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