nsz Git - libfirm/blob - ir/adt/hashset.c

   1 /*
   2  * Copyright (C) 1995-2007 University of Karlsruhe.  All right reserved.
   3  *
   4  * This file is part of libFirm.
   5  *
   6  * This file may be distributed and/or modified under the terms of the
   7  * GNU General Public License version 2 as published by the Free Software
   8  * Foundation and appearing in the file LICENSE.GPL included in the
   9  * packaging of this file.
  10  *
  11  * Licensees holding valid libFirm Professional Edition licenses may use
  12  * this file in accordance with the libFirm Commercial License.
  13  * Agreement provided with the Software.
  14  *
  15  * This file is provided AS IS with NO WARRANTY OF ANY KIND, INCLUDING THE
  16  * WARRANTY OF DESIGN, MERCHANTABILITY AND FITNESS FOR A PARTICULAR
  17  * PURPOSE.
  18  */
  19
  20 /**
  21  * @file
  22  * @brief   Generic hashset implementation
  23  * @author  Matthias Braun, inspiration from densehash from google sparsehash
  24  *          package
  25  * @date    17.03.2007
  26  * @version $Id$
  27  *
  28  *
  29  * You have to specialize this file by defining:
  30  *
  31  * <ul>
  32  *  <li><b>HashSet</b>         The name of the hashset type</li>
  33  *  <li><b>HashSetIterator</b> The name of the hashset iterator type</li>
  34  *  <li><b>ValueType</b>       The type of the stored data values</li>
  35  *  <li><b>NullValue</b>       A special value representing no values</li>
  36  *  <li><b>DeletedValue</b>    A special value representing deleted entries</li>
  37  *  <li><b>Hash(hashset,key)</b> calculates the hash value for a given key</li>
  38  * </ul>
  39  *
  40  * Note that by default it is assumed that the data values themselfes are used
  41  * as keys. However you can change that with additional defines:
  42  *
  43  * <ul>
  44  *  <li><b>KeyType</b>         The type of the keys identifying data values.
  45  *                             Defining this implies, that a data value contains
  46  *                             more than just the key.</li>
  47  *  <li><b>GetKey(value)</b>   Extracts the key from a data value</li>
  48  *  <li><b>KeysEqual(hashset,key1,key2)</b>  Tests wether 2 keys are equal</li>
  49  *  <li><b>DO_REHASH</b>       Instead of storing the hash-values, recalculate
  50  *                             them on demand from the datavalues. (usefull if
  51  *                             calculating the hash-values takes less time than
  52  *                             a memory access)</li>
  53  * </ul>
  54  *
  55  * You can further fine tune your hashset by defining the following:
  56  *
  57  * <ul>
  58  *  <li><b>JUMP(num_probes)</b> The probing method</li>
  59  *  <li><b>Alloc(count)</b>     Allocates count hashset entries (NOT bytes)</li>
  60  *  <li><b>Free(ptr)</b>        Frees a block of memory allocated by Alloc</li>
  61  *  <li><b>SetRangeEmpty(ptr,count)</b> Efficiently sets a range of elements to
  62  *                                      the Null value</li>
  63  *  <li><b>ADDITIONAL_DATA<b>   Additional fields appended to the hashset struct</li>
  64  * </ul>
  65  */
  66 #ifdef HashSet
  67
  68 #include <stdlib.h>
  69 #include <string.h>
  70 #include <assert.h>
  71
  72 #include "bitfiddle.h"
  73 #include "util.h"
  74
  75 /* quadratic probing */
  76 #ifndef JUMP
  77 #define JUMP(num_probes)      (num_probes)
  78 #endif /* JUMP */
  79
  80 #ifndef Hash
  81 #define ID_HASH
  82 #define Hash(self,value)      ((unsigned)(value))
  83 #endif /* Hash */
  84
  85 #ifdef DO_REHASH
  86 #define HashSetEntry                   ValueType
  87 #define EntrySetHash(entry,new_hash)
  88 #define EntryGetHash(self,entry)       Hash(self,entry)
  89 #define EntryGetValue(entry)           (entry)
  90 #else /* ! DO_REHASH */
  91 #define EntryGetHash(self,entry)       (entry).hash
  92 #define EntrySetHash(entry,new_hash)   (entry).hash = (new_hash)
  93 #define EntryGetValue(entry)           (entry).data
  94 #endif /* DO_REHASH */
  95
  96 #ifndef Alloc
  97 #include "xmalloc.h"
  98 #define Alloc(size)    (HashSetEntry*) xmalloc((size) * sizeof(HashSetEntry))
  99 #define Free(ptr)      free(ptr)
 100 #endif /* Alloc */
 101
 102 #ifdef ID_HASH
 103 #define InsertReturnValue               int
 104 #define GetInsertReturnValue(entry,new) (new)
 105 #else /* ! ID_HASH */
 106 #define InsertReturnValue               ValueType
 107 #define GetInsertReturnValue(entry,new) EntryGetValue(entry)
 108 #endif /* ID_HASH */
 109
 110 #ifndef KeyType
 111 #define KeyType                  ValueType
 112 #define GetKey(value)            (value)
 113 #define InitData(self,value,key) (value) = (key)
 114 #endif /* KeyType */
 115
 116 #ifndef ConstKeyType
 117 #define ConstKeyType             const KeyType
 118 #endif /* ConstKeyType */
 119
 120 #ifndef EntrySetEmpty
 121 #define EntrySetEmpty(entry)    EntryGetValue(entry) = NullValue
 122 #endif /* EntrySetEmpty */
 123 #ifndef EntrySetDeleted
 124 #define EntrySetDeleted(entry)  EntryGetValue(entry) = DeletedValue
 125 #endif /* EntrySetDeleted */
 126 #ifndef EntryIsEmpty
 127 #define EntryIsEmpty(entry)     (EntryGetValue(entry) == NullValue)
 128 #endif /* EntryIsEmpty */
 129 #ifndef EntryIsDeleted
 130 #define EntryIsDeleted(entry)   (EntryGetValue(entry) == DeletedValue)
 131 #endif /* EntryIsDeleted */
 132 #ifndef SetRangeEmpty
 133 #define SetRangeEmpty(ptr,size)                \
 134 {                                              \
 135         size_t _i;                                 \
 136         size_t _size = (size);                     \
 137         HashSetEntry *entries = (ptr);             \
 138         for(_i = 0; _i < _size; ++_i) {            \
 139                 HashSetEntry *entry = & entries[_i];   \
 140                 EntrySetEmpty(*entry);                 \
 141         }                                          \
 142 }
 143 #endif /* SetRangeEmpty */
 144
 145 #ifndef HT_OCCUPANCY_FLT
 146 /** how full before we double size */
 147 #define HT_OCCUPANCY_FLT  0.5f
 148 #endif /* HT_OCCUPANCY_FLT */
 149
 150 #ifndef HT_EMPTY_FLT
 151 /** how empty before we half size */
 152 #define HT_EMPTY_FLT      (0.4f * (HT_OCCUPANCY_FLT))
 153 #endif /* HT_EMPTY_FLT */
 154
 155 #ifndef HT_MIN_BUCKETS
 156 /** default smallest bucket size */
 157 #define HT_MIN_BUCKETS    32
 158 #endif /* HT_MIN_BUCKETS */
 159
 160 #define ILLEGAL_POS       ((size_t)-1)
 161
 162 /* check that all needed functions are defined */
 163 #ifndef hashset_init
 164 #error You have to redefine hashset_init
 165 #endif
 166 #ifndef hashset_init_size
 167 #error You have to redefine hashset_init_size
 168 #endif
 169 #ifndef hashset_destroy
 170 #error You have to redefine hashset_destroy
 171 #endif
 172 #ifndef hashset_insert
 173 #error You have to redefine hashset_insert
 174 #endif
 175 #ifndef hashset_remove
 176 #error You have to redefine hashset_remove
 177 #endif
 178 #ifndef hashset_find
 179 #error You have to redefine hashset_find
 180 #endif
 181 #ifndef hashset_size
 182 #error You have to redefine hashset_size
 183 #endif
 184 #ifndef hashset_iterator_init
 185 #error You have to redefine hashset_iterator_init
 186 #endif
 187 #ifndef hashset_iterator_next
 188 #error You have to redefine hashset_iterator_next
 189 #endif
 190 #ifndef hashset_remove_iterator
 191 #error You have to redefine hashset_remove_iterator
 192 #endif
 193
 194 /**
 195  * Returns the number of elements in the hashset
 196  */
 197 size_t hashset_size(const HashSet *self)
 198 {
 199         return self->num_elements - self->num_deleted;
 200 }
 201
 202 /**
 203  * Inserts an element into a hashset without growing the set (you have to make
 204  * sure there's enough room for that.
 205  * @note also see comments for hashset_insert()
 206  * @internal
 207  */
 208 static INLINE
 209 InsertReturnValue insert_nogrow(HashSet *self, KeyType key)
 210 {
 211         size_t num_probes = 0;
 212         size_t num_buckets = self->num_buckets;
 213         size_t hashmask = num_buckets - 1;
 214         unsigned hash = Hash(self, key);
 215         size_t bucknum = hash & hashmask;
 216         size_t insert_pos = ILLEGAL_POS;
 217
 218         assert((num_buckets & (num_buckets - 1)) == 0);
 219
 220         while(1) {
 221                 HashSetEntry *entry = & self->entries[bucknum];
 222
 223                 if(EntryIsEmpty(*entry)) {
 224                         size_t p;
 225                         HashSetEntry *nentry;
 226
 227                         if(insert_pos != ILLEGAL_POS) {
 228                                 p = insert_pos;
 229                         } else {
 230                                 p = bucknum;
 231                         }
 232
 233                         nentry = &self->entries[p];
 234                         InitData(self, EntryGetValue(*nentry), key);
 235                         EntrySetHash(*nentry, hash);
 236                         self->num_elements++;
 237                         return GetInsertReturnValue(*nentry, 1);
 238                 }
 239                 if(EntryIsDeleted(*entry)) {
 240                         if(insert_pos == ILLEGAL_POS)
 241                                 insert_pos = bucknum;
 242                 } else if(EntryGetHash(self, *entry) == hash) {
 243                         if(KeysEqual(self, GetKey(EntryGetValue(*entry)), key)) {
 244                                 // Value already in the set, return it
 245                                 return GetInsertReturnValue(*entry, 0);
 246                         }
 247                 }
 248
 249                 ++num_probes;
 250                 bucknum = (bucknum + JUMP(num_probes)) & hashmask;
 251                 assert(num_probes < num_buckets);
 252         }
 253 }
 254
 255 /**
 256  * Inserts an element into a hashset under the assumption that the hashset
 257  * contains no deleted entries and the element doesn't exist in the hashset yet.
 258  * @internal
 259  */
 260 static
 261 void insert_new(HashSet *self, unsigned hash, ValueType value)
 262 {
 263         size_t num_probes = 0;
 264         size_t num_buckets = self->num_buckets;
 265         size_t hashmask = num_buckets - 1;
 266         size_t bucknum = hash & hashmask;
 267         size_t insert_pos = ILLEGAL_POS;
 268
 269         assert(value != NullValue);
 270
 271         while(1) {
 272                 HashSetEntry *entry = & self->entries[bucknum];
 273
 274                 if(EntryIsEmpty(*entry)) {
 275                         size_t p;
 276                         HashSetEntry *nentry;
 277
 278                         if(insert_pos != ILLEGAL_POS) {
 279                                 p = insert_pos;
 280                         } else {
 281                                 p = bucknum;
 282                         }
 283                         nentry = &self->entries[p];
 284
 285                         EntryGetValue(*nentry) = value;
 286                         EntrySetHash(*nentry, hash);
 287                         self->num_elements++;
 288                         return;
 289                 }
 290                 assert(!EntryIsDeleted(*entry));
 291
 292                 ++num_probes;
 293                 bucknum = (bucknum + JUMP(num_probes)) & hashmask;
 294                 assert(num_probes < num_buckets);
 295         }
 296 }
 297
 298 /**
 299  * calculate shrink and enlarge limits
 300  * @internal
 301  */
 302 static INLINE
 303 void reset_thresholds(HashSet *self)
 304 {
 305         self->enlarge_threshold = (size_t) (self->num_buckets * HT_OCCUPANCY_FLT);
 306         self->shrink_threshold = (size_t) (self->num_buckets * HT_EMPTY_FLT);
 307         self->consider_shrink = 0;
 308 }
 309
 310 /**
 311  * Resize the hashset
 312  * @internal
 313  */
 314 static INLINE
 315 void resize(HashSet *self, size_t new_size)
 316 {
 317         size_t num_buckets = self->num_buckets;
 318         size_t i;
 319         HashSetEntry *old_entries = self->entries;
 320         HashSetEntry *new_entries;
 321
 322         /* allocate a new array with double size */
 323         new_entries = Alloc(new_size);
 324         SetRangeEmpty(new_entries, new_size);
 325
 326         /* use the new array */
 327         self->entries = new_entries;
 328         self->num_buckets = new_size;
 329         self->num_elements = 0;
 330         self->num_deleted = 0;
 331 #ifndef NDEBUG
 332         self->entries_version++;
 333 #endif
 334         reset_thresholds(self);
 335
 336         /* reinsert all elements */
 337         for(i = 0; i < num_buckets; ++i) {
 338                 HashSetEntry *entry = & old_entries[i];
 339                 if(EntryIsEmpty(*entry) || EntryIsDeleted(*entry))
 340                         continue;
 341
 342                 insert_new(self, EntryGetHash(self, *entry), EntryGetValue(*entry));
 343         }
 344
 345         /* now we can free the old array */
 346         Free(old_entries);
 347 }
 348
 349 /**
 350  * grow the hashset if adding 1 more elements would make it too crowded
 351  * @internal
 352  */
 353 static INLINE
 354 void maybe_grow(HashSet *self)
 355 {
 356         size_t resize_to;
 357
 358         if(LIKELY(self->num_elements + 1 <= self->enlarge_threshold))
 359                 return;
 360
 361         /* double table size */
 362         resize_to = self->num_buckets * 2;
 363         resize(self, resize_to);
 364 }
 365
 366 /**
 367  * shrink the hashset if it is only sparsely filled
 368  * @internal
 369  */
 370 static INLINE
 371 void maybe_shrink(HashSet *self)
 372 {
 373         size_t size;
 374         size_t resize_to;
 375
 376         if(!self->consider_shrink)
 377                 return;
 378
 379         self->consider_shrink = 0;
 380         size = hashset_size(self);
 381         if(LIKELY(size > self->shrink_threshold))
 382                 return;
 383
 384         resize_to = ceil_po2(size);
 385
 386         if(resize_to < 4)
 387                 resize_to = 4;
 388
 389         resize(self, resize_to);
 390 }
 391
 392 /**
 393  * Insert an element into the hashset. If no element with key key exists yet,
 394  * then a new one is created and initialized with the InitData function.
 395  * Otherwise the exisiting element is returned (for hashs where key is equal to
 396  * value, nothing is returned.)
 397  *
 398  * @param self   the hashset
 399  * @param key    the key that identifies the data
 400  * @returns      the existing or newly created data element (or nothing in case of hashs where keys are the while value)
 401  */
 402 InsertReturnValue hashset_insert(HashSet *self, KeyType key)
 403 {
 404 #ifndef NDEBUG
 405         self->entries_version++;
 406 #endif
 407
 408         maybe_shrink(self);
 409         maybe_grow(self);
 410         return insert_nogrow(self, key);
 411 }
 412
 413 /**
 414  * Searchs for an element with key @p key.
 415  *
 416  * @param self      the hashset
 417  * @param key       the key to search for
 418  * @returns         the found value or NullValue if nothing was found
 419  */
 420 ValueType hashset_find(const HashSet *self, ConstKeyType key)
 421 {
 422         size_t num_probes = 0;
 423         size_t num_buckets = self->num_buckets;
 424         size_t hashmask = num_buckets - 1;
 425         unsigned hash = Hash(self, key);
 426         size_t bucknum = hash & hashmask;
 427
 428         while(1) {
 429                 HashSetEntry *entry = & self->entries[bucknum];
 430
 431                 if(EntryIsEmpty(*entry)) {
 432                         return NullValue;
 433                 }
 434                 if(EntryIsDeleted(*entry)) {
 435                         // value is deleted
 436                 } else if(EntryGetHash(self, *entry) == hash) {
 437                         if(KeysEqual(self, GetKey(EntryGetValue(*entry)), key)) {
 438                                 // found the value
 439                                 return EntryGetValue(*entry);
 440                         }
 441                 }
 442
 443                 ++num_probes;
 444                 bucknum = (bucknum + JUMP(num_probes)) & hashmask;
 445                 assert(num_probes < num_buckets);
 446         }
 447 }
 448
 449 /**
 450  * Removes an element from a hashset. Does nothing if the set doesn't contain
 451  * the element.
 452  *
 453  * @param self    the hashset
 454  * @param key     key that identifies the data to remove
 455  */
 456 void hashset_remove(HashSet *self, ConstKeyType key)
 457 {
 458         size_t num_probes = 0;
 459         size_t num_buckets = self->num_buckets;
 460         size_t hashmask = num_buckets - 1;
 461         unsigned hash = Hash(self, key);
 462         size_t bucknum = hash & hashmask;
 463
 464 #ifndef NDEBUG
 465         self->entries_version++;
 466 #endif
 467
 468         while(1) {
 469                 HashSetEntry *entry = & self->entries[bucknum];
 470
 471                 if(EntryIsEmpty(*entry)) {
 472                         return;
 473                 }
 474                 if(EntryIsDeleted(*entry)) {
 475                         // entry is deleted
 476                 } else if(EntryGetHash(self, *entry) == hash) {
 477                         if(KeysEqual(self, GetKey(EntryGetValue(*entry)), key)) {
 478                                 EntrySetDeleted(*entry);
 479                                 self->num_deleted++;
 480                                 self->consider_shrink = 1;
 481                                 return;
 482                         }
 483                 }
 484
 485                 ++num_probes;
 486                 bucknum = (bucknum + JUMP(num_probes)) & hashmask;
 487                 assert(num_probes < num_buckets);
 488         }
 489 }
 490
 491 /**
 492  * Initializes hashset with a specific size
 493  * @internal
 494  */
 495 static INLINE
 496 void init_size(HashSet *self, size_t initial_size)
 497 {
 498         if(initial_size < 4)
 499                 initial_size = 4;
 500
 501         self->entries = Alloc(initial_size);
 502         SetRangeEmpty(self->entries, initial_size);
 503         self->num_buckets = initial_size;
 504         self->consider_shrink = 0;
 505         self->num_elements = 0;
 506         self->num_deleted = 0;
 507 #ifndef NDEBUG
 508         self->entries_version = 0;
 509 #endif
 510
 511         reset_thresholds(self);
 512 }
 513
 514 /**
 515  * Initialializes a hashset with the default size. The memory for the set has to
 516  * already allocated.
 517  */
 518 void hashset_init(HashSet *self)
 519 {
 520         init_size(self, HT_MIN_BUCKETS);
 521 }
 522
 523 /**
 524  * Destroys a hashset, freeing all used memory (except the memory for the
 525  * HashSet struct itself).
 526  */
 527 void hashset_destroy(HashSet *self)
 528 {
 529         Free(self->entries);
 530 #ifndef NDEBUG
 531         self->entries = NULL;
 532 #endif
 533 }
 534
 535 /**
 536  * Initializes a hashset expecting expected_element size
 537  */
 538 void hashset_init_size(HashSet *self, size_t expected_elements)
 539 {
 540         size_t needed_size;
 541         size_t po2size;
 542
 543         if(expected_elements >= UINT_MAX/2) {
 544                 abort();
 545         }
 546
 547         needed_size = expected_elements * (int)(1.0 / HT_OCCUPANCY_FLT);
 548         po2size = ceil_po2(needed_size);
 549         init_size(self, po2size);
 550 }
 551
 552 /**
 553  * Initializes a hashset iterator. The memory for the allocator has to be
 554  * already allocated.
 555  * @note it is not allowed to remove or insert elements while iterating
 556  */
 557 void hashset_iterator_init(HashSetIterator *self, const HashSet *hashset)
 558 {
 559         self->current_bucket = hashset->entries - 1;
 560         self->end = hashset->entries + hashset->num_buckets;
 561 #ifndef NDEBUG
 562         self->set = hashset;
 563         self->entries_version = hashset->entries_version;
 564 #endif
 565 }
 566
 567 /**
 568  * Returns the next value in the iterator or NULL if no value is left
 569  * in the hashset.
 570  * @note it is not allowed to remove or insert elements while iterating
 571  */
 572 ValueType hashset_iterator_next(HashSetIterator *self)
 573 {
 574         HashSetEntry *current_bucket = self->current_bucket;
 575         HashSetEntry *end = self->end;
 576
 577         if(current_bucket >= end)
 578                 return NullValue;
 579
 580         /* using hashset_insert or hashset_remove is not allowed while iterating */
 581         assert(self->entries_version == self->set->entries_version);
 582
 583         do {
 584                 current_bucket++;
 585         } while(current_bucket < end &&
 586                         (EntryIsEmpty(*current_bucket) || EntryIsDeleted(*current_bucket)));
 587
 588         if(current_bucket >= end)
 589                 return NullValue;
 590
 591         self->current_bucket = current_bucket;
 592         return EntryGetValue(*current_bucket);
 593 }
 594
 595 /**
 596  * Removes the element the iterator points to. Removing an element a second time
 597  * has no result.
 598  */
 599 void hashset_remove_iterator(HashSet *self, const HashSetIterator *iter)
 600 {
 601         HashSetEntry *entry = iter->current_bucket;
 602
 603         /* iterator_next needs to have been called at least once */
 604         assert(entry >= self->entries);
 605         /* needs to be on a valid element */
 606         assert(entry < self->entries + self->num_buckets);
 607
 608         if(EntryIsDeleted(*entry))
 609                 return;
 610
 611         EntrySetDeleted(*entry);
 612         self->num_deleted++;
 613         self->consider_shrink = 1;
 614 }
 615
 616 #endif