include/mruby/khash.h

   1 /**
   2 ** @file mruby/khash.h - Hash for mruby
   3 **
   4 ** See Copyright Notice in mruby.h
   5 */
   6
   7 #ifndef MRUBY_KHASH_H
   8 #define MRUBY_KHASH_H
   9
  10 #include <string.h>
  11
  12 #include <mruby.h>
  13 #include "common.h"
  14
  15 /**
  16  * khash definitions used in mruby's hash table.
  17  */
  18 MRB_BEGIN_DECL
  19
  20 typedef uint32_t khint_t;
  21 typedef khint_t khiter_t;
  22
  23 #ifndef KHASH_DEFAULT_SIZE
  24 # define KHASH_DEFAULT_SIZE 32
  25 #endif
  26 #define KHASH_MIN_SIZE 8
  27
  28 #define UPPER_BOUND(x) ((x)>>2|(x)>>1)
  29
  30 /* extern uint8_t __m[]; */
  31
  32 /* mask for flags */
  33 static const uint8_t __m_empty[]  = {0x02, 0x08, 0x20, 0x80};
  34 static const uint8_t __m_del[]    = {0x01, 0x04, 0x10, 0x40};
  35 static const uint8_t __m_either[] = {0x03, 0x0c, 0x30, 0xc0};
  36
  37
  38 #define __ac_isempty(ed_flag, i) (ed_flag[(i)/4]&__m_empty[(i)%4])
  39 #define __ac_isdel(ed_flag, i) (ed_flag[(i)/4]&__m_del[(i)%4])
  40 #define __ac_iseither(ed_flag, i) (ed_flag[(i)/4]&__m_either[(i)%4])
  41 #define khash_power2(v) do { \
  42   v--;\
  43   v |= v >> 1;\
  44   v |= v >> 2;\
  45   v |= v >> 4;\
  46   v |= v >> 8;\
  47   v |= v >> 16;\
  48   v++;\
  49 } while (0)
  50 #define khash_mask(h) ((h)->n_buckets-1)
  51 #define khash_upper_bound(h) (UPPER_BOUND((h)->n_buckets))
  52
  53 /* declare struct kh_xxx and kh_xxx_funcs
  54
  55    name: hash name
  56    khkey_t: key data type
  57    khval_t: value data type
  58    kh_is_map: (0: hash set / 1: hash map)
  59 */
  60 #define KHASH_DECLARE(name, khkey_t, khval_t, kh_is_map)                \
  61   typedef struct kh_##name {                                            \
  62     khint_t n_buckets;                                                  \
  63     khint_t size;                                                       \
  64     uint8_t *ed_flags;                                                  \
  65     khkey_t *keys;                                                      \
  66     khval_t *vals;                                                      \
  67   } kh_##name##_t;                                                      \
  68   void kh_alloc_##name(mrb_state *mrb, kh_##name##_t *h);               \
  69   kh_##name##_t *kh_init_##name##_size(mrb_state *mrb, khint_t size);   \
  70   kh_##name##_t *kh_init_##name(mrb_state *mrb);                        \
  71   void kh_destroy_##name(mrb_state *mrb, kh_##name##_t *h);             \
  72   void kh_clear_##name(mrb_state *mrb, kh_##name##_t *h);               \
  73   khint_t kh_get_##name(mrb_state *mrb, kh_##name##_t *h, khkey_t key);           \
  74   khint_t kh_put_##name(mrb_state *mrb, kh_##name##_t *h, khkey_t key, int *ret); \
  75   void kh_resize_##name(mrb_state *mrb, kh_##name##_t *h, khint_t new_n_buckets); \
  76   void kh_del_##name(mrb_state *mrb, kh_##name##_t *h, khint_t x);                \
  77   kh_##name##_t *kh_copy_##name(mrb_state *mrb, kh_##name##_t *h);
  78
  79 static inline void
  80 kh_fill_flags(uint8_t *p, uint8_t c, size_t len)
  81 {
  82   while (len-- > 0) {
  83     *p++ = c;
  84   }
  85 }
  86
  87 /* define kh_xxx_funcs
  88
  89    name: hash name
  90    khkey_t: key data type
  91    khval_t: value data type
  92    kh_is_map: (0: hash set / 1: hash map)
  93    __hash_func: hash function
  94    __hash_equal: hash comparison function
  95 */
  96 #define KHASH_DEFINE(name, khkey_t, khval_t, kh_is_map, __hash_func, __hash_equal) \
  97   mrb_noreturn void mrb_raise_nomemory(mrb_state *mrb);                 \
  98   int kh_alloc_simple_##name(mrb_state *mrb, kh_##name##_t *h)          \
  99   {                                                                     \
 100     khint_t sz = h->n_buckets;                                          \
 101     size_t len = sizeof(khkey_t) + (kh_is_map ? sizeof(khval_t) : 0);   \
 102     uint8_t *p = (uint8_t*)mrb_malloc_simple(mrb, sizeof(uint8_t)*sz/4+len*sz); \
 103     if (!p) { return 1; }                                               \
 104     h->size = 0;                                                        \
 105     h->keys = (khkey_t *)p;                                             \
 106     h->vals = kh_is_map ? (khval_t *)(p+sizeof(khkey_t)*sz) : NULL;     \
 107     h->ed_flags = p+len*sz;                                             \
 108     kh_fill_flags(h->ed_flags, 0xaa, sz/4);                             \
 109     return 0;                                                           \
 110   }                                                                     \
 111   void kh_alloc_##name(mrb_state *mrb, kh_##name##_t *h)                \
 112   {                                                                     \
 113     if (kh_alloc_simple_##name(mrb, h)) {                               \
 114       mrb_raise_nomemory(mrb);                                          \
 115     }                                                                   \
 116   }                                                                     \
 117   kh_##name##_t *kh_init_##name##_size(mrb_state *mrb, khint_t size) {  \
 118     kh_##name##_t *h = (kh_##name##_t*)mrb_calloc(mrb, 1, sizeof(kh_##name##_t)); \
 119     if (size < KHASH_MIN_SIZE)                                          \
 120       size = KHASH_MIN_SIZE;                                            \
 121     khash_power2(size);                                                 \
 122     h->n_buckets = size;                                                \
 123     if (kh_alloc_simple_##name(mrb, h)) {                               \
 124       mrb_free(mrb, h);                                                 \
 125       mrb_raise_nomemory(mrb);                                          \
 126     }                                                                   \
 127     return h;                                                           \
 128   }                                                                     \
 129   kh_##name##_t *kh_init_##name(mrb_state *mrb) {                       \
 130     return kh_init_##name##_size(mrb, KHASH_DEFAULT_SIZE);              \
 131   }                                                                     \
 132   void kh_destroy_##name(mrb_state *mrb, kh_##name##_t *h)              \
 133   {                                                                     \
 134     if (h) {                                                            \
 135       mrb_free(mrb, h->keys);                                           \
 136       mrb_free(mrb, h);                                                 \
 137     }                                                                   \
 138   }                                                                     \
 139   void kh_clear_##name(mrb_state *mrb, kh_##name##_t *h)                \
 140   {                                                                     \
 141     (void)mrb;                                                          \
 142     if (h && h->ed_flags) {                                             \
 143       kh_fill_flags(h->ed_flags, 0xaa, h->n_buckets/4);                 \
 144       h->size = 0;                                                      \
 145     }                                                                   \
 146   }                                                                     \
 147   khint_t kh_get_##name(mrb_state *mrb, kh_##name##_t *h, khkey_t key)  \
 148   {                                                                     \
 149     khint_t k = __hash_func(mrb,key) & khash_mask(h), step = 0;         \
 150     (void)mrb;                                                          \
 151     while (!__ac_isempty(h->ed_flags, k)) {                             \
 152       if (!__ac_isdel(h->ed_flags, k)) {                                \
 153         if (__hash_equal(mrb,h->keys[k], key)) return k;                \
 154       }                                                                 \
 155       k = (k+(++step)) & khash_mask(h);                                 \
 156     }                                                                   \
 157     return kh_end(h);                                                   \
 158   }                                                                     \
 159   void kh_resize_##name(mrb_state *mrb, kh_##name##_t *h, khint_t new_n_buckets) \
 160   {                                                                     \
 161     if (new_n_buckets < KHASH_MIN_SIZE)                                 \
 162       new_n_buckets = KHASH_MIN_SIZE;                                   \
 163     khash_power2(new_n_buckets);                                        \
 164     {                                                                   \
 165       kh_##name##_t hh;                                                 \
 166       uint8_t *old_ed_flags = h->ed_flags;                              \
 167       khkey_t *old_keys = h->keys;                                      \
 168       khval_t *old_vals = h->vals;                                      \
 169       khint_t old_n_buckets = h->n_buckets;                             \
 170       khint_t i;                                                        \
 171       hh.n_buckets = new_n_buckets;                                     \
 172       kh_alloc_##name(mrb, &hh);                                        \
 173       /* relocate */                                                    \
 174       for (i=0 ; i<old_n_buckets ; i++) {                               \
 175         if (!__ac_iseither(old_ed_flags, i)) {                          \
 176           khint_t k = kh_put_##name(mrb, &hh, old_keys[i], NULL);       \
 177           if (kh_is_map) kh_value(&hh,k) = old_vals[i];                 \
 178         }                                                               \
 179       }                                                                 \
 180       /* copy hh to h */                                                \
 181       *h = hh;                                                          \
 182       mrb_free(mrb, old_keys);                                          \
 183     }                                                                   \
 184   }                                                                     \
 185   khint_t kh_put_##name(mrb_state *mrb, kh_##name##_t *h, khkey_t key, int *ret) \
 186   {                                                                     \
 187     khint_t k, del_k, step = 0;                                         \
 188     if (h->size >= khash_upper_bound(h)) {                              \
 189       kh_resize_##name(mrb, h, h->n_buckets*2);                         \
 190     }                                                                   \
 191     k = __hash_func(mrb,key) & khash_mask(h);                           \
 192     del_k = kh_end(h);                                                  \
 193     while (!__ac_isempty(h->ed_flags, k)) {                             \
 194       if (!__ac_isdel(h->ed_flags, k)) {                                \
 195         if (__hash_equal(mrb,h->keys[k], key)) {                        \
 196           if (ret) *ret = 0;                                            \
 197           return k;                                                     \
 198         }                                                               \
 199       }                                                                 \
 200       else if (del_k == kh_end(h)) {                                    \
 201         del_k = k;                                                      \
 202       }                                                                 \
 203       k = (k+(++step)) & khash_mask(h);                                 \
 204     }                                                                   \
 205     if (del_k != kh_end(h)) {                                           \
 206       /* put at del */                                                  \
 207       h->keys[del_k] = key;                                             \
 208       h->ed_flags[del_k/4] &= ~__m_del[del_k%4];                        \
 209       h->size++;                                                        \
 210       if (ret) *ret = 2;                                                \
 211       return del_k;                                                     \
 212     }                                                                   \
 213     else {                                                              \
 214       /* put at empty */                                                \
 215       h->keys[k] = key;                                                 \
 216       h->ed_flags[k/4] &= ~__m_empty[k%4];                              \
 217       h->size++;                                                        \
 218       if (ret) *ret = 1;                                                \
 219       return k;                                                         \
 220     }                                                                   \
 221   }                                                                     \
 222   void kh_del_##name(mrb_state *mrb, kh_##name##_t *h, khint_t x)       \
 223   {                                                                     \
 224     (void)mrb;                                                          \
 225     mrb_assert(x != h->n_buckets && !__ac_iseither(h->ed_flags, x));    \
 226     h->ed_flags[x/4] |= __m_del[x%4];                                   \
 227     h->size--;                                                          \
 228   }                                                                     \
 229   kh_##name##_t *kh_copy_##name(mrb_state *mrb, kh_##name##_t *h)       \
 230   {                                                                     \
 231     kh_##name##_t *h2;                                                  \
 232     khiter_t k, k2;                                                     \
 233                                                                         \
 234     h2 = kh_init_##name(mrb);                                           \
 235     for (k = kh_begin(h); k != kh_end(h); k++) {                        \
 236       if (kh_exist(h, k)) {                                             \
 237         k2 = kh_put_##name(mrb, h2, kh_key(h, k), NULL);                \
 238         if (kh_is_map) kh_value(h2, k2) = kh_value(h, k);               \
 239       }                                                                 \
 240     }                                                                   \
 241     return h2;                                                          \
 242   }
 243
 244
 245 #define khash_t(name) kh_##name##_t
 246
 247 #define kh_init_size(name,mrb,size) kh_init_##name##_size(mrb,size)
 248 #define kh_init(name,mrb) kh_init_##name(mrb)
 249 #define kh_destroy(name, mrb, h) kh_destroy_##name(mrb, h)
 250 #define kh_clear(name, mrb, h) kh_clear_##name(mrb, h)
 251 #define kh_resize(name, mrb, h, s) kh_resize_##name(mrb, h, s)
 252 #define kh_put(name, mrb, h, k) kh_put_##name(mrb, h, k, NULL)
 253 #define kh_put2(name, mrb, h, k, r) kh_put_##name(mrb, h, k, r)
 254 #define kh_get(name, mrb, h, k) kh_get_##name(mrb, h, k)
 255 #define kh_del(name, mrb, h, k) kh_del_##name(mrb, h, k)
 256 #define kh_copy(name, mrb, h) kh_copy_##name(mrb, h)
 257
 258 #define kh_exist(h, x) (!__ac_iseither((h)->ed_flags, (x)))
 259 #define kh_key(h, x) ((h)->keys[x])
 260 #define kh_val(h, x) ((h)->vals[x])
 261 #define kh_value(h, x) ((h)->vals[x])
 262 #define kh_begin(h) (khint_t)(0)
 263 #define kh_end(h) ((h)->n_buckets)
 264 #define kh_size(h) ((h)->size)
 265 #define kh_n_buckets(h) ((h)->n_buckets)
 266
 267 #define kh_int_hash_func(mrb,key) (khint_t)((key)^((key)<<2)^((key)>>2))
 268 #define kh_int_hash_equal(mrb,a, b) (a == b)
 269 #define kh_int64_hash_func(mrb,key) (khint_t)((key)>>33^(key)^(key)<<11)
 270 #define kh_int64_hash_equal(mrb,a, b) (a == b)
 271 static inline khint_t __ac_X31_hash_string(const char *s)
 272 {
 273     khint_t h = *s;
 274     if (h) for (++s ; *s; ++s) h = (h << 5) - h + *s;
 275     return h;
 276 }
 277 #define kh_str_hash_func(mrb,key) __ac_X31_hash_string(key)
 278 #define kh_str_hash_equal(mrb,a, b) (strcmp(a, b) == 0)
 279
 280 typedef const char *kh_cstr_t;
 281
 282 MRB_END_DECL
 283
 284 #endif  /* MRUBY_KHASH_H */