HashTable 解决碰撞(冲突)的方法 —— 分离链接法(separate chaining)
转载
1. ListNode 及 HashTable 的类型声明
-
- 声明
typedef int ElementType;typedef unsigned int Index;
struct ListNode;
typedef struct ListNode* Position;
struct HashTbl;
typedef struct HashTbl* HashTable;
HashTable InitHashTable(int TableSize);
void DestroyHashTable(HashTable H);
Position Find(ElementType Element, HashTable H);
void Insert(ElementType Element, HashTable H);;
ElementType Retrieve(Position P);
- 定义
struct ListNode{ ElementType Element;
Position Next;
};
typedef Position List;
struct HashTbl {
int TableSize;
List* TheLists;
};
-
2. HashTable 的构建
HashTable InitHashTable(int TableSize){
HashTable H;
if (TableSize < MinTableSize){
Error(" ... ");
return NULL;
}
H = (HashTable)malloc(sizeof(struct HashTbl));
if (!H)
FatalError("out of space !!!");
H->TableSize = NextPrime(TableSize);
H->TheLists = (List*)malloc(sizeof(struct ListNode)*H->TableSize);
for (int i = 0; i < H->TableSize; ++i){
H->TheLists[i] = (List)malloc(sizeof(struct ListNode));
if (!H->TheLists[i])
FatalError("");
else
H->TheLists[i]->Next = NULL;
}
}
3. 插入新的结点
void Insert(ElementType Key, HashTable H){
Position Pos, NewCell;
List L;
Pos = Find(key, H);
if (!Pos){
NewCell = (Position)malloc(sizeof(struct ListNode));
if (!NewCell)
FatalError("");
L = H->TheLists[Hash(Key, H->TableSize)];
NewCell->Next = L->Next;
// 插入在首部
NewCell->Element = Key;
L->Next = NewCell->Next;
}
}
本文章为转载内容,我们尊重原作者对文章享有的著作权。如有内容错误或侵权问题,欢迎原作者联系我们进行内容更正或删除文章。