Mocking of API handler generated by swagger Golang - mongodb

Question

i really need some help to test the behaviour of my API, let me explain the context :
This API is written in go generated by swagger using the repository patern. In case of creating a new object i need to check if the name is not already exists in the database so i call repository's function to get Applications by name in the database (see below)
func (a *mongoApplicationRepository) GetApplicationByName(name string, ctx context.Context) (*models.Application, error) {
_, span := otel.Tracer("GetApplicationByName").Start(ctx, "GetApplicationByName")
defer span.End()
var application models.Application
err := a.mgConn.Collection("applications").
FindOne(context.TODO(), bson.M{"name": name}).Decode(&application)
if err != nil {
if err == mongo.ErrNoDocuments {
return &application, nil
}
return nil, err
}
return &application, nil
}
This function is called by the api handler that you can see bellow
type ApplicationCreateHandler struct {
API *operations.KubeesAPI
repo repo.ApplicationRepository
statsRepo repo.StatsRepository
}
func NewApplicationCreateHandler(API *operations.KubeesAPI, repo repo.ApplicationRepository, statsRepo repo.StatsRepository) ApplicationCreateHandler {
return ApplicationCreateHandler{
API: API,
repo: repo,
statsRepo: statsRepo,
}
}
// Handle is the HTTP handler for application creation
func (h *ApplicationCreateHandler) Handle(params application.AppCreateParams, principal *models.Principal) middleware.Responder {
traceName := "application-create"
ctx, span := otel.GetTracerProvider().Tracer(traceName).Start(context.TODO(), traceName)
defer span.End()
if params.Data == nil {
err := "unable to validate input"
logger().Errorln(err)
return application.NewAppCreateBadRequest().WithPayload(&models.APIResponse{
Msg: &err,
})
}
if err := params.Data.Validate(h.API.Formats()); err != nil {
err := fmt.Sprintf("unable to validate input: %v", err)
logger().Errorln(err)
return application.NewAppCreateBadRequest().WithPayload(&models.APIResponse{
Msg: &err,
})
}
app := models.Application{
ID: uuid.NewString(),
Name: params.Data.Name,
Selector: params.Data.Selector,
}
c, err := h.repo.GetApplicationByName(*params.Data.Name, ctx)
if err != nil {
err := fmt.Sprintf("Unable to get Application: %v", err)
logger().Errorln(err)
return application.NewAppCreateInternalServerError().WithPayload(&models.APIResponse{
Msg: &err,
})
}
if c.ID != "" {
r := "name already taken"
logger().Errorln(r)
return application.NewAppCreateInternalServerError().WithPayload(&models.APIResponse{
Msg: &r,
})
}
appDomain, err := app.ToDomainModel()
if err != nil {
err := fmt.Sprintf("Unable to convert application to domain model: %v", err)
logger().Errorln(err)
return application.NewAppCreateInternalServerError().WithPayload(&models.APIResponse{
Msg: &err,
})
}
err = h.repo.CreateApplication(appDomain, ctx)
if err != nil {
err := fmt.Sprintf("Unable to create Application: %v", err)
logger().Errorln(err)
return application.NewAppCreateInternalServerError().WithPayload(&models.APIResponse{
Msg: &err,
})
}
if err := h.statsRepo.UpdateApplicationsHistory(ctx); err != nil {
err := fmt.Sprintf("unable to update applications history: %v", err)
logger().Errorln(err)
return application.NewAppCreateInternalServerError().WithPayload(&models.APIResponse{
Msg: &err,
})
}
return application.NewAppCreateCreated()
}
I just want to check this behaviour and test in case we have already an application called the same as the new created one, which would raise an error and not in opposite case. But i don't know what to mock because the repository have some external dependencies

Answer 1

I would strongly suggest to make sure your ApplicationCreateHandler does only rely on interfaces. When that's the case you can mock the interfaces by writing a mock class, or generating on. You can test the ApplicationCreateHandlerby creating a new instance, and passing the mocked dependencies to it. Then execute theHandle()` method and ensure all dependencies are called.
I would strongly recommend generting your mock-classes. When you write your mocks yourself it's possible to introduce bugs in your test code. You wouldn't want that, right?
One great example of a mocking library is mockgen. You can easily instal it by following their github-readme.
Steps
Create interfaces for your dependencies.
It is best define the interface on the side that relies on the interface. So in you case that would be in the ApplicationCreateHandler file. The interfaces should reflect the needs of the handler.
type (
ApplicationRepo interface {
GetApplicationByName(name string, ctx context.Context)
CreateApplication(domain Domain, ctx context.Context)
}
// other dependencies here...
)
Writing a unit test
Create a new unit test
// youfile_test.go
func TestApplicationHandler(t *testing.T) {
ctrl := gomock.NewController(t)
defer ctrl.Finish()
mockRepo := NewMockRepository(ctrl)
mockRepo.EXPECT().
Handle(gomock.any(), gomock.any()). // name, ctx
Return(nil, nil). // c, err
Once()
// other mocks...
// mockApi
// mockStatsRepo
h := NewApplicationHandler(mockApi, mockStats, mockRepo)
h.Handle(/*your test input*/)
// do checks
}
The test will throw an error if the mockRepo wasn't called exactly once.