yt_dlp/extractor/ciscolive.py

   1 # coding: utf-8
   2 from __future__ import unicode_literals
   3
   4 import itertools
   5
   6 from .common import InfoExtractor
   7 from ..utils import (
   8     clean_html,
   9     float_or_none,
  10     int_or_none,
  11     parse_qs,
  12     try_get,
  13     urlencode_postdata,
  14 )
  15
  16
  17 class CiscoLiveBaseIE(InfoExtractor):
  18     # These appear to be constant across all Cisco Live presentations
  19     # and are not tied to any user session or event
  20     RAINFOCUS_API_URL = 'https://events.rainfocus.com/api/%s'
  21     RAINFOCUS_API_PROFILE_ID = 'Na3vqYdAlJFSxhYTYQGuMbpafMqftalz'
  22     RAINFOCUS_WIDGET_ID = 'n6l4Lo05R8fiy3RpUBm447dZN8uNWoye'
  23     BRIGHTCOVE_URL_TEMPLATE = 'http://players.brightcove.net/5647924234001/SyK2FdqjM_default/index.html?videoId=%s'
  24
  25     HEADERS = {
  26         'Origin': 'https://ciscolive.cisco.com',
  27         'rfApiProfileId': RAINFOCUS_API_PROFILE_ID,
  28         'rfWidgetId': RAINFOCUS_WIDGET_ID,
  29     }
  30
  31     def _call_api(self, ep, rf_id, query, referrer, note=None):
  32         headers = self.HEADERS.copy()
  33         headers['Referer'] = referrer
  34         return self._download_json(
  35             self.RAINFOCUS_API_URL % ep, rf_id, note=note,
  36             data=urlencode_postdata(query), headers=headers)
  37
  38     def _parse_rf_item(self, rf_item):
  39         event_name = rf_item.get('eventName')
  40         title = rf_item['title']
  41         description = clean_html(rf_item.get('abstract'))
  42         presenter_name = try_get(rf_item, lambda x: x['participants'][0]['fullName'])
  43         bc_id = rf_item['videos'][0]['url']
  44         bc_url = self.BRIGHTCOVE_URL_TEMPLATE % bc_id
  45         duration = float_or_none(try_get(rf_item, lambda x: x['times'][0]['length']))
  46         location = try_get(rf_item, lambda x: x['times'][0]['room'])
  47
  48         if duration:
  49             duration = duration * 60
  50
  51         return {
  52             '_type': 'url_transparent',
  53             'url': bc_url,
  54             'ie_key': 'BrightcoveNew',
  55             'title': title,
  56             'description': description,
  57             'duration': duration,
  58             'creator': presenter_name,
  59             'location': location,
  60             'series': event_name,
  61         }
  62
  63
  64 class CiscoLiveSessionIE(CiscoLiveBaseIE):
  65     _VALID_URL = r'https?://(?:www\.)?ciscolive(?:\.cisco)?\.com/[^#]*#/session/(?P<id>[^/?&]+)'
  66     _TESTS = [{
  67         'url': 'https://ciscolive.cisco.com/on-demand-library/?#/session/1423353499155001FoSs',
  68         'md5': 'c98acf395ed9c9f766941c70f5352e22',
  69         'info_dict': {
  70             'id': '5803694304001',
  71             'ext': 'mp4',
  72             'title': '13 Smart Automations to Monitor Your Cisco IOS Network',
  73             'description': 'md5:ec4a436019e09a918dec17714803f7cc',
  74             'timestamp': 1530305395,
  75             'upload_date': '20180629',
  76             'uploader_id': '5647924234001',
  77             'location': '16B Mezz.',
  78         },
  79     }, {
  80         'url': 'https://www.ciscolive.com/global/on-demand-library.html?search.event=ciscoliveemea2019#/session/15361595531500013WOU',
  81         'only_matching': True,
  82     }, {
  83         'url': 'https://www.ciscolive.com/global/on-demand-library.html?#/session/1490051371645001kNaS',
  84         'only_matching': True,
  85     }]
  86
  87     def _real_extract(self, url):
  88         rf_id = self._match_id(url)
  89         rf_result = self._call_api('session', rf_id, {'id': rf_id}, url)
  90         return self._parse_rf_item(rf_result['items'][0])
  91
  92
  93 class CiscoLiveSearchIE(CiscoLiveBaseIE):
  94     _VALID_URL = r'https?://(?:www\.)?ciscolive(?:\.cisco)?\.com/(?:global/)?on-demand-library(?:\.html|/)'
  95     _TESTS = [{
  96         'url': 'https://ciscolive.cisco.com/on-demand-library/?search.event=ciscoliveus2018&search.technicallevel=scpsSkillLevel_aintroductory&search.focus=scpsSessionFocus_designAndDeployment#/',
  97         'info_dict': {
  98             'title': 'Search query',
  99         },
 100         'playlist_count': 5,
 101     }, {
 102         'url': 'https://ciscolive.cisco.com/on-demand-library/?search.technology=scpsTechnology_applicationDevelopment&search.technology=scpsTechnology_ipv6&search.focus=scpsSessionFocus_troubleshootingTroubleshooting#/',
 103         'only_matching': True,
 104     }, {
 105         'url': 'https://www.ciscolive.com/global/on-demand-library.html?search.technicallevel=scpsSkillLevel_aintroductory&search.event=ciscoliveemea2019&search.technology=scpsTechnology_dataCenter&search.focus=scpsSessionFocus_bestPractices#/',
 106         'only_matching': True,
 107     }]
 108
 109     @classmethod
 110     def suitable(cls, url):
 111         return False if CiscoLiveSessionIE.suitable(url) else super(CiscoLiveSearchIE, cls).suitable(url)
 112
 113     @staticmethod
 114     def _check_bc_id_exists(rf_item):
 115         return int_or_none(try_get(rf_item, lambda x: x['videos'][0]['url'])) is not None
 116
 117     def _entries(self, query, url):
 118         query['size'] = 50
 119         query['from'] = 0
 120         for page_num in itertools.count(1):
 121             results = self._call_api(
 122                 'search', None, query, url,
 123                 'Downloading search JSON page %d' % page_num)
 124             sl = try_get(results, lambda x: x['sectionList'][0], dict)
 125             if sl:
 126                 results = sl
 127             items = results.get('items')
 128             if not items or not isinstance(items, list):
 129                 break
 130             for item in items:
 131                 if not isinstance(item, dict):
 132                     continue
 133                 if not self._check_bc_id_exists(item):
 134                     continue
 135                 yield self._parse_rf_item(item)
 136             size = int_or_none(results.get('size'))
 137             if size is not None:
 138                 query['size'] = size
 139             total = int_or_none(results.get('total'))
 140             if total is not None and query['from'] + query['size'] > total:
 141                 break
 142             query['from'] += query['size']
 143
 144     def _real_extract(self, url):
 145         query = parse_qs(url)
 146         query['type'] = 'session'
 147         return self.playlist_result(
 148             self._entries(query, url), playlist_title='Search query')