将待测 PEX 代码限制为单一方法？
PEX 中的待测代码能否限制为单一方法？我知道您可以将其限制为程序集和类，但是成员呢？
我本质上是试图实现以下方法的 100% 代码覆盖率：
public virtual bool GetLastSymbol(string symbolHint, out string symbol)
{
if (symbolHint == null)
{
throw new ArgumentNullException("symbolHint");
}
IEnumerable<string> symbols;
symbol = this.VariableHints.TryGetValue(symbolHint, out symbols) ? symbols.Last() : null;
return symbol != null;
}
以下 PUT 实现 12/15 代码覆盖率，因为我只测试它可以返回的可能的 2 个值中的 1 个
found = symbolManager.GetLastSymbol(symbolHint, out symbol);
PexAssert.IsFalse(found);
：要实现此 PUT 的完全覆盖，我需要更改对象的状态，以便该方法命中两个分支。我可以通过使用工厂方法使用单独的 PUT 来设置不同的状态来满足此要求，但这会给我留下 2 个代码覆盖不完整的 PUT，而不是 1 个完全覆盖的 PUT。我意识到理论上 2 个 PUT 的综合覆盖率将达到 100%，但我在实践中需要 100% 的数字，以便我可以正确设置 CI。
因此，要到达同一个 PUT 中的另一个分支，我必须将以下代码附加到上述 2 行：
symbolManager.CreateSymbol(symbolHint); // Ensure next call returns true.
found = symbolManager.GetLastSymbol(symbolHint, out symbol);
PexAssert.IsTrue(found);
大概 GetLastSymbol 方法的代码覆盖率现在是 100%，但是因为我已经引入了对被测类型的另一个方法调用，代码覆盖率现在下降到20/29。
如何限制 PUT 仅测量单个方法的代码覆盖率？我意识到我可能完全误解了这个指标，所以如果是这种情况请解释原因:)
Can the code-under-test in PEX be constrained to a single method? I am aware you can constrain it to assemblies and classes, but what about members?
I am essentially trying to achieve 100% code coverage for the following method:
public virtual bool GetLastSymbol(string symbolHint, out string symbol)
{
if (symbolHint == null)
{
throw new ArgumentNullException("symbolHint");
}
IEnumerable<string> symbols;
symbol = this.VariableHints.TryGetValue(symbolHint, out symbols) ? symbols.Last() : null;
return symbol != null;
}
The following PUT achieves 12/15 code coverage, because I’m only testing for 1 of the possible 2 values it can return:
found = symbolManager.GetLastSymbol(symbolHint, out symbol);
PexAssert.IsFalse(found);
To achieve full coverage for this PUT, I need to change the object’s state so that the method hits both branches. I could satisfy this by using separate PUTs using a factory method to setup the different states, but this would leave me with 2 PUTs with incomplete code coverage, rather than 1 PUT with full coverage. I realise in theory the 2 PUTs would have a combined coverage of 100%, but I need that 100% figure in practice so I can setup CI properly.
So, to reach the other branch in the same PUT, I must append the following code to the above 2 lines:
symbolManager.CreateSymbol(symbolHint); // Ensure next call returns true.
found = symbolManager.GetLastSymbol(symbolHint, out symbol);
PexAssert.IsTrue(found);
Presumably the code coverage for the GetLastSymbol method is now 100%, but because I’ve introduced another method call to the type under test, the code coverage now drops to 20/29.
How can I constrain a PUT to only measure code coverage for a single method? I realise I may have misunderstood this metric entirely, so please explain why if this is the case :)
如果你对这篇内容有疑问，欢迎到本站社区发帖提问 参与讨论，获取更多帮助，或者扫码二维码加入 Web 技术交流群。
绑定邮箱获取回复消息
由于您还没有绑定你的真实邮箱，如果其他用户或者作者回复了您的评论，将不能在第一时间通知您！
发布评论
评论（1）
随 PEX 一起安装的“PEX API 参考”对于解决此问题非常有用：
Microsoft.Pex.Framework.Coverage 命名空间包含多个过滤器属性，这些属性可以排除影响代码覆盖率的各种方面。我想要的是：
使用此方法，我能够从覆盖率报告中删除
CreateSymbol方法：这现在将我的代码覆盖率增加到13/15；比以前高了，但仍然没有达到我预期的 100%。长话短说，我发现 PEX 也在覆盖率报告中包含了构造函数...
我的代码覆盖率现在收到 13/13 。我是一只快乐的小兔子:)
The 'PEX API Reference' which comes installed with PEX was useful for solving this:
The Microsoft.Pex.Framework.Coverage namespace includes several filter attributes which can exclude various facets from effecting code coverage. The one I wanted was:
Using this method I was able to remove the
CreateSymbolmethod from the coverage report:This now increased my code coverage to 13/15; higher than before, but still not the 100% I was expecting. Long story short, I discovered PEX was including the constructor in the coverage reports too...
I'm now receiving 13/13 for my code coverage. I'm a happy bunny :)