Testing HTTP routes in Golang

Question

I am using Gorilla mux and the net/http package to create some routes as follows

package routes

//some imports

//some stuff

func AddQuestionRoutes(r *mux.Router) {
    s := r.PathPrefix("/questions").Subrouter()
    s.HandleFunc("/{question_id}/{question_type}", getQuestion).Methods("GET")
    s.HandleFunc("/", postQuestion).Methods("POST")
    s.HandleFunc("/", putQuestion).Methods("PUT")
    s.HandleFunc("/{question_id}", deleteQuestion).Methods("DELETE")
}

I am trying to write a test to test these routes. For example, I am trying to test the GET route specifically trying to get a 400 returned so I have the following test code.

package routes

//some imports

var m *mux.Router
var req *http.Request
var err error
var respRec *httptest.ResponseRecorder

func init() {
    //mux router with added question routes
    m = mux.NewRouter()
    AddQuestionRoutes(m)

    //The response recorder used to record HTTP responses
    respRec = httptest.NewRecorder()
}

func TestGet400(t *testing.T) {
    //Testing get of non existent question type
    req, err = http.NewRequest("GET", "/questions/1/SC", nil)
    if err != nil {
        t.Fatal("Creating 'GET /questions/1/SC' request failed!")
    }

    m.ServeHTTP(respRec, req)

    if respRec.Code != http.StatusBadRequest {
        t.Fatal("Server error: Returned ", respRec.Code, " instead of ", http.StatusBadRequest)
    }
}

However, when I run this test, I get a 404 conceivably because the request is not being routed correctly.?

When I test this GET route from the browser, it does return a 400 so I'm certain there is an issue with the way the test is setup.

Answer 1

The use of init() here is suspect. It only executes once as part of program initialization. Instead, perhaps something like:

func setup() {
    //mux router with added question routes
    m = mux.NewRouter()
    AddQuestionRoutes(m)

    //The response recorder used to record HTTP responses
    respRec = httptest.NewRecorder()
}

func TestGet400(t *testing.T) {
    setup()
    //Testing get of non existent question type
    req, err = http.NewRequest("GET", "/questions/1/SC", nil)
    if err != nil {
        t.Fatal("Creating 'GET /questions/1/SC' request failed!")
    }

    m.ServeHTTP(respRec, req)

    if respRec.Code != http.StatusBadRequest {
        t.Fatal("Server error: Returned ", respRec.Code, " instead of ", http.StatusBadRequest)
    }
}

where you call setup() at the beginning of each appropriate test case. Your original code was sharing the same respRec with other tests, which probably polluted your test results.

If you need a testing framework that provides more features like setup/teardown fixtures, see packages such as gocheck.