考虑以下“往返”迭代器,它尝试迭代集合中的所有元素,最终再次迭代第一个元素作为其最后一步:
#include <boost/iterator/iterator_adaptor.hpp>
template<typename IteratorBase>
class roundtrip_iterator
: public boost::iterator_adaptor<
roundtrip_iterator<IteratorBase>, // the derived class overriding iterator behavior
IteratorBase, // the base class providing default behavior
boost::use_default, // iterator value type, will be IteratorBase::value_type
std::forward_iterator_tag, // iterator category
boost::use_default // iterator reference type
>
{
private:
IteratorBase m_itBegin;
IteratorBase m_itEnd;
bool m_complete;
public:
roundtrip_iterator( IteratorBase itBegin, IteratorBase itEnd )
: iterator_adaptor_(itBegin), m_itBegin(itBegin), m_itEnd(itEnd), m_complete(false)
{}
void increment()
{
if( m_complete )
{
base_reference() = m_itEnd;
return;
}
++base_reference();
if(base_reference() == m_itEnd)
{
base_reference() = m_itBegin;
m_complete = true;
}
}
};
我现在只实现了增量。
就目前而言,迭代器似乎在标准的“for”循环中运行良好,但我无法让它与 STL 算法一起使用。例如:
int main(int argc, char *argv[])
{
std::vector<int> v;
v.push_back(1);
v.push_back(2);
v.push_back(3);
roundtrip_iterator<std::vector<int>::iterator> roundtrip(v.begin(), v.end());
for( ; roundtrip.base() != v.end(); ++roundtrip)
std::cout << *roundtrip << std::endl;
std::cout << std::endl;
roundtrip_iterator<std::vector<int>::iterator> roundtrip2(v.begin(), v.end());
std::for_each(
roundtrip2.base(), v.end(),
print);
}
印刷:
1
2
3
1 // First element printed out using standard for loop.
1
2
3 // The for_each algo stops here for some reason.
有人对两者之间的区别有任何想法吗?