Fastdfs源码分析3----哈希管理器设计
记得在4年以前，在下刚刚读完linux 1.0内核源码，当时意气风发，无惧无畏。总感觉看任何c代码都没有问题，如履平地。甚至跟北京一家中科院的博士（某存储公司的cto，是我朋友）掷下豪言说：自己的c语言编程水准，敢于接受任何的挑战。
不料，看余庆大神的哈希表时，就没有完全看明白，虽然当时懂了大部分。
时光荏苒，中间做了一段是时间c后就开始用c++编程。越多了大量的开源库，c已经忘却（在写c++时会觉得c真的垃圾）。4年过去，曾经的少年已经没有那么冲动，多了一份对家庭对亲戚朋友的责任感。
技术准平又有了比较大的提升。也说出了，就算以为自己再牛逼，也不能无知的以为别人都是傻子。在下经过这几年的阅历，已经没有了当年孤高。欣然接受和允许别人比自己牛逼。
− − − − T a n g s h u n c a i \color{#222514}{-}\color{#222514}{-}\color{#222514}{-}\color{#222514}{-}\color{#4285f4}{T}\color{#ea4335}{a}\color{#fbbc05}{n}\color{#4285f4}{g}\color{#34a853}{s}\color{#ea4335}{h}\color{#4285f4}{u}\color{#fbbc05}{n}\color{#34a853}{c}\color{#ea4335}{a}\color{#fbbc05}{i} −−−−Tangshuncai
2020年，这个难忘的春节，这个漫长的假期。 做了三件事情：第一 ，完善一个去块链存储、第二、阅读EOS/bitcoin/ceph源码、第三c的开源项目源码。
本文的fastdfs是因为一次面试，一个公司说如何为IM服务器搭建文件存储。我就告诉他：用mysql存储文字(blob)，表情，存储图片和文件的元数据（doamin:ip:port:dir），图片用fastdfs存储，大文件用ceph（访问量没有那么大，gluster是条带）或gluster存储（访问量比较大时用ceph，因为是分块的）。 不过那个面试官可能是偷到了方案，或者误以为我很菜（我口语表达能力远远不及编程和写文档的能力），或者觉得要价太高，必有一个原因导致最终没有合作。
于是写了Fastdfs集群搭建，2天时间访问量破2000。有继续写了fastdfs的源码分析。本文即是第三篇文章。
注意：余庆的哈希管理器设计的非常牛逼（智能）。阅读需要比较强的c语言功底和逻辑思维能力。
按照我们的传统，直接上源码，自己拉下去读。（已经经过详细的注释）
/**
* Copyright (C) 2008 Happy Fish / YuQing
*
* FastDFS may be copied only under the terms of the GNU General
* Public License V3, which may be found in the FastDFS source kit.
* Please visit the FastDFS Home Page http://www.fastken.com/ for more detail.
**/
#ifndef _HASH_H_
#define _HASH_H_
#include
/**
* Copyright (C) 2008 Happy Fish / YuQing
*
* FastDFS may be copied only under the terms of the GNU General
* Public License V3, which may be found in the FastDFS source kit.
* Please visit the FastDFS Home Page http://www.fastken.com/ for more detail.
**/
#include 0) printf("%d: %d\n", i+1, stats[i]);
}
if (stats[i] > 0) printf(">=%d: %d\n", i+1, stats[i]);
*/
printf("capacity: %d, item_count=%d, bucket_used: %d, " \
"avg length: %.4f, max length: %d, bucket / item = %.2f%%\n",
hs.capacity, hs.item_count, hs.bucket_used,
hs.bucket_avg_length, hs.bucket_max_length,
(double)hs.bucket_used*100.00/(double)hs.capacity);
}
/* 哈希管理器（槽位）扩容/缩容。哈希节点重新映射到新的槽位。
返回值：0，成功； !=0失败，哈希管理器的内容不发生任何变化
*/
static int _rehash1(HashArray *pHash, const int old_capacity,
unsigned int *new_capacity)
{
HashData **old_buckets;
HashData **ppBucket;
HashData **bucket_end;
HashData *hash_data;
HashData *pNext;
int result;
old_buckets = pHash->buckets; // 保存老的槽位
pHash->capacity = new_capacity;
if ((result=_hash_alloc_buckets(pHash, old_capacity)) != 0)
{
pHash->buckets = old_buckets; // 如果申请新的槽位失败，重新指向老的槽位
return result;
}
//printf("old: %d, new: %d\n", old_capacity, *pHash->capacity);
// 将老槽位下非空链表上的哈希节点取下，重新挂载到新槽位
pHash->item_count = 0;
bucket_end = old_buckets + old_capacity;
for (ppBucket=old_buckets; ppBucket<bucket_end; ppBucket++)
{
if (*ppBucket == NULL)
{
continue;
}
hash_data = *ppBucket;
while (hash_data != NULL)
{
pNext = hash_data->next; // 从老槽位取下哈希节点，直接插入新槽位
ADD_TO_BUCKET(pHash, (pHash->buckets + \
(HASH_CODE(pHash, hash_data) % \
(*pHash->capacity))), hash_data)
hash_data = pNext;
}
}
free(old_buckets); // 释放老槽位
return 0;
}
// 哈希管理器扩容到下一个阶梯容量，
static int _rehash(HashArray *pHash)
{
int result;
unsigned int *pOldCapacity;
pOldCapacity = pHash->capacity;
if (pHash->is_malloc_capacity)
{
pHash->capacity = hash_get_prime_capacity(*pOldCapacity);
}
else
{
pHash->capacity++;
}
if ((result=_rehash1(pHash, *pOldCapacity, pHash->capacity)) != 0)
{
pHash->capacity = pOldCapacity; //rollback
}
else
{
if (pHash->is_malloc_capacity)
{
free(pOldCapacity);
pHash->is_malloc_capacity = false;
}
}
/*printf("rehash, old_capacity=%d, new_capacity=%d\n", \
old_capacity, *pHash->capacity);
*/
return result;
}
/* 统计哈希冲突的数量
（槽位上面某条链表的节点的->哈希值不完全一致，即视为该槽位发生了哈希冲突（该槽位临时挂载了本不该映射到该槽位的哈希节点））
*/
static int _hash_conflict_count(HashArray *pHash)
{
HashData **ppBucket;
HashData **bucket_end;
HashData *hash_data;
HashData *pNext;
int conflicted;
int conflict_count;
// 遍历哈希管理器的所有槽位、遍历非NULL槽位下面的节点、对比发现任意一个节点的哈希值与其它节点不一致即视为一条哈希冲突
bucket_end = pHash->buckets + (*pHash->capacity);
conflict_count = 0;
for (ppBucket=pHash->buckets; ppBucket<bucket_end; ppBucket++)
{
if (*ppBucket == NULL || (*ppBucket)->next == NULL)
{
continue;
}
conflicted = 0;
hash_data = *ppBucket;
while (hash_data != NULL)
{
pNext = hash_data->next;
while (pNext != NULL)
{
if (HASH_CODE(pHash, hash_data) != \
HASH_CODE(pHash, pNext))
{
conflicted = 1;
break;
}
pNext = pNext->next;
}
if (conflicted)
{
break;
}
hash_data = hash_data->next;
}
conflict_count += conflicted;
}
return conflict_count;
}
/* 如果发生哈希冲突，用建议的容量尝试解决哈希冲突，如果未解决 则用在建议容量的基础上扩容，并用阶梯容量确定最终真实容量，直到解决哈希冲突 */
int hash_best_op(HashArray *pHash, const int suggest_capacity)
{
int old_capacity;
int conflict_count;
unsigned int *new_capacity;
int result;
// 如果未发生任何哈希冲突，则返回0
if ((conflict_count=_hash_conflict_count(pHash)) == 0)
{
return 0;
}
// 发生哈希冲突则申请变量用于存储新的哈希节点容量，并设置未建议容量
old_capacity = *pHash->capacity;
new_capacity = (unsigned int *)malloc(sizeof(unsigned int));
if (new_capacity == NULL)
{
return -ENOMEM;
}
if ((suggest_capacity > 2) && (suggest_capacity >= pHash->item_count))
{
*new_capacity = suggest_capacity - 2;
if (*new_capacity % 2 == 0)
{
++(*new_capacity);
}
}
else
{
*new_capacity = 2 * (pHash->item_count - 1) + 1;
}
// 用建议容量重新映射哈希( _rehash1() )，直到没有发生任何哈希冲突为止。退出循环
do
{
do
{
*new_capacity += 2;
} while ((*new_capacity % 3 == 0) || (*new_capacity % 5 == 0) \
|| (*new_capacity % 7 == 0));
if ((result=_rehash1(pHash, old_capacity, new_capacity)) != 0)
{
pHash->is_malloc_capacity = \
(pHash->capacity == new_capacity);
*pHash->capacity = old_capacity;
return -1 * result;
}
old_capacity = *new_capacity;
/*printf("rehash, conflict_count=%d, old_capacity=%d, " \
"new_capacity=%d\n", conflict_count, \
old_capacity, *new_capacity);
*/
} while ((conflict_count=_hash_conflict_count(pHash)) > 0);
pHash->is_malloc_capacity = true;
//hash_stat_print(pHash);
return 1;
}
// 查找哈希值相同(key相同)的链表的入口节点的指针
static HashData *_chain_find_entry(HashData **ppBucket, const void *key,
const int key_len, const unsigned int hash_code)
{
HashData *hash_data;
hash_data = *ppBucket;
while (hash_data != NULL)
{
if (key_len == hash_data->key_len && \
memcmp(key, hash_data->key, key_len) == 0)
{
return hash_data;
}
hash_data = hash_data->next;
}
return NULL;
}
// 带互斥锁保障的 查找哈希值入口节点
HashData *hash_find_ex(HashArray *pHash, const void *key, const int key_len)
{
unsigned int hash_code;
HashData **ppBucket;
HashData *hash_data;
hash_code = pHash->hash_func(key, key_len);
ppBucket = pHash->buckets + (hash_code % (*pHash->capacity));
HASH_LOCK(pHash, ppBucket - pHash->buckets)
hash_data = _chain_find_entry(ppBucket, key, key_len, hash_code);
HASH_UNLOCK(pHash, ppBucket - pHash->buckets)
return hash_data;
}
// 在哈希管理器中查找入口哈希节点的存储值的地址
void *hash_find(HashArray *pHash, const void *key, const int key_len)
{
unsigned int hash_code;
HashData **ppBucket;
HashData *hash_data;
hash_code = pHash->hash_func(key, key_len);
ppBucket = pHash->buckets + (hash_code % (*pHash->capacity));
HASH_LOCK(pHash, ppBucket - pHash->buckets)
hash_data = _chain_find_entry(ppBucket, key, key_len, hash_code);
HASH_UNLOCK(pHash, ppBucket - pHash->buckets)
if (hash_data != NULL)
{
return hash_data->value;
}
else
{
return NULL;
}
}
// 在哈希管理器中查找入口哈希节点的存储值，返回值的地址和值的长度
int hash_find2(HashArray *pHash, const string_t *key, string_t *value)
{
HashData *hdata;
if ((hdata=hash_find1_ex(pHash, key)) == NULL)
{
return ENOENT;
}
value->str = hdata->value;
value->len = hdata->value_len;
return 0;
}
// 带互斥锁保障的 查找哈希值入口节点
HashData *hash_find1_ex(HashArray *pHash, const string_t *key)
{
return hash_find_ex(pHash, key->str, key->len);
}
// 获取符合条件的（符合哈希值）哈希的值，写入value[value_len]中
int hash_get(HashArray *pHash, const void *key, const int key_len,
void *value, int *value_len)
{
unsigned int hash_code;
int result;
HashData **ppBucket;
HashData *hash_data;
hash_code = pHash->hash_func(key, key_len);
ppBucket = pHash->buckets + (hash_code % (*pHash->capacity));
HASH_LOCK(pHash, ppBucket - pHash->buckets)
hash_data = _chain_find_entry(ppBucket, key, key_len, hash_code);
if (hash_data != NULL)
{
if (hash_data->value_len <= *value_len)
{
*value_len = hash_data->value_len;
memcpy(value, hash_data->value, hash_data->value_len);
result = 0;
}
else
{
result = ENOSPC;
}
}
else
{
result = ENOENT;
}
HASH_UNLOCK(pHash, ppBucket - pHash->buckets)
return result;
}
// 插入一个哈希节点（存在则覆盖值）到哈希管理器
int hash_insert_ex(HashArray *pHash, const void *key, const int key_len,
void *value, const int value_len, const bool needLock)
{
unsigned int hash_code;
HashData **ppBucket;
HashData *hash_data;
HashData *previous;
char *pBuff;
int bytes;
int malloc_value_size;
hash_code = pHash->hash_func(key, key_len);
ppBucket = pHash->buckets + (hash_code % (*pHash->capacity));
previous = NULL;
if (needLock)
{
HASH_LOCK(pHash, ppBucket - pHash->buckets)
}
// 冲突哈希链表，由两部分组成：A类哈希链表(头插入)(A是临时挂靠) + B类哈希链表(尾插)(B是本家)。中间有一条分界线，不存在混合插入
hash_data = *ppBucket;
while (hash_data != NULL)
{
if (key_len == hash_data->key_len && \
memcmp(key, hash_data->key, key_len) == 0)
{
break;
}
previous = hash_data;
hash_data = hash_data->next;
}
if (hash_data != NULL) //exists 此时previous指向A类哈希链表的尾，hash_data指向B类哈希链表的头，本家节点已经存在（覆盖值即可）
{
if (!pHash->is_malloc_value)
{
hash_data->value_len = value_len;
hash_data->value = (char *)value;
if (needLock)
{
HASH_UNLOCK(pHash, ppBucket - pHash->buckets)
}
return 0;
}
else
{
if (hash_data->malloc_value_size >= value_len && \
(hash_data->malloc_value_size <= 128 ||
hash_data->malloc_value_size / 2 < value_len))
{
hash_data->value_len = value_len;
memcpy(hash_data->value, value, value_len);
if (needLock)
{
HASH_UNLOCK(pHash, ppBucket - pHash->buckets)
}
return 0;
}
DELETE_FROM_BUCKET(pHash, ppBucket, previous, hash_data)
}
}
if (needLock)
{
HASH_UNLOCK(pHash, ppBucket - pHash->buckets)
}
if (!pHash->is_malloc_value)
{
malloc_value_size = 0;
}
else
{
malloc_value_size = MEM_ALIGN(value_len);
}
// 本家节点不存在，则new一个节点挂载在B链表（本家链表）尾巴
bytes = CALC_NODE_MALLOC_BYTES(key_len, malloc_value_size);
if (pHash->max_bytes > 0 && pHash->bytes_used+bytes > pHash->max_bytes)
{
return -ENOSPC;
}
pBuff = (char *)malloc(bytes);
if (pBuff == NULL)
{
return -ENOMEM;
}
pHash->bytes_used += bytes;
hash_data = (HashData *)pBuff;
hash_data->malloc_value_size = malloc_value_size;
hash_data->key_len = key_len;
memcpy(hash_data->key, key, key_len);
#ifdef HASH_STORE_HASH_CODE
hash_data->hash_code = hash_code;
#endif
hash_data->value_len = value_len;
if (!pHash->is_malloc_value)
{
hash_data->value = (char *)value;
}
else
{
hash_data->value = hash_data->key + hash_data->key_len;
memcpy(hash_data->value, value, value_len);
}
if (needLock)
{
HASH_LOCK(pHash, ppBucket - pHash->buckets)
ADD_TO_BUCKET(pHash, ppBucket, hash_data)
HASH_UNLOCK(pHash, ppBucket - pHash->buckets)
}
else
{
ADD_TO_BUCKET(pHash, ppBucket, hash_data)
}
if (pHash->load_factor >= 0.10 && (double)pHash->item_count /
(double)*pHash->capacity >= pHash->load_factor)
{
_rehash(pHash);
}
return 1;
}
/*
如果old_data哈希不为NULL，则返回哈希节点content（本次是只读）值的基础上加inc后的值，并记录在输出参数中
如果old_data是NULL，则没有基础，直接打印inc的值，并记录在输出参数中
这个函数设计的其实有点僵硬，可能是为了格式统一。
*/
int64_t hash_inc_value(const HashData *old_data, const int inc,
char *new_value, int *new_value_len, void *arg)
{
int64_t n;
if (old_data != NULL)
{
if (old_data->value_len < *new_value_len)
{
memcpy(new_value, old_data->value, old_data->value_len);
new_value[old_data->value_len] = '\0';
n = strtoll(new_value, NULL, 10);
n += inc;
}
else
{
n = inc;
}
*new_value_len = sprintf(new_value, "%"PRId64, n);
}
else
{
n = inc;
*new_value_len = sprintf(new_value, "%"PRId64, n);
}
return n;
}
// 插入一个哈希节点（存在则覆盖值）到哈希管理器。convert_func会根据arg参数去改变哈希节点的属性（显然，可以改变的属性只有key）
int hash_inc_ex(HashArray *pHash, const void *key, const int key_len,
const int inc, char *value, int *value_len,
ConvertValueFunc convert_func, void *arg)
{
unsigned int hash_code;
int result;
HashData **ppBucket;
HashData *hash_data;
hash_code = pHash->hash_func(key, key_len);
ppBucket = pHash->buckets + (hash_code % (*pHash->capacity));
HASH_LOCK(pHash, ppBucket - pHash->buckets)
hash_data = _chain_find_entry(ppBucket, key, key_len, hash_code);
convert_func(hash_data, inc, value, value_len, arg);
if (hash_data != NULL)
{
if (!pHash->is_malloc_value)
{
hash_data->value_len = *value_len;
hash_data->value = (char *)value;
HASH_UNLOCK(pHash, ppBucket - pHash->buckets)
return 0;
}
else
{
if (hash_data->malloc_value_size >= *value_len)
{
hash_data->value_len = *value_len;
memcpy(hash_data->value, value, *value_len);
HASH_UNLOCK(pHash, ppBucket - pHash->buckets)
return 0;
}
}
}
result = hash_insert_ex(pHash, key, key_len, value, *value_len, false);
if (result < 0)
{
*value = '\0';
*value_len = 0;
result *= -1;
}
else
{
result = 0;
}
HASH_UNLOCK(pHash, ppBucket - pHash->buckets)
return result;
}
// 设置目标哈希节点的部分/全部value
int hash_partial_set(HashArray *pHash, const void *key, const int key_len,
const char *value, const int offset, const int value_len)
{
unsigned int hash_code;
int result;
HashData **ppBucket;
HashData *hash_data;
char *pNewBuff;
hash_code = pHash->hash_func(key, key_len);
ppBucket = pHash->buckets + (hash_code % (*pHash->capacity));
HASH_LOCK(pHash, ppBucket - pHash->buckets)
hash_data = _chain_find_entry(ppBucket, key, key_len, hash_code);
do
{
if (hash_data != NULL)
{
if (offset < 0 || offset >= hash_data->value_len)
{
result = EINVAL;
break;
}
// 目标哈希节点存在，且原有value的空间足以承载新写入的数据，则覆盖并跳出到函数末尾进行解锁处理
if (offset + value_len <= hash_data->value_len)
{
memcpy(hash_data->value+offset, value, value_len);
result = 0;
break;
}
// 原因目标不足以承载新写入的节点，则new一个新空间存在新写入数据，并插入槽位（value全覆盖）
pNewBuff = (char *)malloc(offset + value_len);
if (pNewBuff == NULL)
{
result = errno != 0 ? errno : ENOMEM;
break;
}
if (offset > 0)
{
memcpy(pNewBuff, hash_data->value, offset);
}
memcpy(pNewBuff + offset, value, value_len);
result = hash_insert_ex(pHash, key, key_len, pNewBuff,
offset + value_len, false);
free(pNewBuff);
}
else
{
if (offset != 0)
{
result = ENOENT;
break;
}
// 目标哈希节点不存在，则设置全值插入槽位
result = hash_insert_ex(pHash, key, key_len, (void *)value,
value_len, false);
}
if (result < 0)
{
result *= -1;
}
else
{
result = 0;
}
} while (0);
HASH_UNLOCK(pHash, ppBucket - pHash->buckets)
return result;
}
/* 删除目标哈希链表。注意不会删除临时挂载的那一截链表 */
int hash_delete(HashArray *pHash, const void *key, const int key_len)
{
HashData **ppBucket;
HashData *hash_data;
HashData *previous;
unsigned int hash_code;
int result;
hash_code = pHash->hash_func(key, key_len);
ppBucket = pHash->buckets + (hash_code % (*pHash->capacity));
result = ENOENT;
previous = NULL;
HASH_LOCK(pHash, ppBucket - pHash->buckets)
hash_data = *ppBucket;
while (hash_data != NULL)
{
if (key_len == hash_data->key_len && \
memcmp(key, hash_data->key, key_len) == 0)
{
DELETE_FROM_BUCKET(pHash, ppBucket, previous, hash_data)
result = 0;
break;
}
previous = hash_data;
hash_data = hash_data->next;
}
HASH_UNLOCK(pHash, ppBucket - pHash->buckets)
return result;
}
// 对全部哈希节点（从第一个槽位上面链表开始遍历，逐步走完全部节点）进行某种操作（例如读，写，打印）
int hash_walk(HashArray *pHash, HashWalkFunc walkFunc, void *args)
{
HashData **ppBucket;
HashData **bucket_end;
HashData *hash_data;
int index;
int result;
index = 0;
bucket_end = pHash->buckets + (*pHash->capacity);
for (ppBucket=pHash->buckets; ppBucket<bucket_end; ppBucket++)
{
hash_data = *ppBucket;
while (hash_data != NULL)
{
result = walkFunc(index, hash_data, args);
if (result != 0)
{
return result;
}
index++;
hash_data = hash_data->next;
}
}
return 0;
}
// 返回哈希节点总数量
int hash_count(HashArray *pHash)
{
return pHash->item_count;
}
// 如果哈希管理器在创建时，有槽位互斥锁，则上锁对应的槽位
int hash_bucket_lock(HashArray *pHash, const unsigned int bucket_index)
{
if (pHash->lock_count <= 0)
{
return 0;
}
return pthread_mutex_lock(pHash->locks + bucket_index %
pHash->lock_count);
}
// 如果哈希管理器在创建时，有槽位互斥锁，则解锁对应的槽位
int hash_bucket_unlock(HashArray *pHash, const unsigned int bucket_index)
{
if (pHash->lock_count <= 0)
{
return 0;
}
return pthread_mutex_unlock(pHash->locks + bucket_index %
pHash->lock_count);
}
// RS Hash Function. RS散列函数
int RSHash(const void *key, const int key_len)
{
unsigned char *pKey;
unsigned char *pEnd;
int a = 63689;
int hash = 0;
pEnd = (unsigned char *)key + key_len;
for (pKey = (unsigned char *)key; pKey != pEnd; pKey++)
{
hash = hash * a + (*pKey);
a *= 378551;
}
return hash;
}
// JS散列函数宏实现
#define JS_HASH_FUNC(init_value) \
unsigned char *pKey; \
unsigned char *pEnd; \
int hash; \
\
hash = init_value; \
pEnd = (unsigned char *)key + key_len; \
for (pKey = (unsigned char *)key; pKey != pEnd; pKey++) \
{ \
hash ^= ((hash << 5) + (*pKey) + (hash >> 2)); \
} \
\
return hash; \
// JS Hash Function. JS散列函数api
int JSHash(const void *key, const int key_len)
{
JS_HASH_FUNC(1315423911)
}
// JS Hash Function. JS散列函数api
int JSHash_ex(const void *key, const int key_len,
const int init_value)
{
JS_HASH_FUNC(init_value)
}
#define BITS_IN_UNIGNED_INT (int)(sizeof(int) * 8)
#define THREE_QUARTERS (int)((BITS_IN_UNIGNED_INT * 3) / 4)
#define HASH_ONE_EIGHTH (int)(BITS_IN_UNIGNED_INT / 8)
#define HASH_HIGH_BITS (int)((unsigned int)(0xFFFFFFFF) << \
(BITS_IN_UNIGNED_INT - HASH_ONE_EIGHTH))
// PJW散列函数
#define PJW_HASH_FUNC(init_value) \
unsigned char *pKey; \
unsigned char *pEnd; \
int hash; \
int test; \
\
hash = init_value; \
pEnd = (unsigned char *)key + key_len; \
for (pKey = (unsigned char *)key; pKey != pEnd; pKey++) \
{ \
hash = (hash << HASH_ONE_EIGHTH) + (*(pKey)); \
if ((test = hash & HASH_HIGH_BITS) != 0) \
{ \
hash = ((hash ^ (test >> THREE_QUARTERS)) & (~HASH_HIGH_BITS)); \
} \
} \
\
return hash; \
// P.J.Weinberger Hash Function, same as ELF Hash
int PJWHash(const void *key, const int key_len)
{
PJW_HASH_FUNC(0)
}
int PJWHash_ex(const void *key, const int key_len,
const int init_value)
{
PJW_HASH_FUNC(init_value)
}
#define ELF_HASH_FUNC(init_value) \
unsigned char *pKey; \
unsigned char *pEnd; \
int hash; \
int x; \
\
hash = init_value; \
pEnd = (unsigned char *)key + key_len; \
for (pKey = (unsigned char *)key; pKey != pEnd; pKey++) \
{ \
hash = (hash << 4) + (*pKey); \
if ((x = hash & 0xF0000000) != 0) \
{ \
hash ^= (x >> 24); \
hash &= ~x; \
} \
} \
\
return hash; \
// ELF Hash Function, same as PJW Hash. ELF散列函数api
int ELFHash(const void *key, const int key_len)
{
ELF_HASH_FUNC(0)
}
// ELF散列函数api
int ELFHash_ex(const void *key, const int key_len,
const int init_value)
{
ELF_HASH_FUNC(init_value)
}
#define BKDR_HASH_FUNC(init_value) \
unsigned char *pKey; \
unsigned char *pEnd; \
int seed = 131; /* 31 131 1313 13131 131313 etc..*/ \
int hash; \
\
hash = init_value; \
pEnd = (unsigned char *)key + key_len; \
for (pKey = (unsigned char *)key; pKey != pEnd; pKey++) \
{ \
hash = hash * seed + (*pKey); \
} \
\
return hash; \
// BKDR Hash Function.
int BKDRHash(const void *key, const int key_len)
{
BKDR_HASH_FUNC(0)
}
int BKDRHash_ex(const void *key, const int key_len,
const int init_value)
{
BKDR_HASH_FUNC(init_value)
}
#define SDBM_HASH_FUNC(init_value) \
unsigned char *pKey; \
unsigned char *pEnd; \
int hash; \
\
hash = init_value; \
pEnd = (unsigned char *)key + key_len; \
for (pKey = (unsigned char *)key; pKey != pEnd; pKey++) \
{ \
hash = (*pKey) + (hash << 6) + (hash << 16) - hash; \
} \
\
return hash; \
// SDBM Hash Function
int SDBMHash(const void *key, const int key_len)
{
SDBM_HASH_FUNC(0)
}
int SDBMHash_ex(const void *key, const int key_len,
const int init_value)
{
SDBM_HASH_FUNC(init_value)
}
#define TIME33_HASH_FUNC(init_value) \
int nHash; \
unsigned char *pKey; \
unsigned char *pEnd; \
\
nHash = init_value; \
pEnd = (unsigned char *)key + key_len; \
for (pKey = (unsigned char *)key; pKey != pEnd; pKey++) \
{ \
nHash += (nHash << 5) + (*pKey); \
} \
\
return nHash; \
int Time33Hash(const void *key, const int key_len)
{
TIME33_HASH_FUNC(0)
}
int Time33Hash_ex(const void *key, const int key_len,
const int init_value)
{
TIME33_HASH_FUNC(init_value)
}
#define DJB_HASH_FUNC(init_value) \
unsigned char *pKey; \
unsigned char *pEnd; \
int hash; \
\
hash = init_value; \
pEnd = (unsigned char *)key + key_len; \
for (pKey = (unsigned char *)key; pKey != pEnd; pKey++) \
{ \
hash += (hash << 5) + (*pKey); \
} \
\
return hash; \
// DJB Hash Function
int DJBHash(const void *key, const int key_len)
{
DJB_HASH_FUNC(5381)
}
int DJBHash_ex(const void *key, const int key_len,
const int init_value)
{
DJB_HASH_FUNC(init_value)
}
#define AP_HASH_FUNC(init_value) \
unsigned char *pKey; \
unsigned char *pEnd; \
int i; \
int hash; \
\
hash = init_value; \
\
pEnd = (unsigned char *)key + key_len; \
for (pKey = (unsigned char *)key, i=0; pKey != pEnd; pKey++, i++) \
{ \
if ((i & 1) == 0) \
{ \
hash ^= ((hash << 7) ^ (*pKey) ^ (hash >> 3)); \
} \
else \
{ \
hash ^= (~((hash << 11) ^ (*pKey) ^ (hash >> 5))); \
} \
} \
\
return hash; \
// AP Hash Function
int APHash(const void *key, const int key_len)
{
AP_HASH_FUNC(0)
}
int APHash_ex(const void *key, const int key_len,
const int init_value)
{
AP_HASH_FUNC(init_value)
}
int calc_hashnr (const void* key, const int key_len)
{
unsigned char *pKey;
unsigned char *pEnd;
int nr = 1, nr2 = 4;
pEnd = (unsigned char *)key + key_len;
for (pKey = (unsigned char *)key; pKey != pEnd; pKey++)
{
nr ^= (((nr & 63) + nr2) * (*pKey)) + (nr << 8);
nr2 += 3;
}
return nr;
}
#define CALC_HASHNR1_FUNC(init_value) \
unsigned char *pKey; \
unsigned char *pEnd; \
int hash; \
\
hash = init_value; \
pEnd = (unsigned char *)key + key_len; \
for (pKey = (unsigned char *)key; pKey != pEnd; pKey++) \
{ \
hash *= 16777619; \
hash ^= *pKey; \
} \
return hash; \
int calc_hashnr1(const void* key, const int key_len)
{
CALC_HASHNR1_FUNC(0)
}
int calc_hashnr1_ex(const void* key, const int key_len,
const int init_value)
{
CALC_HASHNR1_FUNC(init_value)
}
#define SIMPLE_HASH_FUNC(init_value) \
int h; \
unsigned char *p; \
unsigned char *pEnd; \
\
h = init_value; \
pEnd = (unsigned char *)key + key_len; \
for (p = (unsigned char *)key; p!= pEnd; p++) \
{ \
h = 31 * h + *p; \
} \
\
return h; \
int simple_hash(const void* key, const int key_len)
{
SIMPLE_HASH_FUNC(0)
}
int simple_hash_ex(const void* key, const int key_len,
const int init_value)
{
SIMPLE_HASH_FUNC(init_value)
}
static unsigned int crc_table[256] = {
0x00000000, 0x77073096, 0xEE0E612C, 0x990951BA,
0x076DC419, 0x706AF48F, 0xE963A535, 0x9E6495A3,
0x0EDB8832, 0x79DCB8A4, 0xE0D5E91E, 0x97D2D988,
0x09B64C2B, 0x7EB17CBD, 0xE7B82D07, 0x90BF1D91,
0x1DB71064, 0x6AB020F2, 0xF3B97148, 0x84BE41DE,
0x1ADAD47D, 0x6DDDE4EB, 0xF4D4B551, 0x83D385C7,
0x136C9856, 0x646BA8C0, 0xFD62F97A, 0x8A65C9EC,
0x14015C4F, 0x63066CD9, 0xFA0F3D63, 0x8D080DF5,
0x3B6E20C8, 0x4C69105E, 0xD56041E4, 0xA2677172,
0x3C03E4D1, 0x4B04D447, 0xD20D85FD, 0xA50AB56B,
0x35B5A8FA, 0x42B2986C, 0xDBBBC9D6, 0xACBCF940,
0x32D86CE3, 0x45DF5C75, 0xDCD60DCF, 0xABD13D59,
0x26D930AC, 0x51DE003A, 0xC8D75180, 0xBFD06116,
0x21B4F4B5, 0x56B3C423, 0xCFBA9599, 0xB8BDA50F,
0x2802B89E, 0x5F058808, 0xC60CD9B2, 0xB10BE924,
0x2F6F7C87, 0x58684C11, 0xC1611DAB, 0xB6662D3D,
0x76DC4190, 0x01DB7106, 0x98D220BC, 0xEFD5102A,
0x71B18589, 0x06B6B51F, 0x9FBFE4A5, 0xE8B8D433,
0x7807C9A2, 0x0F00F934, 0x9609A88E, 0xE10E9818,
0x7F6A0DBB, 0x086D3D2D, 0x91646C97, 0xE6635C01,
0x6B6B51F4, 0x1C6C6162, 0x856530D8, 0xF262004E,
0x6C0695ED, 0x1B01A57B, 0x8208F4C1, 0xF50FC457,
0x65B0D9C6, 0x12B7E950, 0x8BBEB8EA, 0xFCB9887C,
0x62DD1DDF, 0x15DA2D49, 0x8CD37CF3, 0xFBD44C65,
0x4DB26158, 0x3AB551CE, 0xA3BC0074, 0xD4BB30E2,
0x4ADFA541, 0x3DD895D7, 0xA4D1C46D, 0xD3D6F4FB,
0x4369E96A, 0x346ED9FC, 0xAD678846, 0xDA60B8D0,
0x44042D73, 0x33031DE5, 0xAA0A4C5F, 0xDD0D7CC9,
0x5005713C, 0x270241AA, 0xBE0B1010, 0xC90C2086,
0x5768B525, 0x206F85B3, 0xB966D409, 0xCE61E49F,
0x5EDEF90E, 0x29D9C998, 0xB0D09822, 0xC7D7A8B4,
0x59B33D17, 0x2EB40D81, 0xB7BD5C3B, 0xC0BA6CAD,
0xEDB88320, 0x9ABFB3B6, 0x03B6E20C, 0x74B1D29A,
0xEAD54739, 0x9DD277AF, 0x04DB2615, 0x73DC1683,
0xE3630B12, 0x94643B84, 0x0D6D6A3E, 0x7A6A5AA8,
0xE40ECF0B, 0x9309FF9D, 0x0A00AE27, 0x7D079EB1,
0xF00F9344, 0x8708A3D2, 0x1E01F268, 0x6906C2FE,
0xF762575D, 0x806567CB, 0x196C3671, 0x6E6B06E7,
0xFED41B76, 0x89D32BE0, 0x10DA7A5A, 0x67DD4ACC,
0xF9B9DF6F, 0x8EBEEFF9, 0x17B7BE43, 0x60B08ED5,
0xD6D6A3E8, 0xA1D1937E, 0x38D8C2C4, 0x4FDFF252,
0xD1BB67F1, 0xA6BC5767, 0x3FB506DD, 0x48B2364B,
0xD80D2BDA, 0xAF0A1B4C, 0x36034AF6, 0x41047A60,
0xDF60EFC3, 0xA867DF55, 0x316E8EEF, 0x4669BE79,
0xCB61B38C, 0xBC66831A, 0x256FD2A0, 0x5268E236,
0xCC0C7795, 0xBB0B4703, 0x220216B9, 0x5505262F,
0xC5BA3BBE, 0xB2BD0B28, 0x2BB45A92, 0x5CB36A04,
0xC2D7FFA7, 0xB5D0CF31, 0x2CD99E8B, 0x5BDEAE1D,
0x9B64C2B0, 0xEC63F226, 0x756AA39C, 0x026D930A,
0x9C0906A9, 0xEB0E363F, 0x72076785, 0x05005713,
0x95BF4A82, 0xE2B87A14, 0x7BB12BAE, 0x0CB61B38,
0x92D28E9B, 0xE5D5BE0D, 0x7CDCEFB7, 0x0BDBDF21,
0x86D3D2D4, 0xF1D4E242, 0x68DDB3F8, 0x1FDA836E,
0x81BE16CD, 0xF6B9265B, 0x6FB077E1, 0x18B74777,
0x88085AE6, 0xFF0F6A70, 0x66063BCA, 0x11010B5C,
0x8F659EFF, 0xF862AE69, 0x616BFFD3, 0x166CCF45,
0xA00AE278, 0xD70DD2EE, 0x4E048354, 0x3903B3C2,
0xA7672661, 0xD06016F7, 0x4969474D, 0x3E6E77DB,
0xAED16A4A, 0xD9D65ADC, 0x40DF0B66, 0x37D83BF0,
0xA9BCAE53, 0xDEBB9EC5, 0x47B2CF7F, 0x30B5FFE9,
0xBDBDF21C, 0xCABAC28A, 0x53B39330, 0x24B4A3A6,
0xBAD03605, 0xCDD70693, 0x54DE5729, 0x23D967BF,
0xB3667A2E, 0xC4614AB8, 0x5D681B02, 0x2A6F2B94,
0xB40BBE37, 0xC30C8EA1, 0x5A05DF1B, 0x2D02EF8D
};
#define CRC32_BODY(init_value) \
unsigned char *pKey; \
unsigned char *pEnd; \
int64_t crc; \
\
crc = init_value; \
pEnd = (unsigned char *)key + key_len; \
for (pKey = (unsigned char *)key; pKey != pEnd; pKey++) \
{ \
crc = crc_table[(crc ^ *pKey) & 0xFF] ^ (crc >> 8); \
} \
// crc32 api
int CRC32(const void *key, const int key_len)
{
CRC32_BODY(CRC32_XINIT)
return (int)(crc ^ CRC32_XOROT);
}
// crc32 api
int64_t CRC32_ex(const void *key, const int key_len,
const int64_t init_value)
{
CRC32_BODY(init_value)
return crc;
}