olm/lib/doctest/doc/markdown/testcases.md

8.3 KiB

Test cases

While doctest fully supports the traditional, xUnit, style of class-based fixtures containing test case methods this is not the preferred style. Instead doctest provides a powerful mechanism for nesting subcases within a test case. For a more detailed discussion and examples see the tutorial.

Test cases and subcases are very easy to use in practice:

  • TEST_CASE( test name )
  • SUBCASE( subcase name )

test name and subcase name are free form, quoted, strings. Test names don't have to be unique within the doctest executable. They should also be string literals.

It is possible to write test cases inside of class bodies in C++17 with the help of TEST_CASE_CLASS() - used just like TEST_CASE() - making testing private parts of classes easier.

Keep in mind that even though doctest is thread-safe - using subcases has to be done only in the main test runner thread.

Test cases can also be parameterized - see the documentation

Test cases and subcases can be filtered through the use of the command line

BDD-style test cases

In addition to doctest's take on the classic style of test cases, doctest supports an alternative syntax that allow tests to be written as "executable specifications" (one of the early goals of Behaviour Driven Development). This set of macros map on to TEST_CASEs and SUBCASEs, with a little internal support to make them smoother to work with.

  • SCENARIO( scenario name )

This macro maps onto TEST_CASE and works in the same way, except that the test case name will be prefixed by "Scenario: "

  • SCENARIO_TEMPLATE( scenario name, type, list of types )

This macro maps onto TEST_CASE_TEMPLATE and works in the same way, except that the test case name will be prefixed by "Scenario: "

  • SCENARIO_TEMPLATE_DEFINE( scenario name, type, id )

This macro maps onto TEST_CASE_TEMPLATE_DEFINE and works in the same way, except that the test case name will be prefixed by "Scenario: "

  • GIVEN( something )
  • WHEN( something )
  • THEN( something )

These macros map onto SUBCASEs except that the subcase names are the _something_s prefixed by "given: ", "when: " or "then: " respectively.

  • AND_WHEN( something )
  • AND_THEN( something )

Similar to WHEN and THEN except that the prefixes start with "and ". These are used to chain WHENs and THENs together.

When any of these macros are used the console reporter recognises them and formats the test case header such that the Givens, Whens and Thens are aligned to aid readability.

Other than the additional prefixes and the formatting in the console reporter these macros behave exactly as TEST_CASEs and SUBCASEs. As such there is nothing enforcing the correct sequencing of these macros - that's up to the programmer!

Note that when using the --test-case=<filters> command line option (or --subcase=<filters>) you will have to pass the prefix Scenario: as well.

Test fixtures

Although doctest allows you to group tests together as subcases within a test case, it can still be convenient, sometimes, to group them using a more traditional test fixture. doctest fully supports this too. You define the test fixture as a simple structure:

class UniqueTestsFixture {
private:
    static int uniqueID;
protected:
    DBConnection conn;
public:
    UniqueTestsFixture() : conn(DBConnection::createConnection("myDB")) {}
protected:
    int getID() {
        return ++uniqueID;
    }
};

int UniqueTestsFixture::uniqueID = 0;

TEST_CASE_FIXTURE(UniqueTestsFixture, "Create Employee/No Name") {
    REQUIRE_THROWS(conn.executeSQL("INSERT INTO employee (id, name) VALUES (?, ?)", getID(), ""));
}
TEST_CASE_FIXTURE(UniqueTestsFixture, "Create Employee/Normal") {
    REQUIRE(conn.executeSQL("INSERT INTO employee (id, name) VALUES (?, ?)", getID(), "Joe Bloggs"));
}

The two test cases here will create uniquely-named derived classes of UniqueTestsFixture and thus can access the getID() protected method and conn member variables. This ensures that both the test cases are able to create a DBConnection using the same method (DRY principle) and that any ID's created are unique such that the order that tests are executed does not matter.

Test suites

Test cases can be grouped into test suites. This is done with TEST_SUITE() or TEST_SUITE_BEGIN() / TEST_SUITE_END().

For example:

TEST_CASE("") {} // not part of any test suite

TEST_SUITE("math") {
    TEST_CASE("") {} // part of the math test suite
    TEST_CASE("") {} // part of the math test suite
}

TEST_SUITE_BEGIN("utils");

TEST_CASE("") {} // part of the utils test suite

TEST_SUITE_END();

TEST_CASE("") {} // not part of any test suite

Then test cases from specific test suites can be executed with the help of filters - check out the command line

Decorators

Test cases can be decorated with additional attributes like this:

TEST_CASE("name"
          * doctest::description("shouldn't take more than 500ms")
          * doctest::timeout(0.5)) {
    // asserts
}

Multiple decorators can be used at the same time. These are the currently supported decorators:

  • skip(bool = true) - marks the test case to be skipped from execution - unless the --no-skip option is used
  • no_breaks(bool = true) - no breaking into the debugger for asserts in the test case - useful in combination with may_fail/should_fail/expected_failures
  • no_output(bool = true) - no output from asserts in the test case - useful in combination with may_fail/should_fail/expected_failures
  • may_fail(bool = true) - doesn't fail the test if any given assertion fails (but still reports it) - this can be useful to flag a work-in-progress, or a known issue that you don't want to immediately fix but still want to track in the your tests
  • should_fail(bool = true) - like may_fail() but fails the test if it passes - this can be useful if you want to be notified of accidental, or third-party, fixes
  • expected_failures(int) - defines the number of assertions that are expected to fail within the test case - reported as failure when the number of failed assertions is different than the declared expected number of failures
  • timeout(double) - fails the test case if its execution exceeds this limit (in seconds) - but doesn't terminate it - that would require subprocess support
  • test_suite("name") - can be used on test cases to override (or just set) the test suite they are in
  • description("text") - a description of the test case

The values that the decorators take are computed while registering the test cases (during global initialization) - before entering main() and not just before running them.

Decorators can also be applied to test suite blocks and all test cases in that block inherit them:

TEST_SUITE("some TS" * doctest::description("all tests will have this")) {
    TEST_CASE("has a description from the surrounding test suite") {
        // asserts
    }
}
TEST_SUITE("some TS") {
    TEST_CASE("no description even though in the same test suite as the one above") {
        // asserts
    }
}

Test cases can override the decorators that they inherit from their surrounding test suite:

TEST_SUITE("not longer than 500ms" * doctest::timeout(0.5)) {
    TEST_CASE("500ms limit") {
        // asserts
    }
    TEST_CASE("200ms limit" * doctest::timeout(0.2)) {
        // asserts
    }
}

  • Check out the subcases and BDD example
  • Check out the assertion macros example to see how test suites are used
  • Tests are registered from top to bottom of each processed cpp after the headers have been preprocessed and included but there is no ordering between cpp files.

Home