go-testpredicate
Test predicate style assertions library with extensive diagnostics output.
Package go-testpredicate is an assertions library exposing a test predicate style syntax for use with the built-in Go testing package, producing extensive diagnostics output and reducing the need for debugging failing tests.
The library contains an extensive collection of built-in predicates covering:
- basic tests for nil, true, false
- equality between any type of value
- ordered comparison on numeric, string and sequence values
- regexp match on strings
- sub-sequences match on strings and sequences
- set conditions on unordered collections
- panic conditions on code fragment execution
It also includes a BDD-style bifurcated evaluation context, where each test section is potentially evaluated multiple times in order to evaluate each branch independently.
Installation
go get github.com/maargenton/go-testpredicate
Optionally, you can add predefined code snippets for your text editor or IDE to assist in writing your test code. Snippets for VSCode are available here
Usage
package example_test
import (
"testing"
"github.com/maargenton/go-testpredicate/pkg/bdd"
"github.com/maargenton/go-testpredicate/pkg/require"
"github.com/maargenton/go-testpredicate/pkg/verify"
)
func TestExample(t *testing.T) {
bdd.Given(t, "something", func(t *bdd.T) {
require.That(t, 123).ToString().Length().Eq(3)
t.When("doing something", func(t *bdd.T) {
t.Then("something happens ", func(t *bdd.T) {
verify.That(t, "123").Eq(123)
verify.That(t, 123).ToString().Length().Eq(4)
})
})
})
}
Output:
--- FAIL: TestExample (0.00s)
--- FAIL: TestExample/Given_something (0.00s)
--- FAIL: TestExample/Given_something/when_doing_something (0.00s)
--- FAIL: TestExample/Given_something/when_doing_something/then_something_happens_ (0.00s)
example_test.go:17:
expected: value == 123
error: values of type 'string' and 'int' are never equal
value: "123"
example_test.go:18:
expected: length(value.String()) == 4
value: 123
string: "123"
length: 3
API changes and stability
Older version of this package where exposing a different API that has since been deprecated, and has now been remove for the v1.0.0 release. The latest version supporting the legacy API is v0.6.4.
Predicates are constructed starting with either require.That(t, <value>) or verify.That(t, <value>), where require will abort the test on error, while verify will keep going. Both variants take the testing context t, and the value to test.
Additional diagnostic context can be added to either functions with require.Context{} / verify.Context{} passed as additional arguments.
package example_test
import (
"testing"
"github.com/maargenton/go-testpredicate/pkg/require"
"github.com/maargenton/go-testpredicate/pkg/verify"
)
func TestExample(t *testing.T) {
v := 123
require.That(t, v).ToString().Length().Eq(3)
verify.That(t, v).ToString().Length().Eq(3)
verify.That(t, v,
verify.Context{Name: "double", Value: v * 2},
).ToString().Length().Eq(3)
}
Built-in predicates
All predicates are built through call chaining on the builder object returned by require.That() or verify.That(). For an up-to-date full list of supported predicates and their use, take a look at pkg/internal/builder/builder_api_test.go
func TestCollectionAPI(t *testing.T) {
verify.That(t, []string{"a", "bb", "ccc"}).All(
subexpr.Value().Length().Lt(5))
verify.That(t, []string{"a", "bb", "ccc"}).Any(
subexpr.Value().Length().Ge(3))
}
func TestCompareAPI(t *testing.T) {
verify.That(t, true).IsTrue()
verify.That(t, false).IsFalse()
verify.That(t, nil).IsNil()
verify.That(t, &struct{}{}).IsNotNil()
verify.That(t, 123).IsEqualTo(123)
verify.That(t, 123).IsNotEqualTo(124)
verify.That(t, 123).Eq(123)
verify.That(t, 123).Ne(124)
}
func TestErrorAPI(t *testing.T) {
var sentinel = fmt.Errorf("sentinel")
var err = fmt.Errorf("error: %w", sentinel)
verify.That(t, err).IsError(sentinel)
var err2 = fmt.Errorf("error: %w", &MyError{Code: 123})
var myError *MyError
verify.That(t, err2).AsError(&myError).Field("Code").Eq(123)
}
func TestExtAPI(t *testing.T) {
var customPredicate = func() (desc string, f predicate.PredicateFunc) {
// ...
}
verify.That(t, nil).Is(customPredicate())
var customTransform = func() (desc string, f predicate.TransformFunc) {
// ...
}
verify.That(t, nil).Eval(customTransform()).Is(customPredicate())
verify.That(t, 9).Passes(subexpr.Value().Lt(10))
}
func TestMapAPI(t *testing.T) {
var m = map[string]string{ "aaa": "bbb", "ccc": "ddd" }
verify.That(t, m).MapKeys().IsEqualSet([]string{"aaa", "ccc"})
verify.That(t, m).MapValues().IsEqualSet([]string{"bbb", "ddd"})
}
func TestOrderedAPI(t *testing.T) {
verify.That(t, 123).IsLessThan(124)
verify.That(t, 123).IsLessOrEqualTo(123)
verify.That(t, 123).IsGreaterThan(122)
verify.That(t, 123).IsGreaterOrEqualTo(123)
verify.That(t, 123).IsCloseTo(133, 10)
verify.That(t, 123).Lt(124)
verify.That(t, 123).Le(123)
verify.That(t, 123).Gt(122)
verify.That(t, 123).Ge(123)
}
func TestPanicAPI(t *testing.T) {
verify.That(t, func() {
panic(123)
}).Panics()
verify.That(t, func() {
panic(123)
}).PanicsAndRecoveredValue().Eq(123)
}
func TestSequenceAPI(t *testing.T) {
verify.That(t, make([]int, 3, 5)).Length().Eq(3)
verify.That(t, make([]int, 3, 5)).Capacity().Eq(5)
verify.That(t, []int{}).IsEmpty()
verify.That(t, []int{1, 2, 3, 4, 5}).IsNotEmpty()
verify.That(t, []int{1, 2, 3, 4, 5}).StartsWith([]int{1, 2})
verify.That(t, []int{1, 2, 3, 4, 5}).Contains([]int{2, 3, 4})
verify.That(t, []int{1, 2, 3, 4, 5}).EndsWith([]int{4, 5})
verify.That(t, []int{1, 2, 3, 4, 5}).HasPrefix([]int{1, 2})
verify.That(t, []int{1, 2, 3, 4, 5}).HasSuffix([]int{4, 5})
}
func TestSetAPI(t *testing.T) {
verify.That(t, []int{1, 2, 3, 4, 5}).IsEqualSet([]int{1, 4, 3, 2, 5})
verify.That(t, []int{1, 2, 3, 4, 5}).IsDisjointSetFrom([]int{6, 9, 8, 7})
verify.That(t, []int{1, 2, 3, 4, 5}).IsSubsetOf([]int{1, 4, 3, 2, 5, 6})
verify.That(t, []int{1, 2, 3, 4, 5}).IsSupersetOf([]int{1, 4, 5})
}
func TestStringAPI(t *testing.T) {
verify.That(t, "123").Matches(`\d+`)
verify.That(t, 123).ToString().Eq("123")
verify.That(t, "aBc").ToLower().Eq("abc")
verify.That(t, "aBc").ToUpper().Eq("ABC")
}
BDD-style bifurcated tests
Rationale
First of all, the Go testing package is great and the fact that it is standard, built in and integrated with the Go tooling infrastructure is awesome. This is why the go-testpredicate packages strives to enhance it instead of replacing it, unlike many other testing packages.
If you look at other unit-testing packages, in other languages, you will find either traditional xUnit style packages relying on classes to define test suites and fixtures and test cases, or more recent testing packages (like Catch-2 for C++) that provide, through other means, ways to define setup and test cases than run independently. The common pattern is that setup code, that may be shared by multiple test cases, is usually re-evaluated for every test case so that, despite their potentially mutating interactions with the setup, test cases don't affect each other.
Some great articles and blog posts have explained how the leverage nested t.Run() calls to structure tests in way that is closer to BDD-style given / when / then paradigm. Unfortunately, when using thees approaches, and especially with shared setup sections, the test cases are no longer independent, as all branches are run sequentially, going up and down each branch and into the next branch, without resetting the setup.
The bdd package in go-testpredicate provides a way to write tests with a BDD-style structure, using the built-in testing.T, but evaluating the test cases in a bifurcated fashion, repeating the evaluation of each entire branch for every leaf test case, so that test cases are independent from each other again.
Usage overview
bdd.Wrap() or bdd.Given() are the root level function that setup and iterate through the bifurcated test evaluation context. They define blocks that receive a bdd.T instead of testing.T, but bdd.T is fully compatible with testing.T and can be used with any third party library that expect either the testing.TB interface or a subset of it (including out own verify.That() / require.That()).
Nested and sibling bifurcated branches are defined with t.Run() (on bdd.T) or t.When() / t.Then() for BDD style.
IMPORTANT: In a bifurcated evaluation context, as defined by
bdd.T, test scenarios are run repeatedly in order to evaluate each branch (from root to leaf) independently of each other. When a particular branch is being evaluated, all the other forks and sub-branches are skipped; the other branches are run in separated independent iterations of the scenario.
Usage, traditional style
package example_test
import (
"testing"
"github.com/maargenton/go-testpredicate/pkg/bdd"
)
func TesTraditional(t *testing.T) {
// Global immutable setup code can go here
bdd.Wrap(t, "Given something", func(t *bdd.T) {
// Local mutable setup code goes here
t.Run("something happens", func(t *bdd.T) {
// When this code runs, the code in following `t.Run()` blocks
// will be skipped.
})
t.Run("something else happens", func(t *bdd.T) {
// When this code runs, all code in preceding `t.Run()` blocks
// has been skipped and did not affect the local setup.
})
})
}
Usage, BDD style
package bdd_test
import (
"testing"
"github.com/maargenton/go-testpredicate/pkg/bdd"
)
func TestBDDStyle(t *testing.T) {
// Global immutable setup code can go here
bdd.Given(t, "something", func(t *bdd.T) {
// Local mutable setup code goes here
t.When("doing something", func(t *bdd.T) {
// or here
t.Then("something happens", func(t *bdd.T) {
// When this code runs, the code in the following `t.Then()`
// blocks will be skipped.
})
t.Then("something else happens", func(t *bdd.T) {
// When this code runs, all code in preceding `t.Then()`
// blocks has been skipped and did not affect the local setup.
})
})
})
}
