azutoolkit / schema Goto Github PK

Look into https://dry-rb.org/gems/dry-validation/nested-data/ for ideas

• Rename Definition to Schema
• Provide a default for namespace
• Specify i18n yaml

Source:

class User::Controller < ApplicationController
  def create
    CreateForm.new(params.raw_params.to_h)
  end

  schema CreateForm do
    param email : String, match: /\w+@\w+\.\w{2,3}/, message: "Email must be valid!"
  end
end

Compiler:

instantiating 'User::Controller#create()'
in src/modules/user/controller.cr:3: undefined constant CreateForm (did you mean 'Createform')

    CreateForm.new(params.raw_params.to_h)
    ^~~~~~~~~~

make: *** [build] Error 1

I don't think Schema::Validator is flexible because you can not pass data to it before validating records.

Most other validation libraries store instances of validator classes in the model so that you can pass additional data to it during initialization to customize it's behavior.

Apologies if I've missed the point of this module but I really like the format - I'd been toying with writing my own JSON Schema validator and then stumbled upon this. It may be a bit of a direct question but as a new user checking out this library, what advantages does this have over using the Amber Params module directly?

Some of the higher level ones I can see:

• Less coupled to the Amber controller
• Predicates without patching base types
• Ability to inspect whole record on each validation

The hope would be to be able to create a set of schemas for each Amber controller and have the schema manage it's own validations without having to rely on the in-built param predicate runner. Is this a valid use case in your eyes?