常用的Hash算法

1、RSHash

unsigned int RSHash(const std::string& str)

{

   unsigned int b    = 378551;

   unsigned int a    = 63689;

   unsigned int hash = 0;

 

   for(std::size_t i = 0; i < str.length(); i++)

   {

      hash = hash * a + str[i];

      a    = a * b;

   }

 

   return hash;

}

 

2、JSHash

unsigned int JSHash(const std::string& str)

{

   unsigned int hash = 1315423911;

   for(std::size_t i = 0; i < str.length(); i++)

   {

      hash ^= ((hash << 5) + str[i] + (hash >> 2));

   }

   return hash;

}

 

3、PJWHash

unsigned int PJWHash(const std::string& str)

{

   unsigned int BitsInUnsignedInt = (unsigned int)(sizeof(unsigned int) * 8);

   unsigned int ThreeQuarters     = (unsigned int)((BitsInUnsignedInt  * 3) / 4);

   unsigned int OneEighth         = (unsigned int)(BitsInUnsignedInt / 8);

   unsigned int HighBits          = (unsigned int)(0xFFFFFFFF) << (BitsInUnsignedInt - OneEighth);

   unsigned int hash              = 0;

   unsigned int test              = 0;

 

   for(std::size_t i = 0; i < str.length(); i++)

   {

      hash = (hash << OneEighth) + str[i];

 

      if((test = hash & HighBits)  != 0)

      {

         hash = (( hash ^ (test >> ThreeQuarters)) & (~HighBits));

      }

   }

   return hash;

}

 

4、ELFHash

unsigned int ELFHash(const std::string& str)

{

   unsigned int hash = 0;

   unsigned int x    = 0;

 

   for(std::size_t i = 0; i < str.length(); i++)

   {

      hash = (hash << 4) + str[i];

      if((x = hash & 0xF0000000L) != 0)

      {

         hash ^= (x >> 24);

      }

      hash &= ~x;

   }

 

   return hash;

}

 

5、BKDRHash

unsigned int BKDRHash(const std::string& str)

{

   unsigned int seed = 131; // 31 131 1313 13131 131313 etc..

   unsigned int hash = 0;

 

   for(std::size_t i = 0; i < str.length(); i++)

   {

      hash = (hash * seed) + str[i];

   }

   return hash;

}

 

6、SDBMHash

unsigned int SDBMHash(const std::string& str)

{

   unsigned int hash = 0;

 

   for(std::size_t i = 0; i < str.length(); i++)

   {

      hash = str[i] + (hash << 6) + (hash << 16) - hash;

   }

 

   return hash;

}

 

7、DJBHash(times33)-这个用得非常多，很多库都用它。

unsigned int DJBHash(const std::string& str)

{

   unsigned int hash = 5381;

 

   for(std::size_t i = 0; i < str.length(); i++)

   {

      hash = ((hash << 5) + hash) + str[i];

   }

   return hash;

}

 

8、DEKHash

unsigned int DEKHash(const std::string& str)

{

   unsigned int hash = static_cast(str.length());

 

   for(std::size_t i = 0; i < str.length(); i++)

   {

      hash = ((hash << 5) ^ (hash >> 27)) ^ str[i];

   }

 

   return hash;

}

 

9、BPHash

unsigned int BPHash(const std::string& str)

{

   unsigned int hash = 0;

   for(std::size_t i = 0; i < str.length(); i++)

   {

      hash = hash << 7 ^ str[i];

   }

 

   return hash;

}

 

10、FNVHash

unsigned int FNVHash(const std::string& str)

{

   const unsigned int fnv_prime = 0x811C9DC5;

   unsigned int hash = 0;

   for(std::size_t i = 0; i < str.length(); i++)

   {

      hash *= fnv_prime;

      hash ^= str[i];

   }

 

   return hash;

}

 

11、APHash

unsigned int APHash(const std::string& str)

{

   unsigned int hash = 0xAAAAAAAA;

 

   for(std::size_t i = 0; i < str.length(); i++)

   {

      hash ^= ((i & 1) == 0) ? (  (hash <<  7) ^ str[i] * (hash >> 3)) :

                               (~((hash << 11) + (str[i] ^ (hash >> 5))));

   }

 

   return hash;

}

 

12、MurmurHash - 非常新的一个哈希算法，应该是目前效率最高的一个哈希算法，使用率很高。(伪代码来自维基百科）

Murmur2(key, len, seed)

    m <- 0x5bd1e995

    r <- 24

    seed  0x9747b28c

    hash <- seed XOR len

    for each fourByteChunk of key

        k <- fourByteChunk

        k <- k * m

        k <- k XOR (k >> r)

        k <- k * m

 

        hash <- hash * m

        hash <- hash XOR k

 

    with any remainingBytesInKey

        remainingBytes  SwapEndianOrderOf(remainingBytesInKey)

 

        hash <- hash XOR remainingBytes

        hash <- hash * m

 

    hash <- hash XOR (hash >> 13)

    hash <- hash * m

    hash <- hash XOR (hash >> 15)

 

 

 

MurmurHash算法：高运算性能，低碰撞率，由Austin Appleby创建于2008年，现已应用到Hadoop、libstdc++、nginx、libmemcached等开源系统。2011年Appleby被Google雇佣，随后Google推出其变种的CityHash算法。

官方网站：https://sites.google.com/site/murmurhash/

MurmurHash算法，自称超级快的hash算法，是FNV的4-5倍。官方数据如下：

OneAtATime – 354.163715 mb/sec 
FNV – 443.668038 mb/sec 
SuperFastHash – 985.335173 mb/sec 
lookup3 – 988.080652 mb/sec 
MurmurHash 1.0 – 1363.293480 mb/sec 
MurmurHash 2.0 – 2056.885653 mb/sec

但也有文章声称，只有当key的长度大于10字节的时候，MurmurHash的运算速度才快于DJB。“从计算速度上来看，MurmurHash只适用于已知长度的、长度比较长的字符”。