Allow Code-geneated Moshi Adapters

[macisamuele]

The goal of this PR is pretty straight forward it plugs-in moshi-kotlin-codegen into the generated code.

The main difference in the code generated by swagger-gradle-codegen is the presence of @JsonClass(generateAdapter = true) within the code.

This annotation allows KotlinJsonAdapterFactory to identify the JsonAdapter generated via kotlin annotation processor (kapt) if the kapt directive (look into samples/junit-tests/build.gradle changes) is enabled.

This approach leads to a no-op change for users of the current tool but empowers them to get better runtime performances by enabling kapt.
Disclaimer: I've not verified the performance improvements but considering that this is an opt-in approach I'm not concerned about it

Fixes #68
Fixes #62
Fixes #71

[cortinico]

Generally good 👍 Still we have a couple of things to address:

• You need to remove .add(KotlinJsonAdapterFactory()) from the GeneratedCodeConverters as @filipemp did here: https://github.com/Yelp/swagger-gradle-codegen/pull/68/files

• You need to provide Proguard rules to make sure that type adapters are not removed from the generated artifact.

[cortinico]

Ideally if you could rebase your branch in top of Filipe's would be great so we have his contribution in the graph as well.

…

On Wed, Jan 22, 2020, 13:30 Samuele Maci ***@***.***> wrote: ***@***.**** commented on this pull request. ------------------------------ In plugin/src/main/java/com/yelp/codegen/KotlinGenerator.kt <#82 (comment)> : > + // Import JsonClass annotation to enable Adapters generations via Kotlin Annotation Processor + codegenModel.imports.add("com.squareup.moshi.JsonClass") I totally missed the other PRs that were opened ... sorry :( I'll check if we can move the import into the if statement, I'll let you know soon about the progress. — You are receiving this because your review was requested. Reply to this email directly, view it on GitHub <#82?email_source=notifications&email_token=AAW44ZMM554CU3RMPSL4KYTQ7A375A5CNFSM4KKCMRNKYY3PNVWWK3TUL52HS4DFWFIHK3DMKJSXC5LFON2FEZLWNFSXPKTDN5WW2ZLOORPWSZGOCST7IBA#discussion_r369532244>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AAW44ZL57ZHMV5MADHZWBITQ7A375ANCNFSM4KKCMRNA> .

[macisamuele]

Sure thing, as mentioned I did not see (I should have looked more carefully) PRs already trying to address the same issue.
As soon as I got some extra time I'll try to verify the presence of blockers and ensure that attribution is respected ;)

[macisamuele]

@cortinico, @filipemp: I did re-push the changes as I made the pre-commit hooks configuration less greedy ;)
Instead of ignoring the all samples directory I'm ignoring them into samples/.*/src/main/java/
NOTE: This is done because at the current state the codedgen does not really produce pretty source (according to the lint).
I will eventually look at it, but the reports are mostly related to wrong indentation and/or not needed imports.

[cortinico]

Can we have the re-formatting + pre-commit change on a separate change? This makes checking out the PR Diffs for third party easier.

[macisamuele]

@cortinico just to be sure, before publishing the changes, are you asking me to remove commits
e835456, dcf4e3d and 10dd7df?
If so I'm more than happy to remove it from this PR and publish a follow-up PR once this is merged 😃

Definitely it will simplify reviewing this, after all the commits are applying automated fixes

[cortinico]

Yes correct. Let's keep the "infra" work outside of this PR.

[filipemp]

lgtm but @cortinico has way more context on the global scale than I do so I'd wait for his review