字符串Hash函数评估

Hash查找因为其O(1)的查找性能而著称,被对查找性能要求高的应用所广泛采用。它的基本思想是:
(1) 创建一个定长的线性Hash表,一般可以初始化时指定length;

(2) 设计Hash函数,将关键字key散射到Hash表中。其中hash函数设计是最为关键的,均匀分布、冲突概率小全在它;

(3) 通常采用拉链方法来解决hash冲突问题,即散射到同一个hash表项的关键字,以链表形式来表示(也称为桶backet);

(4) 给定关键字key,就可以在O(1) + O(m)的时间复杂度内定位到目标。其中,m为拉链长度,即桶深。

Hash应用中,字符串是最为常见的关键字,应用非常普通,现在的程序设计语言中基本上都提供了字符串hash表的支持。字符串hash函数非常多,常见的主要有Simple_hash, RS_hash, JS_hash, PJW_hash, ELF_hash, BKDR_hash, SDBM_hash, DJB_hash, AP_hash, CRC_hash等。它们的C语言实现见后面附录代码: hash.h, hash.c。那么这么些字符串hash函数,谁好熟非呢?评估hash函数优劣的基准主要有以下两个指标:

(1) 散列分布性

即桶的使用率backet_usage = (已使用桶数) / (总的桶数),这个比例越高,说明分布性良好,是好的hash设计。

(2) 平均桶长

即avg_backet_len,所有已使用桶的平均长度。理想状态下这个值应该=1,越小说明冲突发生地越少,是好的hash设计。

hash函数计算一般都非常简洁,因此在耗费计算时间复杂性方面判别甚微,这里不作对比。

评估方案设计是这样的:

(1) 以200M的视频文件作为输入源,以4KB的块为大小计算MD5值,并以此作为hash关键字;

(2) 分别应用上面提到的各种字符串hash函数,进行hash散列模拟;

(3) 统计结果,用散列分布性和平均桶长两个指标进行评估分析。

测试程序见附录代码hashtest.c,测试结果如下表所示。从这个结果我们也可以看出,这些字符串hash函数真是不相仲伯,难以决出高低,所以实际应用中可以根据喜好选择。当然,最好实际测试一下,毕竟应用特点不大相同。其他几组测试结果也类似,这里不再给出。

Hash函数 桶数 Hash调用总数 最大桶长 平均桶长 桶使用率%
simple_hash 10240 47198 16 4.63 99.00%
RS_hash 10240 47198 16 4.63 98.91%
JS_hash 10240 47198 15 4.64 98.87%
PJW_hash 10240 47198 16 4.63 99.00%
ELF_hash 10240 47198 16 4.63 99.00%
BKDR_hash 10240 47198 16 4.63 99.00%
SDBM_hash 10240 47198 16 4.63 98.90%
DJB_hash 10240 47198 15 4.64 98.85%
AP_hash 10240 47198 16 4.63 98.96%
CRC_hash 10240 47198 16 4.64 98.77%


附录源代码:

hash.h

view plaincopy to clipboardprint?
#ifndef _HASH_H  
#define _HASH_H  
 
#ifdef __cplusplus  
extern "C" {  
#endif  
 
/* A Simple Hash Function */ 
unsigned int simple_hash(char *str);  
 
/* RS Hash Function */ 
unsigned int RS_hash(char *str);  
 
/* JS Hash Function */ 
unsigned int JS_hash(char *str);  
 
/* P. J. Weinberger Hash Function */ 
unsigned int PJW_hash(char *str);  
 
/* ELF Hash Function */ 
unsigned int ELF_hash(char *str);  
 
/* BKDR Hash Function */ 
unsigned int BKDR_hash(char *str);  
 
/* SDBM Hash Function */ 
unsigned int SDBM_hash(char *str);  
 
/* DJB Hash Function */ 
unsigned int DJB_hash(char *str);  
 
/* AP Hash Function */ 
unsigned int AP_hash(char *str);  
 
/* CRC Hash Function */ 
unsigned int CRC_hash(char *str);  
 
#ifdef __cplusplus  
}  
#endif  
 
#endif 
#ifndef _HASH_H
#define _HASH_H

#ifdef __cplusplus
extern "C" {
#endif

/* A Simple Hash Function */
unsigned int simple_hash(char *str);

/* RS Hash Function */
unsigned int RS_hash(char *str);

/* JS Hash Function */
unsigned int JS_hash(char *str);

/* P. J. Weinberger Hash Function */
unsigned int PJW_hash(char *str);

/* ELF Hash Function */
unsigned int ELF_hash(char *str);

/* BKDR Hash Function */
unsigned int BKDR_hash(char *str);

/* SDBM Hash Function */
unsigned int SDBM_hash(char *str);

/* DJB Hash Function */
unsigned int DJB_hash(char *str);

/* AP Hash Function */
unsigned int AP_hash(char *str);

/* CRC Hash Function */
unsigned int CRC_hash(char *str);

#ifdef __cplusplus
}
#endif

#endif
 

hash.c

view plaincopy to clipboardprint?
#include <string.h>  
#include "hash.h"  
 
/* A Simple Hash Function */ 
unsigned int simple_hash(char *str)  
{  
    register unsigned int hash;  
    register unsigned char *p;  
 
    for(hash = 0, p = (unsigned char *)str; *p ; p++)  
        hash = 31 * hash + *p;  
 
    return (hash & 0x7FFFFFFF);  
}  
 
/* RS Hash Function */ 
unsigned int RS_hash(char *str)  
{  
         unsigned int b = 378551;  
         unsigned int a = 63689;  
         unsigned int hash = 0;  
 
         while (*str)  
         {  
                 hash = hash * a + (*str++);  
                 a *= b;  
         }  
 
         return (hash & 0x7FFFFFFF);  
}  
 
/* JS Hash Function */ 
unsigned int JS_hash(char *str)  
{  
         unsigned int hash = 1315423911;  
 
         while (*str)  
         {  
                 hash ^= ((hash << 5) + (*str++) + (hash >> 2));  
         }  
          
         return (hash & 0x7FFFFFFF);  
}  
 
/* P. J. Weinberger Hash Function */ 
unsigned int PJW_hash(char *str)  
{  
         unsigned int BitsInUnignedInt = (unsigned int)(sizeof(unsigned int) * 8);  
         unsigned int ThreeQuarters     = (unsigned int)((BitsInUnignedInt   * 3) / 4);  
         unsigned int OneEighth         = (unsigned int)(BitsInUnignedInt / 8);  
 
         unsigned int HighBits          = (unsigned int)(0xFFFFFFFF) << (BitsInUnignedInt - OneEighth);  
         unsigned int hash              = 0;  
         unsigned int test              = 0;  
 
         while (*str)  
         {  
                 hash = (hash << OneEighth) + (*str++);  
                 if ((test = hash & HighBits) != 0)  
                 {  
                         hash = ((hash ^ (test >> ThreeQuarters)) & (~HighBits));  
                 }  
         }  
 
         return (hash & 0x7FFFFFFF);  
}  
 
/* ELF Hash Function */ 
unsigned int ELF_hash(char *str)  
{  
         unsigned int hash = 0;  
         unsigned int x     = 0;  
 
         while (*str)  
         {  
                 hash = (hash << 4) + (*str++);  
                 if ((x = hash & 0xF0000000L) != 0)  
                 {  
                         hash ^= (x >> 24);  
                         hash &= ~x;  
                 }  
         }  
 
         return (hash & 0x7FFFFFFF);  
}  
 
/* BKDR Hash Function */ 
unsigned int BKDR_hash(char *str)  
{  
         unsigned int seed = 131; // 31 131 1313 13131 131313 etc..  
         unsigned int hash = 0;  
 
         while (*str)  
         {  
                 hash = hash * seed + (*str++);  
         }  
 
         return (hash & 0x7FFFFFFF);  
}  
 
/* SDBM Hash Function */ 
unsigned int SDBM_hash(char *str)  
{  
         unsigned int hash = 0;  
 
         while (*str)  
         {  
                 hash = (*str++) + (hash << 6) + (hash << 16) - hash;  
         }  
 
         return (hash & 0x7FFFFFFF);  
}  
 
/* DJB Hash Function */ 
unsigned int DJB_hash(char *str)  
{  
         unsigned int hash = 5381;  
 
         while (*str)  
         {  
                 hash += (hash << 5) + (*str++);  
         }  
 
         return (hash & 0x7FFFFFFF);  
}  
 
/* AP Hash Function */ 
unsigned int AP_hash(char *str)  
{  
         unsigned int hash = 0;  
         int i;  
         for (i=0; *str; i++)  
         {  
                 if ((i & 1) == 0)  
                 {  
                         hash ^= ((hash << 7) ^ (*str++) ^ (hash >> 3));  
                 }  
                 else 
                 {  
                         hash ^= (~((hash << 11) ^ (*str++) ^ (hash >> 5)));  
                 }  
         }  
 
         return (hash & 0x7FFFFFFF);  
}  
 
/* CRC Hash Function */ 
unsigned int CRC_hash(char *str)  
{  
    unsigned int        nleft   = strlen(str);  
    unsigned long long  sum     = 0;  
    unsigned short int *w       = (unsigned short int *)str;  
    unsigned short int  answer  = 0;  
 
    /* 
     * Our algorithm is simple, using a 32 bit accumulator (sum), we add 
     * sequential 16 bit words to it, and at the end, fold back all the 
     * carry bits from the top 16 bits into the lower 16 bits. 
     */ 
    while ( nleft > 1 ) {  
        sum += *w++;  
        nleft -= 2;  
    }  
    /* 
     * mop up an odd byte, if necessary 
     */ 
    if ( 1 == nleft ) {  
        *( unsigned char * )( &answer ) = *( unsigned char * )w ;  
        sum += answer;  
    }  
    /* 
     * add back carry outs from top 16 bits to low 16 bits 
     * add hi 16 to low 16 
     */ 
    sum = ( sum >> 16 ) + ( sum & 0xFFFF );  
    /* add carry */ 
    sum += ( sum >> 16 );  
    /* truncate to 16 bits */ 
    answer = ~sum;  
 
    return (answer & 0xFFFFFFFF);  

#include <string.h>
#include "hash.h"

/* A Simple Hash Function */
unsigned int simple_hash(char *str)
{
    register unsigned int hash;
    register unsigned char *p;

    for(hash = 0, p = (unsigned char *)str; *p ; p++)
        hash = 31 * hash + *p;

    return (hash & 0x7FFFFFFF);
}

/* RS Hash Function */
unsigned int RS_hash(char *str)
{
         unsigned int b = 378551;
         unsigned int a = 63689;
         unsigned int hash = 0;

         while (*str)
         {
                 hash = hash * a + (*str++);
                 a *= b;
         }

         return (hash & 0x7FFFFFFF);
}

/* JS Hash Function */
unsigned int JS_hash(char *str)
{
         unsigned int hash = 1315423911;

         while (*str)
         {
                 hash ^= ((hash << 5) + (*str++) + (hash >> 2));
         }
       
         return (hash & 0x7FFFFFFF);
}

/* P. J. Weinberger Hash Function */
unsigned int PJW_hash(char *str)
{
         unsigned int BitsInUnignedInt = (unsigned int)(sizeof(unsigned int) * 8);
         unsigned int ThreeQuarters     = (unsigned int)((BitsInUnignedInt   * 3) / 4);
         unsigned int OneEighth         = (unsigned int)(BitsInUnignedInt / 8);

         unsigned int HighBits          = (unsigned int)(0xFFFFFFFF) << (BitsInUnignedInt - OneEighth);
         unsigned int hash              = 0;
         unsigned int test              = 0;

         while (*str)
         {
                 hash = (hash << OneEighth) + (*str++);
                 if ((test = hash & HighBits) != 0)
                 {
                         hash = ((hash ^ (test >> ThreeQuarters)) & (~HighBits));
                 }
         }

         return (hash & 0x7FFFFFFF);
}

/* ELF Hash Function */
unsigned int ELF_hash(char *str)
{
         unsigned int hash = 0;
         unsigned int x     = 0;

         while (*str)
         {
                 hash = (hash << 4) + (*str++);
                 if ((x = hash & 0xF0000000L) != 0)
                 {
                         hash ^= (x >> 24);
                         hash &= ~x;
                 }
         }

         return (hash & 0x7FFFFFFF);
}

/* BKDR Hash Function */
unsigned int BKDR_hash(char *str)
{
         unsigned int seed = 131; // 31 131 1313 13131 131313 etc..
         unsigned int hash = 0;

         while (*str)
         {
                 hash = hash * seed + (*str++);
         }

         return (hash & 0x7FFFFFFF);
}

/* SDBM Hash Function */
unsigned int SDBM_hash(char *str)
{
         unsigned int hash = 0;

         while (*str)
         {
                 hash = (*str++) + (hash << 6) + (hash << 16) - hash;
         }

         return (hash & 0x7FFFFFFF);
}

/* DJB Hash Function */
unsigned int DJB_hash(char *str)
{
         unsigned int hash = 5381;

         while (*str)
         {
                 hash += (hash << 5) + (*str++);
         }

         return (hash & 0x7FFFFFFF);
}

/* AP Hash Function */
unsigned int AP_hash(char *str)
{
         unsigned int hash = 0;
         int i;
         for (i=0; *str; i++)
         {
                 if ((i & 1) == 0)
                 {
                         hash ^= ((hash << 7) ^ (*str++) ^ (hash >> 3));
                 }
                 else
                 {
                         hash ^= (~((hash << 11) ^ (*str++) ^ (hash >> 5)));
                 }
         }

         return (hash & 0x7FFFFFFF);
}

/* CRC Hash Function */
unsigned int CRC_hash(char *str)
{
    unsigned int        nleft   = strlen(str);
    unsigned long long  sum     = 0;
    unsigned short int *w       = (unsigned short int *)str;
    unsigned short int  answer  = 0;

    /*
     * Our algorithm is simple, using a 32 bit accumulator (sum), we add
     * sequential 16 bit words to it, and at the end, fold back all the
     * carry bits from the top 16 bits into the lower 16 bits.
     */
    while ( nleft > 1 ) {
        sum += *w++;
        nleft -= 2;
    }
    /*
     * mop up an odd byte, if necessary
     */
    if ( 1 == nleft ) {
        *( unsigned char * )( &answer ) = *( unsigned char * )w ;
        sum += answer;
    }
    /*
     * add back carry outs from top 16 bits to low 16 bits
     * add hi 16 to low 16
     */
    sum = ( sum >> 16 ) + ( sum & 0xFFFF );
    /* add carry */
    sum += ( sum >> 16 );
    /* truncate to 16 bits */
    answer = ~sum;

    return (answer & 0xFFFFFFFF);
}

 

hashtest.c

view plaincopy to clipboardprint?
#include <stdio.h>  
#include <stdlib.h>  
#include <sys/types.h>  
#include <sys/stat.h>  
#include <fcntl.h>  
#include <errno.h>  
#include <string.h>  
#include "hash.h"  
#include "md5.h"  
 
struct hash_key {  
    unsigned char *key;  
    struct hash_key *next;   
};  
 
struct hash_counter_entry {  
    unsigned int hit_count;  
    unsigned int entry_count;  
    struct hash_key *keys;  
};  
 
#define BLOCK_LEN   4096  
 
static int backet_len = 10240;  
static int hash_call_count = 0;  
static struct hash_counter_entry *hlist = NULL;  
unsigned int (*hash_func)(char *str);  
 
void choose_hash_func(char *hash_func_name)  
{  
    if (0 == strcmp(hash_func_name, "simple_hash"))  
        hash_func = simple_hash;  
    else if (0 == strcmp(hash_func_name, "RS_hash"))  
        hash_func = RS_hash;  
    else if (0 == strcmp(hash_func_name, "JS_hash"))  
        hash_func = JS_hash;  
    else if (0 == strcmp(hash_func_name, "PJW_hash"))  
        hash_func = PJW_hash;  
    else if (0 == strcmp(hash_func_name, "ELF_hash"))  
        hash_func = ELF_hash;  
    else if (0 == strcmp(hash_func_name, "BKDR_hash"))  
        hash_func = BKDR_hash;  
    else if (0 == strcmp(hash_func_name, "SDBM_hash"))  
        hash_func = SDBM_hash;  
    else if (0 == strcmp(hash_func_name, "DJB_hash"))  
        hash_func = DJB_hash;  
    else if (0 == strcmp(hash_func_name, "AP_hash"))  
        hash_func = AP_hash;  
    else if (0 == strcmp(hash_func_name, "CRC_hash"))  
        hash_func = CRC_hash;  
    else 
        hash_func = NULL;  
}  
 
void insert_hash_entry(unsigned char *key, struct hash_counter_entry *hlist)  
{  
    unsigned int hash_value = hash_func(key) % backet_len;  
    struct hash_key *p;  
 
    p = hlist[hash_value].keys;  
    while(p) {  
        if (0 == strcmp(key, p->key))  
            break;  
        p = p->next;  
    }  
    if (p == NULL)  
    {  
        p = (struct hash_key *)malloc(sizeof(struct hash_key));  
        if (p == NULL)   
        {  
            perror("malloc in insert_hash_entry");  
            return;  
        }  
        p->key = strdup(key);  
        p->next = hlist[hash_value].keys;  
        hlist[hash_value].keys = p;  
        hlist[hash_value].entry_count++;  
    }  
    hlist[hash_value].hit_count++;  
}  
 
void hashtest_init()  
{  
    int i;  
      
    hash_call_count = 0;  
    hlist = (struct hash_counter_entry *) malloc (sizeof(struct hash_counter_entry) *  backet_len);  
    if (NULL == hlist)  
    {  
        perror("malloc in hashtest_init");  
        return;  
    }  
    for (i = 0; i < backet_len; i++)  
    {  
        hlist[i].hit_count = 0;  
        hlist[i].entry_count = 0;  
        hlist[i].keys = NULL;  
    }  
}  
 
void hashtest_clean()  
{  
    int i;  
    struct hash_key *pentry, *p;  
 
    if (NULL == hlist)  
        return;  
 
    for (i = 0; i < backet_len; ++i)  
    {  
        pentry = hlist[i].keys;  
        while(pentry)  
        {  
            p = pentry->next;  
            if (pentry->key) free(pentry->key);  
            free(pentry);  
            pentry = p;  
        }  
    }  
    free(hlist);  
}  
 
void show_hashtest_result()  
{  
    int i, backet = 0, max_link = 0, sum = 0;  
    int conflict_count = 0, hit_count = 0;  
    float avg_link, backet_usage;  
 
    for(i = 0; i < backet_len; i++)  
    {  
        if (hlist[i].hit_count > 0)   
        {  
            backet++;  
            sum += hlist[i].entry_count;  
            if (hlist[i].entry_count > max_link)  
            {  
                max_link = hlist[i].entry_count;  
            }  
            if (hlist[i].entry_count > 1)  
            {  
                conflict_count++;  
            }  
            hit_count += hlist[i].hit_count;  
        }  
    }  
 
    backet_usage = backet/1.0/backet_len * 100;;  
    avg_link = sum/1.0/backet;  
 
    printf("backet_len = %d/n", backet_len);  
    printf("hash_call_count = %d/n", hash_call_count);  
    printf("hit_count = %d/n", hit_count);  
    printf("conflict count = %d/n", conflict_count);  
    printf("longest hash entry = %d/n", max_link);  
    printf("average hash entry length = %.2f/n", avg_link);  
    printf("backet usage = %.2f%/n", backet_usage);  
}  
 
void usage()  
{  
    printf("Usage: hashtest filename hash_func_name [backet_len]/n");  
    printf("hash_func_name:/n");  
    printf("/tsimple_hash/n");  
    printf("/tRS_hash/n");  
    printf("/tJS_hash/n");  
    printf("/tPJW_hash/n");  
    printf("/tELF_hash/n");  
    printf("/tBKDR_hash/n");  
    printf("/tSDBM_hash/n");  
    printf("/tDJB_hash/n");  
    printf("/tAP_hash/n");  
    printf("/tCRC_hash/n");  
}  
 
void md5_to_32(unsigned char *md5_16, unsigned char *md5_32)  
{  
    int i;  
 
    for (i = 0; i < 16; ++i)  
    {  
        sprintf(md5_32 + i * 2, "%02x", md5_16[i]);  
    }  
}  
 
int main(int argc, char *argv[])  
{  
    int fd = -1, rwsize = 0;  
    unsigned char md5_checksum[16 + 1] = {0};  
    unsigned char buf[BLOCK_LEN] = {0};  
 
    if (argc < 3)   
    {  
        usage();  
        return -1;  
    }  
 
    if (-1 == (fd = open(argv[1], O_RDONLY)))  
    {  
        perror("open source file");  
        return errno;  
    }  
 
    if (argc == 4)  
    {  
        backet_len = atoi(argv[3]);  
    }  
 
    hashtest_init();  
    choose_hash_func(argv[2]);  
    while (rwsize = read(fd, buf, BLOCK_LEN))  
    {  
        md5(buf, rwsize, md5_checksum);  
        insert_hash_entry(md5_checksum, hlist);  
        hash_call_count++;  
        memset(buf, 0, BLOCK_LEN);  
        memset(md5_checksum, 0, 16 + 1);  
    }  
    close(fd);  
 
    show_hashtest_result();  
    hashtest_clean();  
    return 0;  




本文出处:http://blog.csdn.net/liuben/archive/2009/12/21/5050697.aspx

  • 1
    点赞
  • 1
    收藏
    觉得还不错? 一键收藏
  • 0
    评论
评论
添加红包

请填写红包祝福语或标题

红包个数最小为10个

红包金额最低5元

当前余额3.43前往充值 >
需支付:10.00
成就一亿技术人!
领取后你会自动成为博主和红包主的粉丝 规则
hope_wisdom
发出的红包
实付
使用余额支付
点击重新获取
扫码支付
钱包余额 0

抵扣说明:

1.余额是钱包充值的虚拟货币,按照1:1的比例进行支付金额的抵扣。
2.余额无法直接购买下载,可以购买VIP、付费专栏及课程。

余额充值