java - JUnit和Mockito的分支覆盖

我正在使用JUnit API编写方法的测试用例。我已经介绍了所有场景，但是让我很难受的场景是在if块中。当我将鼠标悬停在这条线上时，Cobertura会针对每种情况声明50％50％，但我不确定如何获得此覆盖。

被测方法：

protected boolean isDateWithinTimelineRange( Calendar date, ServiceContext ctx ) {
    Calendar end = (Calendar)ctx.getParameter( ServiceConstants.TIMELINE_END );
    Calendar start = (Calendar)ctx.getParameter( ServiceConstants.TIMELINE_BEGIN );

    if( end != null && start != null ) {
        if( date.getTimeInMillis() >= start.getTimeInMillis() && date.getTimeInMillis() <= end.getTimeInMillis() ) {
            return true;
        } else {
            return false;
        }
    }

    return true;
}

JUnit测试用例：

@Test
public void testIsDateWithinTimelineRange() throws Exception {
    ServiceContext context = Mockito.mock(ServiceContext.class);
    Calendar calender = Mockito.mock(Calendar.class);

    Mockito.when(context.getParameter(Mockito.anyString())).thenReturn(calender);

    TestBaseTimelineProvider provider = new TestBaseTimelineProvider();
    boolean answer = provider.isDateWithinTimelineRange(calender, context);

    assertNotNull(answer);
    assertTrue(provider.isDateWithinTimelineRange(calender, context));

    // Testing for NULL condition
    context = Mockito.mock(ServiceContext.class);
    calender = Mockito.mock(Calendar.class);
    Mockito.when(context.getParameter(Mockito.anyString())).thenReturn(null);

    provider = new TestBaseTimelineProvider();
    answer = provider.isDateWithinTimelineRange(calender, context);

    assertNotNull(answer);
    assertTrue(provider.isDateWithinTimelineRange(calender, context));

    // Start date set to null
    context = Mockito.mock(ServiceContext.class);
    calender = Mockito.mock(Calendar.class);
    ServiceConstants constants = new ServiceConstants();

    Mockito.when(context.getParameter(ServiceConstants.TIMELINE_END)).thenReturn(calender);

    provider = new TestBaseTimelineProvider();
    answer = provider.isDateWithinTimelineRange(calender, context);

    assertNotNull(constants);

    // End date set to null
    context = Mockito.mock(ServiceContext.class);
    calender = Mockito.mock(Calendar.class);
    constants = new ServiceConstants();

    Mockito.when(context.getParameter(ServiceConstants.TIMELINE_BEGIN)).thenReturn(calender);

    provider = new TestBaseTimelineProvider();
    answer = provider.isDateWithinTimelineRange(calender, context);

    assertNotNull(constants);
}

让我感到困惑的是我在嘲笑的参数date，它确定了end和start变量的值。

if( date.getTimeInMillis() >= start.getTimeInMillis() && date.getTimeInMillis() <= end.getTimeInMillis() ) {}是我要介绍的行。

谢谢

最佳答案

首先，您从未告诉模拟日历对象在调用getTimeInMillis()时该怎么做。您需要为每个日历条目添加以下内容：

// Assume `long desiredlong` defined;
Mockito.when(calendar.getTimeInMillis()).thenReturn(desiredlong);

您需要对date.getTimeInMillis（）在期望范围内的一组日历对象以及date.getTimeInMillis（）在期望范围之外的另一组日历对象执行此操作。

最终，涵盖该测试真实面的案例将采用以下形式：

@Test
public void validDatesInRange() {
    ServiceContext context = Mockito.mock(ServiceContext.class);
    Calendar calenderstart = Mockito.mock(Calendar.class);
    Mockito.when(calendarstart.getTimeInMillis()).thenReturn(1L);

    Calendar calendertarget = Mockito.mock(Calendar.class);
    Mockito.when(calendartarget.getTimeInMillis()).thenReturn(2L);

    Calendar calenderend = Mockito.mock(Calendar.class);
    Mockito.when(calendarend.getTimeInMillis()).thenReturn(3L);

    Mockito.when(context.getParameter(ServiceConstants.TIMELINE_END)).thenReturn(calenderend);
    Mockito.when(context.getParameter(ServiceConstants.TIMELINE_BEGIN)).thenReturn(calenderstart);

    TestBaseTimelineProvider provider = new TestBaseTimelineProvider();
    boolean answer = provider.isDateWithinTimelineRange(calendertarget, context);

    assertNotNull(answer);
    assertTrue(provider.isDateWithinTimelineRange(calendartarget, context));
}

其次，您实际上从未写过任何测试错误返回的东西。为了覆盖另一面，请复制上面的内容，但是将您的calendartarget.getTimeInMillis()设置为返回诸如1000L之类的荒谬内容，并更改您的断言以反映错误。

您可能还希望将测试用例分解为多个方法，这些方法的名称反映每个测试方法所检查的内容，例如validDatesInRange()，validDatesNotInRange()，startDateIsNull()，endDateIsNull()，contextIsNull()。这样，您的测试变得更小，更易于理解和调试，您的测试运行会产生更清晰，更有用的测试报告，并且一项测试中的失败不会掩盖其他测试中的失败。