logo

oasis-root

Compiled tree of Oasis Linux based on own branch at <https://hacktivis.me/git/oasis/> git clone https://anongit.hacktivis.me/git/oasis-root.git

vidyard.py (18927B)

    

  1. import re
  3. from .common import InfoExtractor
  4. from ..utils import (
  5.     extract_attributes,
  6.     float_or_none,
  7.     int_or_none,
  8.     join_nonempty,
  9.     mimetype2ext,
  10.     parse_resolution,
  11.     str_or_none,
  12.     unescapeHTML,
  13.     url_or_none,
  14. )
  15. from ..utils.traversal import traverse_obj
  18. class VidyardBaseIE(InfoExtractor):
  19.     _HEADERS = {'Referer': 'https://play.vidyard.com/'}
  21.     def _get_formats_and_subtitles(self, sources, video_id):
  22.         formats, subtitles = [], {}
  24.         def add_hls_fmts_and_subs(m3u8_url):
  25.             fmts, subs = self._extract_m3u8_formats_and_subtitles(
  26.                 m3u8_url, video_id, 'mp4', m3u8_id='hls', headers=self._HEADERS, fatal=False)
  27.             formats.extend(fmts)
  28.             self._merge_subtitles(subs, target=subtitles)
  30.         hls_list = isinstance(sources, dict) and sources.pop('hls', None)
  31.         if master_m3u8_url := traverse_obj(
  32.                 hls_list, (lambda _, v: v['profile'] == 'auto', 'url', {url_or_none}, any)):
  33.             add_hls_fmts_and_subs(master_m3u8_url)
  34.         if not formats:  # These are duplicate and unnecesary requests if we got 'auto' hls fmts
  35.             for variant_m3u8_url in traverse_obj(hls_list, (..., 'url', {url_or_none})):
  36.                 add_hls_fmts_and_subs(variant_m3u8_url)
  38.         for source_type, source_list in traverse_obj(sources, ({dict.items}, ...)):
  39.             for source in traverse_obj(source_list, lambda _, v: url_or_none(v['url'])):
  40.                 profile = source.get('profile')
  41.                 formats.append({
  42.                     'url': source['url'],
  43.                     'ext': mimetype2ext(source.get('mimeType'), default=None),
  44.                     'format_id': join_nonempty('http', source_type, profile),
  45.                     **parse_resolution(profile),
  46.                 })
  48.         self._remove_duplicate_formats(formats)
  49.         return formats, subtitles
  51.     def _get_direct_subtitles(self, caption_json):
  52.         subs = {}
  53.         for caption in traverse_obj(caption_json, lambda _, v: url_or_none(v['vttUrl'])):
  54.             subs.setdefault(caption.get('language') or 'und', []).append({
  55.                 'url': caption['vttUrl'],
  56.                 'name': caption.get('name'),
  57.             })
  59.         return subs
  61.     def _fetch_video_json(self, video_id):
  62.         return self._download_json(
  63.             f'https://play.vidyard.com/player/{video_id}.json', video_id)['payload']
  65.     def _process_video_json(self, json_data, video_id):
  66.         formats, subtitles = self._get_formats_and_subtitles(json_data['sources'], video_id)
  67.         self._merge_subtitles(self._get_direct_subtitles(json_data.get('captions')), target=subtitles)
  69.         return {
  70.             **traverse_obj(json_data, {
  71.                 'id': ('facadeUuid', {str}),
  72.                 'display_id': ('videoId', {int}, {str_or_none}),
  73.                 'title': ('name', {str}),
  74.                 'description': ('description', {str}, {unescapeHTML}, filter),
  75.                 'duration': ((
  76.                     ('milliseconds', {float_or_none(scale=1000)}),
  77.                     ('seconds', {int_or_none})), any),
  78.                 'thumbnails': ('thumbnailUrls', ('small', 'normal'), {'url': {url_or_none}}),
  79.                 'tags': ('tags', ..., 'name', {str}),
  80.             }),
  81.             'formats': formats,
  82.             'subtitles': subtitles,
  83.             'http_headers': self._HEADERS,
  84.         }
  87. class VidyardIE(VidyardBaseIE):
  88.     _VALID_URL = [
  89.         r'https?://[\w-]+(?:\.hubs)?\.vidyard\.com/watch/(?P<id>[\w-]+)',
  90.         r'https?://(?:embed|share)\.vidyard\.com/share/(?P<id>[\w-]+)',
  91.         r'https?://play\.vidyard\.com/(?:player/)?(?P<id>[\w-]+)',
  92.     ]
  93.     _EMBED_REGEX = [r'<iframe[^>]* src=["\'](?P<url>(?:https?:)?//play\.vidyard\.com/[\w-]+)']
  94.     _TESTS = [{
  95.         'url': 'https://vyexample03.hubs.vidyard.com/watch/oTDMPlUv--51Th455G5u7Q',
  96.         'info_dict': {
  97.             'id': 'oTDMPlUv--51Th455G5u7Q',
  98.             'display_id': '50347',
  99.             'ext': 'mp4',
  100.             'title': 'Homepage Video',
  101.             'description': 'Look I changed the description.',
  102.             'thumbnail': 'https://cdn.vidyard.com/thumbnails/50347/OUPa5LTKV46849sLYngMqQ_small.jpg',
  103.             'duration': 99,
  104.             'tags': ['these', 'are', 'all', 'tags'],
  105.         },
  106.     }, {
  107.         'url': 'https://share.vidyard.com/watch/PaQzDAT1h8JqB8ivEu2j6Y?',
  108.         'info_dict': {
  109.             'id': 'PaQzDAT1h8JqB8ivEu2j6Y',
  110.             'display_id': '9281024',
  111.             'ext': 'mp4',
  112.             'title': 'Inline Embed',
  113.             'thumbnail': 'https://cdn.vidyard.com/thumbnails/spacer.gif',
  114.             'duration': 41.186,
  115.         },
  116.     }, {
  117.         'url': 'https://embed.vidyard.com/share/oTDMPlUv--51Th455G5u7Q',
  118.         'info_dict': {
  119.             'id': 'oTDMPlUv--51Th455G5u7Q',
  120.             'display_id': '50347',
  121.             'ext': 'mp4',
  122.             'title': 'Homepage Video',
  123.             'description': 'Look I changed the description.',
  124.             'thumbnail': 'https://cdn.vidyard.com/thumbnails/50347/OUPa5LTKV46849sLYngMqQ_small.jpg',
  125.             'duration': 99,
  126.             'tags': ['these', 'are', 'all', 'tags'],
  127.         },
  128.     }, {
  129.         # First video from playlist below
  130.         'url': 'https://embed.vidyard.com/share/SyStyHtYujcBHe5PkZc5DL',
  131.         'info_dict': {
  132.             'id': 'SyStyHtYujcBHe5PkZc5DL',
  133.             'display_id': '41974005',
  134.             'ext': 'mp4',
  135.             'title': 'Prepare the Frame and Track for Palm Beach Polysatin Shutters With BiFold Track',
  136.             'description': r're:In this video, you will learn how to prepare the frame.+',
  137.             'thumbnail': 'https://cdn.vidyard.com/thumbnails/41974005/IJw7oCaJcF1h7WWu3OVZ8A_small.png',
  138.             'duration': 258.666,
  139.         },
  140.     }, {
  141.         # Playlist
  142.         'url': 'https://thelink.hubs.vidyard.com/watch/pwu7pCYWSwAnPxs8nDoFrE',
  143.         'info_dict': {
  144.             'id': 'pwu7pCYWSwAnPxs8nDoFrE',
  145.             'title': 'PLAYLIST - Palm Beach Shutters- Bi-Fold Track System Installation',
  146.             'entries': [{
  147.                 'id': 'SyStyHtYujcBHe5PkZc5DL',
  148.                 'display_id': '41974005',
  149.                 'ext': 'mp4',
  150.                 'title': 'Prepare the Frame and Track for Palm Beach Polysatin Shutters With BiFold Track',
  151.                 'thumbnail': 'https://cdn.vidyard.com/thumbnails/41974005/IJw7oCaJcF1h7WWu3OVZ8A_small.png',
  152.                 'duration': 258.666,
  153.             }, {
  154.                 'id': '1Fw4B84jZTXLXWqkE71RiM',
  155.                 'display_id': '5861113',
  156.                 'ext': 'mp4',
  157.                 'title': 'Palm Beach - Bi-Fold Track System "Frame Installation"',
  158.                 'thumbnail': 'https://cdn.vidyard.com/thumbnails/5861113/29CJ54s5g1_aP38zkKLHew_small.jpg',
  159.                 'duration': 167.858,
  160.             }, {
  161.                 'id': 'DqP3wBvLXSpxrcqpT5kEeo',
  162.                 'display_id': '41976334',
  163.                 'ext': 'mp4',
  164.                 'title': 'Install the Track for Palm Beach Polysatin Shutters With BiFold Track',
  165.                 'thumbnail': 'https://cdn.vidyard.com/thumbnails/5861090/RwG2VaTylUa6KhSTED1r1Q_small.png',
  166.                 'duration': 94.229,
  167.             }, {
  168.                 'id': 'opfybfxpzQArxqtQYB6oBU',
  169.                 'display_id': '41976364',
  170.                 'ext': 'mp4',
  171.                 'title': 'Install the Panel for Palm Beach Polysatin Shutters With BiFold Track',
  172.                 'thumbnail': 'https://cdn.vidyard.com/thumbnails/5860926/JIOaJR08dM4QgXi_iQ2zGA_small.png',
  173.                 'duration': 191.467,
  174.             }, {
  175.                 'id': 'rWrXvkbTNNaNqD6189HJya',
  176.                 'display_id': '41976382',
  177.                 'ext': 'mp4',
  178.                 'title': 'Adjust the Panels for Palm Beach Polysatin Shutters With BiFold Track',
  179.                 'thumbnail': 'https://cdn.vidyard.com/thumbnails/5860687/CwHxBv4UudAhOh43FVB4tw_small.png',
  180.                 'duration': 138.155,
  181.             }, {
  182.                 'id': 'eYPTB521MZ9TPEArSethQ5',
  183.                 'display_id': '41976409',
  184.                 'ext': 'mp4',
  185.                 'title': 'Assemble and Install the Valance for Palm Beach Polysatin Shutters With BiFold Track',
  186.                 'thumbnail': 'https://cdn.vidyard.com/thumbnails/5861425/0y68qlMU4O5VKU7bJ8i_AA_small.png',
  187.                 'duration': 148.224,
  188.             }],
  189.         },
  190.         'playlist_count': 6,
  191.     }, {
  192.         # Non hubs.vidyard.com playlist
  193.         'url': 'https://salesforce.vidyard.com/watch/d4vqPjs7Q5EzVEis5QT3jd',
  194.         'info_dict': {
  195.             'id': 'd4vqPjs7Q5EzVEis5QT3jd',
  196.             'title': 'How To: Service Cloud: Import External Content in Lightning Knowledge',
  197.             'entries': [{
  198.                 'id': 'mcjDpSZir2iSttbvFkx6Rv',
  199.                 'display_id': '29479036',
  200.                 'ext': 'mp4',
  201.                 'title': 'Welcome to this Expert Coaching Series',
  202.                 'thumbnail': 'https://cdn.vidyard.com/thumbnails/ouyQi9WuwyiOupChUWNmjQ/7170d3485ba602e012df05_small.jpg',
  203.                 'duration': 38.205,
  204.             }, {
  205.                 'id': '84bPYwpg243G6xYEfJdYw9',
  206.                 'display_id': '21820704',
  207.                 'ext': 'mp4',
  208.                 'title': 'Chapter 1 - Title + Agenda',
  209.                 'thumbnail': 'https://cdn.vidyard.com/thumbnails/HFPN0ZgQq4Ow8BghGcQSow/bfaa30123c8f6601e7d7f2_small.jpg',
  210.                 'duration': 98.016,
  211.             }, {
  212.                 'id': 'nP17fMuvA66buVHUrzqjTi',
  213.                 'display_id': '21820707',
  214.                 'ext': 'mp4',
  215.                 'title': 'Chapter 2 - Import Options',
  216.                 'thumbnail': 'https://cdn.vidyard.com/thumbnails/rGRIF5nFjPI9OOA2qJ_Dbg/86a8d02bfec9a566845dd4_small.jpg',
  217.                 'duration': 199.136,
  218.             }, {
  219.                 'id': 'm54EcwXdpA5gDBH5rgCYoV',
  220.                 'display_id': '21820710',
  221.                 'ext': 'mp4',
  222.                 'title': 'Chapter 3 - Importing Article Translations',
  223.                 'thumbnail': 'https://cdn.vidyard.com/thumbnails/IVX4XR8zpSsiNIHx45kz-A/1ccbf8a29a33856d06b3ed_small.jpg',
  224.                 'duration': 184.352,
  225.             }, {
  226.                 'id': 'j4nzS42oq4hE9oRV73w3eQ',
  227.                 'display_id': '21820716',
  228.                 'ext': 'mp4',
  229.                 'title': 'Chapter 4 - Best Practices',
  230.                 'thumbnail': 'https://cdn.vidyard.com/thumbnails/BtrRrQpRDLbA4AT95YQyog/1f1e6b8e7fdc3fa95ec8d3_small.jpg',
  231.                 'duration': 296.960,
  232.             }, {
  233.                 'id': 'y28PYfW5pftvers9PXzisC',
  234.                 'display_id': '21820727',
  235.                 'ext': 'mp4',
  236.                 'title': 'Chapter 5 - Migration Steps',
  237.                 'thumbnail': 'https://cdn.vidyard.com/thumbnails/K2CdQOXDfLcrVTF60r0bdw/a09239ada28b6ffce12b1f_small.jpg',
  238.                 'duration': 620.640,
  239.             }, {
  240.                 'id': 'YWU1eQxYvhj29SjYoPw5jH',
  241.                 'display_id': '21820733',
  242.                 'ext': 'mp4',
  243.                 'title': 'Chapter 6 - Demo',
  244.                 'thumbnail': 'https://cdn.vidyard.com/thumbnails/rsmhP-cO8dAa8ilvFGCX0g/7911ef415167cd14032068_small.jpg',
  245.                 'duration': 631.456,
  246.             }, {
  247.                 'id': 'nmEvVqpwdJUgb74zKsLGxn',
  248.                 'display_id': '29479037',
  249.                 'ext': 'mp4',
  250.                 'title': 'Schedule Your Follow-Up',
  251.                 'thumbnail': 'https://cdn.vidyard.com/thumbnails/Rtwc7X4PEkF4Ae5kHi-Jvw/174ebed3f34227b1ffa1d0_small.jpg',
  252.                 'duration': 33.608,
  253.             }],
  254.         },
  255.         'playlist_count': 8,
  256.     }, {
  257.         # URL of iframe embed src
  258.         'url': 'https://play.vidyard.com/iDqTwWGrd36vaLuaCY3nTs.html',
  259.         'info_dict': {
  260.             'id': 'iDqTwWGrd36vaLuaCY3nTs',
  261.             'display_id': '9281009',
  262.             'ext': 'mp4',
  263.             'title': 'Lightbox Embed',
  264.             'thumbnail': 'https://cdn.vidyard.com/thumbnails/spacer.gif',
  265.             'duration': 39.035,
  266.         },
  267.     }, {
  268.         # Player JSON URL
  269.         'url': 'https://play.vidyard.com/player/7GAApnNNbcZZ46k6JqJQSh.json?disable_analytics=0',
  270.         'info_dict': {
  271.             'id': '7GAApnNNbcZZ46k6JqJQSh',
  272.             'display_id': '820026',
  273.             'ext': 'mp4',
  274.             'title': 'The Art of Storytelling: How to Deliver Your Brand Story with Content & Social',
  275.             'thumbnail': 'https://cdn.vidyard.com/thumbnails/MhbE-5sEFQu4x3fI6FkNlA/41eb5717c557cd19456910_small.jpg',
  276.             'duration': 2153.013,
  277.             'tags': ['Summit2017'],
  278.         },
  279.     }, {
  280.         'url': 'http://share.vidyard.com/share/diYeo6YR2yiGgL8odvS8Ri',
  281.         'only_matching': True,
  282.     }, {
  283.         'url': 'https://play.vidyard.com/FFlz3ZpxhIfKQ1fd9DAryA',
  284.         'only_matching': True,
  285.     }, {
  286.         'url': 'https://play.vidyard.com/qhMAu5A76GZVrFzOPgSf9A/type/standalone',
  287.         'only_matching': True,
  288.     }]
  289.     _WEBPAGE_TESTS = [{
  290.         # URL containing inline/lightbox embedded video
  291.         'url': 'https://resources.altium.com/p/2-the-extreme-importance-of-pc-board-stack-up',
  292.         'info_dict': {
  293.             'id': 'GDx1oXrFWj4XHbipfoXaMn',
  294.             'display_id': '3225198',
  295.             'ext': 'mp4',
  296.             'title': 'The Extreme Importance of PC Board Stack Up',
  297.             'thumbnail': 'https://cdn.vidyard.com/thumbnails/73_Q3_hBexWX7Og1sae6cg/9998fa4faec921439e2c04_small.jpg',
  298.             'duration': 3422.742,
  299.         },
  300.     }, {
  301.         # <script ... id="vidyard_embed_code_DXx2sW4WaLA6hTdGFz7ja8" src="//play.vidyard.com/DXx2sW4WaLA6hTdGFz7ja8.js?
  302.         'url': 'http://videos.vivint.com/watch/DXx2sW4WaLA6hTdGFz7ja8',
  303.         'info_dict': {
  304.             'id': 'DXx2sW4WaLA6hTdGFz7ja8',
  305.             'display_id': '2746529',
  306.             'ext': 'mp4',
  307.             'title': 'How To Powercycle the Smart Hub Panel',
  308.             'duration': 30.613,
  309.             'thumbnail': 'https://cdn.vidyard.com/thumbnails/_-6cw8xQUJ3qiCs_JENc_A/b21d7a5e47967f49399d30_small.jpg',
  310.         },
  311.     }, {
  312.         # <script id="vidyard_embed_code_MIBHhiLVTxga7wqLsuoDjQ" src="//embed.vidyard.com/embed/MIBHhiLVTxga7wqLsuoDjQ/inline?v=2.1">
  313.         'url': 'https://www.babypips.com/learn/forex/introduction-to-metatrader4',
  314.         'info_dict': {
  315.             'id': 'MIBHhiLVTxga7wqLsuoDjQ',
  316.             'display_id': '20291',
  317.             'ext': 'mp4',
  318.             'title': 'Lesson 1 - Opening an MT4 Account',
  319.             'description': 'Never heard of MetaTrader4? Here\'s the 411 on the popular trading platform!',
  320.             'duration': 168,
  321.             'thumbnail': 'https://cdn.vidyard.com/thumbnails/20291/IM-G2WXQR9VBLl2Cmzvftg_small.jpg',
  322.         },
  323.     }, {
  324.         # <iframe ... src="//play.vidyard.com/d61w8EQoZv1LDuPxDkQP2Q/type/background?preview=1"
  325.         'url': 'https://www.avaya.com/en/',
  326.         'info_dict': {
  327.             # These values come from the generic extractor and don't matter
  328.             'id': str,
  329.             'title': str,
  330.             'age_limit': 0,
  331.             'upload_date': str,
  332.             'description': str,
  333.             'thumbnail': str,
  334.             'timestamp': float,
  335.         },
  336.         'playlist': [{
  337.             'info_dict': {
  338.                 'id': 'd61w8EQoZv1LDuPxDkQP2Q',
  339.                 'display_id': '42456529',
  340.                 'ext': 'mp4',
  341.                 'title': 'GettyImages-1027',
  342.                 'duration': 6.0,
  343.                 'thumbnail': 'https://cdn.vidyard.com/thumbnails/42061563/p6bY08d2N4e4IDz-7J4_wkgsPq3-qgcx_small.jpg',
  344.             },
  345.         }, {
  346.             'info_dict': {
  347.                 'id': 'VAsYDi7eiqZRbHodUA2meC',
  348.                 'display_id': '42456569',
  349.                 'ext': 'mp4',
  350.                 'title': 'GettyImages-1325598833',
  351.                 'duration': 6.083,
  352.                 'thumbnail': 'https://cdn.vidyard.com/thumbnails/42052358/y3qrbDpn_2quWr_5XBi7yzS3UvEI__ZM_small.jpg',
  353.             },
  354.         }],
  355.         'playlist_count': 2,
  356.     }, {
  357.         # <div class="vidyard-player-embed" data-uuid="vpCWTVHw3qrciLtVY94YkS"
  358.         'url': 'https://www.gogoair.com/',
  359.         'info_dict': {
  360.             # These values come from the generic extractor and don't matter
  361.             'id': str,
  362.             'title': str,
  363.             'description': str,
  364.             'age_limit': 0,
  365.         },
  366.         'playlist': [{
  367.             'info_dict': {
  368.                 'id': 'vpCWTVHw3qrciLtVY94YkS',
  369.                 'display_id': '40780699',
  370.                 'ext': 'mp4',
  371.                 'title': 'Upgrade to AVANCE 100% worth it - Jason Talley, Owner and Pilot, Testimonial',
  372.                 'description': 'md5:f609824839439a51990cef55ffc472aa',
  373.                 'duration': 70.737,
  374.                 'thumbnail': 'https://cdn.vidyard.com/thumbnails/40780699/KzjfYZz5MZl2gHF_e-4i2c6ib1cLDweQ_small.jpg',
  375.             },
  376.         }, {
  377.             'info_dict': {
  378.                 'id': 'xAmV9AsLbnitCw35paLBD8',
  379.                 'display_id': '31130867',
  380.                 'ext': 'mp4',
  381.                 'title': 'Brad Keselowski goes faster with Gogo AVANCE inflight Wi-Fi',
  382.                 'duration': 132.565,
  383.                 'thumbnail': 'https://cdn.vidyard.com/thumbnails/31130867/HknyDtLdm2Eih9JZ4A5XLjhfBX_6HRw5_small.jpg',
  384.             },
  385.         }, {
  386.             'info_dict': {
  387.                 'id': 'RkkrFRNxfP79nwCQavecpF',
  388.                 'display_id': '39009815',
  389.                 'ext': 'mp4',
  390.                 'title': 'Live Demo of Gogo Galileo',
  391.                 'description': 'md5:e2df497236f4e12c3fef8b392b5f23e0',
  392.                 'duration': 112.128,
  393.                 'thumbnail': 'https://cdn.vidyard.com/thumbnails/38144873/CWLlxfUbJ4Gh0ThbUum89IsEM4yupzMb_small.jpg',
  394.             },
  395.         }],
  396.         'playlist_count': 3,
  397.     }]
  399.     @classmethod
  400.     def _extract_embed_urls(cls, url, webpage):
  401.         # Handle protocol-less embed URLs
  402.         for embed_url in super()._extract_embed_urls(url, webpage):
  403.             if embed_url.startswith('//'):
  404.                 embed_url = f'https:{embed_url}'
  405.             yield embed_url
  407.         # Extract inline/lightbox embeds
  408.         for embed_element in re.findall(
  409.                 r'(<(?:img|div)[^>]* class=(["\'])(?:[^>"\']* )?vidyard-player-embed(?: [^>"\']*)?\2[^>]+>)', webpage):
  410.             if video_id := extract_attributes(embed_element[0]).get('data-uuid'):
  411.                 yield f'https://play.vidyard.com/{video_id}'
  413.         for embed_id in re.findall(r'<script[^>]* id=["\']vidyard_embed_code_([\w-]+)["\']', webpage):
  414.             yield f'https://play.vidyard.com/{embed_id}'
  416.     def _real_extract(self, url):
  417.         video_id = self._match_id(url)
  418.         video_json = self._fetch_video_json(video_id)
  420.         if len(video_json['chapters']) == 1:
  421.             return self._process_video_json(video_json['chapters'][0], video_id)
  423.         return self.playlist_result(
  424.             (self._process_video_json(chapter, video_id) for chapter in video_json['chapters']),
  425.             str(video_json['playerUuid']), video_json.get('name'))