java:S3655 false positive when used with assertj

[barclay-reg (Robert Kleinschmager)]

We are running SonarQube DE 8.9.2

There is another rule (java:S5838) to advice to change our test code (which uses assertj)
from
assertThat(optionalValue.isPresent()).isTrue(); to
assertThat(optionalValue).isPresent();

after doing this change, we got another issue:

assertThat(optionalValue).isPresent();
Object value = optionalValue.get(); // here is the FP java:S3655
// do more assertions based on value, f.e. query db again

[barclay-reg (Robert Kleinschmager)]

Hi there - any feedback about this?

[Michael (Michael Gumowski)]

Hey @barclay-reg ,

Thank you for your patience, and sorry, we completely forgot to answer your message.

Okay, so there is a good reason why there is an FP here.

• Rule java:S3655 (Optional value access) is supposed to be executed only on MAIN code, and not on TEST code.
• Rule java:S5838 (Chained AssertJ assertions), is supposed to be executed only on TEST code, and not on MAIN code.

Consequently, and due to their different scope of execution, the implementation of rule S3655 has not been thought to handle assertion frameworks that you would usually find in test cases. It therefore do not consider the isPresent() call as being as legitimate presence test (it is).

Now, I don’t understand how the rule S3655 can be executed on what seems to be test code in your example (and recognized as such, since S5838 is executed). Did you change your default configuration to handle TEST code as MAIN code, and execute all rules on it?

Cheers,
Michael

[barclay-reg (Robert Kleinschmager)]

Hi @Michael ,

thanks for the reply. You were right, the questioned code lies within src/main/java (it’s common/reusable test code) so the problem is caused by our bad code structure.

But actually I didn’t noticed, that SonarQube rules have an association to TEST or MAIN. Is this a general concept or just implementation detail of the specific rules?
What about highlighting this - if general in the docu or issue browser, or if rule-specific within the rule description?

Anyway - many thanks to help me solving this.
kind regards
Robert