A small library to provide given-when-then style testing without a bunch of overhead
Generally, you won't be using this library directly. It is intentionally unbound to any specific test runner so that if your test runner isn't supported, you can support it!
You probably want one of the bound libraries:
Unfortunately, mocha-gwt
was already taken, and is NOT associated with this
project. It is completely different. Don't get confused.
- Install your test library of choice (in this readme we'll use Jest
npm i --save-dev jest-gwt
- In your test files, import your test runner's test function from here
import test from 'jest-gwt';
- Write a test!
describe('test context', () => { test('has no expected errors', { given: { mock_jest_test_function, GOOD_test_case, }, when: { executing_test_case, }, then: { all_GIVENS_called, all_WHENS_called, all_THENS_called, }, }); });
The test function is invoked with two parameters:
name
: the name of the test passed to your test runnerdefinition
: The GWT definition
Name is pretty self-explanatory, so we'll focus on definition
.
The test definition is an object with three keys: given
, when
, and
then
. If you pass any other keys, your test will fail with an error.
All given
clauses are executed first, then all when
clauses, then all
then
clauses.
The default syntax of definition
takes advantage of ES6's shorthand object literal
syntax. In reality, the object looks like this:
{
given: {
mock_jest_test_function: mock_jest_test_function,
GOOD_test_case: GOOD_test_case,
},
when: {
executing_test_case: executing_test_case,
},
then: {
all_GIVENS_called: all_GIVENS_called,
all_WHENS_called: all_WHENS_called,
all_THENS_called: all_THENS_called,
},
}
The key names within the given
, when
, and then
definition do not
matter. For instance, maybe you have a function that takes an expected value:
then: {
the_value_is_correct: the_value_is(15)
}
Generally, you want to use the shorthand syntax as it reads nicer.
The only acceptable clauses in the given
, when
, and then
definitions are
UNBOUND functions. Do NOT use arrow functions, or the context won't work at
all!
If you'd rather use an array instead of a keyed object, you can do that too:
{
given: [
mock_jest_test_function,
GOOD_test_case,
],
when: [
executing_test_case,
],
then: [
all_GIVENS_called,
all_WHENS_called,
all_THENS_called,
],
}
This is more useful if you want to use curried functions:
{
given: {
some_given,
},
when: [
user_enters_data({ some: 'data' }),
],
then: {
the_form_has_data,
}
}
Keep in mind that the functions returned from the curried functions must still be function definitions and can not be array functions:
// GOOD
function user_enters_data(data: any) {
return (this: Context) {
do_the_thing_with_the_data(data);
};
};
// BAD
function user_enters_data(data: any) {
return (this: Context) => {
do_the_thing_with_the_data(data);
};
};
Typical function declarations for usage in the given
, when
, and then
definitions looks like:
function valid_email() {
this.email = 'test@mail.com';
}
async function validating_email() {
this.validation_result = await validateEmailAsync(this.email);
}
function email_is_valid() {
expect(this.validation_result).toBe(true);
}
Clauses can be either regular or asynchronous functions if your test runner
supports it! Every test execution declares a new context
and binds the
clauses in given
, when
, and then
to the test context
so that
accessing this
within the clauses is independent of other test executions.
Passing data between clauses during a test execution should only be done
through the test context
accessed through this
. No arguments are passed
to the clauses.
If your test is expecting an error, pass a function to then
named
expect_error
. gwt-runner
will pick this up, and execute it, passing the
error to your function. Validate the error is what you expect it to be, and
your test will pass.
If however your code throws an error and you do NOT have an expect_error
in
your then clauses, it will fail your test with the error that was thrown.
test('should throw error', {
given: {
an_error_message,
},
when: {
throwing_error,
},
then: {
expect_error,
},
});
function an_error_message() {
this.error_message = 'an error';
}
function throwing_error() {
throw new Error(this.error_message);
}
function expect_error(error) {
expect(error.message).toBe(this.error_message);
}
The following example tests an contrived validateEmailAddress
AJAX call.
// validateEmailAddress.js
import axios from 'axios';
export default email => axios.post('/api/validateEmail', {
email,
}).then(res => {
if (res.data.success) {
return Promise.resolve(true);
}
return Promise.reject(res.data.error);
});
import test from 'jest-gwt';
import axios from 'axios';
import validateEmailAddress from './validateEmailAddress';
jest.mock('axios');
describe('the validate email address api', () => {
beforeEach(() => {
jest.clearAllMocks();
})
test('returns true for valid email addresses', {
given: {
valid_email_address,
},
when: {
validating_email_address,
},
then: {
email_address_is_valid,
},
});
test('extracts error for invalid email address', {
given: {
INVALID_email_address,
},
when: {
validating_email_address,
},
then: {
expect_error: email_address_is_INVALID,
}
});
});
function valid_email_address() {
this.mock_email = 'valid@email.com';
axios.post.mockResolvedValue({
data: { success: true }
});
}
function INVALID_email_address() {
this.mock_email = 'invalid';
axios.post.mockResolvedValue({
data: {
success: false,
error: 'error message from server',
},
});
}
async function validating_email_address() {
this.result = await validateEmailAddress(this.mock_email);
}
function email_address_is_valid() {
expect(this.result).toBe(true);
}
function email_address_is_INVALID(error) {
expect(error).toBe('error message from server');
}
Sometimes a GWT flow doesn't make sense. You might be writing integration tests. Or something that needs to assert something, then do another thing, then assert something else.
In these cases, you can use the scenario definition style which allows
chaining when
and then
, then then_when
and then
blocks together.
{
given: {
mock_jest_test_function,
GOOD_test_case,
},
scenario: [{
when: {
executing_test_case,
},
then: {
assert_something,
},
}, {
then_when: {
user_submits_form,
},
then: {
something_else_happens,
yet_another_thing_is_true,
},
}, {
then_when: {
something_happens,
},
then: {
expect_error: some_check,
and: {
something_is_still_true,
},
}
}]
}
If a step fails, the error will be caught and wrapped with the step index. To make this error more descriptive, you can optionally name each step:
{
given: {
mock_jest_test_function,
GOOD_test_case,
},
scenario: [{
name: 'Adding to cart',
when: {
executing_test_case,
},
then: {
assert_something,
},
}, {
name: 'Checking out',
...
}, {
name: 'Paying',
...
}],
}
The old syntax is still supported, but is DEPRECATED and will be removed in version 3.0
{
given: {
mock_jest_test_function,
GOOD_test_case,
},
scenario: {
when_executing_test_case,
then_assert_something,
when_user_submits_form,
then_something_else_happens,
then_yet_another_thing_is_true,
}
}
You can also expect errors from the scenario using the same function style as a
GWT style expect_error
:
{
given: {
mock_jest_test_function,
GOOD_test_case,
},
scenario: {
when_executing_test_case,
then_assert_something,
when_user_submits_form,
then_something_else_happens,
then_yet_another_thing_is_true,
when_an_error_happens,
},
expect_error: the_error_happened
}