Up to date

This page is up to date for Godot 4.1. If you still find outdated information, please open an issue.

Unit testing

Godot Engine allows to write unit tests directly in C++. The engine integrates the doctest unit testing framework which gives ability to write test suites and test cases next to production code, but since the tests in Godot go through a different main entry point, the tests reside in a dedicated tests/ directory instead, which is located at the root of the engine source code.

Platform and target support

C++ unit tests can be run on Linux, macOS, and Windows operating systems.

Tests can only be run with editor tools enabled, which means that export templates cannot be tested currently.

Running tests

Before tests can be actually run, the engine must be compiled with the tests build option enabled (and any other build option you typically use), as the tests are not compiled as part of the engine by default:

scons tests=yes

Once the build is done, run the tests with a --test command-line option:

./bin/<godot_binary> --test

The test run can be configured with the various doctest-specific command-line options. To retrieve the full list of supported options, run the --test command with the --help option:

./bin/<godot_binary> --test --help

Any other options and arguments after the --test command are treated as arguments for doctest.

Note

Tests are compiled automatically if you use the dev_mode=yes SCons option. dev_mode=yes is recommended if you plan on contributing to the engine development as it will automatically treat compilation warnings as errors. The continuous integration system will fail if any compilation warnings are detected, so you should strive to fix all warnings before opening a pull request.

Filtering tests

By default, all tests are run if you don’t supply any extra arguments after the --test command. But if you’re writing new tests or would like to see the successful assertions output coming from those tests for debugging purposes, you can run the tests of interest with the various filtering options provided by doctest.

The wildcard syntax * is supported for matching any number of characters in test suites, test cases, and source file names:

For instance, to run only the String unit tests, run:

./bin/<godot_binary> --test --test-case="*[String]*"

Successful assertions output can be enabled with the --success (-s) option, and can be combined with any combination of filtering options above, for instance:

./bin/<godot_binary> --test --source-file="*test_color*" --success

Specific tests can be skipped with corresponding -exclude options. As of now, some tests include random stress tests which take a while to execute. In order to skip those kind of tests, run the following command:

./bin/<godot_binary> --test --test-case-exclude="*[Stress]*"

Writing tests

Test suites represent C++ header files which must be included as part of the main test entry point in tests/test_main.cpp. Most test suites are located directly under tests/ directory.

All header files are prefixed with test_, and this is a naming convention which the Godot build system relies on to detect tests throughout the engine.

Here’s a minimal working test suite with a single test case written:

#ifndef TEST_STRING_H
#define TEST_STRING_H
#include "tests/test_macros.h"
namespace TestString {
TEST_CASE("[String] Hello World!") {
    String hello = "Hello World!";
    CHECK(hello == "Hello World!");
}
} // namespace TestString
#endif // TEST_STRING_H

The tests/test_macros.h header encapsulates everything which is needed for writing C++ unit tests in Godot. It includes doctest assertion and logging macros such as CHECK as seen above, and of course the definitions for writing test cases themselves.

See also

tests/test_macros.h source code for currently implemented macros and aliases for them.

Test cases are created using TEST_CASE function-like macro. Each test case must have a brief description written in parentheses, optionally including custom tags which allow to filter the tests at run-time, such as [String], [Stress] etc.

Test cases are written in a dedicated namespace. This is not required, but allows to prevent naming collisions for when other static helper functions are written to accommodate the repeating testing procedures such as populating common test data for each test, or writing parameterized tests.

Godot supports writing tests per C++ module. For instructions on how to write module tests, refer to Writing custom unit tests.

Assertions

A list of all commonly used assertions used throughout the Godot tests, sorted by severity.

All of the above assertions have corresponding *_MESSAGE macros, which allow to print optional message with rationale of what should happen.

Prefer to use CHECK for self-explanatory assertions and CHECK_MESSAGE for more complex ones if you think that it deserves a better explanation.

See also

doctest: Assertion macros.

Logging

The test output is handled by doctest itself, and does not rely on Godot printing or logging functionality at all, so it’s recommended to use dedicated macros which allow to log test output in a format written by doctest.

Different reporters can be chosen at run-time. For instance, here’s how the output can be redirected to a XML file:

./bin/<godot_binary> --test --source-file="*test_validate*" --success --reporters=xml --out=doctest.txt

See also

doctest: Logging macros.

Testing failure paths

Sometimes, it’s not always feasible to test for an expected result. With the Godot development philosophy of that the engine should not crash and should gracefully recover whenever a non-fatal error occurs, it’s important to check that those failure paths are indeed safe to execute without crashing the engine.

Unexpected behavior can be tested in the same way as anything else. The only problem this creates is that the error printing shall unnecessarily pollute the test output with errors coming from the engine itself (even if the end result is successful).

To alleviate this problem, use ERR_PRINT_OFF and ERR_PRINT_ON macros directly within test cases to temporarily disable the error output coming from the engine, for instance:

TEST_CASE("[Color] Constructor methods") {
    ERR_PRINT_OFF;
    Color html_invalid = Color::html("invalid");
    ERR_PRINT_ON; // Don't forget to re-enable!
    CHECK_MESSAGE(html_invalid.is_equal_approx(Color()),
        "Invalid HTML notation should result in a Color with the default values.");
}

Test tools

Test tools are advanced methods which allow you to run arbitrary procedures to facilitate the process of manual testing and debugging the engine internals.

These tools can be run by supplying the name of a tool after the --test command-line option. For instance, the GDScript module implements and registers several tools to help the debugging of the tokenizer, parser, and compiler:

./bin/<godot_binary> --test gdscript-tokenizer test.gd
./bin/<godot_binary> --test gdscript-parser test.gd
./bin/<godot_binary> --test gdscript-compiler test.gd

If any such tool is detected, then the rest of the unit tests are skipped.

Test tools can be registered anywhere throughout the engine as the registering mechanism closely resembles of what doctest provides while registering test cases using dynamic initialization technique, but usually these can be registered at corresponding register_types.cpp sources (per module or core).

Here’s an example of how GDScript registers test tools in modules/gdscript/register_types.cpp:

#ifdef TESTS_ENABLED
void test_tokenizer() {
    TestGDScript::test(TestGDScript::TestType::TEST_TOKENIZER);
}
void test_parser() {
    TestGDScript::test(TestGDScript::TestType::TEST_PARSER);
}
void test_compiler() {
    TestGDScript::test(TestGDScript::TestType::TEST_COMPILER);
}
REGISTER_TEST_COMMAND("gdscript-tokenizer", &test_tokenizer);
REGISTER_TEST_COMMAND("gdscript-parser", &test_parser);
REGISTER_TEST_COMMAND("gdscript-compiler", &test_compiler);
#endif

The custom command-line parsing can be performed by a test tool itself with the help of OS get_cmdline_args method.

Integration tests for GDScript

Godot uses doctest to prevent regressions in GDScript during development. There are several types of test scripts which can be written:

• tests for expected errors;

• tests for warnings;

• tests for features.

Therefore, the process of writing integration tests for GDScript is the following:

1. Pick a type of a test script you’d like to write, and create a new GDScript file under the modules/gdscript/tests/scripts directory within corresponding sub-directory.

2. Write GDScript code. The test script must have a function called test() which takes no arguments. Such function will be called by the test runner. The test should not have any dependency unless it’s part of the test too. Global classes (using class_name) are registered before the runner starts, so those should work if needed.

   Here’s an example test script:

   func test():
       if true # Missing colon here.
           print("true")

3. Generate *.out files to update the expected results from the output:

   ./bin/<godot_binary> --gdscript-generate-tests godot-source/modules/gdscript/tests/scripts

You may add the --print-filenames option to see filenames as their test outputs are generated. If you are working on a new feature that is causing hard crashes, you can use this option to quickly find which test file causes the crash and debug from there.

1. Run GDScript tests with:

   ./bin/<godot_binary> --test --test-suite="*GDScript*"

This also accepts the --print-filenames option (see above).

If no errors are printed and everything goes well, you’re done!

Warning

Make sure the output does have the expected values before submitting a pull request. If --gdscript-generate-tests produces *.out files which are unrelated to newly added tests, you should revert those files back and only commit *.out files for new tests.

Note

The GDScript test runner is meant for testing the GDScript implementation, not for testing user scripts nor testing the engine using scripts. We recommend writing new tests for already resolved issues related to GDScript at GitHub, or writing tests for currently working features.

Note

If your test case requires that there is no test() function present inside the script file, you can disable the runtime section of the test by naming the script file so that it matches the pattern *.notest.gd. For example, “test_empty_file.notest.gd”.