Turn an Assertion into a tasty test case

Like testCase, except in case the test succeeds, the returned string will be shown as the description. If the empty string is returned, it will be ignored.

Create a multi-step unit test.

Example:

main = defaultMain $ testCaseSteps "Multi-step test" $ \step -> do
  step "Preparing..."
  -- do something

  step "Running part 1"
  -- do something

  step "Running part 2"
  -- do something
  assertFailure "BAM!"

  step "Running part 3"
  -- do something

The step calls are mere annotations. They let you see which steps were performed successfully, and which step failed.

You can think of step as putStrLn, except putStrLn would mess up the output with the console reporter and get lost with the others.

For the example above, the output will be

Multi-step test: FAIL
  Preparing...
  Running part 1
  Running part 2
    BAM!

1 out of 1 tests failed (0.00s)

Note that:

• Tasty still treats this as a single test, even though it consists of multiple steps.
• The execution stops after the first failure. When we are looking at a failed test, we know that all displayed steps but the last one were successful, and the last one failed. The steps after the failed one are not displayed, since they didn't run.

Unconditionally signals that a failure has occured. All other assertions can be expressed with the form:

   if conditionIsMet
       then return ()
       else assertFailure msg

Asserts that the specified condition holds.

Asserts that the specified actual value is equal to the expected value. The output message will contain the prefix, the expected value, and the actual value.

If the prefix is the empty string (i.e., ""), then the prefix is omitted and only the expected and actual values are output.

Asserts that the specified actual value is equal to the expected value (with the expected value on the left-hand side).

Asserts that the specified actual value is equal to the expected value (with the actual value on the left-hand side).

An infix and flipped version of assertBool. E.g. instead of

assertBool "Non-empty list" (null [1])

you can write

null [1] @? "Non-empty list"

@? is also overloaded to accept IO Bool predicates, so instead of

do
  e <- doesFileExist "test"
  e @? "File does not exist"

you can write

doesFileExist "test" @? "File does not exist"

An ad-hoc class used to overload the @? operator.

The only intended instances of this class are Bool and IO Bool.

You shouldn't need to interact with this class directly.

An assertion is simply an IO action. Assertion failure is indicated by throwing an exception, typically HUnitFailure.

Instead of throwing the exception directly, you should use functions like assertFailure and assertBool.

Test cases are composed of a sequence of one or more assertions.

Exception thrown by assertFailure etc.

Signals an assertion failure if a non-empty message (i.e., a message other than "") is passed.

Allows the extension of the assertion mechanism.

Since an Assertion can be a sequence of Assertions and IO actions, there is a fair amount of flexibility of what can be achieved. As a rule, the resulting Assertion should be the body of a TestCase or part of a TestCase; it should not be used to assert multiple, independent conditions.

If more complex arrangements of assertions are needed, Test and Testable should be used.

The result of an assertion that hasn't been evaluated yet.

Most test cases follow the following steps:

1. Do some processing or an action.
2. Assert certain conditions.

However, this flow is not always suitable. AssertionPredicate allows for additional steps to be inserted without the initial action to be affected by side effects. Additionally, clean-up can be done before the test case has a chance to end. A potential work flow is:

1. Write data to a file.
2. Read data from a file, evaluate conditions.
3. Clean up the file.
4. Assert that the side effects of the read operation meet certain conditions.
5. Assert that the conditions evaluated in step 2 are met.