Test.Tasty.Checklist

This package provides the ability to run a Checklist of several "checks" during a single test. A "bad" check does not immediately result in a test failure; at the end of the test (passed or failed due to primary testing), all failed checks are reported (and any failed checks will result in an overall test failure at the end.

This type of checking can be very useful when needing to test various aspects of an operation that is complex to setup, has multiple effects, or where the checks are related such that knowing about the multiple failures makes debugging easier.

An alternative approach is to have some sort of common preparation code and use a separate test for each item. This module simply provides a convenient method to collate related items under the aegis of a single test.

This package also provides the checkValues function which can be used to check a number of derived values from a single input value via a checklist. This can be used to independently verify a number of record fields of a data structure or to validate related operations performed from a single input.

See the documentation for check and checkValues for examples of using this library. The tests in the source package also provide additional examples of usage.

Checklist testing context

withChecklist :: (MonadIO m, MonadMask m) => Text -> (CanCheck => m a) -> m a

This should be used to wrap the test that contains checks. This initializes the environment needed for the checks to run, and on exit from the test, reports any (and all) failed checks as a test failure.

type CanCheck = ?checker :: IORef [CheckResult]

A convenient Constraint to apply to functions that will perform checks (i.e. call check one or more times)

Performing or Disabling checks

check :: (CanCheck, TestShow a, MonadIO m) => Text -> (a -> Bool) -> a -> m ()

This is used to run a check within the code. The first argument is the "name" of this check, the second is a function that takes a value and returns True if the value is OK, or False if the value fails the check. The last argument is the value to check.

>>> :set -XOverloadedStrings
>>> import Test.Tasty
>>> import Test.Tasty.HUnit
>>> :{
>>> defaultMain $ testCase "odd numbers" $ withChecklist "odds" $ do
>>> let three = 3 :: Int
>>> check "three is odd" odd three
>>> check "two is odd" odd (2 :: Int)
>>> check "7 + 3 is odd" odd $ 7 + three
>>> check "7 is odd" odd (7 :: Int)
>>> :}
odd numbers: FAIL
  Exception: ERROR: odds
    2 checks failed in this checklist:
    -Failed check 'two is odd' with: 2
    -Failed check '7 + 3 is odd' with: 10

1 out of 1 tests failed (...s)
*** Exception: ExitFailure 1

Any check failures are also printed to stdout (and omitted from the above for clarity). This is so that those failures are reported even if a more standard test assertion is used that prevents completion of the checklist. Thus, if an assertEqual "values" three 7 had been added to the above, that would have been the only actual (and immediate) fail for the test, but any failing checks appearing before that assertEqual would still have printed.

discardCheck :: (CanCheck, MonadIO m) => Text -> m ()

Sometimes checks are provided in common testing code, often in setup/preparation for the main tests. In some cases, the check is not applicable for that particular test. This function can be used to discard any pending failures for the associated named check.

This is especially useful when a common code block is used to perform a set of checks: if a few of the common checks are not appropriate for the current situation, discardCheck can be used to throw away the results of those checks by matching on the check name.

Type-safe multi-check specifications

Implementing a number of discrete check calls can be tedious, especially when they are validating different aspects of the same result value. To facilitate this, the checkValues function can be used along with a type-safe list of checks to perform.

To demonstrate this, first consider the following sample program, which has code that generates a complex Struct value, along with tests for various fields in that Struct.

>>> :set -XPatternSynonyms
>>> :set -XOverloadedStrings
>>> 
>>> import Data.Parameterized.Context ( pattern Empty, pattern (:>) )
>>> import Test.Tasty.Checklist
>>> import Test.Tasty
>>> import Test.Tasty.HUnit
>>> 
>>> :{
>>> data Struct = MyStruct { foo :: Int, bar :: Char, baz :: String }
>>> 
>>> instance Show Struct where
>>> show s = baz s <> " is " <> show (foo s) <> [bar s]
>>> instance TestShow Struct where testShow = show
>>> 
>>> someFun :: Int -> Struct
>>> someFun n = MyStruct (n * 6)
>>> (if n * 6 == 42 then '!' else '?')
>>> "The answer to the universe"
>>> 
>>> oddAnswer :: Struct -> Bool
>>> oddAnswer = odd . foo
>>> 
>>> test = testCase "someFun result" $
>>> withChecklist "results for someFun" $
>>> someFun 3 `checkValues`
>>> (Empty
>>> :> Val "foo" foo 42
>>> :> Val "baz field" baz "The answer to the universe"
>>> :> Val "shown" show "The answer to the universe is 42!"
>>> :> Val "odd answer" oddAnswer False
>>> :> Got "even answer" (not . oddAnswer)
>>> :> Val "double-checking foo" foo 42
>>> )
>>> :}

This code will be used below to demonstrate various advanced checklist capabilities.

checkValues :: CanCheck => TestShow dType => dType -> Assignment (DerivedVal dType) idx -> IO ()

The checkValues is a checklist that tests various values that can be derived from the input value. The input value is provided, along with an Assignment list of extraction functions and the expected result value (and name) of that extraction. Each extraction is performed as a check within the checklist.

This is convenient to gather together a number of validations on a single datatype and represent them economically.

One example is testing the fields of a record structure, given the above code:

>>> defaultMain test
someFun result: FAIL
  Exception: ERROR: results for someFun
    3 checks failed in this checklist:
    --- Input for below: The answer to the universe is 18?
    -Failed check: foo
          expected:    42
          failed with: 18
    -Failed check: shown
          expected:    "The answer to the universe is 42!"
          failed with: "The answer to the universe is 18?"
    -Failed check: double-checking foo
          expected:    42
          failed with: 18

1 out of 1 tests failed (...s)
*** Exception: ExitFailure 1

In this case, several of the values checked were correct, but more than one was wrong. Helpfully, this test output lists all the wrong answers for the single input provided.

data DerivedVal i d where

Each entry in the Assignment list for checkValues should be one of these DerivedVal values.

The i type parameter is the input type, and the d is the value derived from that input type.

Constructors

Val :: (TestShow d, Eq d) => Text -> (i -> d) -> d -> DerivedVal i d

Val allows specification of a description string, an extraction function, and the expected value to be extracted. The checkValues function will add a Failure if the expected value is not obtained.

Got :: Text -> (i -> Bool) -> DerivedVal i Bool

Got allows specification of a description string and an extraction function. The checkValues function will add a Failure if the extraction result is False.

Val "what" f True === Got "what" f

Observe :: Eq d => Text -> (i -> d) -> d -> (d -> d -> String) -> DerivedVal i d

Observe performs the same checking as Val except the TestShow information for the actual and expected values are not as useful (e.g. they are lengthy, multi-line, or gibberish) so instead this allows the specification of a function that will take the supplied expected value and the result of the extraction function (the actual), respectively, and generate its own description of the failure.

Error reporting

data CheckResult

The internal CheckResult captures the failure information for a check

Instances

Show CheckResult

Defined in Test.Tasty.Checklist

Methods

showsPrec :: Int -> CheckResult -> ShowS

show :: CheckResult -> String

showList :: [CheckResult] -> ShowS

data ChecklistFailures

The ChecklistFailures exception is thrown if any checks have failed during testing.

Instances

Exception ChecklistFailures

Defined in Test.Tasty.Checklist

Methods

toException :: ChecklistFailures -> SomeException

fromException :: SomeException -> Maybe ChecklistFailures

displayException :: ChecklistFailures -> String

Show ChecklistFailures

Defined in Test.Tasty.Checklist

Methods

showsPrec :: Int -> ChecklistFailures -> ShowS

show :: ChecklistFailures -> String

showList :: [ChecklistFailures] -> ShowS

Displaying tested values

class TestShow v where

The TestShow class is defined to provide a way for the various data objects tested by this module to be displayed when tests fail. The default testShow will use a Show instance, but this can be overridden if there are alternate ways to display a particular object (e.g. pretty-printing, etc.)

Minimal complete definition

Nothing

Methods

testShow :: v -> String

default testShow :: Show v => v -> String

Instances

TestShow String

Defined in Test.Tasty.Checklist

Methods

testShow :: String -> String

TestShow Integer

Defined in Test.Tasty.Checklist

Methods

testShow :: Integer -> String

TestShow ()

Defined in Test.Tasty.Checklist

Methods

testShow :: () -> String

TestShow Bool

Defined in Test.Tasty.Checklist

Methods

testShow :: Bool -> String

TestShow Char

Defined in Test.Tasty.Checklist

Methods

testShow :: Char -> String

TestShow Float

Defined in Test.Tasty.Checklist

Methods

testShow :: Float -> String

TestShow Int

Defined in Test.Tasty.Checklist

Methods

testShow :: Int -> String

(TestShow a, TestShow b) => TestShow (a, b)

Defined in Test.Tasty.Checklist

Methods

testShow :: (a, b) -> String

(TestShow a, TestShow b, TestShow c) => TestShow (a, b, c)

Defined in Test.Tasty.Checklist

Methods

testShow :: (a, b, c) -> String

testShowList :: TestShow v => [v] -> String

A helper function for defining a testShow for lists of items.

instance TestShow [Int] where testShow = testShowList

multiLineDiff :: Text -> Text -> String

The multiLineDiff is another helper function that can be used to format a line-by-line difference display of two Text representations. This is provided as a convenience function to help format large text regions for easier comparison.