1
votes

I'm developing a custom community connector for the Google Data Studio with the potential goal of publishing it for other users.

Basically it connects to an external REST endpoint, requests data in accordance to what the user has configured in the data source GUI, receives the data and transforms it so that the Google Data Studio can process it.
The connector is using the AuthType USER_PASS. Therefore, when the Google User is creating a data source from that connector, he will be asked for a user/password combination to be used to authenticate at that external REST endpoint. It looks somewhat like this:

Authenticate using <code>USER_PASS</code>

However, consider this scenario:

Google User A creates the data source out of the connector.

  1. He is configuring that data source to authenticate to the external service with the username user and the password password.
  2. He creates a report using that data source.
  3. And then another report.
  4. And then maybe another.
  5. He shares one of these reports with someone else

Now, Google User B receives an E-Mail which tells him there is a report he can view. He clicks on the link. Immediately, the getData() is being called. But it might not be, I didn't quite understand how the caching works. Maybe he is allowed to edit the report. So he does. After a significant change made to that report by B, getData() is being called anyway. But the data source wouldn't know which credentials it should use to authenticate to the external REST endpoint.

I played around with the various CacheServices and PropertiesServices to store that information. I've learned that Cache and Properties are basically the same with the exception that the Cache has a limited lifespan before it expires.

  • The DocumentProperties/DocumentCache is always null, because as I understand it it is not intended to be used from a connector.
  • The ScriptProperties/ScriptCache is shared across all instances of the connector, as in all data sources using that connector. Which is too restricted as maybe a user wishes to use that connector for multiple accounts for that REST API of that external service.
  • The UserProperties/UserCache is too limited, as it is different for Google User A and Google User B.

So the question is: Where should I store user and password to authenticate that instance of the connector to the external REST service?

1

1 Answers

1
votes

I got inspired by Session.getEffectiveUser(), which is different depending on the user under whose authority the script is running.

After some testing with a 2nd Google Account I shared a report with, I come to the conclusion that the UserProperties/UserCache behave differently, depending on the option that the report's creator has chosen on whose credentials access should be used. There is a guide about this here.

Basically, if you choose "Owner's credentials access", which is the default, the UserProperties/UserCache of the creator are being shared with every other viewer. Whereas if you choose "Viewer's credentials access" the UserProperties/UserCache of the current viewer are used.

This means that if you store the credentials in the creator's UserProperties, which I think is the recommended way, they are then shared with every viewer, because the viewers are using the creator's UserProperties and not their own.