在进行集成测试时，如何正确使用 mockito？

Question

我有一个函数，在这个函数中，我想替换一些组件的行为(余弦，正弦，...)，我正在尝试为这些部分创建一个mock，并将它们传递给函数。

class Function1Test {
private static final double accuracy = 10e-5;
static Function1 firstFunction;
static Function2 secondFunction;
static TrFunc.MyCos cos;
static TrFunc.MySin sin;
static TrFunc.MyTan tan;
static TrFunc.MySec sec;
static LgFunc.Log2 log2;
static LgFunc.Log3 log3;
static LgFunc.Log5 log5;
static LgFunc.Log10 log10;
static Func_System sys;
@BeforeAll
static void setUp() {
    cos = mock(TrFunc.MyCos.class);
    sin = mock(TrFunc.MySin.class);
    tan = mock(TrFunc.MyTan.class);
    sec = mock(TrFunc.MySec.class);
    firstFunction = new Function1(cos, sin, tan, sec);
    log2 = mock(LgFunc.Log2.class);
    log3 = mock(LgFunc.Log3.class);
    log5 = mock(LgFunc.Log5.class);
    log10 = mock(LgFunc.Log10.class);
    secondFunction = new Function2(log2, log3, log5, log10);
    sys = new Func_System(firstFunction, secondFunction);
}

static Stream<Arguments> valuesRangeProvider(){
    return Stream.of(
            arguments(-6.282185307179586, 1001001.3330000667),
            arguments(-0.161592653589793, 33.496935752492160)
    );
}
@DisplayName("Integration Test with Mocks")
@ParameterizedTest(name = "{index}: Check range of values, x = {0}")
@MethodSource("valuesRangeProvider")
void test_0(double value, double expected) throws Exception {
    when(firstFunction.sin.mySin(value, accuracy)).thenReturn(sin(value));
    when(firstFunction.cos.myCos(value, accuracy)).thenReturn(cos(value));
    when(firstFunction.tang.myTan(value, accuracy)).thenReturn(sin(value)/cos(value));
    when(firstFunction.sec.mySec(value, accuracy)).thenReturn(1/cos(value));
    assertEquals(expected, firstFunction.calculate(value, accuracy), accuracy);
}

问题是，我收到以下消息：

*org.mockito.exceptions.misusing.MissingMethodInvocationException：当（）需要一个必须是“模拟上的方法调用”的参数。例如：当（mock.getArticles（））. then返回（文章）；此外，这个错误可能会出现，因为：

< li >您stub了final/private/equals()/hashCode()方法之一。这些方法不能被存根化/验证。不支持模仿在非公共父类上声明的方法。 < li >在when()中，不在mock上调用方法，而是在其他对象上调用方法。*

这些方法都不是私有的或最终的

public class TrFunc {
public static class MyCos {
    public static double myCos(double x, double accuracy) throws Exception {
        if (accuracy < 0){
            throw new Exception("The accuracy below zero");
        }
        double theta_norm = Math.abs(x);
        theta_norm -= Math.floor(theta_norm/Math.PI/2)*2*Math.PI;
        double result = 1;
        double step = 1;
        int i;
        for (i = 1; step > accuracy && i != Integer.MAX_VALUE; i++){
            step = step*theta_norm*theta_norm/(2*i-1)/(2*i);
            if (i % 2 == 1){
                result -= step;
            } else {
                result += step;
            }
        }
        if (!Double.isFinite(result) || i == Integer.MAX_VALUE-1){
            throw new Exception("Too many iterations");
        }
        return result;
    }


    private static double factorial(int numer) {
        double factorial = 1.0d;

        while (numer != 0) {
            factorial *= numer--;
        }

        return factorial;
    }
}

public static class MySin {
    public static double mySin(double x, double accuracy) throws Exception {
        return MyCos.myCos(Math.PI/2 - x, accuracy);
    }
}

public static class MyTan {
    public static double myTan(double x, double accuracy) throws Exception {
        double divisor =  MyCos.myCos(x, accuracy);
        if (Math.abs(divisor) <= accuracy) throw new Exception("Division by zero");
        return MySin.mySin(x, accuracy) / divisor;
    }
}

public static class MySec {
    public static double mySec(double x, double accuracy) throws Exception {
        double divisor =  MyCos.myCos(x, accuracy);
        if (Math.abs(divisor) <= accuracy) throw new Exception("Division by zero");
        if ((x + Math.PI / 2) / Math.PI % 1 == 0) throw new Exception("x can't be Pi*n-Pi/2");
        return 1 / divisor;
    }
}
}

Answer 1

在这种情况下，您试图模仿实用程序类上的静态方法，这是Mockito所不允许的。请参见用Mockito模仿静态方法了解更多细节。

如果您试图测试的是数学的有效性，那么您可以简单地使用基本断言对它们进行单元测试。如果您想在测试中使用mock，您可以探索上面文章中建议的替代方案，或者重构代码以依赖非静态方法。

假设Function1只是传递给它的变量的保持器，对于每个要更改的内部类：

public static class MySin {
    public static double mySin(double x, double accuracy) throws Exception {
        return MyCos.myCos(Math.PI/2 - x, accuracy);
    }
}

到

public static class MySin {
    public double mySin(double x, double accuracy) throws Exception {
        return MyCos.myCos(Math.PI/2 - x, accuracy);
    }
}

请注意方法中移除的静态修饰符。