Fix imports in Kotlin model generation

[MariusVolkhart]

The models project generated Kotlin code that didn't have the correct imports. This happened because the imports were not being tracked as imports in the KotlinCodeGenerator, only as code.

[edward3h]

Looks good to me.

If you could add a test case that would catch this, that would be even better.

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Comparison is base (0c83398) 91.20% compared to head (af3b52f) 91.14%.

Additional details and impacted files

@@             Coverage Diff              @@
##               main     #309      +/-   ##
============================================
- Coverage     91.20%   91.14%   -0.07%     
+ Complexity     1204     1203       -1     
============================================
  Files            76       76              
  Lines          3139     3140       +1     
  Branches        489      489              
============================================
- Hits           2863     2862       -1     
- Misses          167      168       +1     
- Partials        109      110       +1     

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

[casid]

Sorry, I just merged a bit too fast. But the change itself is solid!

@edward3h is there a test for the Java way we could adopt for Kotlin?

[MariusVolkhart]

I'd probably go the route of writing a test that parses a template, produces generated files, and then compares the generated files against golden versions. Thoughts on that?

The challenge is that the jte-models project is the only place where the imports are actually read. The KotlinCodeGenerator doesn't care about the list of imports as it generates code directly.

[casid]

Indeed, I think writing a test from the jte-models perspective would be ideal, since that's the project that needs the imports as you said.