src/ucl_hash.c

   1 /* Copyright (c) 2013, Vsevolod Stakhov
   2  * All rights reserved.
   3  *
   4  * Redistribution and use in source and binary forms, with or without
   5  * modification, are permitted provided that the following conditions are met:
   6  *       * Redistributions of source code must retain the above copyright
   7  *         notice, this list of conditions and the following disclaimer.
   8  *       * Redistributions in binary form must reproduce the above copyright
   9  *         notice, this list of conditions and the following disclaimer in the
  10  *         documentation and/or other materials provided with the distribution.
  11  *
  12  * THIS SOFTWARE IS PROVIDED ''AS IS'' AND ANY
  13  * EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED
  14  * WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
  15  * DISCLAIMED. IN NO EVENT SHALL AUTHOR BE LIABLE FOR ANY
  16  * DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES
  17  * (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES;
  18  * LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND
  19  * ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
  20  * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS
  21  * SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
  22  */
  23
  24 #include "ucl_internal.h"
  25 #include "ucl_hash.h"
  26 #include "khash.h"
  27 #include "kvec.h"
  28
  29 #include <time.h>
  30 #include <limits.h>
  31
  32 struct ucl_hash_elt {
  33         const ucl_object_t *obj;
  34         size_t ar_idx;
  35 };
  36
  37 struct ucl_hash_struct {
  38         void *hash;
  39         kvec_t(const ucl_object_t *) ar;
  40         bool caseless;
  41 };
  42
  43 static uint64_t
  44 ucl_hash_seed (void)
  45 {
  46         static uint64_t seed;
  47
  48         if (seed == 0) {
  49 #ifdef UCL_RANDOM_FUNCTION
  50                 seed = UCL_RANDOM_FUNCTION;
  51 #else
  52                 /* Not very random but can be useful for our purposes */
  53                 seed = time (NULL);
  54 #endif
  55         }
  56
  57         return seed;
  58 }
  59
  60 static const unsigned char lc_map[256] = {
  61                 0x00, 0x01, 0x02, 0x03, 0x04, 0x05, 0x06, 0x07,
  62                 0x08, 0x09, 0x0a, 0x0b, 0x0c, 0x0d, 0x0e, 0x0f,
  63                 0x10, 0x11, 0x12, 0x13, 0x14, 0x15, 0x16, 0x17,
  64                 0x18, 0x19, 0x1a, 0x1b, 0x1c, 0x1d, 0x1e, 0x1f,
  65                 0x20, 0x21, 0x22, 0x23, 0x24, 0x25, 0x26, 0x27,
  66                 0x28, 0x29, 0x2a, 0x2b, 0x2c, 0x2d, 0x2e, 0x2f,
  67                 0x30, 0x31, 0x32, 0x33, 0x34, 0x35, 0x36, 0x37,
  68                 0x38, 0x39, 0x3a, 0x3b, 0x3c, 0x3d, 0x3e, 0x3f,
  69                 0x40, 0x61, 0x62, 0x63, 0x64, 0x65, 0x66, 0x67,
  70                 0x68, 0x69, 0x6a, 0x6b, 0x6c, 0x6d, 0x6e, 0x6f,
  71                 0x70, 0x71, 0x72, 0x73, 0x74, 0x75, 0x76, 0x77,
  72                 0x78, 0x79, 0x7a, 0x5b, 0x5c, 0x5d, 0x5e, 0x5f,
  73                 0x60, 0x61, 0x62, 0x63, 0x64, 0x65, 0x66, 0x67,
  74                 0x68, 0x69, 0x6a, 0x6b, 0x6c, 0x6d, 0x6e, 0x6f,
  75                 0x70, 0x71, 0x72, 0x73, 0x74, 0x75, 0x76, 0x77,
  76                 0x78, 0x79, 0x7a, 0x7b, 0x7c, 0x7d, 0x7e, 0x7f,
  77                 0x80, 0x81, 0x82, 0x83, 0x84, 0x85, 0x86, 0x87,
  78                 0x88, 0x89, 0x8a, 0x8b, 0x8c, 0x8d, 0x8e, 0x8f,
  79                 0x90, 0x91, 0x92, 0x93, 0x94, 0x95, 0x96, 0x97,
  80                 0x98, 0x99, 0x9a, 0x9b, 0x9c, 0x9d, 0x9e, 0x9f,
  81                 0xa0, 0xa1, 0xa2, 0xa3, 0xa4, 0xa5, 0xa6, 0xa7,
  82                 0xa8, 0xa9, 0xaa, 0xab, 0xac, 0xad, 0xae, 0xaf,
  83                 0xb0, 0xb1, 0xb2, 0xb3, 0xb4, 0xb5, 0xb6, 0xb7,
  84                 0xb8, 0xb9, 0xba, 0xbb, 0xbc, 0xbd, 0xbe, 0xbf,
  85                 0xc0, 0xc1, 0xc2, 0xc3, 0xc4, 0xc5, 0xc6, 0xc7,
  86                 0xc8, 0xc9, 0xca, 0xcb, 0xcc, 0xcd, 0xce, 0xcf,
  87                 0xd0, 0xd1, 0xd2, 0xd3, 0xd4, 0xd5, 0xd6, 0xd7,
  88                 0xd8, 0xd9, 0xda, 0xdb, 0xdc, 0xdd, 0xde, 0xdf,
  89                 0xe0, 0xe1, 0xe2, 0xe3, 0xe4, 0xe5, 0xe6, 0xe7,
  90                 0xe8, 0xe9, 0xea, 0xeb, 0xec, 0xed, 0xee, 0xef,
  91                 0xf0, 0xf1, 0xf2, 0xf3, 0xf4, 0xf5, 0xf6, 0xf7,
  92                 0xf8, 0xf9, 0xfa, 0xfb, 0xfc, 0xfd, 0xfe, 0xff
  93 };
  94
  95 #if (defined(WORD_BIT) && WORD_BIT == 64) || \
  96         (defined(__WORDSIZE) && __WORDSIZE == 64) || \
  97         defined(__x86_64__) || \
  98         defined(__amd64__)
  99 #define UCL64_BIT_HASH 1
 100 #endif
 101
 102 #ifdef UCL64_BIT_HASH
 103 static inline uint32_t
 104 ucl_hash_func (const ucl_object_t *o)
 105 {
 106         return XXH64 (o->key, o->keylen, ucl_hash_seed ());
 107 }
 108 #else
 109 static inline uint32_t
 110 ucl_hash_func (const ucl_object_t *o)
 111 {
 112         return XXH32 (o->key, o->keylen, ucl_hash_seed ());
 113 }
 114 #endif
 115
 116 static inline int
 117 ucl_hash_equal (const ucl_object_t *k1, const ucl_object_t *k2)
 118 {
 119         if (k1->keylen == k2->keylen) {
 120                 return memcmp (k1->key, k2->key, k1->keylen) == 0;
 121         }
 122
 123         return 0;
 124 }
 125
 126 KHASH_INIT (ucl_hash_node, const ucl_object_t *, struct ucl_hash_elt, 1,
 127                 ucl_hash_func, ucl_hash_equal)
 128
 129 #ifdef UCL64_BIT_HASH
 130 static inline uint32_t
 131 ucl_hash_caseless_func (const ucl_object_t *o)
 132 {
 133         unsigned len = o->keylen;
 134         unsigned leftover = o->keylen % 4;
 135         unsigned fp, i;
 136         const uint8_t* s = (const uint8_t*)o->key;
 137         union {
 138                 struct {
 139                         unsigned char c1, c2, c3, c4;
 140                 } c;
 141                 uint32_t pp;
 142         } u;
 143         XXH64_state_t st;
 144
 145         fp = len - leftover;
 146         XXH64_reset (&st, ucl_hash_seed ());
 147
 148         for (i = 0; i != fp; i += 4) {
 149                 u.c.c1 = s[i], u.c.c2 = s[i + 1], u.c.c3 = s[i + 2], u.c.c4 = s[i + 3];
 150                 u.c.c1 = lc_map[u.c.c1];
 151                 u.c.c2 = lc_map[u.c.c2];
 152                 u.c.c3 = lc_map[u.c.c3];
 153                 u.c.c4 = lc_map[u.c.c4];
 154                 XXH64_update (&st, &u.pp, sizeof (u));
 155         }
 156
 157         u.pp = 0;
 158         switch (leftover) {
 159         case 3:
 160                 u.c.c3 = lc_map[(unsigned char)s[i++]];
 161         case 2:
 162                 u.c.c2 = lc_map[(unsigned char)s[i++]];
 163         case 1:
 164                 u.c.c1 = lc_map[(unsigned char)s[i]];
 165                 XXH64_update (&st, &u.pp, leftover);
 166                 break;
 167         }
 168
 169         return XXH64_digest (&st);
 170 }
 171 #else
 172 static inline uint32_t
 173 ucl_hash_caseless_func (const ucl_object_t *o)
 174 {
 175         unsigned len = o->keylen;
 176         unsigned leftover = o->keylen % 4;
 177         unsigned fp, i;
 178         const uint8_t* s = (const uint8_t*)o->key;
 179         union {
 180                 struct {
 181                         unsigned char c1, c2, c3, c4;
 182                 } c;
 183                 uint32_t pp;
 184         } u;
 185         XXH32_state_t st;
 186
 187         fp = len - leftover;
 188         XXH32_reset (&st, ucl_hash_seed ());
 189
 190         for (i = 0; i != fp; i += 4) {
 191                 u.c.c1 = s[i], u.c.c2 = s[i + 1], u.c.c3 = s[i + 2], u.c.c4 = s[i + 3];
 192                 u.c.c1 = lc_map[u.c.c1];
 193                 u.c.c2 = lc_map[u.c.c2];
 194                 u.c.c3 = lc_map[u.c.c3];
 195                 u.c.c4 = lc_map[u.c.c4];
 196                 XXH32_update (&st, &u.pp, sizeof (u));
 197         }
 198
 199         u.pp = 0;
 200         switch (leftover) {
 201         case 3:
 202                 u.c.c3 = lc_map[(unsigned char)s[i++]];
 203         case 2:
 204                 u.c.c2 = lc_map[(unsigned char)s[i++]];
 205         case 1:
 206                 u.c.c1 = lc_map[(unsigned char)s[i]];
 207                 XXH32_update (&st, &u.pp, leftover);
 208                 break;
 209         }
 210
 211         return XXH32_digest (&st);
 212 }
 213 #endif
 214
 215 static inline int
 216 ucl_hash_caseless_equal (const ucl_object_t *k1, const ucl_object_t *k2)
 217 {
 218         if (k1->keylen == k2->keylen) {
 219                 return memcmp (k1->key, k2->key, k1->keylen) == 0;
 220         }
 221
 222         return 0;
 223 }
 224
 225 KHASH_INIT (ucl_hash_caseless_node, const ucl_object_t *, struct ucl_hash_elt, 1,
 226                 ucl_hash_caseless_func, ucl_hash_caseless_equal)
 227
 228 ucl_hash_t*
 229 ucl_hash_create (bool ignore_case)
 230 {
 231         ucl_hash_t *new;
 232
 233         new = UCL_ALLOC (sizeof (ucl_hash_t));
 234         if (new != NULL) {
 235                 kv_init (new->ar);
 236
 237                 new->caseless = ignore_case;
 238                 if (ignore_case) {
 239                         khash_t(ucl_hash_caseless_node) *h = kh_init (ucl_hash_caseless_node);
 240                         new->hash = (void *)h;
 241                 }
 242                 else {
 243                         khash_t(ucl_hash_node) *h = kh_init (ucl_hash_node);
 244                         new->hash = (void *)h;
 245                 }
 246         }
 247         return new;
 248 }
 249
 250 void ucl_hash_destroy (ucl_hash_t* hashlin, ucl_hash_free_func func)
 251 {
 252         const ucl_object_t *cur, *tmp;
 253
 254         if (hashlin == NULL) {
 255                 return;
 256         }
 257
 258         if (func != NULL) {
 259                 /* Iterate over the hash first */
 260                 khash_t(ucl_hash_node) *h = (khash_t(ucl_hash_node) *)
 261                                 hashlin->hash;
 262                 khiter_t k;
 263
 264                 for (k = kh_begin (h); k != kh_end (h); ++k) {
 265                         if (kh_exist (h, k)) {
 266                                 cur = (kh_value (h, k)).obj;
 267                                 while (cur != NULL) {
 268                                         tmp = cur->next;
 269                                         func (__DECONST (ucl_object_t *, cur));
 270                                         cur = tmp;
 271                                 }
 272                         }
 273                 }
 274         }
 275
 276         if (hashlin->caseless) {
 277                 khash_t(ucl_hash_caseless_node) *h = (khash_t(ucl_hash_caseless_node) *)
 278                         hashlin->hash;
 279                 kh_destroy (ucl_hash_caseless_node, h);
 280         }
 281         else {
 282                 khash_t(ucl_hash_node) *h = (khash_t(ucl_hash_node) *)
 283                         hashlin->hash;
 284                 kh_destroy (ucl_hash_node, h);
 285         }
 286
 287         kv_destroy (hashlin->ar);
 288         UCL_FREE (sizeof (*hashlin), hashlin);
 289 }
 290
 291 void
 292 ucl_hash_insert (ucl_hash_t* hashlin, const ucl_object_t *obj,
 293                 const char *key, unsigned keylen)
 294 {
 295         khiter_t k;
 296         int ret;
 297         struct ucl_hash_elt *elt;
 298
 299         if (hashlin == NULL) {
 300                 return;
 301         }
 302
 303         if (hashlin->caseless) {
 304                 khash_t(ucl_hash_caseless_node) *h = (khash_t(ucl_hash_caseless_node) *)
 305                                 hashlin->hash;
 306                 k = kh_put (ucl_hash_caseless_node, h, obj, &ret);
 307                 if (ret > 0) {
 308                         elt = &kh_value (h, k);
 309                         kv_push (const ucl_object_t *, hashlin->ar, obj);
 310                         elt->obj = obj;
 311                         elt->ar_idx = kv_size (hashlin->ar) - 1;
 312                 }
 313         }
 314         else {
 315                 khash_t(ucl_hash_node) *h = (khash_t(ucl_hash_node) *)
 316                                 hashlin->hash;
 317                 k = kh_put (ucl_hash_node, h, obj, &ret);
 318                 if (ret > 0) {
 319                         elt = &kh_value (h, k);
 320                         kv_push (const ucl_object_t *, hashlin->ar, obj);
 321                         elt->obj = obj;
 322                         elt->ar_idx = kv_size (hashlin->ar) - 1;
 323                 }
 324         }
 325 }
 326
 327 void ucl_hash_replace (ucl_hash_t* hashlin, const ucl_object_t *old,
 328                 const ucl_object_t *new)
 329 {
 330         khiter_t k;
 331         int ret;
 332         struct ucl_hash_elt elt, *pelt;
 333
 334         if (hashlin == NULL) {
 335                 return;
 336         }
 337
 338         if (hashlin->caseless) {
 339                 khash_t(ucl_hash_caseless_node) *h = (khash_t(ucl_hash_caseless_node) *)
 340                                 hashlin->hash;
 341                 k = kh_put (ucl_hash_caseless_node, h, old, &ret);
 342                 if (ret == 0) {
 343                         elt = kh_value (h, k);
 344                         kh_del (ucl_hash_caseless_node, h, k);
 345                         k = kh_put (ucl_hash_caseless_node, h, new, &ret);
 346                         pelt = &kh_value (h, k);
 347                         pelt->obj = new;
 348                         pelt->ar_idx = elt.ar_idx;
 349                         kv_A (hashlin->ar, elt.ar_idx) = new;
 350                 }
 351         }
 352         else {
 353                 khash_t(ucl_hash_node) *h = (khash_t(ucl_hash_node) *)
 354                                 hashlin->hash;
 355                 k = kh_put (ucl_hash_node, h, old, &ret);
 356                 if (ret == 0) {
 357                         elt = kh_value (h, k);
 358                         kh_del (ucl_hash_node, h, k);
 359                         k = kh_put (ucl_hash_node, h, new, &ret);
 360                         pelt = &kh_value (h, k);
 361                         pelt->obj = new;
 362                         pelt->ar_idx = elt.ar_idx;
 363                         kv_A (hashlin->ar, elt.ar_idx) = new;
 364                 }
 365         }
 366 }
 367
 368 struct ucl_hash_real_iter {
 369         const ucl_object_t **cur;
 370         const ucl_object_t **end;
 371 };
 372
 373 const void*
 374 ucl_hash_iterate (ucl_hash_t *hashlin, ucl_hash_iter_t *iter)
 375 {
 376         struct ucl_hash_real_iter *it = (struct ucl_hash_real_iter *)(*iter);
 377         const ucl_object_t *ret = NULL;
 378
 379         if (hashlin == NULL) {
 380                 return NULL;
 381         }
 382
 383         if (it == NULL) {
 384                 it = UCL_ALLOC (sizeof (*it));
 385
 386                 if (it == NULL) {
 387                         return NULL;
 388                 }
 389
 390                 it->cur = &hashlin->ar.a[0];
 391                 it->end = it->cur + hashlin->ar.n;
 392         }
 393
 394         if (it->cur < it->end) {
 395                 ret = *it->cur++;
 396         }
 397         else {
 398                 UCL_FREE (sizeof (*it), it);
 399                 *iter = NULL;
 400                 return NULL;
 401         }
 402
 403         *iter = it;
 404
 405         return ret;
 406 }
 407
 408 bool
 409 ucl_hash_iter_has_next (ucl_hash_t *hashlin, ucl_hash_iter_t iter)
 410 {
 411         struct ucl_hash_real_iter *it = (struct ucl_hash_real_iter *)(iter);
 412
 413         return it->cur < it->end - 1;
 414 }
 415
 416
 417 const ucl_object_t*
 418 ucl_hash_search (ucl_hash_t* hashlin, const char *key, unsigned keylen)
 419 {
 420         khiter_t k;
 421         const ucl_object_t *ret = NULL;
 422         ucl_object_t search;
 423         struct ucl_hash_elt *elt;
 424
 425         search.key = key;
 426         search.keylen = keylen;
 427
 428         if (hashlin == NULL) {
 429                 return NULL;
 430         }
 431
 432         if (hashlin->caseless) {
 433                 khash_t(ucl_hash_caseless_node) *h = (khash_t(ucl_hash_caseless_node) *)
 434                                                 hashlin->hash;
 435
 436                 k = kh_get (ucl_hash_caseless_node, h, &search);
 437                 if (k != kh_end (h)) {
 438                         elt = &kh_value (h, k);
 439                         ret = elt->obj;
 440                 }
 441         }
 442         else {
 443                 khash_t(ucl_hash_node) *h = (khash_t(ucl_hash_node) *)
 444                                                 hashlin->hash;
 445                 k = kh_get (ucl_hash_node, h, &search);
 446                 if (k != kh_end (h)) {
 447                         elt = &kh_value (h, k);
 448                         ret = elt->obj;
 449                 }
 450         }
 451
 452         return ret;
 453 }
 454
 455 void
 456 ucl_hash_delete (ucl_hash_t* hashlin, const ucl_object_t *obj)
 457 {
 458         khiter_t k;
 459         struct ucl_hash_elt *elt;
 460
 461         if (hashlin == NULL) {
 462                 return;
 463         }
 464
 465         if (hashlin->caseless) {
 466                 khash_t(ucl_hash_caseless_node) *h = (khash_t(ucl_hash_caseless_node) *)
 467                         hashlin->hash;
 468
 469                 k = kh_get (ucl_hash_caseless_node, h, obj);
 470                 if (k != kh_end (h)) {
 471                         elt = &kh_value (h, k);
 472                         kv_del (const ucl_object_t *, hashlin->ar, elt->ar_idx);
 473                         kh_del (ucl_hash_caseless_node, h, k);
 474                 }
 475         }
 476         else {
 477                 khash_t(ucl_hash_node) *h = (khash_t(ucl_hash_node) *)
 478                         hashlin->hash;
 479                 k = kh_get (ucl_hash_node, h, obj);
 480                 if (k != kh_end (h)) {
 481                         elt = &kh_value (h, k);
 482                         kv_del (const ucl_object_t *, hashlin->ar, elt->ar_idx);
 483                         kh_del (ucl_hash_node, h, k);
 484                 }
 485         }
 486 }