Fixed a problem not to decrypt user scope data store value in Jinja expression

[userlocalhost]

Summary

This fixes #4633.

Changing Point

This changes the implementation of custom Jinja filter decrypt_kv to be able to handle user scope value.

[CLAassistant]

All committers have signed the CLA.

[Kami]

It would also be a good idea to add "end to end" test which verifies that it works for the following scenarios:

1. API user is provided (if action is scheduled via API, API user should be used)
2. User is not provided - it should default to stanley (if execution is scheduled outside of the API context, no user context is available so stanley is used by default)

[Kami]

After some more digging in - it looks like this an "addition" to this change - #4463 and we already have some tests for value of the user in the context there.

I will go ahead and merge this as is, but going forward it also wouldn't hurt to add some additional end to end tests for this functionality.

[Kami]

While testing and reviewing this, I noticed there are more issues with this functionality.

1. User scoped datastore items are not supped in action-chain workflows.

I've fixed that in 42fe11b. I also updated an action which verifies this functionality - 8e57029.

2. Non-friendly error is throw if decrypt_kv Jinja function is used on non-existing datastore item

Previously Invalid or malformed ciphertext (too short) error was throw in such scenario which is rather confusing and not user-friendly.

Now in such scenario we throw Referenced Datastore item "st2kv.system.my_cmd" doesn't exist error.

I'm still working on test cases.

[Kami]

I've pushed a small refactor and tests for additional fixes I made - ba3c6a5.

[userlocalhost]

I will go ahead and merge this as is, but going forward it also wouldn't hurt to add some additional end to end tests for this functionality.

Thank you very much for your great help!

And sorry for late to reply. It took time to seek the place where is appropriate to extend for doing 'end to end' tests. At last, I reached the test of st2api/tests/unit/controllers/v1/test_executions.py which might be suitable to confirm this change from end to end.

Could you please give me a couple of days to write it?

[Kami]

@userlocalhost

Thank you for the original contribution.

And yes, that's the right place for that test. I will probably go ahead and merge this and you feel free to open a new PR for that tests so take your time - no rush.