大数据处理之Trie树

Tire用于字符串统计，快速查找：

实例：

// example_Trie.cpp : 定义控制台应用程序的入口点。
//


#include "stdafx.h"
#include <stdlib.h>
#include <memory.h>  


/************************************************************************/
/*
利用trie树进行词频统计
*/
/************************************************************************/


const int num_chars = 26;  


typedef struct Trie_node{
int  count;
struct Trie_node *next[26];


}TrieNode, *Trie;




TrieNode* createTrieNode()
{
TrieNode* root =(TrieNode*)malloc(sizeof(TrieNode));
root->count = 0;
memset(root->next,0,sizeof(root->next));
return root;


}


void trie_insert(Trie root,const char* word)
{
TrieNode* node = root;
const char* p = word;
while(*p != '\0')
{
if (NULL == node->next[(*p)-'a'])
{
node->next[(*p)-'a'] = createTrieNode();
}
node = node->next[(*p)-'a'];
p++;
}
node->count +=1; 
}


int trie_search(Trie root,const char* word)
{
TrieNode* node = root;
const char* p = word;


while(*p != '\0')
{
if (node->next[*p-'a'] == NULL)
{
break;
}
node = node->next[*p-'a'];
p++;
}
return ((*p == '\0') && (node->count > 0) );
}


int trie_count(Trie root,const char* word)
{
int ret = 0;
TrieNode* node = root;
const char* p =word;
while(*p != '\0')
{
if (node->next[*p - 'a'] == NULL)
{
break;
}
node = node->next[*p - 'a'];
p++;
}
if (*p == '\0')
{
ret = node->count;
}
return ret;
}

int main(){  
Trie t = createTrieNode();  
char word[][10] = {"test","study","open","show","shit","work","work","test","tea","word","area","word","test","test","test"};  
for(int i = 0;i < 15;i++ ){  
trie_insert(t,word[i]);  
}  
for(int i = 0;i < 15;i++ ){  
printf("the word %s appears %d times in the trie-tree\n",word[i],trie_count(t,word[i]));  
}  
char s[10] = "testit";  
printf("the word %s exist? %d \n",s,trie_search(t,s));  
return 0;  
}  

运行结果：

the word test appears 5 times in the trie-tree
the word study appears 1 times in the trie-tree
the word open appears 1 times in the trie-tree
the word show appears 1 times in the trie-tree
the word shit appears 1 times in the trie-tree
the word work appears 2 times in the trie-tree
the word work appears 2 times in the trie-tree
the word test appears 5 times in the trie-tree
the word tea appears 1 times in the trie-tree
the word word appears 2 times in the trie-tree
the word area appears 1 times in the trie-tree
the word word appears 2 times in the trie-tree
the word test appears 5 times in the trie-tree
the word test appears 5 times in the trie-tree
the word test appears 5 times in the trie-tree
the word testit exist? 0
请按任意键继续. . .