git.oblomov.eu Git - git/blob - khash.h

   1 /* The MIT License
   2
   3    Copyright (c) 2008, 2009, 2011 by Attractive Chaos <attractor@live.co.uk>
   4
   5    Permission is hereby granted, free of charge, to any person obtaining
   6    a copy of this software and associated documentation files (the
   7    "Software"), to deal in the Software without restriction, including
   8    without limitation the rights to use, copy, modify, merge, publish,
   9    distribute, sublicense, and/or sell copies of the Software, and to
  10    permit persons to whom the Software is furnished to do so, subject to
  11    the following conditions:
  12
  13    The above copyright notice and this permission notice shall be
  14    included in all copies or substantial portions of the Software.
  15
  16    THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
  17    EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
  18    MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
  19    NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
  20    BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
  21    ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
  22    CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
  23    SOFTWARE.
  24 */
  25
  26 #ifndef __AC_KHASH_H
  27 #define __AC_KHASH_H
  28
  29 #define AC_VERSION_KHASH_H "0.2.8"
  30
  31 typedef uint32_t khint32_t;
  32 typedef uint64_t khint64_t;
  33
  34 typedef khint32_t khint_t;
  35 typedef khint_t khiter_t;
  36
  37 #define __ac_isempty(flag, i) ((flag[i>>4]>>((i&0xfU)<<1))&2)
  38 #define __ac_isdel(flag, i) ((flag[i>>4]>>((i&0xfU)<<1))&1)
  39 #define __ac_iseither(flag, i) ((flag[i>>4]>>((i&0xfU)<<1))&3)
  40 #define __ac_set_isdel_false(flag, i) (flag[i>>4]&=~(1ul<<((i&0xfU)<<1)))
  41 #define __ac_set_isempty_false(flag, i) (flag[i>>4]&=~(2ul<<((i&0xfU)<<1)))
  42 #define __ac_set_isboth_false(flag, i) (flag[i>>4]&=~(3ul<<((i&0xfU)<<1)))
  43 #define __ac_set_isdel_true(flag, i) (flag[i>>4]|=1ul<<((i&0xfU)<<1))
  44
  45 #define __ac_fsize(m) ((m) < 16? 1 : (m)>>4)
  46
  47 #define kroundup32(x) (--(x), (x)|=(x)>>1, (x)|=(x)>>2, (x)|=(x)>>4, (x)|=(x)>>8, (x)|=(x)>>16, ++(x))
  48
  49 static inline khint_t __ac_X31_hash_string(const char *s)
  50 {
  51         khint_t h = (khint_t)*s;
  52         if (h) for (++s ; *s; ++s) h = (h << 5) - h + (khint_t)*s;
  53         return h;
  54 }
  55
  56 #define kh_str_hash_func(key) __ac_X31_hash_string(key)
  57 #define kh_str_hash_equal(a, b) (strcmp(a, b) == 0)
  58
  59 static const double __ac_HASH_UPPER = 0.77;
  60
  61 #define __KHASH_TYPE(name, khkey_t, khval_t) \
  62         typedef struct { \
  63                 khint_t n_buckets, size, n_occupied, upper_bound; \
  64                 khint32_t *flags; \
  65                 khkey_t *keys; \
  66                 khval_t *vals; \
  67         } kh_##name##_t;
  68
  69 #define __KHASH_PROTOTYPES(name, khkey_t, khval_t)                                              \
  70         extern kh_##name##_t *kh_init_##name(void);                                                     \
  71         extern void kh_destroy_##name(kh_##name##_t *h);                                        \
  72         extern void kh_clear_##name(kh_##name##_t *h);                                          \
  73         extern khint_t kh_get_##name(const kh_##name##_t *h, khkey_t key);      \
  74         extern int kh_resize_##name(kh_##name##_t *h, khint_t new_n_buckets); \
  75         extern khint_t kh_put_##name(kh_##name##_t *h, khkey_t key, int *ret); \
  76         extern void kh_del_##name(kh_##name##_t *h, khint_t x);
  77
  78 #define __KHASH_IMPL(name, SCOPE, khkey_t, khval_t, kh_is_map, __hash_func, __hash_equal) \
  79         SCOPE kh_##name##_t *kh_init_##name(void) {                                                     \
  80                 return (kh_##name##_t*)xcalloc(1, sizeof(kh_##name##_t));               \
  81         }                                                                                                                                       \
  82         SCOPE void kh_destroy_##name(kh_##name##_t *h)                                          \
  83         {                                                                                                                                       \
  84                 if (h) {                                                                                                                \
  85                         free((void *)h->keys); free(h->flags);                                  \
  86                         free((void *)h->vals);                                                                          \
  87                         free(h);                                                                                                        \
  88                 }                                                                                                                               \
  89         }                                                                                                                                       \
  90         SCOPE void kh_clear_##name(kh_##name##_t *h)                                            \
  91         {                                                                                                                                       \
  92                 if (h && h->flags) {                                                                                    \
  93                         memset(h->flags, 0xaa, __ac_fsize(h->n_buckets) * sizeof(khint32_t)); \
  94                         h->size = h->n_occupied = 0;                                                            \
  95                 }                                                                                                                               \
  96         }                                                                                                                                       \
  97         SCOPE khint_t kh_get_##name(const kh_##name##_t *h, khkey_t key)        \
  98         {                                                                                                                                       \
  99                 if (h->n_buckets) {                                                                                             \
 100                         khint_t k, i, last, mask, step = 0; \
 101                         mask = h->n_buckets - 1;                                                                        \
 102                         k = __hash_func(key); i = k & mask;                                                     \
 103                         last = i; \
 104                         while (!__ac_isempty(h->flags, i) && (__ac_isdel(h->flags, i) || !__hash_equal(h->keys[i], key))) { \
 105                                 i = (i + (++step)) & mask; \
 106                                 if (i == last) return h->n_buckets;                                             \
 107                         }                                                                                                                       \
 108                         return __ac_iseither(h->flags, i)? h->n_buckets : i;            \
 109                 } else return 0;                                                                                                \
 110         }                                                                                                                                       \
 111         SCOPE int kh_resize_##name(kh_##name##_t *h, khint_t new_n_buckets) \
 112         { /* This function uses 0.25*n_buckets bytes of working space instead of [sizeof(key_t+val_t)+.25]*n_buckets. */ \
 113                 khint32_t *new_flags = NULL;                                                                            \
 114                 khint_t j = 1;                                                                                                  \
 115                 {                                                                                                                               \
 116                         kroundup32(new_n_buckets);                                                                      \
 117                         if (new_n_buckets < 4) new_n_buckets = 4;                                       \
 118                         if (h->size >= (khint_t)(new_n_buckets * __ac_HASH_UPPER + 0.5)) j = 0; /* requested size is too small */ \
 119                         else { /* hash table size to be changed (shrink or expand); rehash */ \
 120                                 new_flags = (khint32_t*)xmalloc(__ac_fsize(new_n_buckets) * sizeof(khint32_t)); \
 121                                 if (!new_flags) return -1;                                                              \
 122                                 memset(new_flags, 0xaa, __ac_fsize(new_n_buckets) * sizeof(khint32_t)); \
 123                                 if (h->n_buckets < new_n_buckets) {     /* expand */            \
 124                                         khkey_t *new_keys = (khkey_t*)xrealloc((void *)h->keys, new_n_buckets * sizeof(khkey_t)); \
 125                                         if (!new_keys) return -1;                                                       \
 126                                         h->keys = new_keys;                                                                     \
 127                                         if (kh_is_map) {                                                                        \
 128                                                 khval_t *new_vals = (khval_t*)xrealloc((void *)h->vals, new_n_buckets * sizeof(khval_t)); \
 129                                                 if (!new_vals) return -1;                                               \
 130                                                 h->vals = new_vals;                                                             \
 131                                         }                                                                                                       \
 132                                 } /* otherwise shrink */                                                                \
 133                         }                                                                                                                       \
 134                 }                                                                                                                               \
 135                 if (j) { /* rehashing is needed */                                                              \
 136                         for (j = 0; j != h->n_buckets; ++j) {                                           \
 137                                 if (__ac_iseither(h->flags, j) == 0) {                                  \
 138                                         khkey_t key = h->keys[j];                                                       \
 139                                         khval_t val;                                                                            \
 140                                         khint_t new_mask;                                                                       \
 141                                         new_mask = new_n_buckets - 1;                                           \
 142                                         if (kh_is_map) val = h->vals[j];                                        \
 143                                         __ac_set_isdel_true(h->flags, j);                                       \
 144                                         while (1) { /* kick-out process; sort of like in Cuckoo hashing */ \
 145                                                 khint_t k, i, step = 0; \
 146                                                 k = __hash_func(key);                                                   \
 147                                                 i = k & new_mask;                                                               \
 148                                                 while (!__ac_isempty(new_flags, i)) i = (i + (++step)) & new_mask; \
 149                                                 __ac_set_isempty_false(new_flags, i);                   \
 150                                                 if (i < h->n_buckets && __ac_iseither(h->flags, i) == 0) { /* kick out the existing element */ \
 151                                                         { khkey_t tmp = h->keys[i]; h->keys[i] = key; key = tmp; } \
 152                                                         if (kh_is_map) { khval_t tmp = h->vals[i]; h->vals[i] = val; val = tmp; } \
 153                                                         __ac_set_isdel_true(h->flags, i); /* mark it as deleted in the old hash table */ \
 154                                                 } else { /* write the element and jump out of the loop */ \
 155                                                         h->keys[i] = key;                                                       \
 156                                                         if (kh_is_map) h->vals[i] = val;                        \
 157                                                         break;                                                                          \
 158                                                 }                                                                                               \
 159                                         }                                                                                                       \
 160                                 }                                                                                                               \
 161                         }                                                                                                                       \
 162                         if (h->n_buckets > new_n_buckets) { /* shrink the hash table */ \
 163                                 h->keys = (khkey_t*)xrealloc((void *)h->keys, new_n_buckets * sizeof(khkey_t)); \
 164                                 if (kh_is_map) h->vals = (khval_t*)xrealloc((void *)h->vals, new_n_buckets * sizeof(khval_t)); \
 165                         }                                                                                                                       \
 166                         free(h->flags); /* free the working space */                            \
 167                         h->flags = new_flags;                                                                           \
 168                         h->n_buckets = new_n_buckets;                                                           \
 169                         h->n_occupied = h->size;                                                                        \
 170                         h->upper_bound = (khint_t)(h->n_buckets * __ac_HASH_UPPER + 0.5); \
 171                 }                                                                                                                               \
 172                 return 0;                                                                                                               \
 173         }                                                                                                                                       \
 174         SCOPE khint_t kh_put_##name(kh_##name##_t *h, khkey_t key, int *ret) \
 175         {                                                                                                                                       \
 176                 khint_t x;                                                                                                              \
 177                 if (h->n_occupied >= h->upper_bound) { /* update the hash table */ \
 178                         if (h->n_buckets > (h->size<<1)) {                                                      \
 179                                 if (kh_resize_##name(h, h->n_buckets - 1) < 0) { /* clear "deleted" elements */ \
 180                                         *ret = -1; return h->n_buckets;                                         \
 181                                 }                                                                                                               \
 182                         } else if (kh_resize_##name(h, h->n_buckets + 1) < 0) { /* expand the hash table */ \
 183                                 *ret = -1; return h->n_buckets;                                                 \
 184                         }                                                                                                                       \
 185                 } /* TODO: to implement automatically shrinking; resize() already support shrinking */ \
 186                 {                                                                                                                               \
 187                         khint_t k, i, site, last, mask = h->n_buckets - 1, step = 0; \
 188                         x = site = h->n_buckets; k = __hash_func(key); i = k & mask; \
 189                         if (__ac_isempty(h->flags, i)) x = i; /* for speed up */        \
 190                         else {                                                                                                          \
 191                                 last = i; \
 192                                 while (!__ac_isempty(h->flags, i) && (__ac_isdel(h->flags, i) || !__hash_equal(h->keys[i], key))) { \
 193                                         if (__ac_isdel(h->flags, i)) site = i;                          \
 194                                         i = (i + (++step)) & mask; \
 195                                         if (i == last) { x = site; break; }                                     \
 196                                 }                                                                                                               \
 197                                 if (x == h->n_buckets) {                                                                \
 198                                         if (__ac_isempty(h->flags, i) && site != h->n_buckets) x = site; \
 199                                         else x = i;                                                                                     \
 200                                 }                                                                                                               \
 201                         }                                                                                                                       \
 202                 }                                                                                                                               \
 203                 if (__ac_isempty(h->flags, x)) { /* not present at all */               \
 204                         h->keys[x] = key;                                                                                       \
 205                         __ac_set_isboth_false(h->flags, x);                                                     \
 206                         ++h->size; ++h->n_occupied;                                                                     \
 207                         *ret = 1;                                                                                                       \
 208                 } else if (__ac_isdel(h->flags, x)) { /* deleted */                             \
 209                         h->keys[x] = key;                                                                                       \
 210                         __ac_set_isboth_false(h->flags, x);                                                     \
 211                         ++h->size;                                                                                                      \
 212                         *ret = 2;                                                                                                       \
 213                 } else *ret = 0; /* Don't touch h->keys[x] if present and not deleted */ \
 214                 return x;                                                                                                               \
 215         }                                                                                                                                       \
 216         SCOPE void kh_del_##name(kh_##name##_t *h, khint_t x)                           \
 217         {                                                                                                                                       \
 218                 if (x != h->n_buckets && !__ac_iseither(h->flags, x)) {                 \
 219                         __ac_set_isdel_true(h->flags, x);                                                       \
 220                         --h->size;                                                                                                      \
 221                 }                                                                                                                               \
 222         }
 223
 224 #define KHASH_DECLARE(name, khkey_t, khval_t)                                                   \
 225         __KHASH_TYPE(name, khkey_t, khval_t)                                                            \
 226         __KHASH_PROTOTYPES(name, khkey_t, khval_t)
 227
 228 #define KHASH_INIT2(name, SCOPE, khkey_t, khval_t, kh_is_map, __hash_func, __hash_equal) \
 229         __KHASH_TYPE(name, khkey_t, khval_t)                                                            \
 230         __KHASH_IMPL(name, SCOPE, khkey_t, khval_t, kh_is_map, __hash_func, __hash_equal)
 231
 232 #define KHASH_INIT(name, khkey_t, khval_t, kh_is_map, __hash_func, __hash_equal) \
 233         KHASH_INIT2(name, static inline, khkey_t, khval_t, kh_is_map, __hash_func, __hash_equal)
 234
 235 /* Other convenient macros... */
 236
 237 /*! @function
 238   @abstract     Test whether a bucket contains data.
 239   @param  h     Pointer to the hash table [khash_t(name)*]
 240   @param  x     Iterator to the bucket [khint_t]
 241   @return       1 if containing data; 0 otherwise [int]
 242  */
 243 #define kh_exist(h, x) (!__ac_iseither((h)->flags, (x)))
 244
 245 /*! @function
 246   @abstract     Get key given an iterator
 247   @param  h     Pointer to the hash table [khash_t(name)*]
 248   @param  x     Iterator to the bucket [khint_t]
 249   @return       Key [type of keys]
 250  */
 251 #define kh_key(h, x) ((h)->keys[x])
 252
 253 /*! @function
 254   @abstract     Get value given an iterator
 255   @param  h     Pointer to the hash table [khash_t(name)*]
 256   @param  x     Iterator to the bucket [khint_t]
 257   @return       Value [type of values]
 258   @discussion   For hash sets, calling this results in segfault.
 259  */
 260 #define kh_val(h, x) ((h)->vals[x])
 261
 262 /*! @function
 263   @abstract     Alias of kh_val()
 264  */
 265 #define kh_value(h, x) ((h)->vals[x])
 266
 267 /*! @function
 268   @abstract     Get the start iterator
 269   @param  h     Pointer to the hash table [khash_t(name)*]
 270   @return       The start iterator [khint_t]
 271  */
 272 #define kh_begin(h) (khint_t)(0)
 273
 274 /*! @function
 275   @abstract     Get the end iterator
 276   @param  h     Pointer to the hash table [khash_t(name)*]
 277   @return       The end iterator [khint_t]
 278  */
 279 #define kh_end(h) ((h)->n_buckets)
 280
 281 /*! @function
 282   @abstract     Get the number of elements in the hash table
 283   @param  h     Pointer to the hash table [khash_t(name)*]
 284   @return       Number of elements in the hash table [khint_t]
 285  */
 286 #define kh_size(h) ((h)->size)
 287
 288 /*! @function
 289   @abstract     Get the number of buckets in the hash table
 290   @param  h     Pointer to the hash table [khash_t(name)*]
 291   @return       Number of buckets in the hash table [khint_t]
 292  */
 293 #define kh_n_buckets(h) ((h)->n_buckets)
 294
 295 /*! @function
 296   @abstract     Iterate over the entries in the hash table
 297   @param  h     Pointer to the hash table [khash_t(name)*]
 298   @param  kvar  Variable to which key will be assigned
 299   @param  vvar  Variable to which value will be assigned
 300   @param  code  Block of code to execute
 301  */
 302 #define kh_foreach(h, kvar, vvar, code) { khint_t __i;          \
 303         for (__i = kh_begin(h); __i != kh_end(h); ++__i) {              \
 304                 if (!kh_exist(h,__i)) continue;                                         \
 305                 (kvar) = kh_key(h,__i);                                                         \
 306                 (vvar) = kh_val(h,__i);                                                         \
 307                 code;                                                                                           \
 308         } }
 309
 310 /*! @function
 311   @abstract     Iterate over the values in the hash table
 312   @param  h     Pointer to the hash table [khash_t(name)*]
 313   @param  vvar  Variable to which value will be assigned
 314   @param  code  Block of code to execute
 315  */
 316 #define kh_foreach_value(h, vvar, code) { khint_t __i;          \
 317         for (__i = kh_begin(h); __i != kh_end(h); ++__i) {              \
 318                 if (!kh_exist(h,__i)) continue;                                         \
 319                 (vvar) = kh_val(h,__i);                                                         \
 320                 code;                                                                                           \
 321         } }
 322
 323 #define __kh_oid_cmp(a, b) (hashcmp(a, b) == 0)
 324
 325 KHASH_INIT(sha1, const unsigned char *, void *, 1, sha1hash, __kh_oid_cmp)
 326 typedef kh_sha1_t khash_sha1;
 327
 328 KHASH_INIT(sha1_pos, const unsigned char *, int, 1, sha1hash, __kh_oid_cmp)
 329 typedef kh_sha1_pos_t khash_sha1_pos;
 330
 331 #endif /* __AC_KHASH_H */