zoukankan      html  css  js  c++  java
  • [wget学习笔记] 哈希表 hash table (ZZ)

    [wget学习笔记] 哈希表 hash table (lyxint)

    (之前看过相关代码,也记过点东西,不过很很惭愧,没有整理记录哈。。。)

    wget里面有一个hash table的实现. 采用了线性开放地址空间法来解决碰撞, 为了保持效率, 保持哈希表包含元素数目和哈希表总大小比率小于0.75, 否则增长哈希表大小. 哈希表总大小为素数. 在增长哈希表的时候, 会对以前的元素进行rehash, 在删除元素的时候, 也可能会有rehash. 不过放心, 总体性能还是不错的.

    $ gcc hash.c -DSTANDALONE -DTEST -Wall
    $ time ./a.out < /usr/share/dict/words 
    387379 985003
     
    real	0m0.774s
    user	0m0.646s
    sys	0m0.050s
    这里有比较全的hash函数的说明和实现 http://www.partow.net/programming/hashfunctions/index.html
    也可看看 http://bretm.home.comcast.net/~bretm/hash/

    hash.h

    #ifndef HASH_H
    #define HASH_H
     
    struct hash_table;
     
    struct hash_table *hash_table_new (intunsigned long (*) (const void *),
                       int (*) (const void *, const void *));
    void hash_table_destroy (struct hash_table *);
     
    void *hash_table_get (const struct hash_table *, const void *);
    int hash_table_get_pair (const struct hash_table *, const void *,
                             void *, void *);
    int hash_table_contains (const struct hash_table *, const void *);
     
    void hash_table_put (struct hash_table *, const void *, void *);
    int hash_table_remove (struct hash_table *, const void *);
    void hash_table_clear (struct hash_table *);
     
    void hash_table_for_each (struct hash_table *,
                      int (*) (void *, void *, void *), void *);
     
    typedef struct {
      void *key, *value;        /* public members */
      void *pos, *end;      /* private members */
    } hash_table_iterator;
    void hash_table_iterate (struct hash_table *, hash_table_iterator *);
    int hash_table_iter_next (hash_table_iterator *);
     
    int hash_table_count (const struct hash_table *);
     
    struct hash_table *make_string_hash_table (int);
    struct hash_table *make_nocase_string_hash_table (int);
     
    unsigned long hash_pointer (const void *);
     
    #endif /* HASH_H */

    hash.c

    #ifndef STANDALONE
    # include "wget.h"
    #endif
     
    #include <stdio.h>
    #include <stdlib.h>
    #include <assert.h>
    #include <string.h>
    #include <limits.h>
     
    #ifndef STANDALONE
    /* Get Wget's utility headers. */
    # include "utils.h"
    #else
    /* Make do without them. */
    # define xnew(x) xmalloc (sizeof (x))
    # define xnew_array(type, x) xmalloc (sizeof (type) * (x))
    # define xmalloc malloc
    # define xfree free
    # ifndef countof
    #  define countof(x) (sizeof (x) / sizeof ((x)[0]))
    # endif
    # include <ctype.h>
    # define c_tolower(x) tolower ((unsigned char) (x))
    # ifdef HAVE_STDINT_H
    #  include <stdint.h>
    # else
       typedef unsigned long uintptr_t;
    # endif
    #endif
     
    #include "hash.h"
     
    /* INTERFACE:
     
       Hash tables are a technique used to implement mapping between
       objects with near-constant-time access and storage.  The table
       associates keys to values, and a value can be very quickly
       retrieved by providing the key.  Fast lookup tables are typically
       implemented as hash tables.
     
       The entry points are
         hash_table_new       -- creates the table.
         hash_table_destroy   -- destroys the table.
         hash_table_put       -- establishes or updates key->value mapping.
         hash_table_get       -- retrieves value of key.
         hash_table_get_pair  -- get key/value pair for key.
         hash_table_contains  -- test whether the table contains key.
         hash_table_remove    -- remove key->value mapping for given key.
         hash_table_for_each  -- call function for each table entry.
         hash_table_iterate   -- iterate over entries in hash table.
         hash_table_iter_next -- return next element during iteration.
         hash_table_clear     -- clear hash table contents.
         hash_table_count     -- return the number of entries in the table.
     
       The hash table grows internally as new entries are added and is not
       limited in size, except by available memory.  The table doubles
       with each resize, which ensures that the amortized time per
       operation remains constant.
     
       If not instructed otherwise, tables created by hash_table_new
       consider the keys to be equal if their pointer values are the same.
       You can use make_string_hash_table to create tables whose keys are
       considered equal if their string contents are the same.  In the
       general case, the criterion of equality used to compare keys is
       specified at table creation time with two callback functions,
       "hash" and "test".  The hash function transforms the key into an
       arbitrary number that must be the same for two equal keys.  The
       test function accepts two keys and returns non-zero if they are to
       be considered equal.
     
       Note that neither keys nor values are copied when inserted into the
       hash table, so they must exist for the lifetime of the table.  This
       means that e.g. the use of static strings is OK, but objects with a
       shorter life-time probably need to be copied (with strdup() or the
       like in the case of strings) before being inserted.  */
     
    /* IMPLEMENTATION:
     
       The hash table is implemented as an open-addressed table with
       linear probing collision resolution.
     
       The above means that all the cells (each cell containing a key and
       a value pointer) are stored in a contiguous array.  Array position
       of each cell is determined by the hash value of its key and the
       size of the table: location := hash(key) % size.  If two different
       keys end up on the same position (collide), the one that came
       second is stored in the first unoccupied cell that follows it.
       This collision resolution technique is called "linear probing".
     
       There are more advanced collision resolution methods (quadratic
       probing, double hashing), but we don't use them because they incur
       more non-sequential access to the array, which results in worse CPU
       cache behavior.  Linear probing works well as long as the
       count/size ratio (fullness) is kept below 75%.  We make sure to
       grow and rehash the table whenever this threshold is exceeded.
     
       Collisions complicate deletion because simply clearing a cell
       followed by previously collided entries would cause those neighbors
       to not be picked up by find_cell later.  One solution is to leave a
       "tombstone" marker instead of clearing the cell, and another is to
       recalculate the positions of adjacent cells.  We take the latter
       approach because it results in less bookkeeping garbage and faster
       retrieval at the (slight) expense of deletion.  */
     
    /* Maximum allowed fullness: when hash table's fullness exceeds this
       value, the table is resized.  */
    #define HASH_MAX_FULLNESS 0.75
     
    /* The hash table size is multiplied by this factor (and then rounded
       to the next prime) with each resize.  This guarantees infrequent
       resizes.  */
    #define HASH_RESIZE_FACTOR 2
     
    struct cell {
      void *key;
      void *value;
    };
     
    typedef unsigned long (*hashfun_t) (const void *);
    typedef int (*testfun_t) (const void *, const void *);
     
    struct hash_table {
      hashfun_t hash_function; //hash函数指针
      testfun_t test_function; //比较key是否相等的函数指针
     
      struct cell *cells;           /* contiguous array of cells. */
      int size;                     /* size of the array. */
                                      //size是素数, 在求模时能得到更好的效果
     
      int count;                    /* number of occupied entries. */
      int resize_threshold;         /* after size exceeds this number of
                                       entries, resize the table.  */
      int prime_offset;             /* the offset of the current prime in
                                       the prime table. */
    };
     
    /* We use the all-bits-set constant (INVALID_PTR) marker to mean that
       a cell is empty.  It is unaligned and therefore illegal as a
       pointer.  INVALID_PTR_CHAR (0xff) is the single-character constant
       used to initialize the entire cells array as empty.
     
       The all-bits-set value is a better choice than NULL because it
       allows the use of NULL/0 keys.  Since the keys are either integers
       or pointers, the only key that cannot be used is the integer value
       -1.  This is acceptable because it still allows the use of
       nonnegative integer keys.  */
     
    #define INVALID_PTR ((void *) ~(uintptr_t0)
    #ifndef UCHAR_MAX
    # define UCHAR_MAX 0xff
    #endif
    #define INVALID_PTR_CHAR UCHAR_MAX
     
    /* Whether the cell C is occupied (non-empty). */
    #define CELL_OCCUPIED(c) ((c)->key != INVALID_PTR)
     
    /* Clear the cell C, i.e. mark it as empty (unoccupied). */
    #define CLEAR_CELL(c) ((c)->key = INVALID_PTR)
     
    /* "Next" cell is the cell following C, but wrapping back to CELLS
       when C would reach CELLS+SIZE.  */
    #define NEXT_CELL(c, cells, size) (c != cells + (size - 1) ? c + 1 : cells)
     
    /* Loop over occupied cells starting at C, terminating the loop when
       an empty cell is encountered.  */
    #define FOREACH_OCCUPIED_ADJACENT(c, cells, size)                               \
      for (; CELL_OCCUPIED (c); c = NEXT_CELL (c, cells, size))
     
    /* Return the position of KEY in hash table SIZE large, hash function
       being HASHFUN.  */
    #define HASH_POSITION(key, hashfun, size) ((hashfun) (key) % size)
     
    /* Find a prime near, but greather than or equal to SIZE.  The primes
       are looked up from a table with a selection of primes convenient
       for this purpose.
     
       PRIME_OFFSET is a minor optimization: it specifies start position
       for the search for the large enough prime.  The final offset is
       stored in the same variable.  That way the list of primes does not
       have to be scanned from the beginning each time around.  */
     
    static int
    prime_size (int size, int *prime_offset)
    {//得到比size大的下一个素数
      static const int primes[] = {
        1319294159791071491972633474575997871031,
        136117772333303739675167671987371136914783,
        19219249893249142257549417142992861120721156941,
        2040472652713448574483215828217576939850031280519,
        1664681216411128133533657361475459161809898035301,
        104458991357968117653589229496692983460338784989,
        504205516554672985210757110774011144006217187208107,
        243370577316381771411296309534685237695090819903618083,
        1174703521152711461318372991312147483647
      };
      size_t i;
     
      for (i = *prime_offset; i < countof (primes); i++)
        if (primes[i] >= size)
          {
            /* Set the offset to the next prime.  That is safe because,
               next time we are called, it will be with a larger SIZE,
               which means we could never return the same prime anyway.
               (If that is not the case, the caller can simply reset
               *prime_offset.)  */
            *prime_offset = i + 1;
            return primes[i];
          }
     
      abort ();
    }
     
    static int cmp_pointer (const void *, const void *);
     
    /* Create a hash table with hash function HASH_FUNCTION and test
       function TEST_FUNCTION.  The table is empty (its count is 0), but
       pre-allocated to store at least ITEMS items.
     
       ITEMS is the number of items that the table can accept without
       needing to resize.  It is useful when creating a table that is to
       be immediately filled with a known number of items.  In that case,
       the regrows are a waste of time, and specifying ITEMS correctly
       will avoid them altogether.
     
       Note that hash tables grow dynamically regardless of ITEMS.  The
       only use of ITEMS is to preallocate the table and avoid unnecessary
       dynamic regrows.  Don't bother making ITEMS prime because it's not
       used as size unchanged.  To start with a small table that grows as
       needed, simply specify zero ITEMS.
     
       If hash and test callbacks are not specified, identity mapping is
       assumed, i.e. pointer values are used for key comparison.  (Common
       Lisp calls such tables EQ hash tables, and Java calls them
       IdentityHashMaps.)  If your keys require different comparison,
       specify hash and test functions.  For easy use of C strings as hash
       keys, you can use the convenience functions make_string_hash_table
       and make_nocase_string_hash_table.  */
     
    struct hash_table *
    hash_table_new (int items,
                    unsigned long (*hash_function) (const void *),
                    int (*test_function) (const void *, const void *))
    {
      int size;
      struct hash_table *ht = xnew (struct hash_table);
     
      ht->hash_function = hash_function ? hash_function : hash_pointer;
      ht->test_function = test_function ? test_function : cmp_pointer;
     
      /* If the size of struct hash_table ever becomes a concern, this
         field can go.  (Wget doesn't create many hashes.)  */
      ht->prime_offset = 0;
     
      /* Calculate the size that ensures that the table will store at
         least ITEMS keys without the need to resize.  */
      size = 1 + items / HASH_MAX_FULLNESS;
      size = prime_size (size, &ht->prime_offset);
      ht->size = size;
      ht->resize_threshold = size * HASH_MAX_FULLNESS;
      /*assert (ht->resize_threshold >= items);*/
     
      ht->cells = xnew_array (struct cell, ht->size);
     
      /* Mark cells as empty.  We use 0xff rather than 0 to mark empty
         keys because it allows us to use NULL/0 as keys.  */
      memset (ht->cells, INVALID_PTR_CHAR, size * sizeof (struct cell));
     
      ht->count = 0;
     
      return ht;
    }
     
    /* Free the data associated with hash table HT. */
     
    void
    hash_table_destroy (struct hash_table *ht)
    {
      xfree (ht->cells);
      xfree (ht);
    }
     
    /* The heart of most functions in this file -- find the cell whose
       KEY is equal to key, using linear probing.  Returns the cell
       that matches KEY, or the first empty cell if none matches.  */
     
    static inline struct cell *
    find_cell (const struct hash_table *ht, const void *key)
    {//根据hash_table的test_function来寻找一个cell
      struct cell *cells = ht->cells;
      int size = ht->size;
      struct cell *c = cells + HASH_POSITION (key, ht->hash_function, size);
      testfun_t equals = ht->test_function;
     
      FOREACH_OCCUPIED_ADJACENT (c, cells, size)
        if (equals (key, c->key))
          break;
      // count/size >= resize_threshold时会扩展, 所以总是会有空的cell的.
      return c;
    }
     
    /* Get the value that corresponds to the key KEY in the hash table HT.
       If no value is found, return NULL.  Note that NULL is a legal value
       for value; if you are storing NULLs in your hash table, you can use
       hash_table_contains to be sure that a (possibly NULL) value exists
       in the table.  Or, you can use hash_table_get_pair instead of this
       function.  */
     
    void *
    hash_table_get (const struct hash_table *ht, const void *key)
    {
      struct cell *c = find_cell (ht, key);
      if (CELL_OCCUPIED (c))
        return c->value;
      else
        return NULL;
    }
     
    /* Like hash_table_get, but writes out the pointers to both key and
       value.  Returns non-zero on success.  */
     
    int
    hash_table_get_pair (const struct hash_table *ht, const void *lookup_key,
                         void *orig_key, void *value)
    {
      struct cell *c = find_cell (ht, lookup_key);
      if (CELL_OCCUPIED (c))
        {
          if (orig_key)
            *(void **)orig_key = c->key;
          if (value)
            *(void **)value = c->value;
          return 1;
        }
      else
        return 0;
    }
     
    /* Return 1 if HT contains KEY, 0 otherwise. */
     
    int
    hash_table_contains (const struct hash_table *ht, const void *key)
    {
      struct cell *c = find_cell (ht, key);
      return CELL_OCCUPIED (c);
    }
     
    /* Grow hash table HT as necessary, and rehash all the key-value
       mappings.  */
     
    static void
    grow_hash_table (struct hash_table *ht)
    {
      hashfun_t hasher = ht->hash_function;
      struct cell *old_cells = ht->cells;
      struct cell *old_end   = ht->cells + ht->size;
      struct cell *c, *cells;
      int newsize;
     
      newsize = prime_size (ht->size * HASH_RESIZE_FACTOR, &ht->prime_offset);
    #if 0
      printf ("growing from %d to %d; fullness %.2f%% to %.2f%%\n",
              ht->size, newsize,
              100.0 * ht->count / ht->size,
              100.0 * ht->count / newsize);
    #endif
     
      ht->size = newsize;
      ht->resize_threshold = newsize * HASH_MAX_FULLNESS;
     
      cells = xnew_array (struct cell, newsize);
      memset (cells, INVALID_PTR_CHAR, newsize * sizeof (struct cell));
      ht->cells = cells;
     
      for (c = old_cells; c < old_end; c++)
        if (CELL_OCCUPIED (c))
          {
            struct cell *new_c;
            /* We don't need to test for uniqueness of keys because they
               come from the hash table and are therefore known to be
               unique.  */
            new_c = cells + HASH_POSITION (c->key, hasher, newsize);
            FOREACH_OCCUPIED_ADJACENT (new_c, cells, newsize)
              ;
            *new_c = *c;
          }
     
      xfree (old_cells);
    }
     
    /* Put VALUE in the hash table HT under the key KEY.  This regrows the
       table if necessary.  */
     
    void
    hash_table_put (struct hash_table *ht, const void *key, void *value)
    {
      struct cell *c = find_cell (ht, key);
      if (CELL_OCCUPIED (c))
        {
          /* update existing item */
          c->key   = (void *)key; /* const? */
          c->value = value;
          return;
        }
     
      /* If adding the item would make the table exceed max. fullness,
         grow the table first.  */
      if (ht->count >= ht->resize_threshold)
        {
          grow_hash_table (ht);
          c = find_cell (ht, key);
        }
     
      /* add new item */
      ++ht->count;
      c->key   = (void *)key;       /* const? */
      c->value = value;
    }
     
    /* Remove KEY->value mapping from HT.  Return 0 if there was no such
       entry; return 1 if an entry was removed.  */
     
    int
    hash_table_remove (struct hash_table *ht, const void *key)
    {
      struct cell *c = find_cell (ht, key);
      if (!CELL_OCCUPIED (c))
        return 0;
      else
        {
          int size = ht->size;
          struct cell *cells = ht->cells;
          hashfun_t hasher = ht->hash_function;
     
          CLEAR_CELL (c);
          --ht->count;
     
          /* Rehash all the entries following C.  The alternative
             approach is to mark the entry as deleted, i.e. create a
             "tombstone".  That speeds up removal, but leaves a lot of
             garbage and slows down hash_table_get and hash_table_put.  */
     
          c = NEXT_CELL (c, cells, size);
          FOREACH_OCCUPIED_ADJACENT (c, cells, size)
            {
              const void *key2 = c->key;
              struct cell *c_new;
     
              /* Find the new location for the key. */
              c_new = cells + HASH_POSITION (key2, hasher, size);
              FOREACH_OCCUPIED_ADJACENT (c_new, cells, size)
                if (key2 == c_new->key)
                  /* The cell C (key2) is already where we want it (in
                     C_NEW's "chain" of keys.)  */
                  goto next_rehash;
     
              *c_new = *c;
              CLEAR_CELL (c);
     
            next_rehash:
              ;
            }
          return 1;
        }
    }
     
    /* Clear HT of all entries.  After calling this function, the count
       and the fullness of the hash table will be zero.  The size will
       remain unchanged.  */
     
    void
    hash_table_clear (struct hash_table *ht)
    {
      memset (ht->cells, INVALID_PTR_CHAR, ht->size * sizeof (struct cell));
      ht->count = 0;
    }
     
    /* Call FN for each entry in HT.  FN is called with three arguments:
       the key, the value, and ARG.  When FN returns a non-zero value, the
       mapping stops.
     
       It is undefined what happens if you add or remove entries in the
       hash table while hash_table_for_each is running.  The exception is
       the entry you're currently mapping over; you may call
       hash_table_put or hash_table_remove on that entry's key.  That is
       also the reason why this function cannot be implemented in terms of
       hash_table_iterate.  */
     
    void
    hash_table_for_each (struct hash_table *ht,
                         int (*fn) (void *, void *, void *), void *arg)
    {
      struct cell *c = ht->cells;
      struct cell *end = ht->cells + ht->size;
     
      for (; c < end; c++)
        if (CELL_OCCUPIED (c))
          {
            void *key;
          repeat:
            key = c->key;
            if (fn (key, c->value, arg))
              return;
            /* hash_table_remove might have moved the adjacent cells. */
            if (c->key != key && CELL_OCCUPIED (c))
              goto repeat;
          }
    }
     
    /* Initiate iteration over HT.  Entries are obtained with
       hash_table_iter_next, a typical iteration loop looking like this:
     
           hash_table_iterator iter;
           for (hash_table_iterate (ht, &iter); hash_table_iter_next (&iter); )
             ... do something with iter.key and iter.value ...
     
       The iterator does not need to be deallocated after use.  The hash
       table must not be modified while being iterated over.  */
     
    void
    hash_table_iterate (struct hash_table *ht, hash_table_iterator *iter)
    {
      iter->pos = ht->cells;
      iter->end = ht->cells + ht->size;
    }
     
    /* Get the next hash table entry.  ITER is an iterator object
       initialized using hash_table_iterate.  While there are more
       entries, the key and value pointers are stored to ITER->key and
       ITER->value respectively and 1 is returned.  When there are no more
       entries, 0 is returned.
     
       If the hash table is modified between calls to this function, the
       result is undefined.  */
     
    int
    hash_table_iter_next (hash_table_iterator *iter)
    {
      struct cell *c = iter->pos;
      struct cell *end = iter->end;
      for (; c < end; c++)
        if (CELL_OCCUPIED (c))
          {
            iter->key = c->key;
            iter->value = c->value;
            iter->pos = c + 1;
            return 1;
          }
      return 0;
    }
     
    /* Return the number of elements in the hash table.  This is not the
       same as the physical size of the hash table, which is always
       greater than the number of elements.  */
     
    int
    hash_table_count (const struct hash_table *ht)
    {
      return ht->count;
    }
    ^L
    /* Functions from this point onward are meant for convenience and
       don't strictly belong to this file.  However, this is as good a
       place for them as any.  */
     
    /* Guidelines for creating custom hash and test functions:
     
       - The test function returns non-zero for keys that are considered
         "equal", zero otherwise.
     
       - The hash function returns a number that represents the
         "distinctness" of the object.  In more precise terms, it means
         that for any two objects that test "equal" under the test
         function, the hash function MUST produce the same result.
     
         This does not mean that all different objects must produce
         different values (that would be "perfect" hashing), only that
         non-distinct objects must produce the same values!  For instance,
         a hash function that returns 0 for any given object is a
         perfectly valid (albeit extremely bad) hash function.  A hash
         function that hashes a string by adding up all its characters is
         another example of a valid (but still quite bad) hash function.
     
         It is not hard to make hash and test functions agree about
         equality.  For example, if the test function compares strings
         case-insensitively, the hash function can lower-case the
         characters when calculating the hash value.  That ensures that
         two strings differing only in case will hash the same.
     
       - To prevent performance degradation, choose a hash function with
         as good "spreading" as possible.  A good hash function will use
         all the bits of the input when calculating the hash, and will
         react to even small changes in input with a completely different
         output.  But don't make the hash function itself overly slow,
         because you'll be incurring a non-negligible overhead to all hash
         table operations.  */
     
    /*
     * Support for hash tables whose keys are strings.
     *
     */
     
    /* Base 31 hash function.  Taken from Gnome's glib, modified to use
       standard C types.
     
       We used to use the popular hash function from the Dragon Book, but
       this one seems to perform much better, both by being faster and by
       generating less collisions.  */
     
    static unsigned long
    hash_string (const void *key)
    {
      const char *p = key;
      unsigned int h = *p;
     
      if (h)
        for (p += 1; *p != '\0'; p++)
          h = (h << 5) - h + *p;
     
      return h;
    }
     
    /* Frontend for strcmp usable for hash tables. */
     
    static int
    cmp_string (const void *s1, const void *s2)
    {
      return !strcmp ((const char *)s1, (const char *)s2);
    }
     
    /* Return a hash table of preallocated to store at least ITEMS items
       suitable to use strings as keys.  */
     
    struct hash_table *
    make_string_hash_table (int items)
    {
      return hash_table_new (items, hash_string, cmp_string);
    }
     
    /*
     * Support for hash tables whose keys are strings, but which are
     * compared case-insensitively.
     *
     */
     
    /* Like hash_string, but produce the same hash regardless of the case. */
     
    static unsigned long
    hash_string_nocase (const void *key)
    {
      const char *p = key;
      unsigned int h = c_tolower (*p);
     
      if (h)
        for (p += 1; *p != '\0'; p++)
          h = (h << 5) - h + c_tolower (*p);
     
      return h;
    }
     
    /* Like string_cmp, but doing case-insensitive compareison. */
     
    static int
    string_cmp_nocase (const void *s1, const void *s2)
    {
      return !strcasecmp ((const char *)s1, (const char *)s2);
    }
     
    /* Like make_string_hash_table, but uses string_hash_nocase and
       string_cmp_nocase.  */
     
    struct hash_table *
    make_nocase_string_hash_table (int items)
    {
      return hash_table_new (items, hash_string_nocase, string_cmp_nocase);
    }
     
    /* Hashing of numeric values, such as pointers and integers.
     
       This implementation is the Robert Jenkins' 32 bit Mix Function,
       with a simple adaptation for 64-bit values.  According to Jenkins
       it should offer excellent spreading of values.  Unlike the popular
       Knuth's multiplication hash, this function doesn't need to know the
       hash table size to work.  */
     
    unsigned long
    hash_pointer (const void *ptr)
    {
      uintptr_t key = (uintptr_t) ptr;
      key += (key << 12);
      key ^= (key >> 22);
      key += (key << 4);
      key ^= (key >> 9);
      key += (key << 10);
      key ^= (key >> 2);
      key += (key << 7);
      key ^= (key >> 12);
    #if SIZEOF_VOID_P > 4
      key += (key << 44);
      key ^= (key >> 54);
      key += (key << 36);
      key ^= (key >> 41);
      key += (key << 42);
      key ^= (key >> 34);
      key += (key << 39);
      key ^= (key >> 44);
    #endif
      return (unsigned long) key;
    }
     
    static int
    cmp_pointer (const void *ptr1, const void *ptr2)
    {
      return ptr1 == ptr2;
    }
    ^L
    #ifdef TEST
     
    #include <stdio.h>
    #include <string.h>
     
    void
    print_hash (struct hash_table *sht)
    {
      hash_table_iterator iter;
      int count = 0;
     
      for (hash_table_iterate (sht, &iter); hash_table_iter_next (&iter);
           ++count)
        printf ("%s%s\n", (char *)iter.key, (char *)iter.value);
      assert (count == sht->count);
    }
     
    int
    main (void)
    {
      struct hash_table *ht = make_string_hash_table (0);
      char line[80];
      while ((fgets (line, sizeof (line), stdin)))
        {
          int len = strlen (line);
          if (len <= 1)
            continue;
          line[--len] = '\0';
          if (!hash_table_contains (ht, line))
            hash_table_put (ht, strdup (line), "here I am!");
    #if 1
          if (len % 5 == 0)
            {
              char *line_copy;
              if (hash_table_get_pair (ht, line, &line_copy, NULL))
                {
                  hash_table_remove (ht, line);
                  xfree (line_copy);
                }
            }
    #endif
        }
    #if 0
      print_hash (ht);
    #endif
    #if 1
      printf ("%d %d\n", ht->count, ht->size);
    #endif
      return 0;
    }
    #endif /* TEST */

  • 相关阅读:
    DDL
    [笔记]NFC笔记——初始化RF碰撞避免
    [笔记]Java没有C语言的编译开关怎么办?
    [笔记]NFC笔记——通用初始化及单设备检测(SDD)流程
    [笔记]NFC笔记——NFCIP1协议命令集(NFCIP1 Protocol Command Set)
    [笔记]C++代码演示SingletonMap 单类Map实例
    [笔记]NFC笔记——传输帧格式
    [笔记]C++代码演示Singleton单类实例
    [笔记]山寨中文编程语言
    [笔记]NFC笔记——ATR_REQ 消息结构
  • 原文地址:https://www.cnblogs.com/IS2120/p/6745952.html
Copyright © 2011-2022 走看看