Support mock generation via interface Annotations

[chadwpetersen]

Summary

I’d like to request support for generating mocks from interfaces explicitly annotated with a specific comment tag, such as @mockery:generate.

Background

In my current workflow, I use go:generate to trigger mock generation for selected interfaces in a large repository. However, managing inclusion and exclusion through regex patterns has proven difficult due to:

• Not all interfaces require mocks.
• Interface names do not follow consistent patterns that are easily targeted via regex.

Proposed Solution

Introduce an option in the mockery.yml configuration file:

annotation: true

When annotation: true is set, mockery will only generate mocks for interfaces that include a specific comment annotation above them (e.g., // @mockery:generate). This could look like:

// @mockery:generate
type MyInterface interface {
    ...
}

Question

Is something like this currently possible with mockery, or perhaps already being planned? I couldn't find any mention of annotation-based generation in the documentation, so I’m wondering if there's an existing or alternative approach to achieve this use case.

[LandonTClipp]

Interesting! There have been some folks that wanted something similar to this. It's not possible currently. To add this feature, we'd need mockery to unconditionally parse all interfaces in the defined packages and read the ast.GenDecl/ast.TypeSpec comments for the magic string indicating a generation should occur. We could extend this a bit further to allow interface-specific overrides, something like this:

// mockery:generate
// structname: "myMockInterface"
// filename: "my_interface_mock.go"
// template: "matryer"
type MyInterface interface {
    ...
}

This would essentially be equivalent to something like:

packages:
  github.com/user/repo:
    interfaces:
      MyInterface:
        config:
          structname: "myMockInterface"
          filename: "my_interface_mock.go"
          template: "matryer"

I'm very open to this idea! The work itself would be theoretically straightforward, although it will involve a bit of surgery in a few key areas. What do you think of this?