C++(STL源码):31---关联式容器hash_set、hash_map、hash_multiset、hash_multimap源码剖析
程序员文章站
2022-03-23 10:51:06
...
- 这些关联容器底层都是使用hash table实现的,hash table已经在前面一篇文章介绍了:https://blog.csdn.net/qq_41453285/article/details/104260053
一、hash_set
- 由于hash_set底层是以hash table实现的,因此hash_set只是简单的调用hash table的方法即可
-
与set的异同点:
- hash_set与set都是用来快速查找元素的
- 但是set会对元素自动排序,而hash_set没有
- hash_set和set的使用方法相同
- 在介绍hash table的hash functions的时候说过,hash table有一些无法处理的类型(除非用户自己书写hash function)。因此hash_set也无法自己处理
hash_set源码
//以下代码摘录于stl_hash_set.h template <class _Value, class _HashFcn, class _EqualKey, class _Alloc> class hash_set { // requirements: __STL_CLASS_REQUIRES(_Value, _Assignable); __STL_CLASS_UNARY_FUNCTION_CHECK(_HashFcn, size_t, _Value); __STL_CLASS_BINARY_FUNCTION_CHECK(_EqualKey, bool, _Value, _Value); private: //indentity<>定义于<stl_function.h>中 typedef hashtable<_Value, _Value, _HashFcn, _Identity<_Value>, _EqualKey, _Alloc> _Ht; _Ht _M_ht; //底层以hash table实现 public: typedef typename _Ht::key_type key_type; typedef typename _Ht::value_type value_type; typedef typename _Ht::hasher hasher; typedef typename _Ht::key_equal key_equal; typedef typename _Ht::size_type size_type; typedef typename _Ht::difference_type difference_type; typedef typename _Ht::const_pointer pointer; typedef typename _Ht::const_pointer const_pointer; typedef typename _Ht::const_reference reference; typedef typename _Ht::const_reference const_reference; typedef typename _Ht::const_iterator iterator; typedef typename _Ht::const_iterator const_iterator; typedef typename _Ht::allocator_type allocator_type; hasher hash_funct() const { return _M_ht.hash_funct(); } key_equal key_eq() const { return _M_ht.key_eq(); } allocator_type get_allocator() const { return _M_ht.get_allocator(); } public: //缺省使用大小为100的表格,将被hash table调整为最接近且较大的质数 hash_set() : _M_ht(100, hasher(), key_equal(), allocator_type()) {} explicit hash_set(size_type __n) : _M_ht(__n, hasher(), key_equal(), allocator_type()) {} hash_set(size_type __n, const hasher& __hf) : _M_ht(__n, __hf, key_equal(), allocator_type()) {} hash_set(size_type __n, const hasher& __hf, const key_equal& __eql, const allocator_type& __a = allocator_type()) : _M_ht(__n, __hf, __eql, __a) {} #ifdef __STL_MEMBER_TEMPLATES template <class _InputIterator> hash_set(_InputIterator __f, _InputIterator __l) : _M_ht(100, hasher(), key_equal(), allocator_type()) { _M_ht.insert_unique(__f, __l); } template <class _InputIterator> hash_set(_InputIterator __f, _InputIterator __l, size_type __n) : _M_ht(__n, hasher(), key_equal(), allocator_type()) { _M_ht.insert_unique(__f, __l); } template <class _InputIterator> hash_set(_InputIterator __f, _InputIterator __l, size_type __n, const hasher& __hf) : _M_ht(__n, __hf, key_equal(), allocator_type()) { _M_ht.insert_unique(__f, __l); } template <class _InputIterator> hash_set(_InputIterator __f, _InputIterator __l, size_type __n, const hasher& __hf, const key_equal& __eql, const allocator_type& __a = allocator_type()) : _M_ht(__n, __hf, __eql, __a) { _M_ht.insert_unique(__f, __l); } #else hash_set(const value_type* __f, const value_type* __l) : _M_ht(100, hasher(), key_equal(), allocator_type()) { _M_ht.insert_unique(__f, __l); } hash_set(const value_type* __f, const value_type* __l, size_type __n) : _M_ht(__n, hasher(), key_equal(), allocator_type()) { _M_ht.insert_unique(__f, __l); } hash_set(const value_type* __f, const value_type* __l, size_type __n, const hasher& __hf) : _M_ht(__n, __hf, key_equal(), allocator_type()) { _M_ht.insert_unique(__f, __l); } hash_set(const value_type* __f, const value_type* __l, size_type __n, const hasher& __hf, const key_equal& __eql, const allocator_type& __a = allocator_type()) : _M_ht(__n, __hf, __eql, __a) { _M_ht.insert_unique(__f, __l); } hash_set(const_iterator __f, const_iterator __l) : _M_ht(100, hasher(), key_equal(), allocator_type()) { _M_ht.insert_unique(__f, __l); } hash_set(const_iterator __f, const_iterator __l, size_type __n) : _M_ht(__n, hasher(), key_equal(), allocator_type()) { _M_ht.insert_unique(__f, __l); } hash_set(const_iterator __f, const_iterator __l, size_type __n, const hasher& __hf) : _M_ht(__n, __hf, key_equal(), allocator_type()) { _M_ht.insert_unique(__f, __l); } hash_set(const_iterator __f, const_iterator __l, size_type __n, const hasher& __hf, const key_equal& __eql, const allocator_type& __a = allocator_type()) : _M_ht(__n, __hf, __eql, __a) { _M_ht.insert_unique(__f, __l); } #endif /*__STL_MEMBER_TEMPLATES */ public: //所有操作几乎都有hash table的对应版本。传递调用即可 size_type size() const { return _M_ht.size(); } size_type max_size() const { return _M_ht.max_size(); } bool empty() const { return _M_ht.empty(); } void swap(hash_set& __hs) { _M_ht.swap(__hs._M_ht); } #ifdef __STL_MEMBER_TEMPLATES template <class _Val, class _HF, class _EqK, class _Al> friend bool operator== (const hash_set<_Val, _HF, _EqK, _Al>&, const hash_set<_Val, _HF, _EqK, _Al>&); #else /* __STL_MEMBER_TEMPLATES */ friend bool __STD_QUALIFIER operator== __STL_NULL_TMPL_ARGS (const hash_set&, const hash_set&); #endif /* __STL_MEMBER_TEMPLATES */ iterator begin() const { return _M_ht.begin(); } iterator end() const { return _M_ht.end(); } public: pair<iterator, bool> insert(const value_type& __obj) { pair<typename _Ht::iterator, bool> __p = _M_ht.insert_unique(__obj); return pair<iterator,bool>(__p.first, __p.second); } #ifdef __STL_MEMBER_TEMPLATES template <class _InputIterator> void insert(_InputIterator __f, _InputIterator __l) { _M_ht.insert_unique(__f,__l); } #else void insert(const value_type* __f, const value_type* __l) { _M_ht.insert_unique(__f,__l); } void insert(const_iterator __f, const_iterator __l) {_M_ht.insert_unique(__f, __l); } #endif /*__STL_MEMBER_TEMPLATES */ pair<iterator, bool> insert_noresize(const value_type& __obj) { pair<typename _Ht::iterator, bool> __p = _M_ht.insert_unique_noresize(__obj); return pair<iterator, bool>(__p.first, __p.second); } iterator find(const key_type& __key) const { return _M_ht.find(__key); } size_type count(const key_type& __key) const { return _M_ht.count(__key); } pair<iterator, iterator> equal_range(const key_type& __key) const { return _M_ht.equal_range(__key); } size_type erase(const key_type& __key) {return _M_ht.erase(__key); } void erase(iterator __it) { _M_ht.erase(__it); } void erase(iterator __f, iterator __l) { _M_ht.erase(__f, __l); } void clear() { _M_ht.clear(); } public: void resize(size_type __hint) { _M_ht.resize(__hint); } size_type bucket_count() const { return _M_ht.bucket_count(); } size_type max_bucket_count() const { return _M_ht.max_bucket_count(); } size_type elems_in_bucket(size_type __n) const { return _M_ht.elems_in_bucket(__n); } }; template <class _Value, class _HashFcn, class _EqualKey, class _Alloc> inline bool operator==(const hash_set<_Value,_HashFcn,_EqualKey,_Alloc>& __hs1, const hash_set<_Value,_HashFcn,_EqualKey,_Alloc>& __hs2) { return __hs1._M_ht == __hs2._M_ht; }
hash_set使用演示案例
- hash_set并不会对元素进行排序
- 下面演示在hash_set中存储字符串
#include <iostream> #include <hash_set> #include <string.h> using namespace std; struct eqstr { bool operator()(const char* s1, const char* s2) { return strcmp(s1, s2) == 0; } }; void loopup(const hash_set<const char*, hash<const char*>, eqstr>& Set,const char* word) { hash_set<const char*, hash<const char*>, eqstr>::const_iterator it = Set.find(word); std::cout << " " << word << ": " << (it != Set.end() ? "present" : "not present") << std::endl; } int main() { hash_set<const char*, hash<const char*>, eqstr> Set; Set.insert("kiwi"); Set.insert("plum"); Set.insert("apple"); Set.insert("mango"); Set.insert("apricot"); Set.insert("banana"); loopup(Set, "mango"); //mango: present loopup(Set, "apple"); //apple: present loopup(Set, "durian"); //durian: not present //ite1类型为hash_set<const char*, hash<const char*>, eqstr>::iterator auto iter = Set.begin(); for (; iter != Set.end(); ++iter) std::cout << *iter << ""; //banana plum mango apple kiwi apricot std::cout << std::endl; return 0; }
- 下面演示在hash_set中存储int整型
int main() { hash_set<int> Set; //hash_set默认缺省为100。SGI内部采用质数193 Set.insert(3); //实际大小为193 Set.insert(196); Set.insert(1); Set.insert(389); Set.insert(194); Set.insert(387); //ite1类型为hash_set<const char*, hash<const char*>, eqstr>::iterator auto iter = Set.begin(); for (; iter != Set.end(); ++iter) std::cout << *iter << ""; //387 194 1 389 196 3 std::cout << std::endl; return 0; }
- 此时底层的hashtable构造如下:
二、hash_map
- 由于hash_map底层是以hash table实现的,因此hash_map只是简单的调用hash table的方法即可
-
与map的异同点:
- hash_map与map都是用来快速查找元素的
- 但是map会对元素自动排序,而hash_map没有
- hash_map和map的使用方法相同
- 在介绍hash table的hash functions的时候说过,hash table有一些无法处理的类型(除非用户自己书写hash function)。因此hash_map也无法自己处理
hash_map源码
//以下代码摘录于stl_hash_map.h //以下的hash<>是个function object,定义于<stl_hash_fun.h>中 //例如:hash<int>::operator()(int x)const{return x;} template <class _Key, class _Tp, class _HashFcn, class _EqualKey, class _Alloc> class hash_map { // requirements: __STL_CLASS_REQUIRES(_Key, _Assignable); __STL_CLASS_REQUIRES(_Tp, _Assignable); __STL_CLASS_UNARY_FUNCTION_CHECK(_HashFcn, size_t, _Key); __STL_CLASS_BINARY_FUNCTION_CHECK(_EqualKey, bool, _Key, _Key); private: //以下使用的select1st<>定义在<stl_function.h>中 typedef hashtable<pair<const _Key,_Tp>,_Key,_HashFcn, _Select1st<pair<const _Key,_Tp> >,_EqualKey,_Alloc> _Ht; _Ht _M_ht; public: typedef typename _Ht::key_type key_type; typedef _Tp data_type; typedef _Tp mapped_type; typedef typename _Ht::value_type value_type; typedef typename _Ht::hasher hasher; typedef typename _Ht::key_equal key_equal; typedef typename _Ht::size_type size_type; typedef typename _Ht::difference_type difference_type; typedef typename _Ht::pointer pointer; typedef typename _Ht::const_pointer const_pointer; typedef typename _Ht::reference reference; typedef typename _Ht::const_reference const_reference; typedef typename _Ht::iterator iterator; typedef typename _Ht::const_iterator const_iterator; typedef typename _Ht::allocator_type allocator_type; hasher hash_funct() const { return _M_ht.hash_funct(); } key_equal key_eq() const { return _M_ht.key_eq(); } allocator_type get_allocator() const { return _M_ht.get_allocator(); } public: //缺省使用大小为100的表格,将被hash table调整为最接近且较大的质数 hash_map() : _M_ht(100, hasher(), key_equal(), allocator_type()) {} explicit hash_map(size_type __n) : _M_ht(__n, hasher(), key_equal(), allocator_type()) {} hash_map(size_type __n, const hasher& __hf) : _M_ht(__n, __hf, key_equal(), allocator_type()) {} hash_map(size_type __n, const hasher& __hf, const key_equal& __eql, const allocator_type& __a = allocator_type()) : _M_ht(__n, __hf, __eql, __a) {} #ifdef __STL_MEMBER_TEMPLATES template <class _InputIterator> hash_map(_InputIterator __f, _InputIterator __l) : _M_ht(100, hasher(), key_equal(), allocator_type()) { _M_ht.insert_unique(__f, __l); } template <class _InputIterator> hash_map(_InputIterator __f, _InputIterator __l, size_type __n) : _M_ht(__n, hasher(), key_equal(), allocator_type()) { _M_ht.insert_unique(__f, __l); } template <class _InputIterator> hash_map(_InputIterator __f, _InputIterator __l, size_type __n, const hasher& __hf) : _M_ht(__n, __hf, key_equal(), allocator_type()) { _M_ht.insert_unique(__f, __l); } template <class _InputIterator> hash_map(_InputIterator __f, _InputIterator __l, size_type __n, const hasher& __hf, const key_equal& __eql, const allocator_type& __a = allocator_type()) : _M_ht(__n, __hf, __eql, __a) { _M_ht.insert_unique(__f, __l); } #else hash_map(const value_type* __f, const value_type* __l) : _M_ht(100, hasher(), key_equal(), allocator_type()) { _M_ht.insert_unique(__f, __l); } hash_map(const value_type* __f, const value_type* __l, size_type __n) : _M_ht(__n, hasher(), key_equal(), allocator_type()) { _M_ht.insert_unique(__f, __l); } hash_map(const value_type* __f, const value_type* __l, size_type __n, const hasher& __hf) : _M_ht(__n, __hf, key_equal(), allocator_type()) { _M_ht.insert_unique(__f, __l); } hash_map(const value_type* __f, const value_type* __l, size_type __n, const hasher& __hf, const key_equal& __eql, const allocator_type& __a = allocator_type()) : _M_ht(__n, __hf, __eql, __a) { _M_ht.insert_unique(__f, __l); } hash_map(const_iterator __f, const_iterator __l) : _M_ht(100, hasher(), key_equal(), allocator_type()) { _M_ht.insert_unique(__f, __l); } hash_map(const_iterator __f, const_iterator __l, size_type __n) : _M_ht(__n, hasher(), key_equal(), allocator_type()) { _M_ht.insert_unique(__f, __l); } hash_map(const_iterator __f, const_iterator __l, size_type __n, const hasher& __hf) : _M_ht(__n, __hf, key_equal(), allocator_type()) { _M_ht.insert_unique(__f, __l); } hash_map(const_iterator __f, const_iterator __l, size_type __n, const hasher& __hf, const key_equal& __eql, const allocator_type& __a = allocator_type()) : _M_ht(__n, __hf, __eql, __a) { _M_ht.insert_unique(__f, __l); } #endif /*__STL_MEMBER_TEMPLATES */ public: //所有操作几乎都有hash table的对应版本。传递调用即可 size_type size() const { return _M_ht.size(); } size_type max_size() const { return _M_ht.max_size(); } bool empty() const { return _M_ht.empty(); } void swap(hash_map& __hs) { _M_ht.swap(__hs._M_ht); } #ifdef __STL_MEMBER_TEMPLATES template <class _K1, class _T1, class _HF, class _EqK, class _Al> friend bool operator== (const hash_map<_K1, _T1, _HF, _EqK, _Al>&, const hash_map<_K1, _T1, _HF, _EqK, _Al>&); #else /* __STL_MEMBER_TEMPLATES */ friend bool __STD_QUALIFIER operator== __STL_NULL_TMPL_ARGS (const hash_map&, const hash_map&); #endif /* __STL_MEMBER_TEMPLATES */ iterator begin() { return _M_ht.begin(); } iterator end() { return _M_ht.end(); } const_iterator begin() const { return _M_ht.begin(); } const_iterator end() const { return _M_ht.end(); } public: pair<iterator,bool> insert(const value_type& __obj) { return _M_ht.insert_unique(__obj); } #ifdef __STL_MEMBER_TEMPLATES template <class _InputIterator> void insert(_InputIterator __f, _InputIterator __l) { _M_ht.insert_unique(__f,__l); } #else void insert(const value_type* __f, const value_type* __l) { _M_ht.insert_unique(__f,__l); } void insert(const_iterator __f, const_iterator __l) { _M_ht.insert_unique(__f, __l); } #endif /*__STL_MEMBER_TEMPLATES */ pair<iterator,bool> insert_noresize(const value_type& __obj) { return _M_ht.insert_unique_noresize(__obj); } iterator find(const key_type& __key) { return _M_ht.find(__key); } const_iterator find(const key_type& __key) const { return _M_ht.find(__key); } _Tp& operator[](const key_type& __key) { return _M_ht.find_or_insert(value_type(__key, _Tp())).second; } size_type count(const key_type& __key) const { return _M_ht.count(__key); } pair<iterator, iterator> equal_range(const key_type& __key) { return _M_ht.equal_range(__key); } pair<const_iterator, const_iterator> equal_range(const key_type& __key) const { return _M_ht.equal_range(__key); } size_type erase(const key_type& __key) {return _M_ht.erase(__key); } void erase(iterator __it) { _M_ht.erase(__it); } void erase(iterator __f, iterator __l) { _M_ht.erase(__f, __l); } void clear() { _M_ht.clear(); } void resize(size_type __hint) { _M_ht.resize(__hint); } size_type bucket_count() const { return _M_ht.bucket_count(); } size_type max_bucket_count() const { return _M_ht.max_bucket_count(); } size_type elems_in_bucket(size_type __n) const { return _M_ht.elems_in_bucket(__n); } }; template <class _Key, class _Tp, class _HashFcn, class _EqlKey, class _Alloc> inline bool operator==(const hash_map<_Key,_Tp,_HashFcn,_EqlKey,_Alloc>& __hm1, const hash_map<_Key,_Tp,_HashFcn,_EqlKey,_Alloc>& __hm2) { return __hm1._M_ht == __hm2._M_ht; }
hash_map使用演示案例
#include <iostream> #include <hash_map> #include <string.h> using namespace std; struct eqstr { bool operator()(const char* s1, const char* s2) { return strcmp(s1, s2) == 0; } }; int main() { hash_map<const char*, int,hash<const char*>, eqstr> days; days["january"] = 31; days["february"] = 28; days["march"] = 31; days["april"] = 30; days["may"] = 31; days["june"] = 30; days["july"] = 31; days["august"] = 31; days["september"] = 30; days["october"] = 31; days["november"] = 30; days["december"] = 31; std::cout << "september ->" << days["september"] << std::endl;//30 std::cout << "june ->" << days["june"] << std::endl; //30 std::cout << "february ->" << days["february"] << std::endl; //28 std::cout << "december ->" << days["december"] << std::endl; //31 //ite1类型为hash_map<const char*, int,hash<const char*>, eqstr>::iterator auto iter = days.begin(); for (; iter != days.end(); ++iter) std::cout << iter->first << ""; //september june july may january february december //march april november october august std::cout << std::endl; return 0; }
三、hash_multiset
-
hash_multiset与hash_set使用起来相同,只是hash_multiset中允许键值重复
-
在源码中,hash_multiset调用的是insert_equal(),而hash_set调用的是insert_unique()
四、hash_multimap
-
hash_multimap与hash_map使用起来相同,只是hash_multimap中允许键值重复
-
在源码中,hash_multimap调用的是insert_equal(),而hash_map调用的是insert_unique()