枚举位域容器类
我试图编写一个小类来更好地理解 C++ 中的位标志。 但有些事情并没有解决。 它打印出错误的值。 哪里有问题? 我是否误解了如何添加标志? 或者检查位域是否有它们?
代码如下:
#include <iostream>
enum flag
{
A = 1, B = 2, C = 4
};
class Holder
{
public:
Holder() : m_flags(A) {}
~Holder() {}
void add_flag(flag f) { m_flags |= f; }
bool has_flag(flag f) { return ((m_flags&f)==f); }
void remove_flag(flag f)
{
unsigned int flags = 0;
for (int i = 1; i<=(int)C; i *= 2)
{
if ((flag)i!=f && has_flag(f))
flags |= f;
}
m_flags = flags;
}
void print()
{
std::cout << "flags are now: " << m_flags << " | holding: ";
for (int i = 1; i<=(int)C; i *= 2)
{
if (has_flag((flag)i))
std::cout << i << " ";
}
std::cout << std::endl;
}
private:
unsigned int m_flags;
};
int main()
{
Holder h;
h.print(); // should print 1
h.add_flag(B);
h.print(); // should print 1 2
h.remove_flag(A);
h.print(); // should print 2
h.add_flag(C);
h.print(); // should print 2 4
h.remove_flag(B);
h.print(); // should print 4
}
程序输出:
flags are now: 1 | holding: 1
flags are now: 3 | holding: 1 2
flags are now: 1 | holding: 1
flags are now: 5 | holding: 1 4
flags are now: 0 | holding:
Im trying to write a small class to better understand bit flags in c++. But something isnt working out. It prints the wrong values. Where is the problem? Have I misunderstood how to add flags? Or check if the bit field has them?
Heres the code:
#include <iostream>
enum flag
{
A = 1, B = 2, C = 4
};
class Holder
{
public:
Holder() : m_flags(A) {}
~Holder() {}
void add_flag(flag f) { m_flags |= f; }
bool has_flag(flag f) { return ((m_flags&f)==f); }
void remove_flag(flag f)
{
unsigned int flags = 0;
for (int i = 1; i<=(int)C; i *= 2)
{
if ((flag)i!=f && has_flag(f))
flags |= f;
}
m_flags = flags;
}
void print()
{
std::cout << "flags are now: " << m_flags << " | holding: ";
for (int i = 1; i<=(int)C; i *= 2)
{
if (has_flag((flag)i))
std::cout << i << " ";
}
std::cout << std::endl;
}
private:
unsigned int m_flags;
};
int main()
{
Holder h;
h.print(); // should print 1
h.add_flag(B);
h.print(); // should print 1 2
h.remove_flag(A);
h.print(); // should print 2
h.add_flag(C);
h.print(); // should print 2 4
h.remove_flag(B);
h.print(); // should print 4
}
Output of program:
flags are now: 1 | holding: 1
flags are now: 3 | holding: 1 2
flags are now: 1 | holding: 1
flags are now: 5 | holding: 1 4
flags are now: 0 | holding:
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(4)
你的remove_flag()方法有一个错误,它应该是flags |= i;
但是,像这样执行 O(1) 操作:
There's a bug in your remove_flag() method, it should be flags |= i;
But, do it O(1) like this:
has_flag() 和 remove_flag() 是错误的。 他们应该这样:
has_flag() and remove_flag() are wrong. They should go like this:
我个人会使用 std::vector< 布尔> 处理标志,因为它是将布尔值打包成位的专业化。
但是:
我认为您的删除标志有点复杂,请尝试这个
编辑:
有评论问我为什么不
做 m_flags &= ~f
。 我将这个问题视为“学习者”问题,而不是优化问题。 我展示了如何使他的代码正确,而不是快速。personally I would use std::vector< bool > to handle flags, since it is a specialization that packs bools into bit.
However:
I think your remove flag is a bit complex, try this instead
Edit:
A comment asked why I just didn't
do m_flags &= ~f
. I took the question as a 'learner' question not an optimization question. I show how to make his code correct, not fast.每个人都已经明白了这一点:flag &= ~f;
你可以看看我之前的发布。
has_flag():如果 f 中的所有位都已设置,您是否希望返回 true? 或者是否至少设置了其中一项? 这就是flags&f==f 之间的区别 与 标志&f !=0。
您可能会考虑#include和 cout << 十六进制 <
枚举可以位于类 Holder 内部。
然后您可以使用 Holder::A 而不是 A。
您可能希望使用 for(i=0;i
您可能希望您的 add_flag/has_flag/remove_flag 方法采用 int 而不是枚举类型。 这消除了很多铸造。 如果您不想支持所有可能的 int 值,可以使用验证方法和拒绝路径。 顺便说一句,没有什么可以阻止我调用 add_flag(flag(5736))。 而且您已经非常频繁地转换为 enum_flag 了。
您可能想使用 mFlag 而不是 m_flag。 这是你的选择。 但是,当您查看像 m_x*m_y-m_z*m_y-m_x*m_z 这样的代码时,根据您的字体,很容易将 _ 误认为 -。 (反之亦然。)
同样,请考虑 addFlag 而不是 add_flag。 对于这样的事情来说,这并不重要。 但是,当您有一个很长的描述性名称时,这些下划线就会开始累加,耗尽行空间。 然后,人们可能会想缩写名称,从而使代码变得更加晦涩。
只是我的 0.02 美元。
Everyone has already nailed this: flag &= ~f;
You might look at my earlier posting.
has_flag(): Are you looking to return true if all the bits in f are set? Or if at least one of them is set? It's the difference between flags&f==f vs flags&f !=0.
You might consider #include <iomanip> and cout << hex <<m_flag <<dec. (Hex to bit conversion can be more easily done in your head.)
The enum could be inside class Holder.
You would then use Holder::A instead of A.
You may want to use for(i=0;i<N;i++) if has_flag(1<<i) ...
You may want your add_flag/has_flag/remove_flag methods to take an int rather than the enumerated type. That gets rid of a lot of casting. If you don't want to support all possible int values, a validation method and rejection path could be used. By the way, there's nothing stopping me from calling add_flag(flag(5736)). And you are casting to enum_flag quite frequently already.
You may want to use mFlag rather than m_flag. It's your choice. But when you look at code like m_x*m_y-m_z*m_y-m_x*m_z , depending on your font, it can be easy to mistake _ for -. (Or vice versa.)
Similarly, consider addFlag rather than add_flag. For something like this it doesn't matter. But when you have a long descriptive name, those underscores start to add up, using up linespace. The temptation is then to abbreviate the name, making your code more obtuse.
Just my $0.02.