Mercurial > urweb

annotate include/urweb/uthash.h @ 2250:c275bbc41194

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
Start work on pure expression caching.
author Ziv Scully <ziv@mit.edu>
date Sun, 13 Sep 2015 16:02:45 -0400
parents 2f7ed04332a0
children
rev   line source
ziv@2234 1 /*
ziv@2234 2 Copyright (c) 2003-2014, Troy D. Hanson http://troydhanson.github.com/uthash/
ziv@2234 3 All rights reserved.
ziv@2234 4
ziv@2234 5 Redistribution and use in source and binary forms, with or without
ziv@2234 6 modification, are permitted provided that the following conditions are met:
ziv@2234 7
ziv@2234 8 * Redistributions of source code must retain the above copyright
ziv@2234 9 notice, this list of conditions and the following disclaimer.
ziv@2234 10
ziv@2234 11 THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS
ziv@2234 12 IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED
ziv@2234 13 TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A
ziv@2234 14 PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER
ziv@2234 15 OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL,
ziv@2234 16 EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO,
ziv@2234 17 PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
ziv@2234 18 PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
ziv@2234 19 LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING
ziv@2234 20 NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS
ziv@2234 21 SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
ziv@2234 22 */
ziv@2234 23
ziv@2234 24 #ifndef UTHASH_H
ziv@2234 25 #define UTHASH_H
ziv@2234 26
ziv@2234 27 #include <string.h> /* memcmp,strlen */
ziv@2234 28 #include <stddef.h> /* ptrdiff_t */
ziv@2234 29 #include <stdlib.h> /* exit() */
ziv@2234 30
ziv@2234 31 /* These macros use decltype or the earlier __typeof GNU extension.
ziv@2234 32 As decltype is only available in newer compilers (VS2010 or gcc 4.3+
ziv@2234 33 when compiling c++ source) this code uses whatever method is needed
ziv@2234 34 or, for VS2008 where neither is available, uses casting workarounds. */
ziv@2234 35 #if defined(_MSC_VER) /* MS compiler */
ziv@2234 36 #if _MSC_VER >= 1600 && defined(__cplusplus) /* VS2010 or newer in C++ mode */
ziv@2234 37 #define DECLTYPE(x) (decltype(x))
ziv@2234 38 #else /* VS2008 or older (or VS2010 in C mode) */
ziv@2234 39 #define NO_DECLTYPE
ziv@2234 40 #define DECLTYPE(x)
ziv@2234 41 #endif
ziv@2234 42 #elif defined(__BORLANDC__) || defined(__LCC__) || defined(__WATCOMC__)
ziv@2234 43 #define NO_DECLTYPE
ziv@2234 44 #define DECLTYPE(x)
ziv@2234 45 #else /* GNU, Sun and other compilers */
ziv@2234 46 #define DECLTYPE(x) (__typeof(x))
ziv@2234 47 #endif
ziv@2234 48
ziv@2234 49 #ifdef NO_DECLTYPE
ziv@2234 50 #define DECLTYPE_ASSIGN(dst,src) \
ziv@2234 51 do { \
ziv@2234 52 char **_da_dst = (char**)(&(dst)); \
ziv@2234 53 *_da_dst = (char*)(src); \
ziv@2234 54 } while(0)
ziv@2234 55 #else
ziv@2234 56 #define DECLTYPE_ASSIGN(dst,src) \
ziv@2234 57 do { \
ziv@2234 58 (dst) = DECLTYPE(dst)(src); \
ziv@2234 59 } while(0)
ziv@2234 60 #endif
ziv@2234 61
ziv@2234 62 /* a number of the hash function use uint32_t which isn't defined on Pre VS2010 */
ziv@2234 63 #if defined (_WIN32)
ziv@2234 64 #if defined(_MSC_VER) && _MSC_VER >= 1600
ziv@2234 65 #include <stdint.h>
ziv@2234 66 #elif defined(__WATCOMC__)
ziv@2234 67 #include <stdint.h>
ziv@2234 68 #else
ziv@2234 69 typedef unsigned int uint32_t;
ziv@2234 70 typedef unsigned char uint8_t;
ziv@2234 71 #endif
ziv@2234 72 #else
ziv@2234 73 #include <stdint.h>
ziv@2234 74 #endif
ziv@2234 75
ziv@2234 76 #define UTHASH_VERSION 1.9.9
ziv@2234 77
ziv@2234 78 #ifndef uthash_fatal
ziv@2234 79 #define uthash_fatal(msg) exit(-1) /* fatal error (out of memory,etc) */
ziv@2234 80 #endif
ziv@2234 81 #ifndef uthash_malloc
ziv@2234 82 #define uthash_malloc(sz) malloc(sz) /* malloc fcn */
ziv@2234 83 #endif
ziv@2234 84 #ifndef uthash_free
ziv@2234 85 #define uthash_free(ptr,sz) free(ptr) /* free fcn */
ziv@2234 86 #endif
ziv@2234 87
ziv@2234 88 #ifndef uthash_noexpand_fyi
ziv@2234 89 #define uthash_noexpand_fyi(tbl) /* can be defined to log noexpand */
ziv@2234 90 #endif
ziv@2234 91 #ifndef uthash_expand_fyi
ziv@2234 92 #define uthash_expand_fyi(tbl) /* can be defined to log expands */
ziv@2234 93 #endif
ziv@2234 94
ziv@2234 95 /* initial number of buckets */
ziv@2234 96 #define HASH_INITIAL_NUM_BUCKETS 32U /* initial number of buckets */
ziv@2234 97 #define HASH_INITIAL_NUM_BUCKETS_LOG2 5U /* lg2 of initial number of buckets */
ziv@2234 98 #define HASH_BKT_CAPACITY_THRESH 10U /* expand when bucket count reaches */
ziv@2234 99
ziv@2234 100 /* calculate the element whose hash handle address is hhe */
ziv@2234 101 #define ELMT_FROM_HH(tbl,hhp) ((void*)(((char*)(hhp)) - ((tbl)->hho)))
ziv@2234 102
ziv@2234 103 #define HASH_FIND(hh,head,keyptr,keylen,out) \
ziv@2234 104 do { \
ziv@2234 105 out=NULL; \
ziv@2234 106 if (head != NULL) { \
ziv@2234 107 unsigned _hf_bkt,_hf_hashv; \
ziv@2234 108 HASH_FCN(keyptr,keylen, (head)->hh.tbl->num_buckets, _hf_hashv, _hf_bkt); \
ziv@2234 109 if (HASH_BLOOM_TEST((head)->hh.tbl, _hf_hashv) != 0) { \
ziv@2234 110 HASH_FIND_IN_BKT((head)->hh.tbl, hh, (head)->hh.tbl->buckets[ _hf_bkt ], \
ziv@2234 111 keyptr,keylen,out); \
ziv@2234 112 } \
ziv@2234 113 } \
ziv@2234 114 } while (0)
ziv@2234 115
ziv@2234 116 #ifdef HASH_BLOOM
ziv@2234 117 #define HASH_BLOOM_BITLEN (1UL << HASH_BLOOM)
ziv@2234 118 #define HASH_BLOOM_BYTELEN (HASH_BLOOM_BITLEN/8UL) + (((HASH_BLOOM_BITLEN%8UL)!=0UL) ? 1UL : 0UL)
ziv@2234 119 #define HASH_BLOOM_MAKE(tbl) \
ziv@2234 120 do { \
ziv@2234 121 (tbl)->bloom_nbits = HASH_BLOOM; \
ziv@2234 122 (tbl)->bloom_bv = (uint8_t*)uthash_malloc(HASH_BLOOM_BYTELEN); \
ziv@2234 123 if (!((tbl)->bloom_bv)) { uthash_fatal( "out of memory"); } \
ziv@2234 124 memset((tbl)->bloom_bv, 0, HASH_BLOOM_BYTELEN); \
ziv@2234 125 (tbl)->bloom_sig = HASH_BLOOM_SIGNATURE; \
ziv@2234 126 } while (0)
ziv@2234 127
ziv@2234 128 #define HASH_BLOOM_FREE(tbl) \
ziv@2234 129 do { \
ziv@2234 130 uthash_free((tbl)->bloom_bv, HASH_BLOOM_BYTELEN); \
ziv@2234 131 } while (0)
ziv@2234 132
ziv@2234 133 #define HASH_BLOOM_BITSET(bv,idx) (bv[(idx)/8U] |= (1U << ((idx)%8U)))
ziv@2234 134 #define HASH_BLOOM_BITTEST(bv,idx) (bv[(idx)/8U] & (1U << ((idx)%8U)))
ziv@2234 135
ziv@2234 136 #define HASH_BLOOM_ADD(tbl,hashv) \
ziv@2234 137 HASH_BLOOM_BITSET((tbl)->bloom_bv, (hashv & (uint32_t)((1ULL << (tbl)->bloom_nbits) - 1U)))
ziv@2234 138
ziv@2234 139 #define HASH_BLOOM_TEST(tbl,hashv) \
ziv@2234 140 HASH_BLOOM_BITTEST((tbl)->bloom_bv, (hashv & (uint32_t)((1ULL << (tbl)->bloom_nbits) - 1U)))
ziv@2234 141
ziv@2234 142 #else
ziv@2234 143 #define HASH_BLOOM_MAKE(tbl)
ziv@2234 144 #define HASH_BLOOM_FREE(tbl)
ziv@2234 145 #define HASH_BLOOM_ADD(tbl,hashv)
ziv@2234 146 #define HASH_BLOOM_TEST(tbl,hashv) (1)
ziv@2234 147 #define HASH_BLOOM_BYTELEN 0U
ziv@2234 148 #endif
ziv@2234 149
ziv@2234 150 #define HASH_MAKE_TABLE(hh,head) \
ziv@2234 151 do { \
ziv@2234 152 (head)->hh.tbl = (UT_hash_table*)uthash_malloc( \
ziv@2234 153 sizeof(UT_hash_table)); \
ziv@2234 154 if (!((head)->hh.tbl)) { uthash_fatal( "out of memory"); } \
ziv@2234 155 memset((head)->hh.tbl, 0, sizeof(UT_hash_table)); \
ziv@2234 156 (head)->hh.tbl->tail = &((head)->hh); \
ziv@2234 157 (head)->hh.tbl->num_buckets = HASH_INITIAL_NUM_BUCKETS; \
ziv@2234 158 (head)->hh.tbl->log2_num_buckets = HASH_INITIAL_NUM_BUCKETS_LOG2; \
ziv@2234 159 (head)->hh.tbl->hho = (char*)(&(head)->hh) - (char*)(head); \
ziv@2234 160 (head)->hh.tbl->buckets = (UT_hash_bucket*)uthash_malloc( \
ziv@2234 161 HASH_INITIAL_NUM_BUCKETS*sizeof(struct UT_hash_bucket)); \
ziv@2234 162 if (! (head)->hh.tbl->buckets) { uthash_fatal( "out of memory"); } \
ziv@2234 163 memset((head)->hh.tbl->buckets, 0, \
ziv@2234 164 HASH_INITIAL_NUM_BUCKETS*sizeof(struct UT_hash_bucket)); \
ziv@2234 165 HASH_BLOOM_MAKE((head)->hh.tbl); \
ziv@2234 166 (head)->hh.tbl->signature = HASH_SIGNATURE; \
ziv@2234 167 } while(0)
ziv@2234 168
ziv@2234 169 #define HASH_ADD(hh,head,fieldname,keylen_in,add) \
ziv@2234 170 HASH_ADD_KEYPTR(hh,head,&((add)->fieldname),keylen_in,add)
ziv@2234 171
ziv@2234 172 #define HASH_REPLACE(hh,head,fieldname,keylen_in,add,replaced) \
ziv@2234 173 do { \
ziv@2234 174 replaced=NULL; \
ziv@2234 175 HASH_FIND(hh,head,&((add)->fieldname),keylen_in,replaced); \
ziv@2234 176 if (replaced!=NULL) { \
ziv@2234 177 HASH_DELETE(hh,head,replaced); \
ziv@2234 178 } \
ziv@2234 179 HASH_ADD(hh,head,fieldname,keylen_in,add); \
ziv@2234 180 } while(0)
ziv@2234 181
ziv@2234 182 #define HASH_ADD_KEYPTR(hh,head,keyptr,keylen_in,add) \
ziv@2234 183 do { \
ziv@2234 184 unsigned _ha_bkt; \
ziv@2234 185 (add)->hh.next = NULL; \
ziv@2234 186 (add)->hh.key = (char*)(keyptr); \
ziv@2234 187 (add)->hh.keylen = (unsigned)(keylen_in); \
ziv@2234 188 if (!(head)) { \
ziv@2234 189 head = (add); \
ziv@2234 190 (head)->hh.prev = NULL; \
ziv@2234 191 HASH_MAKE_TABLE(hh,head); \
ziv@2234 192 } else { \
ziv@2234 193 (head)->hh.tbl->tail->next = (add); \
ziv@2234 194 (add)->hh.prev = ELMT_FROM_HH((head)->hh.tbl, (head)->hh.tbl->tail); \
ziv@2234 195 (head)->hh.tbl->tail = &((add)->hh); \
ziv@2234 196 } \
ziv@2234 197 (head)->hh.tbl->num_items++; \
ziv@2234 198 (add)->hh.tbl = (head)->hh.tbl; \
ziv@2234 199 HASH_FCN(keyptr,keylen_in, (head)->hh.tbl->num_buckets, \
ziv@2234 200 (add)->hh.hashv, _ha_bkt); \
ziv@2234 201 HASH_ADD_TO_BKT((head)->hh.tbl->buckets[_ha_bkt],&(add)->hh); \
ziv@2234 202 HASH_BLOOM_ADD((head)->hh.tbl,(add)->hh.hashv); \
ziv@2234 203 HASH_EMIT_KEY(hh,head,keyptr,keylen_in); \
ziv@2234 204 HASH_FSCK(hh,head); \
ziv@2234 205 } while(0)
ziv@2234 206
ziv@2234 207 #define HASH_TO_BKT( hashv, num_bkts, bkt ) \
ziv@2234 208 do { \
ziv@2234 209 bkt = ((hashv) & ((num_bkts) - 1U)); \
ziv@2234 210 } while(0)
ziv@2234 211
ziv@2234 212 /* delete "delptr" from the hash table.
ziv@2234 213 * "the usual" patch-up process for the app-order doubly-linked-list.
ziv@2234 214 * The use of _hd_hh_del below deserves special explanation.
ziv@2234 215 * These used to be expressed using (delptr) but that led to a bug
ziv@2234 216 * if someone used the same symbol for the head and deletee, like
ziv@2234 217 * HASH_DELETE(hh,users,users);
ziv@2234 218 * We want that to work, but by changing the head (users) below
ziv@2234 219 * we were forfeiting our ability to further refer to the deletee (users)
ziv@2234 220 * in the patch-up process. Solution: use scratch space to
ziv@2234 221 * copy the deletee pointer, then the latter references are via that
ziv@2234 222 * scratch pointer rather than through the repointed (users) symbol.
ziv@2234 223 */
ziv@2234 224 #define HASH_DELETE(hh,head,delptr) \
ziv@2234 225 do { \
ziv@2234 226 struct UT_hash_handle *_hd_hh_del; \
ziv@2234 227 if ( ((delptr)->hh.prev == NULL) && ((delptr)->hh.next == NULL) ) { \
ziv@2234 228 uthash_free((head)->hh.tbl->buckets, \
ziv@2234 229 (head)->hh.tbl->num_buckets*sizeof(struct UT_hash_bucket) ); \
ziv@2234 230 HASH_BLOOM_FREE((head)->hh.tbl); \
ziv@2234 231 uthash_free((head)->hh.tbl, sizeof(UT_hash_table)); \
ziv@2234 232 head = NULL; \
ziv@2234 233 } else { \
ziv@2234 234 unsigned _hd_bkt; \
ziv@2234 235 _hd_hh_del = &((delptr)->hh); \
ziv@2234 236 if ((delptr) == ELMT_FROM_HH((head)->hh.tbl,(head)->hh.tbl->tail)) { \
ziv@2234 237 (head)->hh.tbl->tail = \
ziv@2234 238 (UT_hash_handle*)((ptrdiff_t)((delptr)->hh.prev) + \
ziv@2234 239 (head)->hh.tbl->hho); \
ziv@2234 240 } \
ziv@2234 241 if ((delptr)->hh.prev != NULL) { \
ziv@2234 242 ((UT_hash_handle*)((ptrdiff_t)((delptr)->hh.prev) + \
ziv@2234 243 (head)->hh.tbl->hho))->next = (delptr)->hh.next; \
ziv@2234 244 } else { \
ziv@2234 245 DECLTYPE_ASSIGN(head,(delptr)->hh.next); \
ziv@2234 246 } \
ziv@2234 247 if (_hd_hh_del->next != NULL) { \
ziv@2234 248 ((UT_hash_handle*)((ptrdiff_t)_hd_hh_del->next + \
ziv@2234 249 (head)->hh.tbl->hho))->prev = \
ziv@2234 250 _hd_hh_del->prev; \
ziv@2234 251 } \
ziv@2234 252 HASH_TO_BKT( _hd_hh_del->hashv, (head)->hh.tbl->num_buckets, _hd_bkt); \
ziv@2234 253 HASH_DEL_IN_BKT(hh,(head)->hh.tbl->buckets[_hd_bkt], _hd_hh_del); \
ziv@2234 254 (head)->hh.tbl->num_items--; \
ziv@2234 255 } \
ziv@2234 256 HASH_FSCK(hh,head); \
ziv@2234 257 } while (0)
ziv@2234 258
ziv@2234 259
ziv@2234 260 /* convenience forms of HASH_FIND/HASH_ADD/HASH_DEL */
ziv@2234 261 #define HASH_FIND_STR(head,findstr,out) \
ziv@2234 262 HASH_FIND(hh,head,findstr,(unsigned)strlen(findstr),out)
ziv@2234 263 #define HASH_ADD_STR(head,strfield,add) \
ziv@2234 264 HASH_ADD(hh,head,strfield[0],(unsigned int)strlen(add->strfield),add)
ziv@2234 265 #define HASH_REPLACE_STR(head,strfield,add,replaced) \
ziv@2234 266 HASH_REPLACE(hh,head,strfield[0],(unsigned)strlen(add->strfield),add,replaced)
ziv@2234 267 #define HASH_FIND_INT(head,findint,out) \
ziv@2234 268 HASH_FIND(hh,head,findint,sizeof(int),out)
ziv@2234 269 #define HASH_ADD_INT(head,intfield,add) \
ziv@2234 270 HASH_ADD(hh,head,intfield,sizeof(int),add)
ziv@2234 271 #define HASH_REPLACE_INT(head,intfield,add,replaced) \
ziv@2234 272 HASH_REPLACE(hh,head,intfield,sizeof(int),add,replaced)
ziv@2234 273 #define HASH_FIND_PTR(head,findptr,out) \
ziv@2234 274 HASH_FIND(hh,head,findptr,sizeof(void *),out)
ziv@2234 275 #define HASH_ADD_PTR(head,ptrfield,add) \
ziv@2234 276 HASH_ADD(hh,head,ptrfield,sizeof(void *),add)
ziv@2234 277 #define HASH_REPLACE_PTR(head,ptrfield,add,replaced) \
ziv@2234 278 HASH_REPLACE(hh,head,ptrfield,sizeof(void *),add,replaced)
ziv@2234 279 #define HASH_DEL(head,delptr) \
ziv@2234 280 HASH_DELETE(hh,head,delptr)
ziv@2234 281
ziv@2234 282 /* HASH_FSCK checks hash integrity on every add/delete when HASH_DEBUG is defined.
ziv@2234 283 * This is for uthash developer only; it compiles away if HASH_DEBUG isn't defined.
ziv@2234 284 */
ziv@2234 285 #ifdef HASH_DEBUG
ziv@2234 286 #define HASH_OOPS(...) do { fprintf(stderr,__VA_ARGS__); exit(-1); } while (0)
ziv@2234 287 #define HASH_FSCK(hh,head) \
ziv@2234 288 do { \
ziv@2234 289 struct UT_hash_handle *_thh; \
ziv@2234 290 if (head) { \
ziv@2234 291 unsigned _bkt_i; \
ziv@2234 292 unsigned _count; \
ziv@2234 293 char *_prev; \
ziv@2234 294 _count = 0; \
ziv@2234 295 for( _bkt_i = 0; _bkt_i < (head)->hh.tbl->num_buckets; _bkt_i++) { \
ziv@2234 296 unsigned _bkt_count = 0; \
ziv@2234 297 _thh = (head)->hh.tbl->buckets[_bkt_i].hh_head; \
ziv@2234 298 _prev = NULL; \
ziv@2234 299 while (_thh) { \
ziv@2234 300 if (_prev != (char*)(_thh->hh_prev)) { \
ziv@2234 301 HASH_OOPS("invalid hh_prev %p, actual %p\n", \
ziv@2234 302 _thh->hh_prev, _prev ); \
ziv@2234 303 } \
ziv@2234 304 _bkt_count++; \
ziv@2234 305 _prev = (char*)(_thh); \
ziv@2234 306 _thh = _thh->hh_next; \
ziv@2234 307 } \
ziv@2234 308 _count += _bkt_count; \
ziv@2234 309 if ((head)->hh.tbl->buckets[_bkt_i].count != _bkt_count) { \
ziv@2234 310 HASH_OOPS("invalid bucket count %u, actual %u\n", \
ziv@2234 311 (head)->hh.tbl->buckets[_bkt_i].count, _bkt_count); \
ziv@2234 312 } \
ziv@2234 313 } \
ziv@2234 314 if (_count != (head)->hh.tbl->num_items) { \
ziv@2234 315 HASH_OOPS("invalid hh item count %u, actual %u\n", \
ziv@2234 316 (head)->hh.tbl->num_items, _count ); \
ziv@2234 317 } \
ziv@2234 318 /* traverse hh in app order; check next/prev integrity, count */ \
ziv@2234 319 _count = 0; \
ziv@2234 320 _prev = NULL; \
ziv@2234 321 _thh = &(head)->hh; \
ziv@2234 322 while (_thh) { \
ziv@2234 323 _count++; \
ziv@2234 324 if (_prev !=(char*)(_thh->prev)) { \
ziv@2234 325 HASH_OOPS("invalid prev %p, actual %p\n", \
ziv@2234 326 _thh->prev, _prev ); \
ziv@2234 327 } \
ziv@2234 328 _prev = (char*)ELMT_FROM_HH((head)->hh.tbl, _thh); \
ziv@2234 329 _thh = ( _thh->next ? (UT_hash_handle*)((char*)(_thh->next) + \
ziv@2234 330 (head)->hh.tbl->hho) : NULL ); \
ziv@2234 331 } \
ziv@2234 332 if (_count != (head)->hh.tbl->num_items) { \
ziv@2234 333 HASH_OOPS("invalid app item count %u, actual %u\n", \
ziv@2234 334 (head)->hh.tbl->num_items, _count ); \
ziv@2234 335 } \
ziv@2234 336 } \
ziv@2234 337 } while (0)
ziv@2234 338 #else
ziv@2234 339 #define HASH_FSCK(hh,head)
ziv@2234 340 #endif
ziv@2234 341
ziv@2234 342 /* When compiled with -DHASH_EMIT_KEYS, length-prefixed keys are emitted to
ziv@2234 343 * the descriptor to which this macro is defined for tuning the hash function.
ziv@2234 344 * The app can #include <unistd.h> to get the prototype for write(2). */
ziv@2234 345 #ifdef HASH_EMIT_KEYS
ziv@2234 346 #define HASH_EMIT_KEY(hh,head,keyptr,fieldlen) \
ziv@2234 347 do { \
ziv@2234 348 unsigned _klen = fieldlen; \
ziv@2234 349 write(HASH_EMIT_KEYS, &_klen, sizeof(_klen)); \
ziv@2234 350 write(HASH_EMIT_KEYS, keyptr, (unsigned long)fieldlen); \
ziv@2234 351 } while (0)
ziv@2234 352 #else
ziv@2234 353 #define HASH_EMIT_KEY(hh,head,keyptr,fieldlen)
ziv@2234 354 #endif
ziv@2234 355
ziv@2234 356 /* default to Jenkin's hash unless overridden e.g. DHASH_FUNCTION=HASH_SAX */
ziv@2234 357 #ifdef HASH_FUNCTION
ziv@2234 358 #define HASH_FCN HASH_FUNCTION
ziv@2234 359 #else
ziv@2234 360 #define HASH_FCN HASH_JEN
ziv@2234 361 #endif
ziv@2234 362
ziv@2234 363 /* The Bernstein hash function, used in Perl prior to v5.6. Note (x<<5+x)=x*33. */
ziv@2234 364 #define HASH_BER(key,keylen,num_bkts,hashv,bkt) \
ziv@2234 365 do { \
ziv@2234 366 unsigned _hb_keylen=(unsigned)keylen; \
ziv@2234 367 const unsigned char *_hb_key=(const unsigned char*)(key); \
ziv@2234 368 (hashv) = 0; \
ziv@2234 369 while (_hb_keylen-- != 0U) { \
ziv@2234 370 (hashv) = (((hashv) << 5) + (hashv)) + *_hb_key++; \
ziv@2234 371 } \
ziv@2234 372 bkt = (hashv) & (num_bkts-1U); \
ziv@2234 373 } while (0)
ziv@2234 374
ziv@2234 375
ziv@2234 376 /* SAX/FNV/OAT/JEN hash functions are macro variants of those listed at
ziv@2234 377 * http://eternallyconfuzzled.com/tuts/algorithms/jsw_tut_hashing.aspx */
ziv@2234 378 #define HASH_SAX(key,keylen,num_bkts,hashv,bkt) \
ziv@2234 379 do { \
ziv@2234 380 unsigned _sx_i; \
ziv@2234 381 const unsigned char *_hs_key=(const unsigned char*)(key); \
ziv@2234 382 hashv = 0; \
ziv@2234 383 for(_sx_i=0; _sx_i < keylen; _sx_i++) { \
ziv@2234 384 hashv ^= (hashv << 5) + (hashv >> 2) + _hs_key[_sx_i]; \
ziv@2234 385 } \
ziv@2234 386 bkt = hashv & (num_bkts-1U); \
ziv@2234 387 } while (0)
ziv@2234 388 /* FNV-1a variation */
ziv@2234 389 #define HASH_FNV(key,keylen,num_bkts,hashv,bkt) \
ziv@2234 390 do { \
ziv@2234 391 unsigned _fn_i; \
ziv@2234 392 const unsigned char *_hf_key=(const unsigned char*)(key); \
ziv@2234 393 hashv = 2166136261U; \
ziv@2234 394 for(_fn_i=0; _fn_i < keylen; _fn_i++) { \
ziv@2234 395 hashv = hashv ^ _hf_key[_fn_i]; \
ziv@2234 396 hashv = hashv * 16777619U; \
ziv@2234 397 } \
ziv@2234 398 bkt = hashv & (num_bkts-1U); \
ziv@2234 399 } while(0)
ziv@2234 400
ziv@2234 401 #define HASH_OAT(key,keylen,num_bkts,hashv,bkt) \
ziv@2234 402 do { \
ziv@2234 403 unsigned _ho_i; \
ziv@2234 404 const unsigned char *_ho_key=(const unsigned char*)(key); \
ziv@2234 405 hashv = 0; \
ziv@2234 406 for(_ho_i=0; _ho_i < keylen; _ho_i++) { \
ziv@2234 407 hashv += _ho_key[_ho_i]; \
ziv@2234 408 hashv += (hashv << 10); \
ziv@2234 409 hashv ^= (hashv >> 6); \
ziv@2234 410 } \
ziv@2234 411 hashv += (hashv << 3); \
ziv@2234 412 hashv ^= (hashv >> 11); \
ziv@2234 413 hashv += (hashv << 15); \
ziv@2234 414 bkt = hashv & (num_bkts-1U); \
ziv@2234 415 } while(0)
ziv@2234 416
ziv@2234 417 #define HASH_JEN_MIX(a,b,c) \
ziv@2234 418 do { \
ziv@2234 419 a -= b; a -= c; a ^= ( c >> 13 ); \
ziv@2234 420 b -= c; b -= a; b ^= ( a << 8 ); \
ziv@2234 421 c -= a; c -= b; c ^= ( b >> 13 ); \
ziv@2234 422 a -= b; a -= c; a ^= ( c >> 12 ); \
ziv@2234 423 b -= c; b -= a; b ^= ( a << 16 ); \
ziv@2234 424 c -= a; c -= b; c ^= ( b >> 5 ); \
ziv@2234 425 a -= b; a -= c; a ^= ( c >> 3 ); \
ziv@2234 426 b -= c; b -= a; b ^= ( a << 10 ); \
ziv@2234 427 c -= a; c -= b; c ^= ( b >> 15 ); \
ziv@2234 428 } while (0)
ziv@2234 429
ziv@2234 430 #define HASH_JEN(key,keylen,num_bkts,hashv,bkt) \
ziv@2234 431 do { \
ziv@2234 432 unsigned _hj_i,_hj_j,_hj_k; \
ziv@2234 433 unsigned const char *_hj_key=(unsigned const char*)(key); \
ziv@2234 434 hashv = 0xfeedbeefu; \
ziv@2234 435 _hj_i = _hj_j = 0x9e3779b9u; \
ziv@2234 436 _hj_k = (unsigned)(keylen); \
ziv@2234 437 while (_hj_k >= 12U) { \
ziv@2234 438 _hj_i += (_hj_key[0] + ( (unsigned)_hj_key[1] << 8 ) \
ziv@2234 439 + ( (unsigned)_hj_key[2] << 16 ) \
ziv@2234 440 + ( (unsigned)_hj_key[3] << 24 ) ); \
ziv@2234 441 _hj_j += (_hj_key[4] + ( (unsigned)_hj_key[5] << 8 ) \
ziv@2234 442 + ( (unsigned)_hj_key[6] << 16 ) \
ziv@2234 443 + ( (unsigned)_hj_key[7] << 24 ) ); \
ziv@2234 444 hashv += (_hj_key[8] + ( (unsigned)_hj_key[9] << 8 ) \
ziv@2234 445 + ( (unsigned)_hj_key[10] << 16 ) \
ziv@2234 446 + ( (unsigned)_hj_key[11] << 24 ) ); \
ziv@2234 447 \
ziv@2234 448 HASH_JEN_MIX(_hj_i, _hj_j, hashv); \
ziv@2234 449 \
ziv@2234 450 _hj_key += 12; \
ziv@2234 451 _hj_k -= 12U; \
ziv@2234 452 } \
ziv@2234 453 hashv += (unsigned)(keylen); \
ziv@2234 454 switch ( _hj_k ) { \
ziv@2234 455 case 11: hashv += ( (unsigned)_hj_key[10] << 24 ); /* FALLTHROUGH */ \
ziv@2234 456 case 10: hashv += ( (unsigned)_hj_key[9] << 16 ); /* FALLTHROUGH */ \
ziv@2234 457 case 9: hashv += ( (unsigned)_hj_key[8] << 8 ); /* FALLTHROUGH */ \
ziv@2234 458 case 8: _hj_j += ( (unsigned)_hj_key[7] << 24 ); /* FALLTHROUGH */ \
ziv@2234 459 case 7: _hj_j += ( (unsigned)_hj_key[6] << 16 ); /* FALLTHROUGH */ \
ziv@2234 460 case 6: _hj_j += ( (unsigned)_hj_key[5] << 8 ); /* FALLTHROUGH */ \
ziv@2234 461 case 5: _hj_j += _hj_key[4]; /* FALLTHROUGH */ \
ziv@2234 462 case 4: _hj_i += ( (unsigned)_hj_key[3] << 24 ); /* FALLTHROUGH */ \
ziv@2234 463 case 3: _hj_i += ( (unsigned)_hj_key[2] << 16 ); /* FALLTHROUGH */ \
ziv@2234 464 case 2: _hj_i += ( (unsigned)_hj_key[1] << 8 ); /* FALLTHROUGH */ \
ziv@2234 465 case 1: _hj_i += _hj_key[0]; \
ziv@2234 466 } \
ziv@2234 467 HASH_JEN_MIX(_hj_i, _hj_j, hashv); \
ziv@2234 468 bkt = hashv & (num_bkts-1U); \
ziv@2234 469 } while(0)
ziv@2234 470
ziv@2234 471 /* The Paul Hsieh hash function */
ziv@2234 472 #undef get16bits
ziv@2234 473 #if (defined(__GNUC__) && defined(__i386__)) || defined(__WATCOMC__) \
ziv@2234 474 || defined(_MSC_VER) || defined (__BORLANDC__) || defined (__TURBOC__)
ziv@2234 475 #define get16bits(d) (*((const uint16_t *) (d)))
ziv@2234 476 #endif
ziv@2234 477
ziv@2234 478 #if !defined (get16bits)
ziv@2234 479 #define get16bits(d) ((((uint32_t)(((const uint8_t *)(d))[1])) << 8) \
ziv@2234 480 +(uint32_t)(((const uint8_t *)(d))[0]) )
ziv@2234 481 #endif
ziv@2234 482 #define HASH_SFH(key,keylen,num_bkts,hashv,bkt) \
ziv@2234 483 do { \
ziv@2234 484 unsigned const char *_sfh_key=(unsigned const char*)(key); \
ziv@2234 485 uint32_t _sfh_tmp, _sfh_len = (uint32_t)keylen; \
ziv@2234 486 \
ziv@2234 487 unsigned _sfh_rem = _sfh_len & 3U; \
ziv@2234 488 _sfh_len >>= 2; \
ziv@2234 489 hashv = 0xcafebabeu; \
ziv@2234 490 \
ziv@2234 491 /* Main loop */ \
ziv@2234 492 for (;_sfh_len > 0U; _sfh_len--) { \
ziv@2234 493 hashv += get16bits (_sfh_key); \
ziv@2234 494 _sfh_tmp = ((uint32_t)(get16bits (_sfh_key+2)) << 11) ^ hashv; \
ziv@2234 495 hashv = (hashv << 16) ^ _sfh_tmp; \
ziv@2234 496 _sfh_key += 2U*sizeof (uint16_t); \
ziv@2234 497 hashv += hashv >> 11; \
ziv@2234 498 } \
ziv@2234 499 \
ziv@2234 500 /* Handle end cases */ \
ziv@2234 501 switch (_sfh_rem) { \
ziv@2234 502 case 3: hashv += get16bits (_sfh_key); \
ziv@2234 503 hashv ^= hashv << 16; \
ziv@2234 504 hashv ^= (uint32_t)(_sfh_key[sizeof (uint16_t)]) << 18; \
ziv@2234 505 hashv += hashv >> 11; \
ziv@2234 506 break; \
ziv@2234 507 case 2: hashv += get16bits (_sfh_key); \
ziv@2234 508 hashv ^= hashv << 11; \
ziv@2234 509 hashv += hashv >> 17; \
ziv@2234 510 break; \
ziv@2234 511 case 1: hashv += *_sfh_key; \
ziv@2234 512 hashv ^= hashv << 10; \
ziv@2234 513 hashv += hashv >> 1; \
ziv@2234 514 } \
ziv@2234 515 \
ziv@2234 516 /* Force "avalanching" of final 127 bits */ \
ziv@2234 517 hashv ^= hashv << 3; \
ziv@2234 518 hashv += hashv >> 5; \
ziv@2234 519 hashv ^= hashv << 4; \
ziv@2234 520 hashv += hashv >> 17; \
ziv@2234 521 hashv ^= hashv << 25; \
ziv@2234 522 hashv += hashv >> 6; \
ziv@2234 523 bkt = hashv & (num_bkts-1U); \
ziv@2234 524 } while(0)
ziv@2234 525
ziv@2234 526 #ifdef HASH_USING_NO_STRICT_ALIASING
ziv@2234 527 /* The MurmurHash exploits some CPU's (x86,x86_64) tolerance for unaligned reads.
ziv@2234 528 * For other types of CPU's (e.g. Sparc) an unaligned read causes a bus error.
ziv@2234 529 * MurmurHash uses the faster approach only on CPU's where we know it's safe.
ziv@2234 530 *
ziv@2234 531 * Note the preprocessor built-in defines can be emitted using:
ziv@2234 532 *
ziv@2234 533 * gcc -m64 -dM -E - < /dev/null (on gcc)
ziv@2234 534 * cc -## a.c (where a.c is a simple test file) (Sun Studio)
ziv@2234 535 */
ziv@2234 536 #if (defined(__i386__) || defined(__x86_64__) || defined(_M_IX86))
ziv@2234 537 #define MUR_GETBLOCK(p,i) p[i]
ziv@2234 538 #else /* non intel */
ziv@2234 539 #define MUR_PLUS0_ALIGNED(p) (((unsigned long)p & 3UL) == 0UL)
ziv@2234 540 #define MUR_PLUS1_ALIGNED(p) (((unsigned long)p & 3UL) == 1UL)
ziv@2234 541 #define MUR_PLUS2_ALIGNED(p) (((unsigned long)p & 3UL) == 2UL)
ziv@2234 542 #define MUR_PLUS3_ALIGNED(p) (((unsigned long)p & 3UL) == 3UL)
ziv@2234 543 #define WP(p) ((uint32_t*)((unsigned long)(p) & ~3UL))
ziv@2234 544 #if (defined(__BIG_ENDIAN__) || defined(SPARC) || defined(__ppc__) || defined(__ppc64__))
ziv@2234 545 #define MUR_THREE_ONE(p) ((((*WP(p))&0x00ffffff) << 8) | (((*(WP(p)+1))&0xff000000) >> 24))
ziv@2234 546 #define MUR_TWO_TWO(p) ((((*WP(p))&0x0000ffff) <<16) | (((*(WP(p)+1))&0xffff0000) >> 16))
ziv@2234 547 #define MUR_ONE_THREE(p) ((((*WP(p))&0x000000ff) <<24) | (((*(WP(p)+1))&0xffffff00) >> 8))
ziv@2234 548 #else /* assume little endian non-intel */
ziv@2234 549 #define MUR_THREE_ONE(p) ((((*WP(p))&0xffffff00) >> 8) | (((*(WP(p)+1))&0x000000ff) << 24))
ziv@2234 550 #define MUR_TWO_TWO(p) ((((*WP(p))&0xffff0000) >>16) | (((*(WP(p)+1))&0x0000ffff) << 16))
ziv@2234 551 #define MUR_ONE_THREE(p) ((((*WP(p))&0xff000000) >>24) | (((*(WP(p)+1))&0x00ffffff) << 8))
ziv@2234 552 #endif
ziv@2234 553 #define MUR_GETBLOCK(p,i) (MUR_PLUS0_ALIGNED(p) ? ((p)[i]) : \
ziv@2234 554 (MUR_PLUS1_ALIGNED(p) ? MUR_THREE_ONE(p) : \
ziv@2234 555 (MUR_PLUS2_ALIGNED(p) ? MUR_TWO_TWO(p) : \
ziv@2234 556 MUR_ONE_THREE(p))))
ziv@2234 557 #endif
ziv@2234 558 #define MUR_ROTL32(x,r) (((x) << (r)) | ((x) >> (32 - (r))))
ziv@2234 559 #define MUR_FMIX(_h) \
ziv@2234 560 do { \
ziv@2234 561 _h ^= _h >> 16; \
ziv@2234 562 _h *= 0x85ebca6bu; \
ziv@2234 563 _h ^= _h >> 13; \
ziv@2234 564 _h *= 0xc2b2ae35u; \
ziv@2234 565 _h ^= _h >> 16; \
ziv@2234 566 } while(0)
ziv@2234 567
ziv@2234 568 #define HASH_MUR(key,keylen,num_bkts,hashv,bkt) \
ziv@2234 569 do { \
ziv@2234 570 const uint8_t *_mur_data = (const uint8_t*)(key); \
ziv@2234 571 const int _mur_nblocks = (int)(keylen) / 4; \
ziv@2234 572 uint32_t _mur_h1 = 0xf88D5353u; \
ziv@2234 573 uint32_t _mur_c1 = 0xcc9e2d51u; \
ziv@2234 574 uint32_t _mur_c2 = 0x1b873593u; \
ziv@2234 575 uint32_t _mur_k1 = 0; \
ziv@2234 576 const uint8_t *_mur_tail; \
ziv@2234 577 const uint32_t *_mur_blocks = (const uint32_t*)(_mur_data+(_mur_nblocks*4)); \
ziv@2234 578 int _mur_i; \
ziv@2234 579 for(_mur_i = -_mur_nblocks; _mur_i!=0; _mur_i++) { \
ziv@2234 580 _mur_k1 = MUR_GETBLOCK(_mur_blocks,_mur_i); \
ziv@2234 581 _mur_k1 *= _mur_c1; \
ziv@2234 582 _mur_k1 = MUR_ROTL32(_mur_k1,15); \
ziv@2234 583 _mur_k1 *= _mur_c2; \
ziv@2234 584 \
ziv@2234 585 _mur_h1 ^= _mur_k1; \
ziv@2234 586 _mur_h1 = MUR_ROTL32(_mur_h1,13); \
ziv@2234 587 _mur_h1 = (_mur_h1*5U) + 0xe6546b64u; \
ziv@2234 588 } \
ziv@2234 589 _mur_tail = (const uint8_t*)(_mur_data + (_mur_nblocks*4)); \
ziv@2234 590 _mur_k1=0; \
ziv@2234 591 switch((keylen) & 3U) { \
ziv@2234 592 case 3: _mur_k1 ^= (uint32_t)_mur_tail[2] << 16; /* FALLTHROUGH */ \
ziv@2234 593 case 2: _mur_k1 ^= (uint32_t)_mur_tail[1] << 8; /* FALLTHROUGH */ \
ziv@2234 594 case 1: _mur_k1 ^= (uint32_t)_mur_tail[0]; \
ziv@2234 595 _mur_k1 *= _mur_c1; \
ziv@2234 596 _mur_k1 = MUR_ROTL32(_mur_k1,15); \
ziv@2234 597 _mur_k1 *= _mur_c2; \
ziv@2234 598 _mur_h1 ^= _mur_k1; \
ziv@2234 599 } \
ziv@2234 600 _mur_h1 ^= (uint32_t)(keylen); \
ziv@2234 601 MUR_FMIX(_mur_h1); \
ziv@2234 602 hashv = _mur_h1; \
ziv@2234 603 bkt = hashv & (num_bkts-1U); \
ziv@2234 604 } while(0)
ziv@2234 605 #endif /* HASH_USING_NO_STRICT_ALIASING */
ziv@2234 606
ziv@2234 607 /* key comparison function; return 0 if keys equal */
ziv@2234 608 #define HASH_KEYCMP(a,b,len) memcmp(a,b,(unsigned long)(len))
ziv@2234 609
ziv@2234 610 /* iterate over items in a known bucket to find desired item */
ziv@2234 611 #define HASH_FIND_IN_BKT(tbl,hh,head,keyptr,keylen_in,out) \
ziv@2234 612 do { \
ziv@2234 613 if (head.hh_head != NULL) { DECLTYPE_ASSIGN(out,ELMT_FROM_HH(tbl,head.hh_head)); } \
ziv@2234 614 else { out=NULL; } \
ziv@2234 615 while (out != NULL) { \
ziv@2234 616 if ((out)->hh.keylen == (keylen_in)) { \
ziv@2234 617 if ((HASH_KEYCMP((out)->hh.key,keyptr,keylen_in)) == 0) { break; } \
ziv@2234 618 } \
ziv@2234 619 if ((out)->hh.hh_next != NULL) { DECLTYPE_ASSIGN(out,ELMT_FROM_HH(tbl,(out)->hh.hh_next)); } \
ziv@2234 620 else { out = NULL; } \
ziv@2234 621 } \
ziv@2234 622 } while(0)
ziv@2234 623
ziv@2234 624 /* add an item to a bucket */
ziv@2234 625 #define HASH_ADD_TO_BKT(head,addhh) \
ziv@2234 626 do { \
ziv@2234 627 head.count++; \
ziv@2234 628 (addhh)->hh_next = head.hh_head; \
ziv@2234 629 (addhh)->hh_prev = NULL; \
ziv@2234 630 if (head.hh_head != NULL) { (head).hh_head->hh_prev = (addhh); } \
ziv@2234 631 (head).hh_head=addhh; \
ziv@2234 632 if ((head.count >= ((head.expand_mult+1U) * HASH_BKT_CAPACITY_THRESH)) \
ziv@2234 633 && ((addhh)->tbl->noexpand != 1U)) { \
ziv@2234 634 HASH_EXPAND_BUCKETS((addhh)->tbl); \
ziv@2234 635 } \
ziv@2234 636 } while(0)
ziv@2234 637
ziv@2234 638 /* remove an item from a given bucket */
ziv@2234 639 #define HASH_DEL_IN_BKT(hh,head,hh_del) \
ziv@2234 640 (head).count--; \
ziv@2234 641 if ((head).hh_head == hh_del) { \
ziv@2234 642 (head).hh_head = hh_del->hh_next; \
ziv@2234 643 } \
ziv@2234 644 if (hh_del->hh_prev) { \
ziv@2234 645 hh_del->hh_prev->hh_next = hh_del->hh_next; \
ziv@2234 646 } \
ziv@2234 647 if (hh_del->hh_next) { \
ziv@2234 648 hh_del->hh_next->hh_prev = hh_del->hh_prev; \
ziv@2234 649 }
ziv@2234 650
ziv@2234 651 /* Bucket expansion has the effect of doubling the number of buckets
ziv@2234 652 * and redistributing the items into the new buckets. Ideally the
ziv@2234 653 * items will distribute more or less evenly into the new buckets
ziv@2234 654 * (the extent to which this is true is a measure of the quality of
ziv@2234 655 * the hash function as it applies to the key domain).
ziv@2234 656 *
ziv@2234 657 * With the items distributed into more buckets, the chain length
ziv@2234 658 * (item count) in each bucket is reduced. Thus by expanding buckets
ziv@2234 659 * the hash keeps a bound on the chain length. This bounded chain
ziv@2234 660 * length is the essence of how a hash provides constant time lookup.
ziv@2234 661 *
ziv@2234 662 * The calculation of tbl->ideal_chain_maxlen below deserves some
ziv@2234 663 * explanation. First, keep in mind that we're calculating the ideal
ziv@2234 664 * maximum chain length based on the *new* (doubled) bucket count.
ziv@2234 665 * In fractions this is just n/b (n=number of items,b=new num buckets).
ziv@2234 666 * Since the ideal chain length is an integer, we want to calculate
ziv@2234 667 * ceil(n/b). We don't depend on floating point arithmetic in this
ziv@2234 668 * hash, so to calculate ceil(n/b) with integers we could write
ziv@2234 669 *
ziv@2234 670 * ceil(n/b) = (n/b) + ((n%b)?1:0)
ziv@2234 671 *
ziv@2234 672 * and in fact a previous version of this hash did just that.
ziv@2234 673 * But now we have improved things a bit by recognizing that b is
ziv@2234 674 * always a power of two. We keep its base 2 log handy (call it lb),
ziv@2234 675 * so now we can write this with a bit shift and logical AND:
ziv@2234 676 *
ziv@2234 677 * ceil(n/b) = (n>>lb) + ( (n & (b-1)) ? 1:0)
ziv@2234 678 *
ziv@2234 679 */
ziv@2234 680 #define HASH_EXPAND_BUCKETS(tbl) \
ziv@2234 681 do { \
ziv@2234 682 unsigned _he_bkt; \
ziv@2234 683 unsigned _he_bkt_i; \
ziv@2234 684 struct UT_hash_handle *_he_thh, *_he_hh_nxt; \
ziv@2234 685 UT_hash_bucket *_he_new_buckets, *_he_newbkt; \
ziv@2234 686 _he_new_buckets = (UT_hash_bucket*)uthash_malloc( \
ziv@2234 687 2UL * tbl->num_buckets * sizeof(struct UT_hash_bucket)); \
ziv@2234 688 if (!_he_new_buckets) { uthash_fatal( "out of memory"); } \
ziv@2234 689 memset(_he_new_buckets, 0, \
ziv@2234 690 2UL * tbl->num_buckets * sizeof(struct UT_hash_bucket)); \
ziv@2234 691 tbl->ideal_chain_maxlen = \
ziv@2234 692 (tbl->num_items >> (tbl->log2_num_buckets+1U)) + \
ziv@2234 693 (((tbl->num_items & ((tbl->num_buckets*2U)-1U)) != 0U) ? 1U : 0U); \
ziv@2234 694 tbl->nonideal_items = 0; \
ziv@2234 695 for(_he_bkt_i = 0; _he_bkt_i < tbl->num_buckets; _he_bkt_i++) \
ziv@2234 696 { \
ziv@2234 697 _he_thh = tbl->buckets[ _he_bkt_i ].hh_head; \
ziv@2234 698 while (_he_thh != NULL) { \
ziv@2234 699 _he_hh_nxt = _he_thh->hh_next; \
ziv@2234 700 HASH_TO_BKT( _he_thh->hashv, tbl->num_buckets*2U, _he_bkt); \
ziv@2234 701 _he_newbkt = &(_he_new_buckets[ _he_bkt ]); \
ziv@2234 702 if (++(_he_newbkt->count) > tbl->ideal_chain_maxlen) { \
ziv@2234 703 tbl->nonideal_items++; \
ziv@2234 704 _he_newbkt->expand_mult = _he_newbkt->count / \
ziv@2234 705 tbl->ideal_chain_maxlen; \
ziv@2234 706 } \
ziv@2234 707 _he_thh->hh_prev = NULL; \
ziv@2234 708 _he_thh->hh_next = _he_newbkt->hh_head; \
ziv@2234 709 if (_he_newbkt->hh_head != NULL) { _he_newbkt->hh_head->hh_prev = \
ziv@2234 710 _he_thh; } \
ziv@2234 711 _he_newbkt->hh_head = _he_thh; \
ziv@2234 712 _he_thh = _he_hh_nxt; \
ziv@2234 713 } \
ziv@2234 714 } \
ziv@2234 715 uthash_free( tbl->buckets, tbl->num_buckets*sizeof(struct UT_hash_bucket) ); \
ziv@2234 716 tbl->num_buckets *= 2U; \
ziv@2234 717 tbl->log2_num_buckets++; \
ziv@2234 718 tbl->buckets = _he_new_buckets; \
ziv@2234 719 tbl->ineff_expands = (tbl->nonideal_items > (tbl->num_items >> 1)) ? \
ziv@2234 720 (tbl->ineff_expands+1U) : 0U; \
ziv@2234 721 if (tbl->ineff_expands > 1U) { \
ziv@2234 722 tbl->noexpand=1; \
ziv@2234 723 uthash_noexpand_fyi(tbl); \
ziv@2234 724 } \
ziv@2234 725 uthash_expand_fyi(tbl); \
ziv@2234 726 } while(0)
ziv@2234 727
ziv@2234 728
ziv@2234 729 /* This is an adaptation of Simon Tatham's O(n log(n)) mergesort */
ziv@2234 730 /* Note that HASH_SORT assumes the hash handle name to be hh.
ziv@2234 731 * HASH_SRT was added to allow the hash handle name to be passed in. */
ziv@2234 732 #define HASH_SORT(head,cmpfcn) HASH_SRT(hh,head,cmpfcn)
ziv@2234 733 #define HASH_SRT(hh,head,cmpfcn) \
ziv@2234 734 do { \
ziv@2234 735 unsigned _hs_i; \
ziv@2234 736 unsigned _hs_looping,_hs_nmerges,_hs_insize,_hs_psize,_hs_qsize; \
ziv@2234 737 struct UT_hash_handle *_hs_p, *_hs_q, *_hs_e, *_hs_list, *_hs_tail; \
ziv@2234 738 if (head != NULL) { \
ziv@2234 739 _hs_insize = 1; \
ziv@2234 740 _hs_looping = 1; \
ziv@2234 741 _hs_list = &((head)->hh); \
ziv@2234 742 while (_hs_looping != 0U) { \
ziv@2234 743 _hs_p = _hs_list; \
ziv@2234 744 _hs_list = NULL; \
ziv@2234 745 _hs_tail = NULL; \
ziv@2234 746 _hs_nmerges = 0; \
ziv@2234 747 while (_hs_p != NULL) { \
ziv@2234 748 _hs_nmerges++; \
ziv@2234 749 _hs_q = _hs_p; \
ziv@2234 750 _hs_psize = 0; \
ziv@2234 751 for ( _hs_i = 0; _hs_i < _hs_insize; _hs_i++ ) { \
ziv@2234 752 _hs_psize++; \
ziv@2234 753 _hs_q = (UT_hash_handle*)((_hs_q->next != NULL) ? \
ziv@2234 754 ((void*)((char*)(_hs_q->next) + \
ziv@2234 755 (head)->hh.tbl->hho)) : NULL); \
ziv@2234 756 if (! (_hs_q) ) { break; } \
ziv@2234 757 } \
ziv@2234 758 _hs_qsize = _hs_insize; \
ziv@2234 759 while ((_hs_psize > 0U) || ((_hs_qsize > 0U) && (_hs_q != NULL))) {\
ziv@2234 760 if (_hs_psize == 0U) { \
ziv@2234 761 _hs_e = _hs_q; \
ziv@2234 762 _hs_q = (UT_hash_handle*)((_hs_q->next != NULL) ? \
ziv@2234 763 ((void*)((char*)(_hs_q->next) + \
ziv@2234 764 (head)->hh.tbl->hho)) : NULL); \
ziv@2234 765 _hs_qsize--; \
ziv@2234 766 } else if ( (_hs_qsize == 0U) || (_hs_q == NULL) ) { \
ziv@2234 767 _hs_e = _hs_p; \
ziv@2234 768 if (_hs_p != NULL){ \
ziv@2234 769 _hs_p = (UT_hash_handle*)((_hs_p->next != NULL) ? \
ziv@2234 770 ((void*)((char*)(_hs_p->next) + \
ziv@2234 771 (head)->hh.tbl->hho)) : NULL); \
ziv@2234 772 } \
ziv@2234 773 _hs_psize--; \
ziv@2234 774 } else if (( \
ziv@2234 775 cmpfcn(DECLTYPE(head)(ELMT_FROM_HH((head)->hh.tbl,_hs_p)), \
ziv@2234 776 DECLTYPE(head)(ELMT_FROM_HH((head)->hh.tbl,_hs_q))) \
ziv@2234 777 ) <= 0) { \
ziv@2234 778 _hs_e = _hs_p; \
ziv@2234 779 if (_hs_p != NULL){ \
ziv@2234 780 _hs_p = (UT_hash_handle*)((_hs_p->next != NULL) ? \
ziv@2234 781 ((void*)((char*)(_hs_p->next) + \
ziv@2234 782 (head)->hh.tbl->hho)) : NULL); \
ziv@2234 783 } \
ziv@2234 784 _hs_psize--; \
ziv@2234 785 } else { \
ziv@2234 786 _hs_e = _hs_q; \
ziv@2234 787 _hs_q = (UT_hash_handle*)((_hs_q->next != NULL) ? \
ziv@2234 788 ((void*)((char*)(_hs_q->next) + \
ziv@2234 789 (head)->hh.tbl->hho)) : NULL); \
ziv@2234 790 _hs_qsize--; \
ziv@2234 791 } \
ziv@2234 792 if ( _hs_tail != NULL ) { \
ziv@2234 793 _hs_tail->next = ((_hs_e != NULL) ? \
ziv@2234 794 ELMT_FROM_HH((head)->hh.tbl,_hs_e) : NULL); \
ziv@2234 795 } else { \
ziv@2234 796 _hs_list = _hs_e; \
ziv@2234 797 } \
ziv@2234 798 if (_hs_e != NULL) { \
ziv@2234 799 _hs_e->prev = ((_hs_tail != NULL) ? \
ziv@2234 800 ELMT_FROM_HH((head)->hh.tbl,_hs_tail) : NULL); \
ziv@2234 801 } \
ziv@2234 802 _hs_tail = _hs_e; \
ziv@2234 803 } \
ziv@2234 804 _hs_p = _hs_q; \
ziv@2234 805 } \
ziv@2234 806 if (_hs_tail != NULL){ \
ziv@2234 807 _hs_tail->next = NULL; \
ziv@2234 808 } \
ziv@2234 809 if ( _hs_nmerges <= 1U ) { \
ziv@2234 810 _hs_looping=0; \
ziv@2234 811 (head)->hh.tbl->tail = _hs_tail; \
ziv@2234 812 DECLTYPE_ASSIGN(head,ELMT_FROM_HH((head)->hh.tbl, _hs_list)); \
ziv@2234 813 } \
ziv@2234 814 _hs_insize *= 2U; \
ziv@2234 815 } \
ziv@2234 816 HASH_FSCK(hh,head); \
ziv@2234 817 } \
ziv@2234 818 } while (0)
ziv@2234 819
ziv@2234 820 /* This function selects items from one hash into another hash.
ziv@2234 821 * The end result is that the selected items have dual presence
ziv@2234 822 * in both hashes. There is no copy of the items made; rather
ziv@2234 823 * they are added into the new hash through a secondary hash
ziv@2234 824 * hash handle that must be present in the structure. */
ziv@2234 825 #define HASH_SELECT(hh_dst, dst, hh_src, src, cond) \
ziv@2234 826 do { \
ziv@2234 827 unsigned _src_bkt, _dst_bkt; \
ziv@2234 828 void *_last_elt=NULL, *_elt; \
ziv@2234 829 UT_hash_handle *_src_hh, *_dst_hh, *_last_elt_hh=NULL; \
ziv@2234 830 ptrdiff_t _dst_hho = ((char*)(&(dst)->hh_dst) - (char*)(dst)); \
ziv@2234 831 if (src != NULL) { \
ziv@2234 832 for(_src_bkt=0; _src_bkt < (src)->hh_src.tbl->num_buckets; _src_bkt++) { \
ziv@2234 833 for(_src_hh = (src)->hh_src.tbl->buckets[_src_bkt].hh_head; \
ziv@2234 834 _src_hh != NULL; \
ziv@2234 835 _src_hh = _src_hh->hh_next) { \
ziv@2234 836 _elt = ELMT_FROM_HH((src)->hh_src.tbl, _src_hh); \
ziv@2234 837 if (cond(_elt)) { \
ziv@2234 838 _dst_hh = (UT_hash_handle*)(((char*)_elt) + _dst_hho); \
ziv@2234 839 _dst_hh->key = _src_hh->key; \
ziv@2234 840 _dst_hh->keylen = _src_hh->keylen; \
ziv@2234 841 _dst_hh->hashv = _src_hh->hashv; \
ziv@2234 842 _dst_hh->prev = _last_elt; \
ziv@2234 843 _dst_hh->next = NULL; \
ziv@2234 844 if (_last_elt_hh != NULL) { _last_elt_hh->next = _elt; } \
ziv@2234 845 if (dst == NULL) { \
ziv@2234 846 DECLTYPE_ASSIGN(dst,_elt); \
ziv@2234 847 HASH_MAKE_TABLE(hh_dst,dst); \
ziv@2234 848 } else { \
ziv@2234 849 _dst_hh->tbl = (dst)->hh_dst.tbl; \
ziv@2234 850 } \
ziv@2234 851 HASH_TO_BKT(_dst_hh->hashv, _dst_hh->tbl->num_buckets, _dst_bkt); \
ziv@2234 852 HASH_ADD_TO_BKT(_dst_hh->tbl->buckets[_dst_bkt],_dst_hh); \
ziv@2234 853 (dst)->hh_dst.tbl->num_items++; \
ziv@2234 854 _last_elt = _elt; \
ziv@2234 855 _last_elt_hh = _dst_hh; \
ziv@2234 856 } \
ziv@2234 857 } \
ziv@2234 858 } \
ziv@2234 859 } \
ziv@2234 860 HASH_FSCK(hh_dst,dst); \
ziv@2234 861 } while (0)
ziv@2234 862
ziv@2234 863 #define HASH_CLEAR(hh,head) \
ziv@2234 864 do { \
ziv@2234 865 if (head != NULL) { \
ziv@2234 866 uthash_free((head)->hh.tbl->buckets, \
ziv@2234 867 (head)->hh.tbl->num_buckets*sizeof(struct UT_hash_bucket)); \
ziv@2234 868 HASH_BLOOM_FREE((head)->hh.tbl); \
ziv@2234 869 uthash_free((head)->hh.tbl, sizeof(UT_hash_table)); \
ziv@2234 870 (head)=NULL; \
ziv@2234 871 } \
ziv@2234 872 } while(0)
ziv@2234 873
ziv@2234 874 #define HASH_OVERHEAD(hh,head) \
ziv@2234 875 ((head != NULL) ? ( \
ziv@2234 876 (size_t)(((head)->hh.tbl->num_items * sizeof(UT_hash_handle)) + \
ziv@2234 877 ((head)->hh.tbl->num_buckets * sizeof(UT_hash_bucket)) + \
ziv@2234 878 sizeof(UT_hash_table) + \
ziv@2234 879 (HASH_BLOOM_BYTELEN))) : 0U)
ziv@2234 880
ziv@2234 881 #ifdef NO_DECLTYPE
ziv@2234 882 #define HASH_ITER(hh,head,el,tmp) \
ziv@2234 883 for(((el)=(head)), ((*(char**)(&(tmp)))=(char*)((head!=NULL)?(head)->hh.next:NULL)); \
ziv@2234 884 (el) != NULL; ((el)=(tmp)), ((*(char**)(&(tmp)))=(char*)((tmp!=NULL)?(tmp)->hh.next:NULL)))
ziv@2234 885 #else
ziv@2234 886 #define HASH_ITER(hh,head,el,tmp) \
ziv@2234 887 for(((el)=(head)), ((tmp)=DECLTYPE(el)((head!=NULL)?(head)->hh.next:NULL)); \
ziv@2234 888 (el) != NULL; ((el)=(tmp)), ((tmp)=DECLTYPE(el)((tmp!=NULL)?(tmp)->hh.next:NULL)))
ziv@2234 889 #endif
ziv@2234 890
ziv@2234 891 /* obtain a count of items in the hash */
ziv@2234 892 #define HASH_COUNT(head) HASH_CNT(hh,head)
ziv@2234 893 #define HASH_CNT(hh,head) ((head != NULL)?((head)->hh.tbl->num_items):0U)
ziv@2234 894
ziv@2234 895 typedef struct UT_hash_bucket {
ziv@2234 896 struct UT_hash_handle *hh_head;
ziv@2234 897 unsigned count;
ziv@2234 898
ziv@2234 899 /* expand_mult is normally set to 0. In this situation, the max chain length
ziv@2234 900 * threshold is enforced at its default value, HASH_BKT_CAPACITY_THRESH. (If
ziv@2234 901 * the bucket's chain exceeds this length, bucket expansion is triggered).
ziv@2234 902 * However, setting expand_mult to a non-zero value delays bucket expansion
ziv@2234 903 * (that would be triggered by additions to this particular bucket)
ziv@2234 904 * until its chain length reaches a *multiple* of HASH_BKT_CAPACITY_THRESH.
ziv@2234 905 * (The multiplier is simply expand_mult+1). The whole idea of this
ziv@2234 906 * multiplier is to reduce bucket expansions, since they are expensive, in
ziv@2234 907 * situations where we know that a particular bucket tends to be overused.
ziv@2234 908 * It is better to let its chain length grow to a longer yet-still-bounded
ziv@2234 909 * value, than to do an O(n) bucket expansion too often.
ziv@2234 910 */
ziv@2234 911 unsigned expand_mult;
ziv@2234 912
ziv@2234 913 } UT_hash_bucket;
ziv@2234 914
ziv@2234 915 /* random signature used only to find hash tables in external analysis */
ziv@2234 916 #define HASH_SIGNATURE 0xa0111fe1u
ziv@2234 917 #define HASH_BLOOM_SIGNATURE 0xb12220f2u
ziv@2234 918
ziv@2234 919 typedef struct UT_hash_table {
ziv@2234 920 UT_hash_bucket *buckets;
ziv@2234 921 unsigned num_buckets, log2_num_buckets;
ziv@2234 922 unsigned num_items;
ziv@2234 923 struct UT_hash_handle *tail; /* tail hh in app order, for fast append */
ziv@2234 924 ptrdiff_t hho; /* hash handle offset (byte pos of hash handle in element */
ziv@2234 925
ziv@2234 926 /* in an ideal situation (all buckets used equally), no bucket would have
ziv@2234 927 * more than ceil(#items/#buckets) items. that's the ideal chain length. */
ziv@2234 928 unsigned ideal_chain_maxlen;
ziv@2234 929
ziv@2234 930 /* nonideal_items is the number of items in the hash whose chain position
ziv@2234 931 * exceeds the ideal chain maxlen. these items pay the penalty for an uneven
ziv@2234 932 * hash distribution; reaching them in a chain traversal takes >ideal steps */
ziv@2234 933 unsigned nonideal_items;
ziv@2234 934
ziv@2234 935 /* ineffective expands occur when a bucket doubling was performed, but
ziv@2234 936 * afterward, more than half the items in the hash had nonideal chain
ziv@2234 937 * positions. If this happens on two consecutive expansions we inhibit any
ziv@2234 938 * further expansion, as it's not helping; this happens when the hash
ziv@2234 939 * function isn't a good fit for the key domain. When expansion is inhibited
ziv@2234 940 * the hash will still work, albeit no longer in constant time. */
ziv@2234 941 unsigned ineff_expands, noexpand;
ziv@2234 942
ziv@2234 943 uint32_t signature; /* used only to find hash tables in external analysis */
ziv@2234 944 #ifdef HASH_BLOOM
ziv@2234 945 uint32_t bloom_sig; /* used only to test bloom exists in external analysis */
ziv@2234 946 uint8_t *bloom_bv;
ziv@2234 947 uint8_t bloom_nbits;
ziv@2234 948 #endif
ziv@2234 949
ziv@2234 950 } UT_hash_table;
ziv@2234 951
ziv@2234 952 typedef struct UT_hash_handle {
ziv@2234 953 struct UT_hash_table *tbl;
ziv@2234 954 void *prev; /* prev element in app order */
ziv@2234 955 void *next; /* next element in app order */
ziv@2234 956 struct UT_hash_handle *hh_prev; /* previous hh in bucket order */
ziv@2234 957 struct UT_hash_handle *hh_next; /* next hh in bucket order */
ziv@2234 958 void *key; /* ptr to enclosing struct's key */
ziv@2234 959 unsigned keylen; /* enclosing struct's key len */
ziv@2234 960 unsigned hashv; /* result of hash-fcn(key) */
ziv@2234 961 } UT_hash_handle;
ziv@2234 962
ziv@2234 963 #endif /* UTHASH_H */