_ Git - youtube-dl/blob - youtube_dl/extractor/ciscolive.py

   1 # coding: utf-8
   2 from __future__ import unicode_literals
   3
   4 import re
   5 from .common import InfoExtractor
   6 from ..compat import (
   7     compat_urllib_parse_urlparse,
   8     compat_parse_qs
   9 )
  10 from ..utils import (
  11     clean_html,
  12     int_or_none,
  13     try_get,
  14     urlencode_postdata,
  15 )
  16
  17
  18 class CiscoLiveIE(InfoExtractor):
  19     IE_NAME = 'ciscolive'
  20     _VALID_URL = r'(?:https?://)?ciscolive\.cisco\.com/on-demand-library/\??(?P<query>[^#]+)#/(?:session/(?P<id>.+))?$'
  21     _TESTS = [
  22         {
  23             'url': 'https://ciscolive.cisco.com/on-demand-library/?#/session/1423353499155001FoSs',
  24             'md5': 'c98acf395ed9c9f766941c70f5352e22',
  25             'info_dict': {
  26                 'id': '5803694304001',
  27                 'ext': 'mp4',
  28                 'title': '13 Smart Automations to Monitor Your Cisco IOS Network',
  29                 'description': 'md5:ec4a436019e09a918dec17714803f7cc',
  30                 'timestamp': 1530305395,
  31                 'uploader_id': '5647924234001',
  32                 'upload_date': '20180629',
  33                 'location': '16B Mezz.',
  34             },
  35         },
  36         {
  37             'url': 'https://ciscolive.cisco.com/on-demand-library/?search.event=ciscoliveus2018&search.technicallevel=scpsSkillLevel_aintroductory&search.focus=scpsSessionFocus_designAndDeployment#/',
  38             'md5': '993d4cf051f6174059328b1dce8e94bd',
  39             'info_dict': {
  40                 'upload_date': '20180629',
  41                 'title': 'DevNet Panel-Applying Design Thinking to Building Products in Cisco',
  42                 'timestamp': 1530316421,
  43                 'uploader_id': '5647924234001',
  44                 'id': '5803751616001',
  45                 'description': 'md5:5f144575cd6848117fe2f756855b038b',
  46                 'location': 'WoS, DevNet Theater',
  47                 'ext': 'mp4',
  48             },
  49         },
  50         {
  51             'url': 'https://ciscolive.cisco.com/on-demand-library/?search.technology=scpsTechnology_applicationDevelopment&search.technology=scpsTechnology_ipv6&search.focus=scpsSessionFocus_troubleshootingTroubleshooting#/',
  52             'md5': '80e0c3b87e373fe3a3316b934b8915bf',
  53             'info_dict': {
  54                 'upload_date': '20180629',
  55                 'title': 'Beating the CCIE Routing & Switching',
  56                 'timestamp': 1530311842,
  57                 'uploader_id': '5647924234001',
  58                 'id': '5803735679001',
  59                 'description': 'md5:e71970799e92d7f5ff57ae23f64b0929',
  60                 'location': 'Tulúm 02',
  61                 'ext': 'mp4',
  62             },
  63         }
  64     ]
  65
  66     # These appear to be constant across all Cisco Live presentations
  67     # and are not tied to any user session or event
  68     RAINFOCUS_API_URL = 'https://events.rainfocus.com/api/%s'
  69     RAINFOCUS_APIPROFILEID = 'Na3vqYdAlJFSxhYTYQGuMbpafMqftalz'
  70     RAINFOCUS_WIDGETID = 'n6l4Lo05R8fiy3RpUBm447dZN8uNWoye'
  71     BRIGHTCOVE_URL_TEMPLATE = 'http://players.brightcove.net/5647924234001/SyK2FdqjM_default/index.html?videoId=%s'
  72
  73     def _parse_rf_item(self, rf_item):
  74         ''' Parses metadata and passes to Brightcove extractor '''
  75         event_name = rf_item.get('eventName')
  76         title = rf_item['title']
  77         description = clean_html(rf_item.get('abstract'))
  78         presenter_name = try_get(rf_item, lambda x: x['participants'][0]['fullName'])
  79         bc_id = rf_item['videos'][0]['url']
  80         bc_url = self.BRIGHTCOVE_URL_TEMPLATE % bc_id
  81         duration = int_or_none(try_get(rf_item, lambda x: x['times'][0]['length']))
  82         location = try_get(rf_item, lambda x: x['times'][0]['room'])
  83
  84         if duration:
  85             duration = duration * 60
  86
  87         return {
  88             '_type': 'url_transparent',
  89             'creator': presenter_name,
  90             'description': description,
  91             'duration': duration,
  92             'ie_key': 'BrightcoveNew',
  93             'location': location,
  94             'series': event_name,
  95             'title': title,
  96             'url': bc_url,
  97         }
  98
  99     def _check_bc_id_exists(self, rf_item):
 100         ''' Checks for the existence of a Brightcove URL in an API result '''
 101         bc_id = try_get(rf_item, lambda x: x['videos'][0]['url'])
 102         if bc_id:
 103             if bc_id.strip().isdigit():
 104                 return rf_item
 105
 106     def _real_extract(self, url):
 107         mobj = re.match(self._VALID_URL, url)
 108         HEADERS = {
 109             'Origin': 'https://ciscolive.cisco.com',
 110             'rfApiProfileId': self.RAINFOCUS_APIPROFILEID,
 111             'rfWidgetId': self.RAINFOCUS_WIDGETID,
 112             'Referer': url,
 113         }
 114         # Single session URL (single video)
 115         if mobj.group('id'):
 116             rf_id = mobj.group('id')
 117             request = self.RAINFOCUS_API_URL % 'session'
 118             data = urlencode_postdata({'id': rf_id})
 119             rf_result = self._download_json(request, rf_id, data=data, headers=HEADERS)
 120             rf_item = self._check_bc_id_exists(rf_result['items'][0])
 121             return self._parse_rf_item(rf_item)
 122         else:
 123             # Filter query URL (multiple videos)
 124             rf_query = compat_parse_qs((compat_urllib_parse_urlparse(url).query))
 125             rf_query['type'] = 'session'
 126             rf_query['size'] = 1000
 127             data = urlencode_postdata(rf_query)
 128             request = self.RAINFOCUS_API_URL % 'search'
 129             rf_results = self._download_json(request, 'Filter query', data=data, headers=HEADERS)
 130             entries = [
 131                 self._parse_rf_item(rf_item)
 132                 for rf_item
 133                 in rf_results['sectionList'][0]['items']
 134                 if self._check_bc_id_exists(rf_item)
 135             ]
 136             return self.playlist_result(entries, 'Filter query')