
 424                 'description': ('description', {str}),
 425                 'uploader': ('creator', 'name', {str}),
 426                 'uploader_id': ('creator', 'username', {str}),
 427                 'timestamp': ('createdOn', {parse_iso8601}),
 428                 'release_date': ('releaseDate', {lambda x: x.replace('-', '')}, filter),
 429                 'thumbnail': ('picture', ('original', 'url'), {url_or_none}, any),
 430                 'like_count': ('counters', 'likes', {int_or_none}),
 431                 'comment_count': ('counters', 'comments', {int_or_none}),
 432                 'view_count': ('counters', 'plays', {int_or_none}),
 433             }),
 434             **(traverse_obj(playlist_data, {
 435                 'album': ('name', {str}),
 436                 'album_type': ('type', {str}),
 437             }) if playlist_type == 'albums' else {}))