yt_dlp/extractor/slideslive.py

   1 import re
   2 import urllib.parse
   3 import xml.etree.ElementTree
   4
   5 from .common import InfoExtractor
   6 from ..utils import (
   7     ExtractorError,
   8     int_or_none,
   9     parse_qs,
  10     smuggle_url,
  11     traverse_obj,
  12     unified_timestamp,
  13     update_url_query,
  14     url_or_none,
  15     xpath_text,
  16 )
  17
  18
  19 class SlidesLiveIE(InfoExtractor):
  20     _VALID_URL = r'https?://slideslive\.com/(?:embed/(?:presentation/)?)?(?P<id>[0-9]+)'
  21     _TESTS = [{
  22         # service_name = yoda, only XML slides info
  23         'url': 'https://slideslive.com/38902413/gcc-ia16-backend',
  24         'info_dict': {
  25             'id': '38902413',
  26             'ext': 'mp4',
  27             'title': 'GCC IA16 backend',
  28             'timestamp': 1648189972,
  29             'upload_date': '20220325',
  30             'thumbnail': r're:^https?://.*\.jpg',
  31             'thumbnails': 'count:42',
  32             'chapters': 'count:41',
  33             'duration': 1638,
  34         },
  35         'params': {
  36             'skip_download': 'm3u8',
  37         },
  38     }, {
  39         # service_name = yoda, /v7/ slides
  40         'url': 'https://slideslive.com/38935785',
  41         'info_dict': {
  42             'id': '38935785',
  43             'ext': 'mp4',
  44             'title': 'Offline Reinforcement Learning: From Algorithms to Practical Challenges',
  45             'upload_date': '20211115',
  46             'timestamp': 1636996003,
  47             'thumbnail': r're:^https?://.*\.(?:jpg|png)',
  48             'thumbnails': 'count:640',
  49             'chapters': 'count:639',
  50             'duration': 9832,
  51         },
  52         'params': {
  53             'skip_download': 'm3u8',
  54         },
  55     }, {
  56         # service_name = yoda, /v1/ slides
  57         'url': 'https://slideslive.com/38973182/how-should-a-machine-learning-researcher-think-about-ai-ethics',
  58         'info_dict': {
  59             'id': '38973182',
  60             'ext': 'mp4',
  61             'title': 'How Should a Machine Learning Researcher Think About AI Ethics?',
  62             'upload_date': '20220201',
  63             'thumbnail': r're:^https?://.*\.jpg',
  64             'timestamp': 1643728135,
  65             'thumbnails': 'count:3',
  66             'chapters': 'count:2',
  67             'duration': 5889,
  68         },
  69         'params': {
  70             'skip_download': 'm3u8',
  71         },
  72     }, {
  73         # service_name = youtube, only XML slides info
  74         'url': 'https://slideslive.com/38897546/special-metaprednaska-petra-ludwiga-hodnoty-pro-lepsi-spolecnost',
  75         'md5': '8a79b5e3d700837f40bd2afca3c8fa01',
  76         'info_dict': {
  77             'id': 'jmg02wCJD5M',
  78             'display_id': '38897546',
  79             'ext': 'mp4',
  80             'title': 'SPECIÁL: Meta-přednáška Petra Ludwiga - Hodnoty pro lepší společnost',
  81             'description': 'Watch full version of this video at https://slideslive.com/38897546.',
  82             'channel_url': 'https://www.youtube.com/channel/UCZWdAkNYFncuX0khyvhqnxw',
  83             'channel': 'SlidesLive Videos - G1',
  84             'channel_id': 'UCZWdAkNYFncuX0khyvhqnxw',
  85             'uploader_id': 'UCZWdAkNYFncuX0khyvhqnxw',
  86             'uploader': 'SlidesLive Videos - G1',
  87             'uploader_url': 'http://www.youtube.com/channel/UCZWdAkNYFncuX0khyvhqnxw',
  88             'live_status': 'not_live',
  89             'upload_date': '20160710',
  90             'timestamp': 1618786715,
  91             'duration': 6827,
  92             'like_count': int,
  93             'view_count': int,
  94             'comment_count': int,
  95             'channel_follower_count': int,
  96             'age_limit': 0,
  97             'thumbnail': r're:^https?://.*\.(?:jpg|webp)',
  98             'thumbnails': 'count:169',
  99             'playable_in_embed': True,
 100             'availability': 'unlisted',
 101             'tags': [],
 102             'categories': ['People & Blogs'],
 103             'chapters': 'count:168',
 104         },
 105     }, {
 106         # embed-only presentation, only XML slides info
 107         'url': 'https://slideslive.com/embed/presentation/38925850',
 108         'info_dict': {
 109             'id': '38925850',
 110             'ext': 'mp4',
 111             'title': 'Towards a Deep Network Architecture for Structured Smoothness',
 112             'thumbnail': r're:^https?://.*\.jpg',
 113             'thumbnails': 'count:8',
 114             'timestamp': 1629671508,
 115             'upload_date': '20210822',
 116             'chapters': 'count:7',
 117             'duration': 326,
 118         },
 119         'params': {
 120             'skip_download': 'm3u8',
 121         },
 122     }, {
 123         # embed-only presentation, only JSON slides info, /v5/ slides (.png)
 124         'url': 'https://slideslive.com/38979920/',
 125         'info_dict': {
 126             'id': '38979920',
 127             'ext': 'mp4',
 128             'title': 'MoReL: Multi-omics Relational Learning',
 129             'thumbnail': r're:^https?://.*\.(?:jpg|png)',
 130             'thumbnails': 'count:7',
 131             'timestamp': 1654714970,
 132             'upload_date': '20220608',
 133             'chapters': 'count:6',
 134             'duration': 171,
 135         },
 136         'params': {
 137             'skip_download': 'm3u8',
 138         },
 139     }, {
 140         # /v2/ slides (.jpg)
 141         'url': 'https://slideslive.com/38954074',
 142         'info_dict': {
 143             'id': '38954074',
 144             'ext': 'mp4',
 145             'title': 'Decentralized Attribution of Generative Models',
 146             'thumbnail': r're:^https?://.*\.jpg',
 147             'thumbnails': 'count:16',
 148             'timestamp': 1622806321,
 149             'upload_date': '20210604',
 150             'chapters': 'count:15',
 151             'duration': 306,
 152         },
 153         'params': {
 154             'skip_download': 'm3u8',
 155         },
 156     }, {
 157         # /v4/ slides (.png)
 158         'url': 'https://slideslive.com/38979570/',
 159         'info_dict': {
 160             'id': '38979570',
 161             'ext': 'mp4',
 162             'title': 'Efficient Active Search for Combinatorial Optimization Problems',
 163             'thumbnail': r're:^https?://.*\.(?:jpg|png)',
 164             'thumbnails': 'count:9',
 165             'timestamp': 1654714896,
 166             'upload_date': '20220608',
 167             'chapters': 'count:8',
 168             'duration': 295,
 169         },
 170         'params': {
 171             'skip_download': 'm3u8',
 172         },
 173     }, {
 174         # /v10/ slides
 175         'url': 'https://slideslive.com/embed/presentation/38979880?embed_parent_url=https%3A%2F%2Fedit.videoken.com%2F',
 176         'info_dict': {
 177             'id': '38979880',
 178             'ext': 'mp4',
 179             'title': 'The Representation Power of Neural Networks',
 180             'timestamp': 1654714962,
 181             'thumbnail': r're:^https?://.*\.(?:jpg|png)',
 182             'thumbnails': 'count:22',
 183             'upload_date': '20220608',
 184             'chapters': 'count:21',
 185             'duration': 294,
 186         },
 187         'params': {
 188             'skip_download': 'm3u8',
 189         },
 190     }, {
 191         # /v7/ slides, 2 video slides
 192         'url': 'https://slideslive.com/embed/presentation/38979682?embed_container_origin=https%3A%2F%2Fedit.videoken.com',
 193         'playlist_count': 3,
 194         'info_dict': {
 195             'id': '38979682-playlist',
 196             'title': 'LoRA: Low-Rank Adaptation of Large Language Models',
 197         },
 198         'playlist': [{
 199             'info_dict': {
 200                 'id': '38979682',
 201                 'ext': 'mp4',
 202                 'title': 'LoRA: Low-Rank Adaptation of Large Language Models',
 203                 'timestamp': 1654714920,
 204                 'thumbnail': r're:^https?://.*\.(?:jpg|png)',
 205                 'thumbnails': 'count:30',
 206                 'upload_date': '20220608',
 207                 'chapters': 'count:31',
 208                 'duration': 272,
 209             },
 210         }, {
 211             'info_dict': {
 212                 'id': '38979682-021',
 213                 'ext': 'mp4',
 214                 'title': 'LoRA: Low-Rank Adaptation of Large Language Models - Slide 021',
 215                 'duration': 3,
 216                 'timestamp': 1654714920,
 217                 'upload_date': '20220608',
 218             },
 219         }, {
 220             'info_dict': {
 221                 'id': '38979682-024',
 222                 'ext': 'mp4',
 223                 'title': 'LoRA: Low-Rank Adaptation of Large Language Models - Slide 024',
 224                 'duration': 4,
 225                 'timestamp': 1654714920,
 226                 'upload_date': '20220608',
 227             },
 228         }],
 229         'params': {
 230             'skip_download': 'm3u8',
 231         },
 232     }, {
 233         # /v6/ slides, 1 video slide, edit.videoken.com embed
 234         'url': 'https://slideslive.com/38979481/',
 235         'playlist_count': 2,
 236         'info_dict': {
 237             'id': '38979481-playlist',
 238             'title': 'How to Train Your MAML to Excel in Few-Shot Classification',
 239         },
 240         'playlist': [{
 241             'info_dict': {
 242                 'id': '38979481',
 243                 'ext': 'mp4',
 244                 'title': 'How to Train Your MAML to Excel in Few-Shot Classification',
 245                 'timestamp': 1654714877,
 246                 'thumbnail': r're:^https?://.*\.(?:jpg|png)',
 247                 'thumbnails': 'count:43',
 248                 'upload_date': '20220608',
 249                 'chapters': 'count:43',
 250                 'duration': 315,
 251             },
 252         }, {
 253             'info_dict': {
 254                 'id': '38979481-013',
 255                 'ext': 'mp4',
 256                 'title': 'How to Train Your MAML to Excel in Few-Shot Classification - Slide 013',
 257                 'duration': 3,
 258                 'timestamp': 1654714877,
 259                 'upload_date': '20220608',
 260             },
 261         }],
 262         'params': {
 263             'skip_download': 'm3u8',
 264         },
 265     }, {
 266         # /v3/ slides, .jpg and .png, service_name = youtube
 267         'url': 'https://slideslive.com/embed/38932460/',
 268         'info_dict': {
 269             'id': 'RTPdrgkyTiE',
 270             'display_id': '38932460',
 271             'ext': 'mp4',
 272             'title': 'Active Learning for Hierarchical Multi-Label Classification',
 273             'description': 'Watch full version of this video at https://slideslive.com/38932460.',
 274             'channel': 'SlidesLive Videos - A',
 275             'channel_id': 'UC62SdArr41t_-_fX40QCLRw',
 276             'channel_url': 'https://www.youtube.com/channel/UC62SdArr41t_-_fX40QCLRw',
 277             'uploader': 'SlidesLive Videos - A',
 278             'uploader_id': 'UC62SdArr41t_-_fX40QCLRw',
 279             'uploader_url': 'http://www.youtube.com/channel/UC62SdArr41t_-_fX40QCLRw',
 280             'upload_date': '20200903',
 281             'timestamp': 1602599092,
 282             'duration': 942,
 283             'age_limit': 0,
 284             'live_status': 'not_live',
 285             'playable_in_embed': True,
 286             'availability': 'unlisted',
 287             'categories': ['People & Blogs'],
 288             'tags': [],
 289             'channel_follower_count': int,
 290             'like_count': int,
 291             'view_count': int,
 292             'thumbnail': r're:^https?://.*\.(?:jpg|png|webp)',
 293             'thumbnails': 'count:21',
 294             'chapters': 'count:20',
 295         },
 296         'params': {
 297             'skip_download': 'm3u8',
 298         },
 299     }, {
 300         # /v3/ slides, .png only, service_name = yoda
 301         'url': 'https://slideslive.com/38983994',
 302         'info_dict': {
 303             'id': '38983994',
 304             'ext': 'mp4',
 305             'title': 'Zero-Shot AutoML with Pretrained Models',
 306             'timestamp': 1662384834,
 307             'upload_date': '20220905',
 308             'thumbnail': r're:^https?://.*\.(?:jpg|png)',
 309             'thumbnails': 'count:23',
 310             'chapters': 'count:22',
 311             'duration': 295,
 312         },
 313         'params': {
 314             'skip_download': 'm3u8',
 315         },
 316     }, {
 317         # service_name = yoda
 318         'url': 'https://slideslive.com/38903721/magic-a-scientific-resurrection-of-an-esoteric-legend',
 319         'only_matching': True,
 320     }, {
 321         # dead link, service_name = url
 322         'url': 'https://slideslive.com/38922070/learning-transferable-skills-1',
 323         'only_matching': True,
 324     }, {
 325         # dead link, service_name = vimeo
 326         'url': 'https://slideslive.com/38921896/retrospectives-a-venue-for-selfreflection-in-ml-research-3',
 327         'only_matching': True,
 328     }]
 329
 330     _WEBPAGE_TESTS = [{
 331         # only XML slides info
 332         'url': 'https://iclr.cc/virtual_2020/poster_Hklr204Fvr.html',
 333         'info_dict': {
 334             'id': '38925850',
 335             'ext': 'mp4',
 336             'title': 'Towards a Deep Network Architecture for Structured Smoothness',
 337             'thumbnail': r're:^https?://.*\.jpg',
 338             'thumbnails': 'count:8',
 339             'timestamp': 1629671508,
 340             'upload_date': '20210822',
 341             'chapters': 'count:7',
 342             'duration': 326,
 343         },
 344         'params': {
 345             'skip_download': 'm3u8',
 346         },
 347     }]
 348
 349     @classmethod
 350     def _extract_embed_urls(cls, url, webpage):
 351         # Reference: https://slideslive.com/embed_presentation.js
 352         for embed_id in re.findall(r'(?s)new\s+SlidesLiveEmbed\s*\([^)]+\bpresentationId:\s*["\'](\d+)["\']', webpage):
 353             url_parsed = urllib.parse.urlparse(url)
 354             origin = f'{url_parsed.scheme}://{url_parsed.netloc}'
 355             yield update_url_query(
 356                 f'https://slideslive.com/embed/presentation/{embed_id}', {
 357                     'embed_parent_url': url,
 358                     'embed_container_origin': origin,
 359                 })
 360
 361     def _download_embed_webpage_handle(self, video_id, headers):
 362         return self._download_webpage_handle(
 363             f'https://slideslive.com/embed/presentation/{video_id}', video_id,
 364             headers=headers, query=traverse_obj(headers, {
 365                 'embed_parent_url': 'Referer',
 366                 'embed_container_origin': 'Origin',
 367             }))
 368
 369     def _extract_custom_m3u8_info(self, m3u8_data):
 370         m3u8_dict = {}
 371
 372         lookup = {
 373             'PRESENTATION-TITLE': 'title',
 374             'PRESENTATION-UPDATED-AT': 'timestamp',
 375             'PRESENTATION-THUMBNAIL': 'thumbnail',
 376             'PLAYLIST-TYPE': 'playlist_type',
 377             'VOD-VIDEO-SERVICE-NAME': 'service_name',
 378             'VOD-VIDEO-ID': 'service_id',
 379             'VOD-VIDEO-SERVERS': 'video_servers',
 380             'VOD-SUBTITLES': 'subtitles',
 381             'VOD-SLIDES-JSON-URL': 'slides_json_url',
 382             'VOD-SLIDES-XML-URL': 'slides_xml_url',
 383         }
 384
 385         for line in m3u8_data.splitlines():
 386             if not line.startswith('#EXT-SL-'):
 387                 continue
 388             tag, _, value = line.partition(':')
 389             key = lookup.get(tag[8:])
 390             if not key:
 391                 continue
 392             m3u8_dict[key] = value
 393
 394         # Some values are stringified JSON arrays
 395         for key in ('video_servers', 'subtitles'):
 396             if key in m3u8_dict:
 397                 m3u8_dict[key] = self._parse_json(m3u8_dict[key], None, fatal=False) or []
 398
 399         return m3u8_dict
 400
 401     def _extract_formats_and_duration(self, cdn_hostname, path, video_id, skip_duration=False):
 402         formats, duration = [], None
 403
 404         hls_formats = self._extract_m3u8_formats(
 405             f'https://{cdn_hostname}/{path}/master.m3u8',
 406             video_id, 'mp4', m3u8_id='hls', fatal=False, live=True)
 407         if hls_formats:
 408             if not skip_duration:
 409                 duration = self._extract_m3u8_vod_duration(
 410                     hls_formats[0]['url'], video_id, note='Extracting duration from HLS manifest')
 411             formats.extend(hls_formats)
 412
 413         dash_formats = self._extract_mpd_formats(
 414             f'https://{cdn_hostname}/{path}/master.mpd', video_id, mpd_id='dash', fatal=False)
 415         if dash_formats:
 416             if not duration and not skip_duration:
 417                 duration = self._extract_mpd_vod_duration(
 418                     f'https://{cdn_hostname}/{path}/master.mpd', video_id,
 419                     note='Extracting duration from DASH manifest')
 420             formats.extend(dash_formats)
 421
 422         return formats, duration
 423
 424     def _real_extract(self, url):
 425         video_id = self._match_id(url)
 426         webpage, urlh = self._download_embed_webpage_handle(
 427             video_id, headers=traverse_obj(parse_qs(url), {
 428                 'Referer': ('embed_parent_url', -1),
 429                 'Origin': ('embed_container_origin', -1)}))
 430         redirect_url = urlh.url
 431         if 'domain_not_allowed' in redirect_url:
 432             domain = traverse_obj(parse_qs(redirect_url), ('allowed_domains[]', ...), get_all=False)
 433             if not domain:
 434                 raise ExtractorError(
 435                     'This is an embed-only presentation. Try passing --referer', expected=True)
 436             webpage, _ = self._download_embed_webpage_handle(video_id, headers={
 437                 'Referer': f'https://{domain}/',
 438                 'Origin': f'https://{domain}',
 439             })
 440
 441         player_token = self._search_regex(r'data-player-token="([^"]+)"', webpage, 'player token')
 442         player_data = self._download_webpage(
 443             f'https://ben.slideslive.com/player/{video_id}', video_id,
 444             note='Downloading player info', query={'player_token': player_token})
 445         player_info = self._extract_custom_m3u8_info(player_data)
 446
 447         service_name = player_info['service_name'].lower()
 448         assert service_name in ('url', 'yoda', 'vimeo', 'youtube')
 449         service_id = player_info['service_id']
 450
 451         slide_url_template = 'https://slides.slideslive.com/%s/slides/original/%s%s'
 452         slides, slides_info = {}, []
 453
 454         if player_info.get('slides_json_url'):
 455             slides = self._download_json(
 456                 player_info['slides_json_url'], video_id, fatal=False,
 457                 note='Downloading slides JSON', errnote=False) or {}
 458             slide_ext_default = '.png'
 459             slide_quality = traverse_obj(slides, ('slide_qualities', 0))
 460             if slide_quality:
 461                 slide_ext_default = '.jpg'
 462                 slide_url_template = f'https://cdn.slideslive.com/data/presentations/%s/slides/{slide_quality}/%s%s'
 463             for slide_id, slide in enumerate(traverse_obj(slides, ('slides', ...), expected_type=dict), 1):
 464                 slides_info.append((
 465                     slide_id, traverse_obj(slide, ('image', 'name')),
 466                     traverse_obj(slide, ('image', 'extname'), default=slide_ext_default),
 467                     int_or_none(slide.get('time'), scale=1000)))
 468
 469         if not slides and player_info.get('slides_xml_url'):
 470             slides = self._download_xml(
 471                 player_info['slides_xml_url'], video_id, fatal=False,
 472                 note='Downloading slides XML', errnote='Failed to download slides info')
 473             if isinstance(slides, xml.etree.ElementTree.Element):
 474                 slide_url_template = 'https://cdn.slideslive.com/data/presentations/%s/slides/big/%s%s'
 475                 for slide_id, slide in enumerate(slides.findall('./slide')):
 476                     slides_info.append((
 477                         slide_id, xpath_text(slide, './slideName', 'name'), '.jpg',
 478                         int_or_none(xpath_text(slide, './timeSec', 'time'))))
 479
 480         chapters, thumbnails = [], []
 481         if url_or_none(player_info.get('thumbnail')):
 482             thumbnails.append({'id': 'cover', 'url': player_info['thumbnail']})
 483         for slide_id, slide_path, slide_ext, start_time in slides_info:
 484             if slide_path:
 485                 thumbnails.append({
 486                     'id': f'{slide_id:03d}',
 487                     'url': slide_url_template % (video_id, slide_path, slide_ext),
 488                 })
 489             chapters.append({
 490                 'title': f'Slide {slide_id:03d}',
 491                 'start_time': start_time,
 492             })
 493
 494         subtitles = {}
 495         for sub in traverse_obj(player_info, ('subtitles', ...), expected_type=dict):
 496             webvtt_url = url_or_none(sub.get('webvtt_url'))
 497             if not webvtt_url:
 498                 continue
 499             subtitles.setdefault(sub.get('language') or 'en', []).append({
 500                 'url': webvtt_url,
 501                 'ext': 'vtt',
 502             })
 503
 504         info = {
 505             'id': video_id,
 506             'title': player_info.get('title') or self._html_search_meta('title', webpage, default=''),
 507             'timestamp': unified_timestamp(player_info.get('timestamp')),
 508             'is_live': player_info.get('playlist_type') != 'vod',
 509             'thumbnails': thumbnails,
 510             'chapters': chapters,
 511             'subtitles': subtitles,
 512         }
 513
 514         if service_name == 'url':
 515             info['url'] = service_id
 516         elif service_name == 'yoda':
 517             formats, duration = self._extract_formats_and_duration(
 518                 player_info['video_servers'][0], service_id, video_id)
 519             info.update({
 520                 'duration': duration,
 521                 'formats': formats,
 522             })
 523         else:
 524             info.update({
 525                 '_type': 'url_transparent',
 526                 'url': service_id,
 527                 'ie_key': service_name.capitalize(),
 528                 'display_id': video_id,
 529             })
 530             if service_name == 'vimeo':
 531                 info['url'] = smuggle_url(
 532                     f'https://player.vimeo.com/video/{service_id}',
 533                     {'referer': url})
 534
 535         video_slides = traverse_obj(slides, ('slides', ..., 'video', 'id'))
 536         if not video_slides:
 537             return info
 538
 539         def entries():
 540             yield info
 541
 542             service_data = self._download_json(
 543                 f'https://ben.slideslive.com/player/{video_id}/slides_video_service_data',
 544                 video_id, fatal=False, query={
 545                     'player_token': player_token,
 546                     'videos': ','.join(video_slides),
 547                 }, note='Downloading video slides info', errnote='Failed to download video slides info') or {}
 548
 549             for slide_id, slide in enumerate(traverse_obj(slides, ('slides', ...)), 1):
 550                 if not traverse_obj(slide, ('video', 'service')) == 'yoda':
 551                     continue
 552                 video_path = traverse_obj(slide, ('video', 'id'))
 553                 cdn_hostname = traverse_obj(service_data, (
 554                     video_path, 'video_servers', ...), get_all=False)
 555                 if not cdn_hostname or not video_path:
 556                     continue
 557                 formats, _ = self._extract_formats_and_duration(
 558                     cdn_hostname, video_path, video_id, skip_duration=True)
 559                 if not formats:
 560                     continue
 561                 yield {
 562                     'id': f'{video_id}-{slide_id:03d}',
 563                     'title': f'{info["title"]} - Slide {slide_id:03d}',
 564                     'timestamp': info['timestamp'],
 565                     'duration': int_or_none(traverse_obj(slide, ('video', 'duration_ms')), scale=1000),
 566                     'formats': formats,
 567                 }
 568
 569         return self.playlist_result(entries(), f'{video_id}-playlist', info['title'])