Incredibly simple Go snapshot testing: cupaloy takes a snapshot of your test output and compares it to a snapshot committed alongside your tests. If the values don't match then the test will be failed.

There's no need to manually manage snapshot files: just use the cupaloy.SnapshotT(t, value) function in your tests and cupaloy will automatically find the relevant snapshot file (based on the test name) and compare it with the given value.

Firstly, write a test case generating some output and pass this output to cupaloy.SnapshotT:

func TestParsing(t *testing.T) {
    ast := ParseFile("test_input")

    // check that the result is the same as the last time the snapshot was updated
    // if the result has changed (e.g. because the behaviour of the parser has changed)
    // then the test will be failed with an error containing a diff of the changes
    cupaloy.SnapshotT(t, ast)
}

The first time this test is run, a snapshot will be automatically created (using the github.com/davecgh/go-spew package).

When the behaviour of your software changes causing the snapshot to change, this test will begin to fail with an error showing the difference between the old and new snapshots. Once you are happy that the new snapshot is correct (and hasn't just changed unexpectedly), you can save the new snapshot by setting the UPDATE_SNAPSHOTS environment and re-running your tests:

UPDATE_SNAPSHOTS=true go test ./...

This will fail all tests where the snapshot was updated (to stop you accidentally updating snapshots in CI) but your snapshot files will now have been updated to reflect the current output of your code.

Snapshots of test output are generated using the github.com/davecgh/go-spew package which uses reflection to deep pretty-print your test result and so will support almost all the basic types (from simple strings, slices, and maps to deeply nested structs) without issue. The only types whose contents cannot be fully pretty-printed are functions and channels.

The most important property of your test output is that it is deterministic: if your output contains timestamps or other fields which will change on every run, then cupaloy will detect this as a change and so fail the test.

var testCases = map[string][]string{
    "TestCaseOne": []string{......},
    "AnotherTestCase": []string{......},
    ....
}

func TestCases(t *testing.T) {
    for testName, args := range testCases {
        t.Run(testName, func(t *testing.T) {
            result := functionUnderTest(args...)
            cupaloy.SnapshotT(t, result)
        })
    }
}

func TestSubdirectory(t *testing.T) {
    result := someFunction()
    snapshotter := cupaloy.New(cupaloy.SnapshotSubdirectory("testdata"))
    err := snapshotter.Snapshot(result)
    if err != nil {
        t.Fatalf("error: %s", err)
    }
}

For further usage examples see basic_test.go and advanced_test.go in the examples/ directory which are both kept up to date and run on CI.

It is important to note that git on Windows might be configured in a way that \n is replaced by \r\n during checkout (it is the case on GitHub actions). In such a case, the snapshot appears to be the same, but the test fails. Please ensure that git is configured correctly everywhere. There are multiple ways to do it, please check actions/checkout#135 and #73 for more details.