test

A lightweight test helper package 🧪

Project Description

test is my take on a handy, lightweight Go test helper package. Inspired by matryer/is, carlmjohnson/be and others.

It provides a lightweight, but useful, extension to the std lib testing package with a friendlier and hopefully intuitive API. You definitely don't need it, but might find it useful anyway 🙂

Installation

go get github.com/FollowTheProcess/test@latest

Usage

test is as easy as...

func TestSomething(t *testing.T) {
    test.Equal(t, "hello", "hello") // Obviously fine
    test.Equal(t, "hello", "there") // Fails

    test.NotEqual(t, 42, 27) // Passes, these are not equal
    test.NotEqual(t, 42, 42) // Fails

    test.NearlyEqual(t, 3.0000000001, 3.0) // Look, floats handled easily!

    err := doSomething()
    test.Ok(t, err) // Fails if err != nil
    test.Err(t, err) // Fails if err == nil

    test.True(t, true) // Passes
    test.False(t, true) // Fails

    // Get $CWD/testdata easily
    test.Data(t) // /Users/you/project/package/testdata

    // Check against contents of a file including line ending normalisation
    file := filepath.Join(test.Data(t), "expected.txt")
    test.File(t, "hello\n", file)

    // Just like the good old reflect.DeepEqual, but with a nicer format
    test.DeepEqual(t, []string{"hello"}, []string{"world"}) // Fails
}

Tip

Line comments on the line you call most test functions on will be shown in failure messages as additional context

Non Comparable Types

test uses Go 1.18+ generics under the hood for most of the comparison, which is great, but what if your types don't satisfy comparable. We also provide test.EqualFunc and test.NotEqualFunc for those exact situations!

These allow you to pass in a custom comparator function for your type, if your comparator function returns true, the types are considered equal.

func TestNonComparableTypes(t *testing.T) {
    // Slices do not satisfy comparable
    a := []string{"hello", "there"}
    b := []string{"hello", "there"}
    c := []string{"general", "kenobi"}

    // Custom function, returns true if things should be considered equal
    sliceEqual := func(a, b, []string) { return true } // Cheating

    test.EqualFunc(t, a, b, sliceEqual) // Passes

    // Can also use e.g. the new slices package
    test.EqualFunc(t, a, b, slices.Equal[string]) // Also passes :)

    test.EqualFunc(t, a, c, slices.Equal[string]) // Fails
}

You can also use this same pattern for custom user defined types, structs etc.

Rich Comparison

Large structs or long slices can often be difficult to compare using reflect.DeepEqual, you have to scan for the difference yourself. test provides a test.Diff function that produces a rich text diff for you on failure:

func TestDiff(t *testing.T) {
    // Pretend these are very long, or are large structs
    a := []string{"hello", "world"}
    b := []string{"hello", "there"}

    test.Diff(t, a, b)
}

Will give you:

--- FAIL: TestDiff (0.00s)
    main_test.go:14: Mismatch (-want, +got):
          []string{
                "hello",
        -       "there",
        +       "world",
          }

Table Driven Tests

Table driven tests are great! But when you test errors too it can get a bit awkward, you have to do the if (err != nil) != tt.wantErr thing and I personally always have to do the boolean logic in my head to make sure I got that right. Enter test.WantErr:

func TestTableThings(t *testing.T) {
    tests := []struct {
        name    string
        want    int
        wantErr bool
    }{
        {
            name:    "no error",
            want:    4,
            wantErr: false,
        },
        {
            name:    "yes error",
            want:    4,
            wantErr: true,
        },
    }
    
    for _, tt := range tests {
        t.Run(tt.name, func(t *testing.T) {
            got, err := SomeFunction()
    
            test.WantErr(t, err, tt.wantErr)
            test.Equal(t, got, tt.want)
        })
    }
}

Which is basically semantically equivalent to:

func TestTableThings(t *testing.T) {
    tests := []struct {
        name    string
        want    int
        wantErr bool
    }{
        {
            name:    "no error",
            want:    4,
            wantErr: false,
        },
        {
            name:    "yes error",
            want:    4,
            wantErr: true,
        },
    }
    
    for _, tt := range tests {
        t.Run(tt.name, func(t *testing.T) {
            got, err := SomeFunction()
    
            if tt.wantErr {
                test.Err(t, err)
            } else {
                test.Ok(t, err)
            }
            test.Equal(t, got, tt.want)
        })
    }
}

Capturing Stdout and Stderr

We've all been there, trying to test a function that prints but doesn't accept an io.Writer as a destination 🙄.

That's where test.CaptureOutput comes in!

func TestOutput(t *testing.T) {
    // Function that prints to stdout and stderr, but imagine this is defined somewhere else
    // maybe a 3rd party library that you don't control, it just prints and you can't tell it where
    fn := func() error {
        fmt.Fprintln(os.Stdout, "hello stdout")
        fmt.Fprintln(os.Stderr, "hello stderr")

        return nil
    }

    // CaptureOutput to the rescue!
    stdout, stderr := test.CaptureOutput(t, fn)

    test.Equal(t, stdout, "hello stdout\n")
    test.Equal(t, stderr, "hello stderr\n")
}

Under the hood CaptureOutput temporarily captures both streams, copies the data to a buffer and returns the output back to you, before cleaning everything back up again.

Golden Files

test has great support for golden files:

func TestFile(t *testing.T) {
    got := "some contents\n"
    want := filepath.Join(test.Data(t), "golden.txt")

    test.File(t, got, want)
}

This wil read the file, normalise line endings and then generate an output almost like a git diff:

--- want
+++ got
@@ -1 +1 @@
-some file contents
+some contents

Credits

This package was created with copier and the FollowTheProcess/go_copier project template.