TestCaseSource
TestCaseSourceAttribute
is used on a parameterized test method to identify the source from which the required
arguments will be provided. The attribute additionally identifies the method as a test method. The data is kept separate
from the test itself and may be used by multiple test methods. See Parameterized Tests for a
general introduction to tests with arguments.
Usage
Consider a test of the divide operation, taking three arguments: the numerator, the denominator and the expected result. We can specify the test and its data using one of the forms of TestCaseSourceAttribute:
Form 1 - [TestCaseSource(string sourceName)]
Note
We use the nameof
operator
to avoid introducing magic strings into code, which offers better
resilience when refactoring. While nameof
is recommended, you could also use the string "DivideCases" to achieve the
same outcome.
public class BasicTestCaseSourceFixture
{
[TestCaseSource(nameof(DivideCases))]
public void DivideTest(int n, int d, int q)
{
ClassicAssert.AreEqual(q, n / d);
}
public static object[] DivideCases =
{
new object[] { 12, 3, 4 },
new object[] { 12, 2, 6 },
new object[] { 12, 4, 3 }
};
}
The single attribute argument in this form is a string representing the name of the source used to provide test cases. It has the following characteristics:
- It may be a field, property or method in the test class.
- It must be static. This is a change from NUnit 2.x.
- It must return an
IEnumerable
or a type that implementsIEnumerable
. For fields an array is generally used. For properties and methods, you may return an array or implement your own iterator.- Methods may also return an
IAsyncEnumerable
or a type that implementsIAsyncEnumerable
. (NUnit 4+) - Methods may be async by wrapping the return type in a
Task<T>
. (NUnit 3.14+)
- Methods may also return an
- The individual items returned by the enumerator must be compatible with the signature of the method on which the attribute appears. See the Test Case Construction section below for details.
Sometimes we would like to parameterize the source, e.g. if we use the same source for multiple tests, to this end it is possible to pass parameters to the source, if the source is a method. The parameters are specified as an array of parameters that are passed to the source method.
public class ParameterizedSourceExampleFixture
{
[TestCaseSource(nameof(TestStrings), new object[] { true })]
public void LongNameWithEvenNumberOfCharacters(string name)
{
Assert.That(name.Length, Is.GreaterThan(5));
bool hasEvenNumOfCharacters = (name.Length % 2) == 0;
Assert.That(hasEvenNumOfCharacters, Is.True);
}
[TestCaseSource(nameof(TestStrings), new object[] { false })]
public void ShortNameWithEvenNumberOfCharacters(string name)
{
Assert.That(name.Length, Is.LessThan(15));
bool hasEvenNumOfCharacters = (name.Length % 2) == 0;
Assert.That(hasEvenNumOfCharacters, Is.True);
}
static IEnumerable<string> TestStrings(bool generateLongTestCase)
{
if (generateLongTestCase)
{
yield return "ThisIsAVeryLongNameThisIsAVeryLongName";
yield return "SomeName";
yield return "YetAnotherName";
}
else
{
yield return "AA";
yield return "BB";
yield return "CC";
}
}
}
Form 2 - [TestCaseSource(Type sourceType, string sourceName)]
public class TestFixtureThatUsesClassMethodAsTestCaseSource
{
[TestCaseSource(typeof(AnotherClassWithTestFixtures), nameof(AnotherClassWithTestFixtures.DivideCases))]
public void DivideTest(int n, int d, int q)
{
ClassicAssert.AreEqual(q, n / d);
}
}
public class AnotherClassWithTestFixtures
{
public static object[] DivideCases =
{
new object[] { 12, 3, 4 },
new object[] { 12, 2, 6 },
new object[] { 12, 4, 3 }
};
}
The first argument of the attribute in this form is a Type representing the class that will provide the test cases.
The second argument is a string representing the name of the source used to provide test cases. It has the following characteristics:
- It may be a field, property or method in the test class.
- It must be static. This is a change from NUnit 2.x.
- It must return an
IEnumerable
or a type that implementsIEnumerable
. For fields an array is generally used. For properties and methods, you may return an array or implement your own iterator.- Methods may also return an
IAsyncEnumerable
or a type that implementsIAsyncEnumerable
. (NUnit 4+) - Methods may be async by wrapping the return type in a
Task<T>
. (NUnit 3.14+)
- Methods may also return an
- The individual items returned by the enumerator must be compatible with the signature of the method on which the attribute appears. See the Test Case Construction section below for details.
Similar to Form 1 it is possible to pass parameters to the source, if the source is a method.
Form 3 - [TestCaseSource(Type sourceType)]
public class TestFixtureThatUsesClassAsTestCaseSource
{
[TestCaseSource(typeof(DivideCasesClass))]
public void DivideTest(int n, int d, int q)
{
ClassicAssert.AreEqual(q, n / d);
}
}
public class DivideCasesClass : IEnumerable
{
public IEnumerator GetEnumerator()
{
yield return new object[] { 12, 3, 4 };
yield return new object[] { 12, 2, 6 };
yield return new object[] { 12, 4, 3 };
}
}
The Type argument in this form represents the class that provides test cases. It must have a default constructor and
implement IEnumerable
. The enumerator should return test case data compatible with the signature of the test on which
the attribute appears. See the Test Case Construction section below for details.
Note that it is not possible to pass parameters to the source, even if the source is a method.
Sources with expected result using TestCaseData
As of NUnit 3.12, it is possible to use a typed source with an expected result. This is done by using the
TestCaseSource
attribute on a method that returns a TestCaseData
object. The TestCaseData
object can be
constructed with the expected result as a parameter.
[TestFixture]
public class MyTests
{
[TestCaseSource(typeof(MyDataClass), nameof(MyDataClass.TestCases))]
public int DivideTest(int n, int d)
{
return n / d;
}
}
public class MyDataClass
{
public static IEnumerable TestCases
{
get
{
yield return new TestCaseData(12, 3).Returns(4);
yield return new TestCaseData(12, 2).Returns(6);
yield return new TestCaseData(12, 4).Returns(3);
}
}
}
See TestCaseData for more information on the TestCaseData
class.
Sources for generic methods using TestCaseData
As of NUnit 4.1, it is possible to explicitly specify the generic types to be used for a generic method. This
may be useful when any of the test case arguments differ from the desired generic types. When omitted, NUnit will
infer the generic type arguments based on the passed values from the TestCaseSource
.
[TestFixture]
public class MyExplicitlyTypedTests
{
[TestCaseSource(nameof(ExplicitTypeArgsTestCases))]
public void ExplicitTypeArgs<T>(T input)
{
Assert.That(typeof(T), Is.EqualTo(typeof(long)));
}
private static IEnumerable<TestCaseData> ExplicitTypeArgsTestCases()
{
yield return new TestCaseData(2) { TypeArgs = new[] { typeof(long) } };
yield return new TestCaseData(2L) { TypeArgs = new[] { typeof(long) } };
}
}
See TestCaseData for more information on the TestCaseData
class.
Examples using TestCaseSource with Typed data and expected results
It may seem from the examples above that TestCaseSource can only be used with simple data types or the base Object type. This is not the case. TestCaseSource can be used with typed data and also including expected results, also without using TestCaseData.
In the example below the test method takes a single argument of a an anonymous tuple type with Person
and an expected
value of type bool
. It can of course be any type, if that makes sense for the test. The TestCaseSource method returns
an IEnumerable<>
of the anonymous tuple type.
public class TypedValuesWithExpectedAsAnonymousTuple
{
[TestCaseSource(nameof(TestCases))]
public void TestOfPersonAge((Person P, bool Expected) td)
{
var res = td.P.IsOldEnoughToBuyAlcohol();
Assert.That(res, Is.EqualTo(td.Expected));
}
public static IEnumerable<(Person, bool)> TestCases()
{
yield return (new Person { Name = "John", Age = 10 }, false);
yield return (new Person { Name = "Jane", Age = 30 }, true);
}
}
public class Person
{
public string Name { get; set; } = "";
public int Age { get; set; }
public bool IsOldEnoughToBuyAlcohol()
{
return Age >= 18;
}
}
It is also possible to use a generic wrapper (or any custom wrapper) for the testcase data and the expected result, as shown in the example below.
public class TypedValuesWithExpectedInWrapperClass
{
[TestCaseSource(nameof(TestCases))]
public void TestOfPersonAge(TestDataWrapper<Person, bool> td)
{
var res = td.Value?.IsOldEnoughToBuyAlcohol();
Assert.That(res, Is.EqualTo(td.Expected));
}
public static IEnumerable<TestDataWrapper<Person, bool>> TestCases()
{
yield return new TestDataWrapper<Person, bool> { Value = new Person { Name = "John", Age = 10 }, Expected = false };
yield return new TestDataWrapper<Person, bool> { Value = new Person { Name = "Jane", Age = 30 }, Expected = true };
}
}
public class TestDataWrapper<T, TExp>
{
public T? Value { get; set; }
public TExp? Expected { get; set; }
}
Named Parameters
TestCaseSourceAttribute supports one named parameter:
- Category is used to assign one or more categories to every test case returned from this source.
Test Case Construction
In constructing tests, NUnit uses each item returned by the enumerator as follows:
- If it is an object derived from the
TestCaseParameters
class, its properties are used to provide the test case. NUnit provides the TestCaseData type for this purpose. - If the test has a single argument and the returned value matches the type of that argument it is used directly. This can eliminate a bit of extra typing by the programmer, as in this example:
private static int[] _evenNumbers = { 2, 4, 6, 8 };
[Test, TestCaseSource(nameof(_evenNumbers))]
public void TestMethod(int num)
{
Assert.That(num % 2, Is.Zero);
}
- If it is an
object[]
, its members are used to provide the arguments for the method. This is the approach taken in the three first examples above. - If it is an array of some other type, NUnit can use it provided that the arguments to the method are all of that type.
For example, the above examples could be modified to make the three nested arrays of type
int[]
. - If anything else is returned, it is used directly as the sole argument to the method. Because every returned value is used, NUnit is able to give an error message in cases where the method requires a different number of arguments or an argument of a different type.
Notes
- It is recommended that the SourceType not be the same as the test fixture class. It may be a nested class, however, and probably should be if the data is only used within that fixture.
- A generic
IEnumerable
andIEnumerator
may be used but NUnit will actually deal with the underlyingIEnumerator
in the current release. - The GetEnumerator method may use yield statements or simply return the enumerator for an array or other collection held by the class.
Order of Execution
Individual test cases are executed in the order in which NUnit discovers them. This order does not follow the lexical order of the attributes and will often vary between different compilers or different versions of the CLR.
As a result, when TestCaseSourceAttribute appears multiple times on a method or when other data-providing attributes are used in combination with TestCaseSourceAttribute, the order of the test cases is undefined.
However, when a single TestCaseSourceAttribute is used by itself, the order of the tests follows exactly the order in which the test cases are returned from the source.
Object Construction
NUnit locates the test cases at the time the tests are loaded. It creates instances of each class used with the third form of the attribute and builds a list of tests to be executed. Each data source class is only created once at this time and is destroyed after all tests are loaded. By design, no communication is possible between the load and execution phases except through the tests that are created.