STL源码学习(vector篇)

#include <concept_checks.h>
#include<stl_allocate.h>

/// The vector base class's constructor and destructor allocate
///(but don't initialize) storage.  This makes exception safety easier.
template <class _Tp, class _Alloc>
class _Vector_base {
public:
  typedef _Alloc allocator_type;
  allocator_type get_allocator() const { return allocator_type(); }

  _Vector_base(const _Alloc&)
    : _M_start(0), _M_finish(0), _M_end_of_storage(0) {}
  _Vector_base(size_t __n, const _Alloc&)
    : _M_start(0), _M_finish(0), _M_end_of_storage(0)
  {
    _M_start = _M_allocate(__n);     ///分配内存
    _M_finish = _M_start;
    _M_end_of_storage = _M_start + __n;
  }

  ~_Vector_base()
   {
       _M_deallocate(_M_start, _M_end_of_storage - _M_start);   ///回收内存
   }

protected:
  _Tp* _M_start;
  _Tp* _M_finish;
  _Tp* _M_end_of_storage;

  typedef simple_alloc<_Tp, _Alloc> _M_data_allocator;     ///将_Alloc用simple_alloc包裹,更方便使用
  _Tp* _M_allocate(size_t __n)
    { return _M_data_allocator::allocate(__n); }
  void _M_deallocate(_Tp* __p, size_t __n)
    { _M_data_allocator::deallocate(__p, __n); }
};


template <class _Tp, class _Alloc = Stl_Default_Alloc >
class vector : protected _Vector_base<_Tp, _Alloc>
{
  __STL_CLASS_REQUIRES(_Tp, _Assignable);   ///vector中的元素必须可赋值,在此处检查

private:
  typedef _Vector_base<_Tp, _Alloc> _Base;
public:
  typedef _Tp value_type;
  typedef value_type* pointer;
  typedef const value_type* const_pointer;
  typedef value_type* iterator;
  typedef const value_type* const_iterator;    ///const_iterator实际是指向const value_type类型的指针
  typedef value_type& reference;
  typedef const value_type& const_reference;
  typedef size_t size_type;
  typedef ptrdiff_t difference_type;

  typedef typename _Base::allocator_type allocator_type;
  allocator_type get_allocator() const { return _Base::get_allocator(); }

  ///先定义const_reverse_iterator,再定义reverse_iterator,否则先定义的reverse_iterator
  ///会将全局空间内的reverse_iterator覆盖掉,从而在定义const_reverse_iterator时,出现错误
  typedef reverse_iterator<const_iterator, value_type, const_reference,
                           difference_type>  const_reverse_iterator;
  typedef reverse_iterator<iterator, value_type, reference, difference_type>
          reverse_iterator;

protected:
  using _Base::_M_allocate;
  using _Base::_M_deallocate;
  using _Base::_M_start;
  using _Base::_M_finish;
  using _Base::_M_end_of_storage;
protected:
  void _M_insert_aux(iterator __position, const _Tp& __x);
  void _M_insert_aux(iterator __position);

public:
  iterator begin() { return _M_start; }
  const_iterator begin() const { return _M_start; }

  iterator end() { return _M_finish; }
  const_iterator end() const { return _M_finish; }

  reverse_iterator rbegin()
    { return reverse_iterator(end()); }
  const_reverse_iterator rbegin() const
    { return const_reverse_iterator(end()); }

  reverse_iterator rend()
    { return reverse_iterator(begin()); }
  const_reverse_iterator rend() const
    { return const_reverse_iterator(begin()); }

  size_type size() const
    { return size_type(end() - begin()); }
  size_type max_size() const
    { return size_type(-1) / sizeof(_Tp); }
  size_type capacity() const
    { return size_type(_M_end_of_storage - begin()); }
  bool empty() const
    { return begin() == end(); }

  ///对operator[]的重载,使得该函数的使用性更广
  reference operator[](size_type __n) { return *(begin() + __n); }
  const_reference operator[](size_type __n) const { return *(begin() + __n); }

  explicit vector(const allocator_type& __a = allocator_type())
    : _Base(__a) {}

  vector(size_type __n, const _Tp& __value,
         const allocator_type& __a = allocator_type())
    : _Base(__n, __a)
    {
        ///分配内存在基类的构造函数中完成,此处仅进行初始化
        _M_finish = uninitialized_fill_n(_M_start, __n, __value);
    }

  explicit vector(size_type __n)
    : _Base(__n, allocator_type())
    { _M_finish = uninitialized_fill_n(_M_start, __n, _Tp()); }

  vector(const vector<_Tp, _Alloc>& __x)
    : _Base(__x.size(), __x.get_allocator())
    { _M_finish = uninitialized_copy(__x.begin(), __x.end(), _M_start); }

  /// Check whether it's an integral type.  If so, it's not an iterator.
  template <class _InputIterator>
  vector(_InputIterator __first, _InputIterator __last,
         const allocator_type& __a = allocator_type()) : _Base(__a) {
    typedef typename _Is_integer<_InputIterator>::_Integral _Integral;
    _M_initialize_aux(__first, __last, _Integral());
  }

  template <class _Integer>
  void _M_initialize_aux(_Integer __n, _Integer __value, __true_type) {
    _M_start = _M_allocate(__n);
    _M_end_of_storage = _M_start + __n;
    _M_finish = uninitialized_fill_n(_M_start, __n, __value);
  }

  template <class _InputIterator>
  void _M_initialize_aux(_InputIterator __first, _InputIterator __last,
                         __false_type) {
    _M_range_initialize(__first, __last, __ITERATOR_CATEGORY(__first));
  }

  ~vector() {
      destroy(_M_start, _M_finish); ///析构所有对象,回收内存在基类析构函数中完成
    }

  vector<_Tp, _Alloc>& operator=(const vector<_Tp, _Alloc>& __x);
  void reserve(size_type __n) {
    if (capacity() < __n) {
      const size_type __old_size = size();

      ///重新分配内存,并将原内存数据copy到新分配的内存
      iterator __tmp = _M_allocate_and_copy(__n, _M_start, _M_finish);

      destroy(_M_start, _M_finish);            ///一一析构原内存中的对象
      _M_deallocate(_M_start, _M_end_of_storage - _M_start);      ///归还原内存
      _M_start = __tmp;
      _M_finish = __tmp + __old_size;
      _M_end_of_storage = _M_start + __n;
    }
  }

  /// assign(), a generalized assignment member function.  Two
  /// versions: one that takes a count, and one that takes a range.
  /// The range version is a member template, so we dispatch on whether
  /// or not the type is an integer.

  void assign(size_type __n, const _Tp& __val) { _M_fill_assign(__n, __val); }
  void _M_fill_assign(size_type __n, const _Tp& __val);


  template <class _InputIterator>
  void assign(_InputIterator __first, _InputIterator __last) {
    typedef typename _Is_integer<_InputIterator>::_Integral _Integral;
    _M_assign_dispatch(__first, __last, _Integral());
  }

  template <class _Integer>
  void _M_assign_dispatch(_Integer __n, _Integer __val, __true_type)
    { _M_fill_assign((size_type) __n, (_Tp) __val); }

  template <class _InputIter>
  void _M_assign_dispatch(_InputIter __first, _InputIter __last, __false_type)
    { _M_assign_aux(__first, __last, __ITERATOR_CATEGORY(__first)); }

  template <class _InputIterator>
  void _M_assign_aux(_InputIterator __first, _InputIterator __last,
                     input_iterator_tag);

  template <class _ForwardIterator>
  void _M_assign_aux(_ForwardIterator __first, _ForwardIterator __last,
                     forward_iterator_tag);

  reference front() { return *begin(); }
  const_reference front() const { return *begin(); }
  reference back() { return *(end() - 1); }
  const_reference back() const { return *(end() - 1); }

  void push_back(const _Tp& __x) {
    if (_M_finish != _M_end_of_storage) {    ///vector尚有空间
      construct(_M_finish, __x);
      ++_M_finish;
    }
    else
      _M_insert_aux(end(), __x);
  }

  void push_back() {

    if (_M_finish != _M_end_of_storage) {
      construct(_M_finish);
      ++_M_finish;
    }
    else
      _M_insert_aux(end());
  }

  void swap(vector<_Tp, _Alloc>& __x) {
    ///vector的swap方法,只是将其三个标志指针交换,因此会有很高的效率
    __STD::swap(_M_start, __x._M_start);
    __STD::swap(_M_finish, __x._M_finish);
    __STD::swap(_M_end_of_storage, __x._M_end_of_storage);
  }

  iterator insert(iterator __position, const _Tp& __x) {
    size_type __n = __position - begin();
    if (_M_finish != _M_end_of_storage && __position == end()) {   ///vector尚有空间,而且插入位置为末尾
      construct(_M_finish, __x);
      ++_M_finish;
    }
    else
      _M_insert_aux(__position, __x);
    return begin() + __n;
  }

  iterator insert(iterator __position) {
    size_type __n = __position - begin();
    if (_M_finish != _M_end_of_storage && __position == end()) {
      construct(_M_finish);
      ++_M_finish;
    }
    else
      _M_insert_aux(__position);
    return begin() + __n;
  }

  /// Check whether it's an integral type.  If so, it's not an iterator.
  template <class _InputIterator>
  void insert(iterator __pos, _InputIterator __first, _InputIterator __last) {
    typedef typename _Is_integer<_InputIterator>::_Integral _Integral;
    _M_insert_dispatch(__pos, __first, __last, _Integral());
  }

  template <class _Integer>
  void _M_insert_dispatch(iterator __pos, _Integer __n, _Integer __val,
                          __true_type)
    { _M_fill_insert(__pos, (size_type) __n, (_Tp) __val); }

  template <class _InputIterator>
  void _M_insert_dispatch(iterator __pos,
                          _InputIterator __first, _InputIterator __last,
                          __false_type) {
    _M_range_insert(__pos, __first, __last, __ITERATOR_CATEGORY(__first));
  }

  void insert (iterator __pos, size_type __n, const _Tp& __x)
    { _M_fill_insert(__pos, __n, __x); }

  void _M_fill_insert (iterator __pos, size_type __n, const _Tp& __x);

  void pop_back() {
    --_M_finish;
    destroy(_M_finish);
  }
  iterator erase(iterator __position) {

    ///将__position以后的每个对象依次向前移动一位,再将最后一个元素析构,效率不高(删除尾部元素除外)
    if (__position + 1 != end())
      copy(__position + 1, _M_finish, __position);
    --_M_finish;
    destroy(_M_finish);
    return __position;
  }
  iterator erase(iterator __first, iterator __last) {
    ///还是通过大量移动来完成的
    iterator __i = copy(__last, _M_finish, __first);
    destroy(__i, _M_finish);
    _M_finish = _M_finish - (__last - __first);
    return __first;
  }

  void resize(size_type __new_size, const _Tp& __x) {
    if (__new_size < size())
      erase(begin() + __new_size, end());
    else
      insert(end(), __new_size - size(), __x);
  }
  void resize(size_type __new_size) { resize(__new_size, _Tp()); }
  void clear() { erase(begin(), end()); }

protected:

  template <class _ForwardIterator>
  iterator _M_allocate_and_copy(size_type __n, _ForwardIterator __first,
                                               _ForwardIterator __last)
{
    iterator __result = _M_allocate(__n);
    try {
      uninitialized_copy(__first, __last, __result);
      return __result;
    }catch(...){
        _M_deallocate(__result, __n);   ///若失败,归还分配内存,防止内存泄露
    }

  }


  template <class _InputIterator>
  void _M_range_initialize(_InputIterator __first,
                           _InputIterator __last, input_iterator_tag)
  {
      ///_InputIterator属性决定无法计算其区间内的元素个数,因此只能对每个对象都调用push_back插入
      ///此行为有可能导致过多次数的内存分配
    for ( ; __first != __last; ++__first)
      push_back(*__first);
  }

  /// This function is only called by the constructor.
  template <class _ForwardIterator>
  void _M_range_initialize(_ForwardIterator __first,
                           _ForwardIterator __last, forward_iterator_tag)
  {
    size_type __n = 0;
    distance(__first, __last, __n);   ///先计算元素个数

    ///再根据元素个数分配合适的内存,防止了多次内存分配以及由其导致的大量对象的构造和析构
    _M_start = _M_allocate(__n);
    _M_end_of_storage = _M_start + __n;
    _M_finish = uninitialized_copy(__first, __last, _M_start);
  }

  template <class _InputIterator>
  void _M_range_insert(iterator __pos,
                       _InputIterator __first, _InputIterator __last,
                       input_iterator_tag);

  template <class _ForwardIterator>
  void _M_range_insert(iterator __pos,
                       _ForwardIterator __first, _ForwardIterator __last,
                       forward_iterator_tag);

};

template <class _Tp, class _Alloc>
inline bool
operator==(const vector<_Tp, _Alloc>& __x, const vector<_Tp, _Alloc>& __y)
{
    ///元素个数相同且所有相同位置的元素值相同
  return __x.size() == __y.size() &&
         equal(__x.begin(), __x.end(), __y.begin());
}

template <class _Tp, class _Alloc>
inline bool
operator<(const vector<_Tp, _Alloc>& __x, const vector<_Tp, _Alloc>& __y)
{
    ///按字典序比较
  return lexicographical_compare(__x.begin(), __x.end(),
                                 __y.begin(), __y.end());
}

template <class _Tp, class _Alloc>
vector<_Tp,_Alloc>&
vector<_Tp,_Alloc>::operator=(const vector<_Tp, _Alloc>& __x)
{
  if (&__x != this) {            ///如果是同一个对象,则不需要执行下面的操作

    const size_type __xlen = __x.size();
    if (__xlen > capacity()) {           ///比本vector的容量还大,需要分配更大内存
      iterator __tmp = _M_allocate_and_copy(__xlen, __x.begin(), __x.end());

      destroy(_M_start, _M_finish);     ///析构原有内存空间的对象
      _M_deallocate(_M_start, _M_end_of_storage - _M_start);   ///归还原有内存
      _M_start = __tmp;
      _M_end_of_storage = _M_start + __xlen;
    }
    else if (size() >= __xlen) {     ///比本vecotr的元素个数还少

        ///直接复制,然后析构多余对象
      iterator __i = copy(__x.begin(), __x.end(), begin());
      destroy(__i, _M_finish);
    }
    else {                                               ///比本vector的元素个数多,单比本vector的容量小

     ///此段内存已创建有对象,可直接通过copy函数调用其赋值运算符将提供的对象赋值给原有对象
      copy(__x.begin(), __x.begin() + size(), _M_start);

      ///此段内存空白,需通过 uninitialized_copy调用其复制构造函数来创建新的对象
      uninitialized_copy(__x.begin() + size(), __x.end(), _M_finish);
    }
    _M_finish = _M_start + __xlen;
  }
  return *this;
}

template <class _Tp, class _Alloc>
void vector<_Tp, _Alloc>::_M_fill_assign(size_t __n, const value_type& __val)
{
  if (__n > capacity()) {

    ///通过局部对象的创建和对象的交换完成了内存的重新分配和旧内存对象的析构
    vector<_Tp, _Alloc> __tmp(__n, __val, get_allocator());
    __tmp.swap(*this);
  }
  else if (__n > size()) {
    fill(begin(), end(), __val);
    _M_finish = uninitialized_fill_n(_M_finish, __n - size(), __val);
  }
  else
    erase(fill_n(begin(), __n, __val), end());
}


template <class _Tp, class _Alloc> template <class _InputIter>
void vector<_Tp, _Alloc>::_M_assign_aux(_InputIter __first, _InputIter __last,
                                        input_iterator_tag) {
  iterator __cur = begin();

  for ( ; __first != __last && __cur != end(); ++__cur, ++__first)   ///先对已有对象依次赋值
    *__cur = *__first;

  if (__first == __last)       ///赋值区间比原有区间小,删除多余元素
    erase(__cur, end());
  else
    insert(end(), __first, __last);   ///赋值区间比原有区间大,插入剩余元素,此处有可能遭遇多次内存分配
}

template <class _Tp, class _Alloc> template <class _ForwardIter>
void
vector<_Tp, _Alloc>::_M_assign_aux(_ForwardIter __first, _ForwardIter __last,
                                   forward_iterator_tag) {
  size_type __len = 0;
  distance(__first, __last, __len);

  ///通过计算赋值区间内元素个数,有计划地进行处理
  if (__len > capacity()) {
    iterator __tmp = _M_allocate_and_copy(__len, __first, __last);
    destroy(_M_start, _M_finish);
    _M_deallocate(_M_start, _M_end_of_storage - _M_start);
    _M_start = __tmp;
    _M_end_of_storage = _M_finish = _M_start + __len;
  }
  else if (size() >= __len) {
    iterator __new_finish = copy(__first, __last, _M_start);
    destroy(__new_finish, _M_finish);
    _M_finish = __new_finish;
  }
  else {
    _ForwardIter __mid = __first;
    advance(__mid, size());
    copy(__first, __mid, _M_start);
    _M_finish = uninitialized_copy(__mid, __last, _M_finish);
  }
}

///这个函数用来处理vector容量不足或者需要在vector非尾部插入
template <class _Tp, class _Alloc>
void
vector<_Tp, _Alloc>::_M_insert_aux(iterator __position, const _Tp& __x)
{
  if (_M_finish != _M_end_of_storage) {       ///vector容量足够

    ///从最后一个对象开始到__position依次后移,为新插入的对象腾位置
    construct(_M_finish, *(_M_finish - 1));    ///此处属空白内存,需调用复制构造函数创建对象
    ++_M_finish;
    _Tp __x_copy = __x;
    copy_backward(__position, _M_finish - 2, _M_finish - 1);

   ///将需要插入的值赋值给合适位置
    *__position = __x_copy;
  }
  else {         ///vector容量不足

    const size_type __old_size = size();
    const size_type __len = __old_size != 0 ? 2 * __old_size : 1;  ///此举是为了防止多次内存分配

    iterator __new_start = _M_allocate(__len);
    iterator __new_finish = __new_start;
    try {
      __new_finish = uninitialized_copy(_M_start, __position, __new_start);
      construct(__new_finish, __x);
      ++__new_finish;
      __new_finish = uninitialized_copy(__position, _M_finish, __new_finish);
    }catch(...){
        ///一旦失败,析构创建的对象,归还分配的内存
        destroy(__new_start,__new_finish);
        _M_deallocate(__new_start,__len);
    }

    destroy(begin(), end());
    _M_deallocate(_M_start, _M_end_of_storage - _M_start);

    _M_start = __new_start;
    _M_finish = __new_finish;
    _M_end_of_storage = __new_start + __len;
  }
}

///实现完全同上
template <class _Tp, class _Alloc>
void
vector<_Tp, _Alloc>::_M_insert_aux(iterator __position)
{
  if (_M_finish != _M_end_of_storage) {
    construct(_M_finish, *(_M_finish - 1));
    ++_M_finish;
    copy_backward(__position, _M_finish - 2, _M_finish - 1);
    *__position = _Tp();
  }
  else {
    const size_type __old_size = size();
    const size_type __len = __old_size != 0 ? 2 * __old_size : 1;
    iterator __new_start = _M_allocate(__len);
    iterator __new_finish = __new_start;
    try {
      __new_finish = uninitialized_copy(_M_start, __position, __new_start);
      construct(__new_finish);
      ++__new_finish;
      __new_finish = uninitialized_copy(__position, _M_finish, __new_finish);
    }catch(...){
        destroy(__new_start,__new_finish);
        _M_deallocate(__new_start,__len);
    }

    destroy(begin(), end());
    _M_deallocate(_M_start, _M_end_of_storage - _M_start);
    _M_start = __new_start;
    _M_finish = __new_finish;
    _M_end_of_storage = __new_start + __len;
  }
}

///在position处插入n个x
template <class _Tp, class _Alloc>
void vector<_Tp, _Alloc>::_M_fill_insert(iterator __position, size_type __n,
                                         const _Tp& __x)
{
  if (__n != 0) {
    if (size_type(_M_end_of_storage - _M_finish) >= __n) {    ///vector剩余容量足以容纳n个对象
      _Tp __x_copy = __x;
      const size_type __elems_after = _M_finish - __position;
      iterator __old_finish = _M_finish;

      if (__elems_after > __n) {      ///__position后续元素数量多于n

        uninitialized_copy(_M_finish - __n, _M_finish, _M_finish);   ///将后n个元素移动至finish后
        _M_finish += __n;   ///修改finish
        copy_backward(__position, __old_finish - __n, __old_finish);     ///再将__position后尚未移完的元素后移

        ///至此,恰好腾出n个空间,充入即可
        fill(__position, __position + __n, __x_copy);
      }
      else {                ///__position后续对象数量少于n

        ///由于充入对象区间将超出原有区间范围,因此需分两次充入,一次是通过直接赋值充入
        ///一次是创建新对象充入

        ///在空白内存充入__n-__elems_after个__x_copy
        uninitialized_fill_n(_M_finish, __n - __elems_after, __x_copy);
        _M_finish += __n - __elems_after;

        ///将__position后的原有对象移动到已充入对象之后
        uninitialized_copy(__position, __old_finish, _M_finish);
        _M_finish += __elems_after;

        ///在腾出的空间充入__x_copy
        fill(__position, __old_finish, __x_copy);
      }
    }
    else {                                 ///vector剩余容量不足以容纳n个对象,需重新分配内存

      const size_type __old_size = size();
      const size_type __len = __old_size + max(__old_size, __n);
      iterator __new_start = _M_allocate(__len);
      iterator __new_finish = __new_start;
      try{
        __new_finish = uninitialized_copy(_M_start, __position, __new_start);
        __new_finish = uninitialized_fill_n(__new_finish, __n, __x);
        __new_finish
          = uninitialized_copy(__position, _M_finish, __new_finish);
      }catch(...){
          destroy(__new_start,__new_finish);
          _M_deallocate(__new_start,__len);

      }

      destroy(_M_start, _M_finish);
      _M_deallocate(_M_start, _M_end_of_storage - _M_start);
      _M_start = __new_start;
      _M_finish = __new_finish;
      _M_end_of_storage = __new_start + __len;
    }
  }
}


template <class _Tp, class _Alloc> template <class _InputIterator>
void
vector<_Tp, _Alloc>::_M_range_insert(iterator __pos,
                                     _InputIterator __first,
                                     _InputIterator __last,
                                     input_iterator_tag)
{
    ///效率太低
  for ( ; __first != __last; ++__first) {
    __pos = insert(__pos, *__first);
    ++__pos;
  }
}

///和_M_fill_insert大同小异
template <class _Tp, class _Alloc> template <class _ForwardIterator>
void
vector<_Tp, _Alloc>::_M_range_insert(iterator __position,
                                     _ForwardIterator __first,
                                     _ForwardIterator __last,
                                     forward_iterator_tag)
{
  if (__first != __last) {
    size_type __n = 0;
    distance(__first, __last, __n);
    if (size_type(_M_end_of_storage - _M_finish) >= __n) {
      const size_type __elems_after = _M_finish - __position;
      iterator __old_finish = _M_finish;
      if (__elems_after > __n) {
        uninitialized_copy(_M_finish - __n, _M_finish, _M_finish);
        _M_finish += __n;
        copy_backward(__position, __old_finish - __n, __old_finish);
        copy(__first, __last, __position);
      }
      else {
        _ForwardIterator __mid = __first;
        advance(__mid, __elems_after);
        uninitialized_copy(__mid, __last, _M_finish);
        _M_finish += __n - __elems_after;
        uninitialized_copy(__position, __old_finish, _M_finish);
        _M_finish += __elems_after;
        copy(__first, __mid, __position);
      }
    }
    else {
      const size_type __old_size = size();
      const size_type __len = __old_size + max(__old_size, __n);
      iterator __new_start = _M_allocate(__len);
      iterator __new_finish = __new_start;
      __STL_TRY {
        __new_finish = uninitialized_copy(_M_start, __position, __new_start);
        __new_finish = uninitialized_copy(__first, __last, __new_finish);
        __new_finish
          = uninitialized_copy(__position, _M_finish, __new_finish);
      }
      __STL_UNWIND((destroy(__new_start,__new_finish),
                    _M_deallocate(__new_start,__len)));
      destroy(_M_start, _M_finish);
      _M_deallocate(_M_start, _M_end_of_storage - _M_start);
      _M_start = __new_start;
      _M_finish = __new_finish;
      _M_end_of_storage = __new_start + __len;
    }
  }
}

大概总结一下，vector其实就是一个动态分配的数组而已，比数组好一点的是vector在每次扩充内存的时候都会实际需要的内存多分配一些（2倍),这样减少了vector的过多的重新分配内存。其实vector的重新分配内存是相当耗时的操作，因为他要对每个原有内存中的对象执行复制构造函数调用。除此之外，vector位于中间的删除和插入也是很耗时的操作，因为对插入/删除后的每个对象都要执行复制构造函数/赋值函数的调用。