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                                         REALLOC_ARRAY(h->keys, new_n_buckets); \
 125                                         if (kh_is_map) {                                                                        \
 126                                                 REALLOC_ARRAY(h->vals, new_n_buckets); \
 127                                         }                                                                                                       \
 128                                 } /* otherwise shrink */                                                                \
 129                         }                                                                                                                       \
 130                 }                                                                                                                               \
 131                 if (j) { /* rehashing is needed */                                                              \
 132                         for (j = 0; j != h->n_buckets; ++j) {                                           \
 133                                 if (__ac_iseither(h->flags, j) == 0) {                                  \
 134                                         khkey_t key = h->keys[j];                                                       \
 135                                         khval_t val;                                                                            \
 136                                         khint_t new_mask;                                                                       \
 137                                         new_mask = new_n_buckets - 1;                                           \
 138                                         if (kh_is_map) val = h->vals[j];                                        \
 139                                         __ac_set_isdel_true(h->flags, j);                                       \
 140                                         while (1) { /* kick-out process; sort of like in Cuckoo hashing */ \
 141                                                 khint_t k, i, step = 0; \
 142                                                 k = __hash_func(key);                                                   \
 143                                                 i = k & new_mask;                                                               \
 144                                                 while (!__ac_isempty(new_flags, i)) i = (i + (++step)) & new_mask; \
 145                                                 __ac_set_isempty_false(new_flags, i);                   \
 146                                                 if (i < h->n_buckets && __ac_iseither(h->flags, i) == 0) { /* kick out the existing element */ \
 147                                                         { khkey_t tmp = h->keys[i]; h->keys[i] = key; key = tmp; } \
 148                                                         if (kh_is_map) { khval_t tmp = h->vals[i]; h->vals[i] = val; val = tmp; } \
 149                                                         __ac_set_isdel_true(h->flags, i); /* mark it as deleted in the old hash table */ \
 150                                                 } else { /* write the element and jump out of the loop */ \
 151                                                         h->keys[i] = key;                                                       \
 152                                                         if (kh_is_map) h->vals[i] = val;                        \
 153                                                         break;                                                                          \
 154                                                 }                                                                                               \
 155                                         }                                                                                                       \
 156                                 }                                                                                                               \
 157                         }                                                                                                                       \
 158                         if (h->n_buckets > new_n_buckets) { /* shrink the hash table */ \
 159                                 REALLOC_ARRAY(h->keys, new_n_buckets); \
 160                                 if (kh_is_map) REALLOC_ARRAY(h->vals, new_n_buckets); \
 161                         }                                                                                                                       \
 162                         free(h->flags); /* free the working space */                            \
 163                         h->flags = new_flags;                                                                           \
 164                         h->n_buckets = new_n_buckets;                                                           \
 165                         h->n_occupied = h->size;                                                                        \
 166                         h->upper_bound = (khint_t)(h->n_buckets * __ac_HASH_UPPER + 0.5); \
 167                 }                                                                                                                               \
 168                 return 0;                                                                                                               \
 169         }                                                                                                                                       \
 170         SCOPE khint_t kh_put_##name(kh_##name##_t *h, khkey_t key, int *ret) \
 171         {                                                                                                                                       \
 172                 khint_t x;                                                                                                              \
 173                 if (h->n_occupied >= h->upper_bound) { /* update the hash table */ \
 174                         if (h->n_buckets > (h->size<<1)) {                                                      \
 175                                 if (kh_resize_##name(h, h->n_buckets - 1) < 0) { /* clear "deleted" elements */ \
 176                                         *ret = -1; return h->n_buckets;                                         \
 177                                 }                                                                                                               \
 178                         } else if (kh_resize_##name(h, h->n_buckets + 1) < 0) { /* expand the hash table */ \
 179                                 *ret = -1; return h->n_buckets;                                                 \
 180                         }                                                                                                                       \
 181                 } /* TODO: to implement automatically shrinking; resize() already support shrinking */ \
 182                 {                                                                                                                               \
 183                         khint_t k, i, site, last, mask = h->n_buckets - 1, step = 0; \
 184                         x = site = h->n_buckets; k = __hash_func(key); i = k & mask; \
 185                         if (__ac_isempty(h->flags, i)) x = i; /* for speed up */        \
 186                         else {                                                                                                          \
 187                                 last = i; \
 188                                 while (!__ac_isempty(h->flags, i) && (__ac_isdel(h->flags, i) || !__hash_equal(h->keys[i], key))) { \
 189                                         if (__ac_isdel(h->flags, i)) site = i;                          \
 190                                         i = (i + (++step)) & mask; \
 191                                         if (i == last) { x = site; break; }                                     \
 192                                 }                                                                                                               \
 193                                 if (x == h->n_buckets) {                                                                \
 194                                         if (__ac_isempty(h->flags, i) && site != h->n_buckets) x = site; \
 195                                         else x = i;                                                                                     \
 196                                 }                                                                                                               \
 197                         }                                                                                                                       \
 198                 }                                                                                                                               \
 199                 if (__ac_isempty(h->flags, x)) { /* not present at all */               \
 200                         h->keys[x] = key;                                                                                       \
 201                         __ac_set_isboth_false(h->flags, x);                                                     \
 202                         ++h->size; ++h->n_occupied;                                                                     \
 203                         *ret = 1;                                                                                                       \
 204                 } else if (__ac_isdel(h->flags, x)) { /* deleted */                             \
 205                         h->keys[x] = key;                                                                                       \
 206                         __ac_set_isboth_false(h->flags, x);                                                     \
 207                         ++h->size;                                                                                                      \
 208                         *ret = 2;                                                                                                       \
 209                 } else *ret = 0; /* Don't touch h->keys[x] if present and not deleted */ \
 210                 return x;                                                                                                               \
 211         }                                                                                                                                       \
 212         SCOPE void kh_del_##name(kh_##name##_t *h, khint_t x)                           \
 213         {                                                                                                                                       \
 214                 if (x != h->n_buckets && !__ac_iseither(h->flags, x)) {                 \
 215                         __ac_set_isdel_true(h->flags, x);                                                       \
 216                         --h->size;                                                                                                      \
 217                 }                                                                                                                               \
 218         }
 219
 220 #define KHASH_DECLARE(name, khkey_t, khval_t)                                                   \
 221         __KHASH_TYPE(name, khkey_t, khval_t)                                                            \
 222         __KHASH_PROTOTYPES(name, khkey_t, khval_t)
 223
 224 #define KHASH_INIT2(name, SCOPE, khkey_t, khval_t, kh_is_map, __hash_func, __hash_equal) \
 225         __KHASH_TYPE(name, khkey_t, khval_t)                                                            \
 226         __KHASH_IMPL(name, SCOPE, khkey_t, khval_t, kh_is_map, __hash_func, __hash_equal)
 227
 228 #define KHASH_INIT(name, khkey_t, khval_t, kh_is_map, __hash_func, __hash_equal) \
 229         KHASH_INIT2(name, static inline, khkey_t, khval_t, kh_is_map, __hash_func, __hash_equal)
 230
 231 /* Other convenient macros... */
 232
 233 /*! @function
 234   @abstract     Test whether a bucket contains data.
 235   @param  h     Pointer to the hash table [khash_t(name)*]
 236   @param  x     Iterator to the bucket [khint_t]
 237   @return       1 if containing data; 0 otherwise [int]
 238  */
 239 #define kh_exist(h, x) (!__ac_iseither((h)->flags, (x)))
 240
 241 /*! @function
 242   @abstract     Get key given an iterator
 243   @param  h     Pointer to the hash table [khash_t(name)*]
 244   @param  x     Iterator to the bucket [khint_t]
 245   @return       Key [type of keys]
 246  */
 247 #define kh_key(h, x) ((h)->keys[x])
 248
 249 /*! @function
 250   @abstract     Get value given an iterator
 251   @param  h     Pointer to the hash table [khash_t(name)*]
 252   @param  x     Iterator to the bucket [khint_t]
 253   @return       Value [type of values]
 254   @discussion   For hash sets, calling this results in segfault.
 255  */
 256 #define kh_val(h, x) ((h)->vals[x])
 257
 258 /*! @function
 259   @abstract     Alias of kh_val()
 260  */
 261 #define kh_value(h, x) ((h)->vals[x])
 262
 263 /*! @function
 264   @abstract     Get the start iterator
 265   @param  h     Pointer to the hash table [khash_t(name)*]
 266   @return       The start iterator [khint_t]
 267  */
 268 #define kh_begin(h) (khint_t)(0)
 269
 270 /*! @function
 271   @abstract     Get the end iterator
 272   @param  h     Pointer to the hash table [khash_t(name)*]
 273   @return       The end iterator [khint_t]
 274  */
 275 #define kh_end(h) ((h)->n_buckets)
 276
 277 /*! @function
 278   @abstract     Get the number of elements in the hash table
 279   @param  h     Pointer to the hash table [khash_t(name)*]
 280   @return       Number of elements in the hash table [khint_t]
 281  */
 282 #define kh_size(h) ((h)->size)
 283
 284 /*! @function
 285   @abstract     Get the number of buckets in the hash table
 286   @param  h     Pointer to the hash table [khash_t(name)*]
 287   @return       Number of buckets in the hash table [khint_t]
 288  */
 289 #define kh_n_buckets(h) ((h)->n_buckets)
 290
 291 /*! @function
 292   @abstract     Iterate over the entries in the hash table
 293   @param  h     Pointer to the hash table [khash_t(name)*]
 294   @param  kvar  Variable to which key will be assigned
 295   @param  vvar  Variable to which value will be assigned
 296   @param  code  Block of code to execute
 297  */
 298 #define kh_foreach(h, kvar, vvar, code) { khint_t __i;          \
 299         for (__i = kh_begin(h); __i != kh_end(h); ++__i) {              \
 300                 if (!kh_exist(h,__i)) continue;                                         \
 301                 (kvar) = kh_key(h,__i);                                                         \
 302                 (vvar) = kh_val(h,__i);                                                         \
 303                 code;                                                                                           \
 304         } }
 305
 306 /*! @function
 307   @abstract     Iterate over the values in the hash table
 308   @param  h     Pointer to the hash table [khash_t(name)*]
 309   @param  vvar  Variable to which value will be assigned
 310   @param  code  Block of code to execute
 311  */
 312 #define kh_foreach_value(h, vvar, code) { khint_t __i;          \
 313         for (__i = kh_begin(h); __i != kh_end(h); ++__i) {              \
 314                 if (!kh_exist(h,__i)) continue;                                         \
 315                 (vvar) = kh_val(h,__i);                                                         \
 316                 code;                                                                                           \
 317         } }
 318
 319 #define __kh_oid_cmp(a, b) (hashcmp(a, b) == 0)
 320
 321 KHASH_INIT(sha1, const unsigned char *, void *, 1, sha1hash, __kh_oid_cmp)
 322 typedef kh_sha1_t khash_sha1;
 323
 324 KHASH_INIT(sha1_pos, const unsigned char *, int, 1, sha1hash, __kh_oid_cmp)
 325 typedef kh_sha1_pos_t khash_sha1_pos;
 326
 327 #endif /* __AC_KHASH_H */