Collection of Credentials
objects to be used alongside applications leveraging
the Cortex Data Lake API See
https://github.com/PaloAltoNetworks/pan-cortex-data-lake-nodejs
It also provides the CortexHubHelper
abstract class as well as a couple of
implementations for quick prototyping SaaS Components to interface with Cortex hub.
Incorporate the pan-cortex-hub
NodeJS package in your project with the following
bash command:
npm i @paloaltonetworks/pan-cortex-hub
You can also install the package from its GITHUB repo
npm i git://github.com/PaloAltoNetworks/pan-cortex-hub-nodejs
You can now import the package into your NodeJS code.
const cortex = require('@paloaltonetworks/pan-cortex-hub');
Source code is written in TypeScript and the build process productes type definition files which means you can leverage strongly type and code auto-complete features.
import * as cortex from '@paloaltonetworks/pan-cortex-hub'
Quick overview of available classes
The most basic of them all. It just wraps a static access_token
value
const cortex = require('@paloaltonetworks/pan-cortex-hub');
const ACCESS_TOKEN = 'eyaa...4t8t';
const cred = new cortex.StaticCredentials(ACCESS_TOKEN, cortex.cortexConstants.APIEPMAP.americas);
A credentials object that provides a refreshed access_token
from a known
OAuth2 refresh_token
(plus client_id
and client_secret
)
Best practise to keep secrets secure is to provide them using environmental variables.
PAN_CLIENT_ID=<client_id> \
PAN_CLIENT_SECRET=<client_secret> \
PAN_REFRESH_TOKEN=<refresj_token> \
node application.js
const cortex = require('@paloaltonetworks/pan-cortex-hub');
async function main() {
const cred = await cortex.SimpleCredentialsProvider.factory();
}
But, if needed, you can provide the secrets programatically.
const cortex = require('@paloaltonetworks/pan-cortex-hub');
async function main() {
const cred = await cortex.SimpleCredentialsProvider.factory({
clientId: CLIENT_ID,
clientSecret: CLIENT_SECRET,
refreshToken: REFRESH_TOKEN
});
}
Leverages a Token Redemption service (i.e. API Explorer)
Best practise is to provide the developer token using an environmental variable:
PAN_DEVELOPER_TOKEN=<developer_token> \
node application.js
const cortex = require('@paloaltonetworks/pan-cortex-hub');
const cred = cortex.DevTokenCredentials.factory({
developerTokenProvider: TOKEN_PROVIDER_URL
});
You can pass the developer token programatically if needed
const cortex = require('@paloaltonetworks/pan-cortex-hub');
const cred = cortex.DevTokenCredentials.factory({
developerToken: DEVELOPER_TOKEN,
developerTokenProvider: TOKEN_PROVIDER_URL
});
The following code snippet shows how to leverage this package's Credentials
collection with QueryServiceClient
objects from the pan-cortex-data-lake
package.
const dl = require("@paloaltonetworks/pan-cortex-data-lake");
const hub = require("@paloaltonetworks/pan-cortex-hub");
const SQLCMD = 'SELECT * FROM `<instance_id>.firewall.traffic` LIMIT 100';
const DEVELOPER_TOKEN = 'eyJh...BE9A';
const DEVELOPER_TOKEN_PROVIDER = 'https://app.apiexplorer.rocks/request_token';
async function main() {
const credentials = hub.DevTokenCredentials.factory({
developerToken: DEVELOPER_TOKEN,
developerTokenProvider: DEVELOPER_TOKEN_PROVIDER
});
const qsc = dl.QueryServiceClient.factory({ cortexDefCredentials: credentials });
for await (const page of qsc.iterator(SQLCMD)) {
console.log(page);
}
}
main().catch(console.error);
If your application grows to the point it needs to interface with multiple data
lake instances then you'll face the need to store multiple refresh_token
's.
This is the moment when you can leverage the CortexCredentialProvider
abstract
class. This class provides methods to cover the full life-cycle of a OAuth2
secret:
addWithRefreshToken()
: To register a new data lake instanceaddWithCode()
: To register a new data lake instance using the OAuth2 code (from the code grant flow)revokeDatalake()
: To revoke already issued refresh tokengetCredentialsObject(datalakeId)
: Retrieves aCredentials
object bound to the data lake identifier.
CortexCredentialProvider
is meant to be subclassed. Developer doing so must
implement the following storage methods that will be triggered when needed.
upsertStoreItem(dlid, item)
: to storeitem
as the valuer for data lake instancedlid
deleteStoreItem(dlid)
: remove the item for the data lake instancedlid
getStoreItem(dlid)
: retrieve the item for the data lake instancedlid
loadDb()
: perform initial database load
Subclass must call super(opts)
with an object with configuration options. The
only two mandatory options are:
clientId
: OAuth2 application client_id valueclientSecret
: OAuth2 application client_secret value
The library provides a CortexCredentialProvier
implementation that stores the
secrets in a local file using AES encryption of sensitive values. You can leverage this class for initial prototyping.
Secrets must me provided as environmental variables:
PAN_CLIENT_ID=<OAuth2 client_id> \
PAN_CLIENT_SECRET=<OAuth2 client_secret> \
PAN_SCRET=<AES Encryption key> \
node application.js
const cortex = require('@paloaltonetworks/pan-cortex-hub');
async function main() {
const credProvider = await cortex.FsCredProvider.factory();
}
Now you can register a refresh_token
you've received (i.e. at the end of a
OAuth2 code grant flow)
const cred = await credProvider.addWithRefreshToken(
'datalake-id',
cortex.cortexConstants.APIEPMAP.americas,
REFESH_TOKEN);
Or, if you want, you can use the CredentialProvider object to complete the OAuth2 code grant flow for you.
const cred = await credProvider.addWithCode(
'datalake-id',
cortex.cortexConstants.APIEPMAP.americas,
{ code: CODE, idpCallbackUrl: CALLBACK_URL });
In any case you receive at the end of the process a valid Credentials
object
bound to the provided OAuth2 instance.
Secrets keep stored in a file named PANCLOUD_CONFIG.json
(you can use another
file name using the option configFile
in the static factory()
method).
The static factory()
methods attempts to locate the database file and, if
found, then its content is loaded as initial data. That means that you retrieve
a credentials object for a data lake instance that was registered in another
work session.
const cred = await credProvider.getCredentialsObject('datalake-id');
CortexHubHelper
is an abstract class that provides two main features:
- Hooks to help onboard customers that are consuming applications through the
Cortex hub:
- Initial
params
parsing - Generation of the IDP Authentication Request URL
- Completing the OAuth2 code grant flow
- Initial
- Multi-tenancy: It automates a
CortexCredentialProvider
leveraging its metadada capability to organize data lakes into tenants.
See code examples in the /examples
folder