Raphaël G. Git Repositories - youtubedl/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?://(?:www\.)?ciscolive(?:\.cisco)?\.com/[^#]*#/session/(?P<id>[^/?&]+)'
  69     _TESTS = [{
  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         'url': 'https://www.ciscolive.com/global/on-demand-library.html?search.event=ciscoliveemea2019#/session/15361595531500013WOU',
  84         'only_matching': True,
  85     }, {
  86         'url': 'https://www.ciscolive.com/global/on-demand-library.html?#/session/1490051371645001kNaS',
  87         'only_matching': True,
  88     }]
  89
  90     def _real_extract(self, url):
  91         rf_id = self._match_id(url)
  92         rf_result = self._call_api('session', rf_id, {'id': rf_id}, url)
  93         return self._parse_rf_item(rf_result['items'][0])
  94
  95
  96 class CiscoLiveSearchIE(CiscoLiveBaseIE):
  97     _VALID_URL = r'https?://(?:www\.)?ciscolive(?:\.cisco)?\.com/(?:global/)?on-demand-library(?:\.html|/)'
  98     _TESTS = [{
  99         'url': 'https://ciscolive.cisco.com/on-demand-library/?search.event=ciscoliveus2018&search.technicallevel=scpsSkillLevel_aintroductory&search.focus=scpsSessionFocus_designAndDeployment#/',
 100         'info_dict': {
 101             'title': 'Search query',
 102         },
 103         'playlist_count': 5,
 104     }, {
 105         'url': 'https://ciscolive.cisco.com/on-demand-library/?search.technology=scpsTechnology_applicationDevelopment&search.technology=scpsTechnology_ipv6&search.focus=scpsSessionFocus_troubleshootingTroubleshooting#/',
 106         'only_matching': True,
 107     }, {
 108         '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#/',
 109         'only_matching': True,
 110     }]
 111
 112     @classmethod
 113     def suitable(cls, url):
 114         return False if CiscoLiveSessionIE.suitable(url) else super(CiscoLiveSearchIE, cls).suitable(url)
 115
 116     @staticmethod
 117     def _check_bc_id_exists(rf_item):
 118         return int_or_none(try_get(rf_item, lambda x: x['videos'][0]['url'])) is not None
 119
 120     def _entries(self, query, url):
 121         query['size'] = 50
 122         query['from'] = 0
 123         for page_num in itertools.count(1):
 124             results = self._call_api(
 125                 'search', None, query, url,
 126                 'Downloading search JSON page %d' % page_num)
 127             sl = try_get(results, lambda x: x['sectionList'][0], dict)
 128             if sl:
 129                 results = sl
 130             items = results.get('items')
 131             if not items or not isinstance(items, list):
 132                 break
 133             for item in items:
 134                 if not isinstance(item, dict):
 135                     continue
 136                 if not self._check_bc_id_exists(item):
 137                     continue
 138                 yield self._parse_rf_item(item)
 139             size = int_or_none(results.get('size'))
 140             if size is not None:
 141                 query['size'] = size
 142             total = int_or_none(results.get('total'))
 143             if total is not None and query['from'] + query['size'] > total:
 144                 break
 145             query['from'] += query['size']
 146
 147     def _real_extract(self, url):
 148         query = compat_parse_qs(compat_urllib_parse_urlparse(url).query)
 149         query['type'] = 'session'
 150         return self.playlist_result(
 151             self._entries(query, url), playlist_title='Search query')