refactor: split up huma.Register

[b-kamphorst]

The current huma.Register function is 900+ lines of code and has up to 9 levels of nesting. While working on #694 I found it really hard to understand the logic. In my effort to understand the function, I performed several refactorings and I thought that you may appreciate them as well.

Surely some parts are a matter of taste -- I don't mind changing them.

Looking forward to your thoughts.

[b-kamphorst]

The diff is quite large, so I made an effort to clean up the branch history. I think it may be easiest to review this PR by traversing the commits. Every commit passes the tests (git rebase main -x 'go test ./...'), as should be expected in a refactoring.

Before I cleaned up the history I made a change, passed the tests, and later figured out that it broke the flow of logic for the TextUnmarshaler. I added a test to prevent that from happening again. This just reminds us that even refactor efforts that pass all tests should be reviewed with care :)

[danielgtaylor]

@b-kamphorst this has been on my ToDo list for ages, so 1. Thank you so much for taking the time to do this and 2. Sorry I didn't get to it earlier!

Looks like there is a small conflict I may be able to resolve, but I'll need to take some time and review. Originally I mashed much of this code together in an effort to keep it performant, so I will run some benchmarks too. Give me a couple days please - I do want to merge this cleanup!

[b-kamphorst]

No worries, I appreciate your devotion to this wonderful project and I am glad that this may be of value. I'll rebase on main to resolve the minor conflict tomorrow.

Please take your time. Looking forward to your feedback.

[codecov]

Codecov Report

Attention: Patch coverage is 94.17476% with 42 lines in your changes missing coverage. Please review.

Project coverage is 92.90%. Comparing base (fe78329) to head (1ce45cb).

Files with missing lines	Patch %	Lines
huma.go	94.00%	31 Missing and 9 partials ⚠️
formdata.go	96.29%	1 Missing and 1 partial ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##             main     #705      +/-   ##
==========================================
+ Coverage   92.78%   92.90%   +0.12%     
==========================================
  Files          22       22              
  Lines        4917     4964      +47     
==========================================
+ Hits         4562     4612      +50     
+ Misses        308      305       -3     
  Partials       47       47              

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

[Copilot]

Copilot reviewed 3 out of 3 changed files in this pull request and generated 1 comment.

Comments suppressed due to low confidence (2)

formdata.go:180

• The error message 'Failed to open file' is not very descriptive. Consider providing more details to help with debugging.

return FormFile{}, &ErrorDetail{Message: "Failed to open file", Location: location}

formdata.go:94

• The new behavior introduced in the Decode function should be covered by tests.

func (m *MultipartFormFiles[T]) Decode(opMediaType *MediaType) []error {

[danielgtaylor]

@b-kamphorst thanks for all this work! I've reviewed the code and it looks good. The one issue I found is that benchmarks showed an increase in dynamic memory allocations per request and I was able to git bisect it to 78b0cc4, specifically the change in splitting query param strings before they might be needed. I've pushed a fix to move that logic to only be applied if the underlying parameter is actually an array of values. The performance is now equal to before the PR, so I am going to merge. Thanks again 🥇

[b-kamphorst]

That's nice! If you have any recommended sources on (that kind of) Go benchmarking then I would be interested. I'm a couple of months into the language.

Glad to hear you valued the refactoring, happy to help!