I am trying to write a unit test for http.Pusher. I tried using httptest.NewRecorder() as http.ResponseWriter but the type conversion failed. How else can I write the test?
//My function
func push(w http.ResponseWriter, resource string) error {
pusher, ok := w.(http.Pusher)
if ok {
return pusher.Push(resource, nil)
}
return fmt.Errorf("Pusher not supported")
}
//My test
func TestPush(t *testing.T) {
resource := "static/css/main.css"
response := httptest.NewRecorder()
got := push(response, resource)
if got != nil {
t.Errorf("Error : %v", got)
}
}
The output is "Pusher not supported", which I assume that w.(http.Pusher) failed.
You may create a mocked http.ResponseWriter
that also implements http.Pusher
, and pass that during testing.
Here's a simple implementation that suits your testable function:
type pusher struct {
http.ResponseWriter
err error // err to return from Push()
target string
opts *http.PushOptions
}
func (p *pusher) Push(target string, opts *http.PushOptions) error {
// record passed arguments for later inspection
p.target = target
p.opts = opts
return p.err
}
Example test function:
func TestPush(t *testing.T) {
resource := "static/css/main.css"
p := &pusher{}
err := push(p, resource)
if err != p.err {
t.Errorf("Expected: %v, got: %v", p.err, err)
}
if resource != p.target {
t.Errorf("Expected: %v, got: %v", p.target, resource)
}
}
Note that this simple pusher
embeds the http.ResponseWriter
type which will make it itself an http.ResponseWriter
(it will make pusher
implement http.ResponseWriter
). During the test we left this field nil
because the testable push()
function did not use anything from it. If your real function would call methods of it such as ResponseWriter.Header()
, that would cause a runtime panic of course. In that case you have to provide a valid http.ResponseWriter
too, e.g.:
p := &pusher{ResponseWriter: httptest.NewRecorder()}