DJBX33A (Daniel J. Bernstein, Times 33 with Addition) APR哈希默认算法

APR_DECLARE_NONSTD(unsigned   int ) apr_hashfunc_default( const    char   *char_key,   
                                                                                apr_ssize_t *klen)   
{   
     unsigned int   hash = 0;   
     const   unsigned   char   *key = ( const   unsigned   char   *)char_key;   
     const   unsigned   char   *p;   
     apr_ssize_t i;   

     /*   
      * This is the popular `times 33' hash algorithm which is used by   
      * perl and also appears in Berkeley DB. This is one of the best   
      * known hash functions for strings because it is both computed   
      * very fast and distributes very well.   
      *   
      * The originator may be Dan Bernstein but the code in Berkeley DB   
      * cites Chris Torek as the source. The best citation I have found   
      * is "Chris Torek, Hash function for text in C, Usenet message   
      * <27038@mimsy.umd.edu> in comp.lang.c , October, 1990." in Rich   
      * Salz's USENIX 1992 paper about INN which can be found at   
      * .   
      *   
      * The magic of number 33, i.e. why it works better than many other   
      * constants, prime or not, has never been adequately explained by   
      * anyone. So I try an explanation: if one experimentally tests all   
      * multipliers between 1 and 256 (as I did while writing a low-level   
      * data structure library some time ago) one detects that even   
      * numbers are not useable at all. The remaining 128 odd numbers   
      * (except for the number 1) work more or less all equally well.   
      * They all distribute in an acceptable way and this way fill a hash   
      * table with an average percent of approx. 86%.   
      *   
      * If one compares the chi^2 values of the variants (see   
      * Bob Jenkins ``Hashing Frequently Asked Questions'' at   
      * http://burtleburtle.net/bob/hash/hashfaq.html for a description   
      * of chi^2), the number 33 not even has the best value. But the   
      * number 33 and a few other equally good numbers like 17, 31, 63,   
      * 127 and 129 have nevertheless a great advantage to the remaining   
      * numbers in the large set of possible multipliers: their multiply   
      * operation can be replaced by a faster operation based on just one   
      * shift plus either a single addition or subtraction operation. And   
      * because a hash function has to both distribute good _and_ has to   
      * be very fast to compute, those few numbers should be preferred.   
      *   
      *                           -- Ralf S. Engelschall   
      */   

     if   (*klen == APR_HASH_KEY_STRING) {   
           for   (p = key; *p; p++) {   
                 hash = hash * 33 + *p;   
           }   
           *klen = p - key;   
     }   
     else   {   
           for   (p = key, i = *klen; i; i--, p++) {   
                 hash = hash * 33 + *p;   
           }   
     }   
     return   hash;   
}   

APR_DECLARE_NONSTD(unsigned int) apr_hashfunc_default(const char *char_key,   
                                                                                apr_ssize_t *klen)   
{   
     unsigned int hash = 0;   
     const unsigned char *key = (const unsigned char *)char_key;   
     const unsigned char *p;   
     apr_ssize_t i;   

     /*
      * This is the popular `times 33' hash algorithm which is used by
      * perl and also appears in Berkeley DB. This is one of the best
      * known hash functions for strings because it is both computed
      * very fast and distributes very well.
      *
      * The originator may be Dan Bernstein but the code in Berkeley DB
      * cites Chris Torek as the source. The best citation I have found
      * is "Chris Torek, Hash function for text in C, Usenet message
      * <27038@mimsy.umd.edu> in comp.lang.c , October, 1990." in Rich
      * Salz's USENIX 1992 paper about INN which can be found at
      * .
      *
      * The magic of number 33, i.e. why it works better than many other
      * constants, prime or not, has never been adequately explained by
      * anyone. So I try an explanation: if one experimentally tests all
      * multipliers between 1 and 256 (as I did while writing a low-level
      * data structure library some time ago) one detects that even
      * numbers are not useable at all. The remaining 128 odd numbers
      * (except for the number 1) work more or less all equally well.
      * They all distribute in an acceptable way and this way fill a hash
      * table with an average percent of approx. 86%.
      *
      * If one compares the chi^2 values of the variants (see
      * Bob Jenkins ``Hashing Frequently Asked Questions'' at
      * http://burtleburtle.net/bob/hash/hashfaq.html for a description
      * of chi^2), the number 33 not even has the best value. But the
      * number 33 and a few other equally good numbers like 17, 31, 63,
      * 127 and 129 have nevertheless a great advantage to the remaining
      * numbers in the large set of possible multipliers: their multiply
      * operation can be replaced by a faster operation based on just one
      * shift plus either a single addition or subtraction operation. And
      * because a hash function has to both distribute good _and_ has to
      * be very fast to compute, those few numbers should be preferred.
      *
      *                           -- Ralf S. Engelschall
      */   

     if (*klen == APR_HASH_KEY_STRING) {   
           for (p = key; *p; p++) {   
                 hash = hash * 33 + *p;   
           }   
           *klen = p - key;   
     }   
     else {   
           for (p = key, i = *klen; i; i--, p++) {   
                 hash = hash * 33 + *p;   
           }   
     }   
     return hash;   
}   

对函数注释部分的翻译:
这是很出名的times33哈希算法,此算法被perl语言采用并在Berkeley DB中出现.它是已知的最好的哈希算法之一,在处理以字符串为键值的哈希时,有着极快的计算效率和很好哈希分布.最早提出这个算法的是Dan Bernstein,但是源代码确实由Clris Torek在Berkeley DB出实作的.我找到的最确切的引文中这样说"Chris Torek,C语言文本哈希函数,Usenet消息< in comp.lang.c ,1990年十月."在Rich Salz于1992年在USENIX报上发表的讨论INN的文章中提到.这篇文章可以在上找到.

33这个奇妙的数字,为什么它能够比其他数值效果更好呢?无论重要与否,却从来没有人能够充分说明其中的原因.因此在这里,我来试着解释一下.如果某人试 着测试1到256之间的每个数字(就像我前段时间写的一个底层数据结构库那样),他会发现,没有哪一个数字的表现是特别突出的.其中的128个奇数(1除 外)的表现都差不多,都能够达到一个能接受的哈希分布,平均分布率大概是86%.

如果比较这128个奇数中的方差值(gibbon:统计术语,表示随机变量与它的数学期望之间的平均偏离程度)的话(见Bob Jenkins的http://burtleburtle.net/bob/hash/hashfaq.html,中对平方 差的描述),数字33并不是表现最好的一个.(gibbon:这里按照我的理解,照常理,应该是方差越小稳定,但是由于这里不清楚作者方差的计算公式,以 及在哈希离散表,是不是离散度越大越好,所以不得而知这里的表现好是指方差值大还是指方差值小),但是数字33以及其他一些同样好的数字比如 17,31,63,127和129对于其他剩下的数字,在面对大量的哈希运算时,仍然有一个大大的优势,就是这些数字能够将乘法用位运算配合加减法来替 换,这样的运算速度会提高.毕竟一个好的哈希算法要求既有好的分布,也要有高的计算速度,能同时达到这两点的数字很少

From: http://blog.csdn.net/zuiaituantuan/article/details/6057586