如何自复制向量？ [复制]

Question

假设我有一个包含“a”和“b”的vector<string>，我想复制自己 2 次，以便向量现在包含“a”、“b”、“a”、“b”、“a ", "b"

有什么比使用for 和push_back 更好的方法？

Answer 1

我最初的想法：

myvec.reserve(myvec.size()*3);  //reserve not only speeds things upt it also protects us ftom iterator invalidation
vector<string>::iterator it = myvec.end();    //we ant to add two times the origional onto the end so we save the end of the origional end
myvec.insert(myvec.end(), myvec.begin(), it);
myvec.insert(myvec.end(), myvec.begin(), it);

感谢 Emilio Garavaglia 首先指出了这方面的问题，这里有很多问题的原因：Does std::vector::insert() invalidate iterators if the vector has enough room (created through reserve)?

第二次尝试：

std::size_t size = myvec.size();
myvec.resize(size*3);  //resize must protects us from iterator invalidation
vector<string>::iterator it = myvec.begin()+size;
std::copy(myvec.begin(),it,it);
std::copy(myvec.begin(),it,it+size);

由于没有实现将实现 std::string 其默认构造函数在堆上分配一些东西，这应该会导致更少的堆访问，因此比其他示例更快。

另一个堆访问最小化是将向量复制到另一个插入它然后移入原件，我偷了 Emilio Garavaglia 代码并拉皮条：

{
vector<string> m = { "a", "b" };
auto n = m; // keep initial value safe
m.reserve(3 * m.size()); // preallocate, to avoid consecutive allocations
m.insert(m.end, n.begin(), n.end());
std::for_each(n.begin(),n.end(),[&n](std::string &in){n.emplace_back(std::move(in));});
}

Answer 2

vector<string> m = { "a", "b" };

auto n = m; // keep initial value safe

m.reserve(3 * m.size()); // preallocate, to avoid consecutive allocations
m.insert(m.end, n.begin(), n.end());
m.insert(m.end, n.begin(), n.end());

Answer 3

我的第一个想法是避免迭代器无效的问题。

{   // note: nested scope
    vector<string> temp(vec); // 1st copy
    temp.insert(temp.end(), vec.begin(), vec.end()); // 2nd copy
    temp.insert(temp.end(), vec.begin(), vec.end()); // 3rd copy
    temp.swap(vec); // swap contents before temp is destroyed
}

经过审核，我认为 PorkyBrain 和 Emilio Garavaglia 的回答可能更有意义。

Answer 4

这是一个简单的方法：

template<typename T, typename A1, typename A2>
std::vector<T, A1> operator+( std::vector<T, A1> left, std::vector<T, A2> const& right ) {
  left.insert( left.end(), right.begin(), right.end() );
  return left;
}


int main() {
  std::vector<string> m = { "a", "b" );

  m = m + m + m;
}

但正如@ChristianAmmer 指出的那样，在std::vector 上覆盖operator+ 是模棱两可的。那是错误的。

因此，您可以编写完整的中缀命名运算符语法，并使用 C++ 的魔力将其嵌入 C++ 语言中，以消除这种歧义。有点像这样：

#include <utility>

template<typename Operation, short op>
struct InfixOp {
  Operation* self() { return static_cast<Operation*>(this); }
  Operation const* self() const { return static_cast<Operation const*>(this); }
};

template<typename first_type, typename Infix, short op>
struct PartialForm {
  Infix const* infix;

  first_type a;

  template<typename T>
  PartialForm(T&& first, Infix const* i):infix(i), a(std::forward<T>(first)) {}
};

#define OVERRIDE_OPERATORS(OP, CODE) \
template<\
  typename Left,\
  typename Operation\
>\
PartialForm<typename std::remove_reference<Left>::type, Operation, CODE> operator OP( Left&& left, InfixOp<Operation, CODE> const& op ) {\
  return PartialForm<typename std::remove_reference<Left>::type, Operation, CODE>( std::forward<Left>(left), op.self() );\
}\
\
template<\
  typename Right,\
  typename First,\
  typename Operation\
>\
auto operator OP( PartialForm<First, Operation, CODE>&& left, Right&& right )\
  ->decltype( (*left.infix)( std::move( left.a ), std::forward<Right>(right)) )\
{\
  return (*left.infix)( std::move( left.a ), std::forward<Right>(right) );\
}

OVERRIDE_OPERATORS(+, '+')
OVERRIDE_OPERATORS(*, '*')
OVERRIDE_OPERATORS(%, '%')
OVERRIDE_OPERATORS(^, '^')
OVERRIDE_OPERATORS(/, '/')
OVERRIDE_OPERATORS(==, '=')
OVERRIDE_OPERATORS(<, '<')
OVERRIDE_OPERATORS(>, '>')
OVERRIDE_OPERATORS(&, '&')
OVERRIDE_OPERATORS(|, '|')
//OVERRIDE_OPERATORS(!=, '!=')
//OVERRIDE_OPERATORS(<=, '<=')
//OVERRIDE_OPERATORS(>=, '>=')


template<typename Functor, char... ops>
struct Infix:InfixOp<Infix<Functor, ops...>, ops>...
{
  Functor f;
  template<typename F>
  explicit Infix(F&& f_in):f(std::forward<F>(f_in)) {}
  Infix(Infix<Functor, ops...> const& o):f(o.f) {}
  Infix(Infix<Functor, ops...>&& o):f(std::move(o.f)) {}
  Infix(Infix<Functor, ops...>& o):f(o.f) {}
  template<typename L, typename R>
  auto operator()( L&& left, R&& right ) const
    -> decltype( f(std::forward<L>(left), std::forward<R>(right)))
  {
    return f(std::forward<L>(left), std::forward<R>(right));
  }
};

template<char... ops, typename Functor>
Infix<Functor, ops...> make_infix( Functor&& f )
{
  return Infix<Functor, ops...>(std::forward<Functor>(f));
}

#include <vector>

struct append_vectors {
  template<typename T>
  std::vector<T> operator()(std::vector<T> left, std::vector<T>const& right) const {
    left.insert(left.end(), right.begin(), right.end());
    return std::move(left);
  }
};

struct sum_elements {
  template<typename T>
  std::vector<T> operator()(std::vector<T> left, std::vector<T>const& right) const {
    for(auto it = left.begin(), it2 = right.begin(); it != left.end() && it2 != right.end(); ++it, ++it2) {
      *it = *it + *it2;
    }
    return left;
  }
};
struct prod_elements {
  template<typename T>
  std::vector<T> operator()(std::vector<T> left, std::vector<T>const& right) const {
    for(auto it = left.begin(), it2 = right.begin(); it != left.end() && it2 != right.end(); ++it, ++it2) {
      *it = *it * *it2;
    }
    return left;
  }
};

#include <iostream>

int main() {
  auto append = make_infix<'+'>(append_vectors());
  auto sum = make_infix<'+'>(sum_elements());
  auto prod = make_infix<'*'>(prod_elements());

  std::vector<int> a = {1,2,3};
  a = a +append+ a +append+ a;
  a = a +sum+ a;
  a = a *prod* a;

  std::cout << a.size() << "\n";
  for (auto&& x:a) {
    std::cout << x << ",";
  }
  std::cout << "\n";
}

它的优点是在你使用它的时候很清楚（我的意思是，a = a +append+ a 很清楚它打算做什么），但代价是理解它的工作原理有点棘手，而且有点对于这样一个简单的问题，冗长。

但至少歧义消失了，这是一件好事，对吧？