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