Raphaël G. Git Repositories - youtubedl/blob - youtube_dl/extractor/peertube.py

   1 # coding: utf-8
   2 from __future__ import unicode_literals
   3
   4 import re
   5
   6 from .common import InfoExtractor
   7 from ..compat import compat_str
   8 from ..utils import (
   9     int_or_none,
  10     parse_resolution,
  11     try_get,
  12     unified_timestamp,
  13     url_or_none,
  14     urljoin,
  15 )
  16
  17
  18 class PeerTubeIE(InfoExtractor):
  19     _INSTANCES_RE = r'''(?:
  20                             # Taken from https://instances.joinpeertube.org/instances
  21                             peertube\.rainbowswingers\.net|
  22                             tube\.stanisic\.nl|
  23                             peer\.suiri\.us|
  24                             medias\.libox\.fr|
  25                             videomensoif\.ynh\.fr|
  26                             peertube\.travelpandas\.eu|
  27                             peertube\.rachetjay\.fr|
  28                             peertube\.montecsys\.fr|
  29                             tube\.eskuero\.me|
  30                             peer\.tube|
  31                             peertube\.umeahackerspace\.se|
  32                             tube\.nx-pod\.de|
  33                             video\.monsieurbidouille\.fr|
  34                             tube\.openalgeria\.org|
  35                             vid\.lelux\.fi|
  36                             video\.anormallostpod\.ovh|
  37                             tube\.crapaud-fou\.org|
  38                             peertube\.stemy\.me|
  39                             lostpod\.space|
  40                             exode\.me|
  41                             peertube\.snargol\.com|
  42                             vis\.ion\.ovh|
  43                             videosdulib\.re|
  44                             v\.mbius\.io|
  45                             videos\.judrey\.eu|
  46                             peertube\.osureplayviewer\.xyz|
  47                             peertube\.mathieufamily\.ovh|
  48                             www\.videos-libr\.es|
  49                             fightforinfo\.com|
  50                             peertube\.fediverse\.ru|
  51                             peertube\.oiseauroch\.fr|
  52                             video\.nesven\.eu|
  53                             v\.bearvideo\.win|
  54                             video\.qoto\.org|
  55                             justporn\.cc|
  56                             video\.vny\.fr|
  57                             peervideo\.club|
  58                             tube\.taker\.fr|
  59                             peertube\.chantierlibre\.org|
  60                             tube\.ipfixe\.info|
  61                             tube\.kicou\.info|
  62                             tube\.dodsorf\.as|
  63                             videobit\.cc|
  64                             video\.yukari\.moe|
  65                             videos\.elbinario\.net|
  66                             hkvideo\.live|
  67                             pt\.tux\.tf|
  68                             www\.hkvideo\.live|
  69                             FIGHTFORINFO\.com|
  70                             pt\.765racing\.com|
  71                             peertube\.gnumeria\.eu\.org|
  72                             nordenmedia\.com|
  73                             peertube\.co\.uk|
  74                             tube\.darfweb\.eu|
  75                             tube\.kalah-france\.org|
  76                             0ch\.in|
  77                             vod\.mochi\.academy|
  78                             film\.node9\.org|
  79                             peertube\.hatthieves\.es|
  80                             video\.fitchfamily\.org|
  81                             peertube\.ddns\.net|
  82                             video\.ifuncle\.kr|
  83                             video\.fdlibre\.eu|
  84                             tube\.22decembre\.eu|
  85                             peertube\.harmoniescreatives\.com|
  86                             tube\.fabrigli\.fr|
  87                             video\.thedwyers\.co|
  88                             video\.bruitbruit\.com|
  89                             peertube\.foxfam\.club|
  90                             peer\.philoxweb\.be|
  91                             videos\.bugs\.social|
  92                             peertube\.malbert\.xyz|
  93                             peertube\.bilange\.ca|
  94                             libretube\.net|
  95                             diytelevision\.com|
  96                             peertube\.fedilab\.app|
  97                             libre\.video|
  98                             video\.mstddntfdn\.online|
  99                             us\.tv|
 100                             peertube\.sl-network\.fr|
 101                             peertube\.dynlinux\.io|
 102                             peertube\.david\.durieux\.family|
 103                             peertube\.linuxrocks\.online|
 104                             peerwatch\.xyz|
 105                             v\.kretschmann\.social|
 106                             tube\.otter\.sh|
 107                             yt\.is\.nota\.live|
 108                             tube\.dragonpsi\.xyz|
 109                             peertube\.boneheadmedia\.com|
 110                             videos\.funkwhale\.audio|
 111                             watch\.44con\.com|
 112                             peertube\.gcaillaut\.fr|
 113                             peertube\.icu|
 114                             pony\.tube|
 115                             spacepub\.space|
 116                             tube\.stbr\.io|
 117                             v\.mom-gay\.faith|
 118                             tube\.port0\.xyz|
 119                             peertube\.simounet\.net|
 120                             play\.jergefelt\.se|
 121                             peertube\.zeteo\.me|
 122                             tube\.danq\.me|
 123                             peertube\.kerenon\.com|
 124                             tube\.fab-l3\.org|
 125                             tube\.calculate\.social|
 126                             peertube\.mckillop\.org|
 127                             tube\.netzspielplatz\.de|
 128                             vod\.ksite\.de|
 129                             peertube\.laas\.fr|
 130                             tube\.govital\.net|
 131                             peertube\.stephenson\.cc|
 132                             bistule\.nohost\.me|
 133                             peertube\.kajalinifi\.de|
 134                             video\.ploud\.jp|
 135                             video\.omniatv\.com|
 136                             peertube\.ffs2play\.fr|
 137                             peertube\.leboulaire\.ovh|
 138                             peertube\.tronic-studio\.com|
 139                             peertube\.public\.cat|
 140                             peertube\.metalbanana\.net|
 141                             video\.1000i100\.fr|
 142                             peertube\.alter-nativ-voll\.de|
 143                             tube\.pasa\.tf|
 144                             tube\.worldofhauru\.xyz|
 145                             pt\.kamp\.site|
 146                             peertube\.teleassist\.fr|
 147                             videos\.mleduc\.xyz|
 148                             conf\.tube|
 149                             media\.privacyinternational\.org|
 150                             pt\.forty-two\.nl|
 151                             video\.halle-leaks\.de|
 152                             video\.grosskopfgames\.de|
 153                             peertube\.schaeferit\.de|
 154                             peertube\.jackbot\.fr|
 155                             tube\.extinctionrebellion\.fr|
 156                             peertube\.f-si\.org|
 157                             video\.subak\.ovh|
 158                             videos\.koweb\.fr|
 159                             peertube\.zergy\.net|
 160                             peertube\.roflcopter\.fr|
 161                             peertube\.floss-marketing-school\.com|
 162                             vloggers\.social|
 163                             peertube\.iriseden\.eu|
 164                             videos\.ubuntu-paris\.org|
 165                             peertube\.mastodon\.host|
 166                             armstube\.com|
 167                             peertube\.s2s\.video|
 168                             peertube\.lol|
 169                             tube\.open-plug\.eu|
 170                             open\.tube|
 171                             peertube\.ch|
 172                             peertube\.normandie-libre\.fr|
 173                             peertube\.slat\.org|
 174                             video\.lacaveatonton\.ovh|
 175                             peertube\.uno|
 176                             peertube\.servebeer\.com|
 177                             peertube\.fedi\.quebec|
 178                             tube\.h3z\.jp|
 179                             tube\.plus200\.com|
 180                             peertube\.eric\.ovh|
 181                             tube\.metadocs\.cc|
 182                             tube\.unmondemeilleur\.eu|
 183                             gouttedeau\.space|
 184                             video\.antirep\.net|
 185                             nrop\.cant\.at|
 186                             tube\.ksl-bmx\.de|
 187                             tube\.plaf\.fr|
 188                             tube\.tchncs\.de|
 189                             video\.devinberg\.com|
 190                             hitchtube\.fr|
 191                             peertube\.kosebamse\.com|
 192                             yunopeertube\.myddns\.me|
 193                             peertube\.varney\.fr|
 194                             peertube\.anon-kenkai\.com|
 195                             tube\.maiti\.info|
 196                             tubee\.fr|
 197                             videos\.dinofly\.com|
 198                             toobnix\.org|
 199                             videotape\.me|
 200                             voca\.tube|
 201                             video\.heromuster\.com|
 202                             video\.lemediatv\.fr|
 203                             video\.up\.edu\.ph|
 204                             balafon\.video|
 205                             video\.ivel\.fr|
 206                             thickrips\.cloud|
 207                             pt\.laurentkruger\.fr|
 208                             video\.monarch-pass\.net|
 209                             peertube\.artica\.center|
 210                             video\.alternanet\.fr|
 211                             indymotion\.fr|
 212                             fanvid\.stopthatimp\.net|
 213                             video\.farci\.org|
 214                             v\.lesterpig\.com|
 215                             video\.okaris\.de|
 216                             tube\.pawelko\.net|
 217                             peertube\.mablr\.org|
 218                             tube\.fede\.re|
 219                             pytu\.be|
 220                             evertron\.tv|
 221                             devtube\.dev-wiki\.de|
 222                             raptube\.antipub\.org|
 223                             video\.selea\.se|
 224                             peertube\.mygaia\.org|
 225                             video\.oh14\.de|
 226                             peertube\.livingutopia\.org|
 227                             peertube\.the-penguin\.de|
 228                             tube\.thechangebook\.org|
 229                             tube\.anjara\.eu|
 230                             pt\.pube\.tk|
 231                             video\.samedi\.pm|
 232                             mplayer\.demouliere\.eu|
 233                             widemus\.de|
 234                             peertube\.me|
 235                             peertube\.zapashcanon\.fr|
 236                             video\.latavernedejohnjohn\.fr|
 237                             peertube\.pcservice46\.fr|
 238                             peertube\.mazzonetto\.eu|
 239                             video\.irem\.univ-paris-diderot\.fr|
 240                             video\.livecchi\.cloud|
 241                             alttube\.fr|
 242                             video\.coop\.tools|
 243                             video\.cabane-libre\.org|
 244                             peertube\.openstreetmap\.fr|
 245                             videos\.alolise\.org|
 246                             irrsinn\.video|
 247                             video\.antopie\.org|
 248                             scitech\.video|
 249                             tube2\.nemsia\.org|
 250                             video\.amic37\.fr|
 251                             peertube\.freeforge\.eu|
 252                             video\.arbitrarion\.com|
 253                             video\.datsemultimedia\.com|
 254                             stoptrackingus\.tv|
 255                             peertube\.ricostrongxxx\.com|
 256                             docker\.videos\.lecygnenoir\.info|
 257                             peertube\.togart\.de|
 258                             tube\.postblue\.info|
 259                             videos\.domainepublic\.net|
 260                             peertube\.cyber-tribal\.com|
 261                             video\.gresille\.org|
 262                             peertube\.dsmouse\.net|
 263                             cinema\.yunohost\.support|
 264                             tube\.theocevaer\.fr|
 265                             repro\.video|
 266                             tube\.4aem\.com|
 267                             quaziinc\.com|
 268                             peertube\.metawurst\.space|
 269                             videos\.wakapo\.com|
 270                             video\.ploud\.fr|
 271                             video\.freeradical\.zone|
 272                             tube\.valinor\.fr|
 273                             refuznik\.video|
 274                             pt\.kircheneuenburg\.de|
 275                             peertube\.asrun\.eu|
 276                             peertube\.lagob\.fr|
 277                             videos\.side-ways\.net|
 278                             91video\.online|
 279                             video\.valme\.io|
 280                             video\.taboulisme\.com|
 281                             videos-libr\.es|
 282                             tv\.mooh\.fr|
 283                             nuage\.acostey\.fr|
 284                             video\.monsieur-a\.fr|
 285                             peertube\.librelois\.fr|
 286                             videos\.pair2jeux\.tube|
 287                             videos\.pueseso\.club|
 288                             peer\.mathdacloud\.ovh|
 289                             media\.assassinate-you\.net|
 290                             vidcommons\.org|
 291                             ptube\.rousset\.nom\.fr|
 292                             tube\.cyano\.at|
 293                             videos\.squat\.net|
 294                             video\.iphodase\.fr|
 295                             peertube\.makotoworkshop\.org|
 296                             peertube\.serveur\.slv-valbonne\.fr|
 297                             vault\.mle\.party|
 298                             hostyour\.tv|
 299                             videos\.hack2g2\.fr|
 300                             libre\.tube|
 301                             pire\.artisanlogiciel\.net|
 302                             videos\.numerique-en-commun\.fr|
 303                             video\.netsyms\.com|
 304                             video\.die-partei\.social|
 305                             video\.writeas\.org|
 306                             peertube\.swarm\.solvingmaz\.es|
 307                             tube\.pericoloso\.ovh|
 308                             watching\.cypherpunk\.observer|
 309                             videos\.adhocmusic\.com|
 310                             tube\.rfc1149\.net|
 311                             peertube\.librelabucm\.org|
 312                             videos\.numericoop\.fr|
 313                             peertube\.koehn\.com|
 314                             peertube\.anarchmusicall\.net|
 315                             tube\.kampftoast\.de|
 316                             vid\.y-y\.li|
 317                             peertube\.xtenz\.xyz|
 318                             diode\.zone|
 319                             tube\.egf\.mn|
 320                             peertube\.nomagic\.uk|
 321                             visionon\.tv|
 322                             videos\.koumoul\.com|
 323                             video\.rastapuls\.com|
 324                             video\.mantlepro\.com|
 325                             video\.deadsuperhero\.com|
 326                             peertube\.musicstudio\.pro|
 327                             peertube\.we-keys\.fr|
 328                             artitube\.artifaille\.fr|
 329                             peertube\.ethernia\.net|
 330                             tube\.midov\.pl|
 331                             peertube\.fr|
 332                             watch\.snoot\.tube|
 333                             peertube\.donnadieu\.fr|
 334                             argos\.aquilenet\.fr|
 335                             tube\.nemsia\.org|
 336                             tube\.bruniau\.net|
 337                             videos\.darckoune\.moe|
 338                             tube\.traydent\.info|
 339                             dev\.videos\.lecygnenoir\.info|
 340                             peertube\.nayya\.org|
 341                             peertube\.live|
 342                             peertube\.mofgao\.space|
 343                             video\.lequerrec\.eu|
 344                             peertube\.amicale\.net|
 345                             aperi\.tube|
 346                             tube\.ac-lyon\.fr|
 347                             video\.lw1\.at|
 348                             www\.yiny\.org|
 349                             videos\.pofilo\.fr|
 350                             tube\.lou\.lt|
 351                             choob\.h\.etbus\.ch|
 352                             tube\.hoga\.fr|
 353                             peertube\.heberge\.fr|
 354                             video\.obermui\.de|
 355                             videos\.cloudfrancois\.fr|
 356                             betamax\.video|
 357                             video\.typica\.us|
 358                             tube\.piweb\.be|
 359                             video\.blender\.org|
 360                             peertube\.cat|
 361                             tube\.kdy\.ch|
 362                             pe\.ertu\.be|
 363                             peertube\.social|
 364                             videos\.lescommuns\.org|
 365                             tv\.datamol\.org|
 366                             videonaute\.fr|
 367                             dialup\.express|
 368                             peertube\.nogafa\.org|
 369                             megatube\.lilomoino\.fr|
 370                             peertube\.tamanoir\.foucry\.net|
 371                             peertube\.devosi\.org|
 372                             peertube\.1312\.media|
 373                             tube\.bootlicker\.party|
 374                             skeptikon\.fr|
 375                             video\.blueline\.mg|
 376                             tube\.homecomputing\.fr|
 377                             tube\.ouahpiti\.info|
 378                             video\.tedomum\.net|
 379                             video\.g3l\.org|
 380                             fontube\.fr|
 381                             peertube\.gaialabs\.ch|
 382                             tube\.kher\.nl|
 383                             peertube\.qtg\.fr|
 384                             video\.migennes\.net|
 385                             tube\.p2p\.legal|
 386                             troll\.tv|
 387                             videos\.iut-orsay\.fr|
 388                             peertube\.solidev\.net|
 389                             videos\.cemea\.org|
 390                             video\.passageenseine\.fr|
 391                             videos\.festivalparminous\.org|
 392                             peertube\.touhoppai\.moe|
 393                             sikke\.fi|
 394                             peer\.hostux\.social|
 395                             share\.tube|
 396                             peertube\.walkingmountains\.fr|
 397                             videos\.benpro\.fr|
 398                             peertube\.parleur\.net|
 399                             peertube\.heraut\.eu|
 400                             tube\.aquilenet\.fr|
 401                             peertube\.gegeweb\.eu|
 402                             framatube\.org|
 403                             thinkerview\.video|
 404                             tube\.conferences-gesticulees\.net|
 405                             peertube\.datagueule\.tv|
 406                             video\.lqdn\.fr|
 407                             tube\.mochi\.academy|
 408                             media\.zat\.im|
 409                             video\.colibris-outilslibres\.org|
 410                             tube\.svnet\.fr|
 411                             peertube\.video|
 412                             peertube3\.cpy\.re|
 413                             peertube2\.cpy\.re|
 414                             videos\.tcit\.fr|
 415                             peertube\.cpy\.re
 416                         )'''
 417     _UUID_RE = r'[\da-fA-F]{8}-[\da-fA-F]{4}-[\da-fA-F]{4}-[\da-fA-F]{4}-[\da-fA-F]{12}'
 418     _VALID_URL = r'''(?x)
 419                     (?:
 420                         peertube:(?P<host>[^:]+):|
 421                         https?://(?P<host_2>%s)/(?:videos/(?:watch|embed)|api/v\d/videos)/
 422                     )
 423                     (?P<id>%s)
 424                     ''' % (_INSTANCES_RE, _UUID_RE)
 425     _TESTS = [{
 426         'url': 'https://peertube.cpy.re/videos/watch/2790feb0-8120-4e63-9af3-c943c69f5e6c',
 427         'md5': '80f24ff364cc9d333529506a263e7feb',
 428         'info_dict': {
 429             'id': '2790feb0-8120-4e63-9af3-c943c69f5e6c',
 430             'ext': 'mp4',
 431             'title': 'wow',
 432             'description': 'wow such video, so gif',
 433             'thumbnail': r're:https?://.*\.(?:jpg|png)',
 434             'timestamp': 1519297480,
 435             'upload_date': '20180222',
 436             'uploader': 'Luclu7',
 437             'uploader_id': '7fc42640-efdb-4505-a45d-a15b1a5496f1',
 438             'uploder_url': 'https://peertube.nsa.ovh/accounts/luclu7',
 439             'license': 'Unknown',
 440             'duration': 3,
 441             'view_count': int,
 442             'like_count': int,
 443             'dislike_count': int,
 444             'tags': list,
 445             'categories': list,
 446         }
 447     }, {
 448         'url': 'https://peertube.tamanoir.foucry.net/videos/watch/0b04f13d-1e18-4f1d-814e-4979aa7c9c44',
 449         'only_matching': True,
 450     }, {
 451         # nsfw
 452         'url': 'https://tube.22decembre.eu/videos/watch/9bb88cd3-9959-46d9-9ab9-33d2bb704c39',
 453         'only_matching': True,
 454     }, {
 455         'url': 'https://tube.22decembre.eu/videos/embed/fed67262-6edb-4d1c-833b-daa9085c71d7',
 456         'only_matching': True,
 457     }, {
 458         'url': 'https://tube.openalgeria.org/api/v1/videos/c1875674-97d0-4c94-a058-3f7e64c962e8',
 459         'only_matching': True,
 460     }, {
 461         'url': 'peertube:video.blender.org:b37a5b9f-e6b5-415c-b700-04a5cd6ec205',
 462         'only_matching': True,
 463     }]
 464
 465     @staticmethod
 466     def _extract_peertube_url(webpage, source_url):
 467         mobj = re.match(
 468             r'https?://(?P<host>[^/]+)/videos/(?:watch|embed)/(?P<id>%s)'
 469             % PeerTubeIE._UUID_RE, source_url)
 470         if mobj and any(p in webpage for p in (
 471                 '<title>PeerTube<',
 472                 'There will be other non JS-based clients to access PeerTube',
 473                 '>We are sorry but it seems that PeerTube is not compatible with your web browser.<')):
 474             return 'peertube:%s:%s' % mobj.group('host', 'id')
 475
 476     @staticmethod
 477     def _extract_urls(webpage, source_url):
 478         entries = re.findall(
 479             r'''(?x)<iframe[^>]+\bsrc=["\'](?P<url>(?:https?:)?//%s/videos/embed/%s)'''
 480             % (PeerTubeIE._INSTANCES_RE, PeerTubeIE._UUID_RE), webpage)
 481         if not entries:
 482             peertube_url = PeerTubeIE._extract_peertube_url(webpage, source_url)
 483             if peertube_url:
 484                 entries = [peertube_url]
 485         return entries
 486
 487     def _real_extract(self, url):
 488         mobj = re.match(self._VALID_URL, url)
 489         host = mobj.group('host') or mobj.group('host_2')
 490         video_id = mobj.group('id')
 491
 492         video = self._download_json(
 493             'https://%s/api/v1/videos/%s' % (host, video_id), video_id)
 494
 495         title = video['name']
 496
 497         formats = []
 498         for file_ in video['files']:
 499             if not isinstance(file_, dict):
 500                 continue
 501             file_url = url_or_none(file_.get('fileUrl'))
 502             if not file_url:
 503                 continue
 504             file_size = int_or_none(file_.get('size'))
 505             format_id = try_get(
 506                 file_, lambda x: x['resolution']['label'], compat_str)
 507             f = parse_resolution(format_id)
 508             f.update({
 509                 'url': file_url,
 510                 'format_id': format_id,
 511                 'filesize': file_size,
 512             })
 513             formats.append(f)
 514         self._sort_formats(formats)
 515
 516         def account_data(field):
 517             return try_get(video, lambda x: x['account'][field], compat_str)
 518
 519         category = try_get(video, lambda x: x['category']['label'], compat_str)
 520         categories = [category] if category else None
 521
 522         nsfw = video.get('nsfw')
 523         if nsfw is bool:
 524             age_limit = 18 if nsfw else 0
 525         else:
 526             age_limit = None
 527
 528         return {
 529             'id': video_id,
 530             'title': title,
 531             'description': video.get('description'),
 532             'thumbnail': urljoin(url, video.get('thumbnailPath')),
 533             'timestamp': unified_timestamp(video.get('publishedAt')),
 534             'uploader': account_data('displayName'),
 535             'uploader_id': account_data('uuid'),
 536             'uploder_url': account_data('url'),
 537             'license': try_get(
 538                 video, lambda x: x['licence']['label'], compat_str),
 539             'duration': int_or_none(video.get('duration')),
 540             'view_count': int_or_none(video.get('views')),
 541             'like_count': int_or_none(video.get('likes')),
 542             'dislike_count': int_or_none(video.get('dislikes')),
 543             'age_limit': age_limit,
 544             'tags': try_get(video, lambda x: x['tags'], list),
 545             'categories': categories,
 546             'formats': formats,
 547         }