Akkariin/youtube-dl
1
0
Fork 0
mirror of https://github.com/l1ving/youtube-dl synced 2025-03-11 02:37:18 +08:00
Code Issues Releases Wiki Activity

[ciscolive] Changes requested by upstream maintainers

Browse Source
This commit is contained in:
Austin de Coup-Crank 2018-10-29 14:03:17 -07:00
parent 8f7465ef54
commit 1b394685e8
1 changed files with 95 additions and 119 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

214
youtube_dl/extractor/ciscolive.py
View File

@ -4,179 +4,155 @@ from __future__ import unicode_literals
import re import re
from ..compat import compat_urllib_parse_urlencode from ..compat import compat_urllib_parse_urlencode
from .common import InfoExtractor from .common import InfoExtractor
from ..utils import smuggle_url
class CiscoLiveIE(InfoExtractor): class CiscoLiveIE(InfoExtractor):
IE_NAME = "ciscolive" IE_NAME = 'ciscolive'
_VALID_URL = r"https://ciscolive.cisco.com/on-demand-library/\??(?P<query>[^#]+)#/(?:session/(?P<id>.+))?$" _VALID_URL = r'https://ciscolive.cisco.com/on-demand-library/\??(?P<query>[^#]+)#/(?:session/(?P<id>.+))?$'
_TESTS = [ _TESTS = [
{ {
"url": "https://ciscolive.cisco.com/on-demand-library/?#/session/1423353499155001FoSs", 'url': 'https://ciscolive.cisco.com/on-demand-library/?#/session/1423353499155001FoSs',
"md5": "c98acf395ed9c9f766941c70f5352e22", 'md5': 'c98acf395ed9c9f766941c70f5352e22',
"info_dict": { 'info_dict': {
"id": "5803694304001", 'id': '5803694304001',
"ext": "mp4", 'ext': 'mp4',
"title": "13 Smart Automations to Monitor Your Cisco IOS Network", 'title': '13 Smart Automations to Monitor Your Cisco IOS Network [BRKNMS-2465]',
"timestamp": 1530305395, 'description': 'md5:9c8b286dea1e3cb479c4562f1c3e5000',
"uploader_id": "5647924234001", 'timestamp': 1530305395,
"upload_date": "20180629", 'uploader_id': '5647924234001',
'upload_date': '20180629',
'location': '16B Mezz.',
}, },
}, },
{ {
"url": "https://ciscolive.cisco.com/on-demand-library/?search.event=ciscoliveus2018&search.technicallevel=scpsSkillLevel_aintroductory&search.focus=scpsSessionFocus_designAndDeployment#/", 'url': 'https://ciscolive.cisco.com/on-demand-library/?search.event=ciscoliveus2018&search.technicallevel=scpsSkillLevel_aintroductory&search.focus=scpsSessionFocus_designAndDeployment#/',
"md5": "993d4cf051f6174059328b1dce8e94bd", 'md5': '993d4cf051f6174059328b1dce8e94bd',
"info_dict": { 'info_dict': {
"id": "5803751616001", 'upload_date': '20180629',
"ext": "mp4", 'title': 'DevNet Panel-Applying Design Thinking to Building Products in Cisco [DEVNET-1794]',
"timestamp": 1530316421, 'timestamp': 1530316421,
"title": "DevNet Panel-Applying Design Thinking to Building Products in Cisco", 'uploader_id': '5647924234001',
"uploader_id": "5647924234001", 'id': '5803751616001',
"upload_date": "20180629", 'description': 'md5:df02755cc961cc38950c36f53849ff1b',
'location': 'WoS, DevNet Theater',
'ext': 'mp4',
}, },
}, },
{ {
"url": "https://ciscolive.cisco.com/on-demand-library/?search.technology=scpsTechnology_applicationDevelopment&search.technology=scpsTechnology_ipv6&search.focus=scpsSessionFocus_troubleshootingTroubleshooting#/", 'url': 'https://ciscolive.cisco.com/on-demand-library/?search.technology=scpsTechnology_applicationDevelopment&search.technology=scpsTechnology_ipv6&search.focus=scpsSessionFocus_troubleshootingTroubleshooting#/',
"md5": "80e0c3b87e373fe3a3316b934b8915bf", 'md5': '80e0c3b87e373fe3a3316b934b8915bf',
"info_dict": { 'info_dict': {
"id": "5803735679001", 'upload_date': '20180629',
"ext": "mp4", 'title': 'Beating the CCIE Routing & Switching [BRKCCIE-9162]',
"timestamp": 1530311842, 'timestamp': 1530311842,
"title": "Beating the CCIE Routing & Switching", 'uploader_id': '5647924234001',
"uploader_id": "5647924234001", 'id': '5803735679001',
"upload_date": "20180629", 'description': 'md5:9e05b6772263276a5b8feef6f04887a1',
'location': 'Tulúm 02',
'ext': 'mp4',
}, },
}, }
{
"url": "https://ciscolive.cisco.com/on-demand-library/?search.solutions=scpsSolutions_cleanair#/",
"md5": "80e0c3b87e373fe3a3316b934b8915bf",
"info_dict": {
"id": "5803735679001",
"ext": "mp4",
"timestamp": 1530311842,
"title": "Beating the CCIE Routing & Switching",
"uploader_id": "5647924234001",
"upload_date": "20180629",
},
},
] ]
# These appear to be constant across all Cisco Live presentations # These appear to be constant across all Cisco Live presentations
# and are not tied to any user session or event # and are not tied to any user session or event
RAINFOCUS_API_URL = "https://events.rainfocus.com/api/%s" RAINFOCUS_API_URL = 'https://events.rainfocus.com/api/%s'
RAINFOCUS_APIPROFILEID = "Na3vqYdAlJFSxhYTYQGuMbpafMqftalz" RAINFOCUS_APIPROFILEID = 'Na3vqYdAlJFSxhYTYQGuMbpafMqftalz'
RAINFOCUS_WIDGETID = "n6l4Lo05R8fiy3RpUBm447dZN8uNWoye" RAINFOCUS_WIDGETID = 'n6l4Lo05R8fiy3RpUBm447dZN8uNWoye'
BRIGHTCOVE_URL_TEMPLATE = "http://players.brightcove.net/5647924234001/SyK2FdqjM_default/index.html?videoId=%s" BRIGHTCOVE_URL_TEMPLATE = 'http://players.brightcove.net/5647924234001/SyK2FdqjM_default/index.html?videoId=%s'
def _parse_rf_item(self, rf_item): def _parse_rf_item(self, rf_item):
""" Parses metadata and passes to Brightcove extractor ''' Parses metadata and passes to Brightcove extractor
""" '''
# Metadata parsed from Rainfocus API result # Metadata parsed from Rainfocus API result
# Not all of which is appropriate to pass to Brightcove extractor # Not all of which is appropriate to pass to Brightcove extractor
# but might be nice to print to output # but might be nice to print to output
event_name = rf_item.get("eventName") event_name = rf_item.get('eventName')
# Full event name [Cisco Live EMEA 2016] # Full event name [Cisco Live EMEA 2016]
# rf_id = rf_item.get('eventId') # rf_id = rf_item.get('eventId')
# Rainfocus ID [14382715417240cleu16] # Rainfocus ID [14382715417240cleu16]
cl_id = rf_item.get("abbreviation") cl_id = rf_item.get('abbreviation')
# Cisco Live ID - Shorthand session ID [BRKCRS-2501] # Cisco Live ID - Shorthand session ID [BRKCRS-2501]
title = rf_item.get("title") title = rf_item.get('title')
# Full session title [Campus QoS Design-Simplified] # Full session title [Campus QoS Design-Simplified]
description = rf_item.get("abstract") description = rf_item.get('abstract')
# Description [This session will apply Cisco's QoS strategy for rich media...] # Description [This session will apply Cisco's QoS strategy for rich media...]
# TODO: Needs safe get() method presenter_name = rf_item.get('participants')[0].get('fullName')
presenter_name = rf_item["participants"][0]["fullName"]
# Presenter's full name [Tim Szigeti] # Presenter's full name [Tim Szigeti]
presenter_title = rf_item["participants"][0]["jobTitle"] presenter_title = rf_item.get('participants')[0].get('jobTitle')
# Presenter's job title [Principal Engineer - Technical Marketing] # Presenter's job title [Principal Engineer - Technical Marketing]
pdf_url = rf_item["files"][0]["url"] pdf_url = rf_item.get('files')[0].get('url')
# Presentation PDF URL [https://clnv.s3.amazonaws.com/2016/eur/pdf/BRKCRS-2501.pdf] # Presentation PDF URL [https://clnv.s3.amazonaws.com/2016/eur/pdf/BRKCRS-2501.pdf]
bc_id = rf_item["videos"][0]["url"] bc_id = rf_item.get('videos')[0].get('url')
# Brightcove video ID [5803710412001] # Brightcove video ID [5803710412001]
bc_url = self.BRIGHTCOVE_URL_TEMPLATE % bc_id bc_url = self.BRIGHTCOVE_URL_TEMPLATE % bc_id
# Brightcove video URL [http://players.brightcove.net/5647924234001/SyK2FdqjM_default/index.html?videoId=5803710412001] # Brightcove video URL [http://players.brightcove.net/5647924234001/SyK2FdqjM_default/index.html?videoId=5803710412001]
duration = rf_item["times"][0]["length"] * 60 duration = rf_item.get('times')[0].get('length') * 60
# Duration. Provided in minutes * 60 = seconds [7200] # Duration. Provided in minutes * 60 = seconds [7200]
location = rf_item["times"][0]["room"] location = rf_item.get('times')[0].get('room')
# Location [Hall 7.3 Breakout Room 732] # Location [Hall 7.3 Breakout Room 732]
# Metadata passed to final Brightcove extractor return {
# TODO: Only title is passed--need to work on how to best merge smuggled metadata '_type': 'url_transparent',
metadata = { 'url': bc_url,
"id": cl_id, 'id': cl_id,
"title": title, 'title': '%s [%s]' % (title, cl_id),
"creator": presenter_name, 'creator': '%s, %s' % (presenter_name, presenter_title),
"description": description, 'description': '%s\nSlide Deck: %s' % (description, pdf_url),
"series": event_name, 'series': event_name,
"duration": duration, 'duration': duration,
"location": location, 'location': location,
'ie_key': 'BrightcoveNew',
} }
self.to_screen("Event Name: %s" % event_name)
self.to_screen("Session ID: %s" % cl_id)
self.to_screen("Session Title: %s" % title)
self.to_screen("Presenter: %s, %s" % (presenter_name, presenter_title))
self.to_screen("Slide Deck URL: %s" % pdf_url)
self.to_screen("Video URL: %s" % bc_url)
return self.url_result(
smuggle_url(bc_url, metadata), "BrightcoveNew", bc_id, title
)
def _check_bc_url_exists(self, rf_item): def _check_bc_url_exists(self, rf_item):
""" Checks for the existence of a Brightcove URL in a ''' Checks for the existence of a Brightcove URL in a
RainFocus result item RainFocus result item
""" '''
msg = (
"Skipping session that does not include a valid video URL: %s"
% rf_item.get("title", "Unknown title")
)
try: try:
bc_id = rf_item["videos"][0]["url"] bc_id = rf_item['videos'][0]['url']
mobj = re.match(r"\d+", bc_id) mobj = re.match(r'\d+', bc_id)
if mobj: if mobj:
return rf_item return rf_item
else: else:
self.report_warning(msg) pass
except IndexError: except IndexError:
self.report_warning(msg) pass
def _real_extract(self, url): def _real_extract(self, url):
mobj = re.match(self._VALID_URL, url) mobj = re.match(self._VALID_URL, url)
rf_headers = { headers = {
"Origin": "https://ciscolive.cisco.com", 'Origin': 'https://ciscolive.cisco.com',
"rfApiProfileId": self.RAINFOCUS_APIPROFILEID, 'rfApiProfileId': self.RAINFOCUS_APIPROFILEID,
"rfWidgetId": self.RAINFOCUS_WIDGETID, 'rfWidgetId': self.RAINFOCUS_WIDGETID,
"Referer": url, 'Referer': url,
} }
rf_args = {"video_id": None, "headers": rf_headers}
# Single session URL (single video) # Single session URL (single video)
if mobj.group("id"): if mobj.group('id'):
rf_id = mobj.group("id") rf_id = mobj.group('id')
rf_args["url_or_request"] = self.RAINFOCUS_API_URL % "session" request = self.RAINFOCUS_API_URL % 'session'
rf_args["video_id"] = rf_id data = compat_urllib_parse_urlencode({'id': rf_id})
rf_args["data"] = compat_urllib_parse_urlencode({"id": rf_id}) rf_result = self._download_json(request, rf_id, data=data,
rf_result = self._download_json(**rf_args) headers=headers)
rf_item = self._check_bc_url_exists(rf_result["items"][0]) rf_item = self._check_bc_url_exists(rf_result.get('items')[0])
return self._parse_rf_item(rf_item) return self._parse_rf_item(rf_item)
else: else:
# Filter query URL (multiple videos) # Filter query URL (multiple videos)
if mobj.group("query"): rf_query = mobj.group('query')
rf_query = mobj.group("query") rf_query = str(rf_query + '&type=session&size=1000')
rf_query = str(rf_query + "&type=session&size=1000") request = self.RAINFOCUS_API_URL % 'search'
rf_args["url_or_request"] = self.RAINFOCUS_API_URL % "search" # Query JSON results offer no obvious way to ID the search
rf_args["data"] = rf_query rf_results = self._download_json(request, 'Filter query',
# Query JSON results offer no obvious way to ID the search data=rf_query, headers=headers)
rf_args["video_id"] = "Filter query" # Not all sessions have videos; filter them out before moving on
self.to_screen('Video collection for filter query "%s"' % rf_query) rf_video_results = [
rf_results = self._download_json(**rf_args) rf_item
# Not all sessions have videos; filter them out before moving on for rf_item in rf_results.get('sectionList')[0].get('items')
rf_video_results = [ if self._check_bc_url_exists(rf_item)
rf_item ]
for rf_item in rf_results["sectionList"][0]["items"] entries = [self._parse_rf_item(rf_item) for rf_item in rf_video_results]
if self._check_bc_url_exists(rf_item) return self.playlist_result(entries)
]
entries = [self._parse_rf_item(rf_item) for rf_item in rf_video_results]
return self.playlist_result(entries)