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

   1 /*
   2  * Copyright (C) 1995-2008 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,key)        ((unsigned)(((char *)key) - (char *)0))
  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, GetKey(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,found) (found)
 105 #define NullReturnValue                   0
 106 #else /* ! ID_HASH */
 107 #ifdef SCALAR_RETURN
 108 #define InsertReturnValue                 ValueType
 109 #define GetInsertReturnValue(entry,found) EntryGetValue(entry)
 110 #define NullReturnValue                   NullValue
 111 #else
 112 #define InsertReturnValue                 ValueType*
 113 #define GetInsertReturnValue(entry,found) & EntryGetValue(entry)
 114 #define NullReturnValue                   & NullValue
 115 #endif
 116 #endif /* ID_HASH */
 117
 118 #ifndef KeyType
 119 #define KeyType                  ValueType
 120 #define GetKey(value)            (value)
 121 #define InitData(self,value,key) (value) = (key)
 122 #endif /* KeyType */
 123
 124 #ifndef ConstKeyType
 125 #define ConstKeyType             const KeyType
 126 #endif /* ConstKeyType */
 127
 128 #ifndef EntrySetEmpty
 129 #define EntrySetEmpty(entry)    EntryGetValue(entry) = NullValue
 130 #endif /* EntrySetEmpty */
 131 #ifndef EntrySetDeleted
 132 #define EntrySetDeleted(entry)  EntryGetValue(entry) = DeletedValue
 133 #endif /* EntrySetDeleted */
 134 #ifndef EntryIsEmpty
 135 #define EntryIsEmpty(entry)     (EntryGetValue(entry) == NullValue)
 136 #endif /* EntryIsEmpty */
 137 #ifndef EntryIsDeleted
 138 #define EntryIsDeleted(entry)   (EntryGetValue(entry) == DeletedValue)
 139 #endif /* EntryIsDeleted */
 140 #ifndef SetRangeEmpty
 141 #define SetRangeEmpty(ptr,size)                \
 142 {                                              \
 143         size_t _i;                                 \
 144         size_t _size = (size);                     \
 145         HashSetEntry *entries = (ptr);             \
 146         for(_i = 0; _i < _size; ++_i) {            \
 147                 HashSetEntry *entry = & entries[_i];   \
 148                 EntrySetEmpty(*entry);                 \
 149         }                                          \
 150 }
 151 #endif /* SetRangeEmpty */
 152
 153 #ifndef HT_OCCUPANCY_FLT
 154 /** how full before we double size */
 155 #define HT_OCCUPANCY_FLT(x) ((x)/2)
 156 #endif /* HT_OCCUPANCY_FLT */
 157 #ifndef HT_1_DIV_OCCUPANCY_FLT
 158 #define HT_1_DIV_OCCUPANCY_FLT 2
 159 #endif
 160
 161 #ifndef HT_EMPTY_FLT
 162 /** how empty before we half size */
 163 #define HT_EMPTY_FLT(x)     ((x)/5)
 164 #endif /* HT_EMPTY_FLT */
 165
 166 #ifndef HT_MIN_BUCKETS
 167 /** default smallest bucket size */
 168 #define HT_MIN_BUCKETS    32
 169 #endif /* HT_MIN_BUCKETS */
 170
 171 #define ILLEGAL_POS       ((size_t)-1)
 172
 173 /* check that all needed functions are defined */
 174 #ifndef hashset_init
 175 #error You have to redefine hashset_init
 176 #endif
 177 #ifndef hashset_init_size
 178 #error You have to redefine hashset_init_size
 179 #endif
 180 #ifndef hashset_destroy
 181 #error You have to redefine hashset_destroy
 182 #endif
 183 #ifndef hashset_insert
 184 #error You have to redefine hashset_insert
 185 #endif
 186 #ifndef hashset_remove
 187 #error You have to redefine hashset_remove
 188 #endif
 189 #ifndef hashset_find
 190 #error You have to redefine hashset_find
 191 #endif
 192 #ifndef hashset_size
 193 #error You have to redefine hashset_size
 194 #endif
 195
 196 #ifndef NO_ITERATOR
 197 #ifndef hashset_iterator_init
 198 #error You have to redefine hashset_iterator_init
 199 #endif
 200 #ifndef hashset_iterator_next
 201 #error You have to redefine hashset_iterator_next
 202 #endif
 203 #ifndef hashset_remove_iterator
 204 #error You have to redefine hashset_remove_iterator
 205 #endif
 206 #endif
 207
 208 /**
 209  * Returns the number of elements in the hashset
 210  */
 211 size_t hashset_size(const HashSet *self)
 212 {
 213         return self->num_elements - self->num_deleted;
 214 }
 215
 216 /**
 217  * Inserts an element into a hashset without growing the set (you have to make
 218  * sure there's enough room for that.
 219  * @note also see comments for hashset_insert()
 220  * @internal
 221  */
 222 static INLINE
 223 InsertReturnValue insert_nogrow(HashSet *self, KeyType key)
 224 {
 225         size_t   num_probes  = 0;
 226         size_t   num_buckets = self->num_buckets;
 227         size_t   hashmask    = num_buckets - 1;
 228         unsigned hash        = Hash(self, key);
 229         size_t   bucknum     = hash & hashmask;
 230         size_t   insert_pos  = ILLEGAL_POS;
 231
 232         assert((num_buckets & (num_buckets - 1)) == 0);
 233
 234         while(1) {
 235                 HashSetEntry *entry = & self->entries[bucknum];
 236
 237                 if(EntryIsEmpty(*entry)) {
 238                         size_t p;
 239                         HashSetEntry *nentry;
 240
 241                         if(insert_pos != ILLEGAL_POS) {
 242                                 p = insert_pos;
 243                         } else {
 244                                 p = bucknum;
 245                         }
 246
 247                         nentry = &self->entries[p];
 248                         InitData(self, EntryGetValue(*nentry), key);
 249                         EntrySetHash(*nentry, hash);
 250                         self->num_elements++;
 251                         return GetInsertReturnValue(*nentry, 0);
 252                 }
 253                 if(EntryIsDeleted(*entry)) {
 254                         if(insert_pos == ILLEGAL_POS)
 255                                 insert_pos = bucknum;
 256                 } else if(EntryGetHash(self, *entry) == hash) {
 257                         if(KeysEqual(self, GetKey(EntryGetValue(*entry)), key)) {
 258                                 // Value already in the set, return it
 259                                 return GetInsertReturnValue(*entry, 1);
 260                         }
 261                 }
 262
 263                 ++num_probes;
 264                 bucknum = (bucknum + JUMP(num_probes)) & hashmask;
 265                 assert(num_probes < num_buckets);
 266         }
 267 }
 268
 269 /**
 270  * Inserts an element into a hashset under the assumption that the hashset
 271  * contains no deleted entries and the element doesn't exist in the hashset yet.
 272  * @internal
 273  */
 274 static
 275 void insert_new(HashSet *self, unsigned hash, ValueType value)
 276 {
 277         size_t num_probes  = 0;
 278         size_t num_buckets = self->num_buckets;
 279         size_t hashmask    = num_buckets - 1;
 280         size_t bucknum     = hash & hashmask;
 281         size_t insert_pos  = ILLEGAL_POS;
 282
 283         //assert(value != NullValue);
 284
 285         while(1) {
 286                 HashSetEntry *entry = & self->entries[bucknum];
 287
 288                 if(EntryIsEmpty(*entry)) {
 289                         size_t        p;
 290                         HashSetEntry *nentry;
 291
 292                         if(insert_pos != ILLEGAL_POS) {
 293                                 p = insert_pos;
 294                         } else {
 295                                 p = bucknum;
 296                         }
 297                         nentry = &self->entries[p];
 298
 299                         EntryGetValue(*nentry) = value;
 300                         EntrySetHash(*nentry, hash);
 301                         self->num_elements++;
 302                         return;
 303                 }
 304                 assert(!EntryIsDeleted(*entry));
 305
 306                 ++num_probes;
 307                 bucknum = (bucknum + JUMP(num_probes)) & hashmask;
 308                 assert(num_probes < num_buckets);
 309         }
 310 }
 311
 312 /**
 313  * calculate shrink and enlarge limits
 314  * @internal
 315  */
 316 static INLINE
 317 void reset_thresholds(HashSet *self)
 318 {
 319         self->enlarge_threshold = (size_t) HT_OCCUPANCY_FLT(self->num_buckets);
 320         self->shrink_threshold  = (size_t) HT_EMPTY_FLT(self->num_buckets);
 321         self->consider_shrink   = 0;
 322 }
 323
 324 #ifndef HAVE_OWN_RESIZE
 325 /**
 326  * Resize the hashset
 327  * @internal
 328  */
 329 static INLINE
 330 void resize(HashSet *self, size_t new_size)
 331 {
 332         size_t num_buckets = self->num_buckets;
 333         size_t i;
 334         HashSetEntry *old_entries = self->entries;
 335         HashSetEntry *new_entries;
 336
 337         /* allocate a new array with double size */
 338         new_entries = Alloc(new_size);
 339         SetRangeEmpty(new_entries, new_size);
 340
 341         /* use the new array */
 342         self->entries      = new_entries;
 343         self->num_buckets  = new_size;
 344         self->num_elements = 0;
 345         self->num_deleted  = 0;
 346 #ifndef NDEBUG
 347         self->entries_version++;
 348 #endif
 349         reset_thresholds(self);
 350
 351         /* reinsert all elements */
 352         for(i = 0; i < num_buckets; ++i) {
 353                 HashSetEntry *entry = & old_entries[i];
 354                 if(EntryIsEmpty(*entry) || EntryIsDeleted(*entry))
 355                         continue;
 356
 357                 insert_new(self, EntryGetHash(self, *entry), EntryGetValue(*entry));
 358         }
 359
 360         /* now we can free the old array */
 361         Free(old_entries);
 362 }
 363 #else
 364
 365 /* resize must be defined outside */
 366 static void resize(HashSet *self, size_t new_size);
 367
 368 #endif
 369
 370 /**
 371  * grow the hashset if adding 1 more elements would make it too crowded
 372  * @internal
 373  */
 374 static INLINE
 375 void maybe_grow(HashSet *self)
 376 {
 377         size_t resize_to;
 378
 379         if(LIKELY(self->num_elements + 1 <= self->enlarge_threshold))
 380                 return;
 381
 382         /* double table size */
 383         resize_to = self->num_buckets * 2;
 384         resize(self, resize_to);
 385 }
 386
 387 /**
 388  * shrink the hashset if it is only sparsely filled
 389  * @internal
 390  */
 391 static INLINE
 392 void maybe_shrink(HashSet *self)
 393 {
 394         size_t size;
 395         size_t resize_to;
 396
 397         if(!self->consider_shrink)
 398                 return;
 399
 400         self->consider_shrink = 0;
 401         size                  = hashset_size(self);
 402         if(size <= HT_MIN_BUCKETS)
 403                 return;
 404
 405         if(LIKELY(size > self->shrink_threshold))
 406                 return;
 407
 408         resize_to = ceil_po2(size);
 409
 410         if(resize_to < 4)
 411                 resize_to = 4;
 412
 413         resize(self, resize_to);
 414 }
 415
 416 /**
 417  * Insert an element into the hashset. If no element with the given key exists yet,
 418  * then a new one is created and initialized with the InitData function.
 419  * Otherwise the existing element is returned (for hashs where key is equal to
 420  * value, nothing is returned.)
 421  *
 422  * @param self   the hashset
 423  * @param key    the key that identifies the data
 424  * @returns      the existing or newly created data element (or nothing in case of hashs where keys are the while value)
 425  */
 426 InsertReturnValue hashset_insert(HashSet *self, KeyType key)
 427 {
 428 #ifndef NDEBUG
 429         self->entries_version++;
 430 #endif
 431
 432         maybe_shrink(self);
 433         maybe_grow(self);
 434         return insert_nogrow(self, key);
 435 }
 436
 437 /**
 438  * Searches for an element with key @p key.
 439  *
 440  * @param self      the hashset
 441  * @param key       the key to search for
 442  * @returns         the found value or NullValue if nothing was found
 443  */
 444 InsertReturnValue hashset_find(const HashSet *self, ConstKeyType key)
 445 {
 446         size_t   num_probes  = 0;
 447         size_t   num_buckets = self->num_buckets;
 448         size_t   hashmask    = num_buckets - 1;
 449         unsigned hash        = Hash(self, key);
 450         size_t   bucknum     = hash & hashmask;
 451
 452         while(1) {
 453                 HashSetEntry *entry = & self->entries[bucknum];
 454
 455                 if(EntryIsEmpty(*entry)) {
 456                         return NullReturnValue;
 457                 }
 458                 if(EntryIsDeleted(*entry)) {
 459                         // value is deleted
 460                 } else if(EntryGetHash(self, *entry) == hash) {
 461                         if(KeysEqual(self, GetKey(EntryGetValue(*entry)), key)) {
 462                                 // found the value
 463                                 return GetInsertReturnValue(*entry, 1);
 464                         }
 465                 }
 466
 467                 ++num_probes;
 468                 bucknum = (bucknum + JUMP(num_probes)) & hashmask;
 469                 assert(num_probes < num_buckets);
 470         }
 471 }
 472
 473 /**
 474  * Removes an element from a hashset. Does nothing if the set doesn't contain
 475  * the element.
 476  *
 477  * @param self    the hashset
 478  * @param key     key that identifies the data to remove
 479  */
 480 void hashset_remove(HashSet *self, ConstKeyType key)
 481 {
 482         size_t   num_probes  = 0;
 483         size_t   num_buckets = self->num_buckets;
 484         size_t   hashmask    = num_buckets - 1;
 485         unsigned hash        = Hash(self, key);
 486         size_t   bucknum     = hash & hashmask;
 487
 488 #ifndef NDEBUG
 489         self->entries_version++;
 490 #endif
 491
 492         while(1) {
 493                 HashSetEntry *entry = & self->entries[bucknum];
 494
 495                 if(EntryIsEmpty(*entry)) {
 496                         return;
 497                 }
 498                 if(EntryIsDeleted(*entry)) {
 499                         // entry is deleted
 500                 } else if(EntryGetHash(self, *entry) == hash) {
 501                         if(KeysEqual(self, GetKey(EntryGetValue(*entry)), key)) {
 502                                 EntrySetDeleted(*entry);
 503                                 self->num_deleted++;
 504                                 self->consider_shrink = 1;
 505                                 return;
 506                         }
 507                 }
 508
 509                 ++num_probes;
 510                 bucknum = (bucknum + JUMP(num_probes)) & hashmask;
 511                 assert(num_probes < num_buckets);
 512         }
 513 }
 514
 515 /**
 516  * Initializes hashset with a specific size
 517  * @internal
 518  */
 519 static INLINE
 520 void init_size(HashSet *self, size_t initial_size)
 521 {
 522         if(initial_size < 4)
 523                 initial_size = 4;
 524
 525         self->entries         = Alloc(initial_size);
 526         SetRangeEmpty(self->entries, initial_size);
 527         self->num_buckets     = initial_size;
 528         self->consider_shrink = 0;
 529         self->num_elements    = 0;
 530         self->num_deleted     = 0;
 531 #ifndef NDEBUG
 532         self->entries_version = 0;
 533 #endif
 534 #ifdef ADDITIONAL_INIT
 535         ADDITIONAL_INIT
 536 #endif
 537
 538         reset_thresholds(self);
 539 }
 540
 541 /**
 542  * Initializes a hashset with the default size. The memory for the set has to
 543  * already allocated.
 544  */
 545 void hashset_init(HashSet *self)
 546 {
 547         init_size(self, HT_MIN_BUCKETS);
 548 }
 549
 550 /**
 551  * Destroys a hashset, freeing all used memory (except the memory for the
 552  * HashSet struct itself).
 553  */
 554 void hashset_destroy(HashSet *self)
 555 {
 556 #ifdef ADDITIONAL_TERM
 557         ADDITIONAL_TERM
 558 #endif
 559         Free(self->entries);
 560 #ifndef NDEBUG
 561         self->entries = NULL;
 562 #endif
 563 }
 564
 565 /**
 566  * Initializes a hashset expecting expected_element size.
 567  */
 568 void hashset_init_size(HashSet *self, size_t expected_elements)
 569 {
 570         size_t needed_size;
 571         size_t po2size;
 572
 573         if(expected_elements >= UINT_MAX/2) {
 574                 abort();
 575         }
 576
 577         needed_size = expected_elements * HT_1_DIV_OCCUPANCY_FLT;
 578         po2size     = ceil_po2(needed_size);
 579         init_size(self, po2size);
 580 }
 581
 582 #ifndef NO_ITERATOR
 583 /**
 584  * Initializes a hashset iterator. The memory for the allocator has to be
 585  * already allocated.
 586  * @note it is not allowed to remove or insert elements while iterating
 587  */
 588 void hashset_iterator_init(HashSetIterator *self, const HashSet *hashset)
 589 {
 590         self->current_bucket = hashset->entries - 1;
 591         self->end            = hashset->entries + hashset->num_buckets;
 592 #ifndef NDEBUG
 593         self->set             = hashset;
 594         self->entries_version = hashset->entries_version;
 595 #endif
 596 }
 597
 598 /**
 599  * Returns the next value in the iterator or NULL if no value is left
 600  * in the hashset.
 601  * @note it is not allowed to remove or insert elements while iterating
 602  */
 603 ValueType hashset_iterator_next(HashSetIterator *self)
 604 {
 605         HashSetEntry *current_bucket = self->current_bucket;
 606         HashSetEntry *end            = self->end;
 607
 608         /* using hashset_insert or hashset_remove is not allowed while iterating */
 609         assert(self->entries_version == self->set->entries_version);
 610
 611         do {
 612                 current_bucket++;
 613                 if(current_bucket >= end)
 614                         return NullValue;
 615         } while(EntryIsEmpty(*current_bucket) || EntryIsDeleted(*current_bucket));
 616
 617         self->current_bucket = current_bucket;
 618         return EntryGetValue(*current_bucket);
 619 }
 620
 621 /**
 622  * Removes the element the iterator points to. Removing an element a second time
 623  * has no result.
 624  */
 625 void hashset_remove_iterator(HashSet *self, const HashSetIterator *iter)
 626 {
 627         HashSetEntry *entry = iter->current_bucket;
 628
 629         /* iterator_next needs to have been called at least once */
 630         assert(entry >= self->entries);
 631         /* needs to be on a valid element */
 632         assert(entry < self->entries + self->num_buckets);
 633
 634         if(EntryIsDeleted(*entry))
 635                 return;
 636
 637         EntrySetDeleted(*entry);
 638         self->num_deleted++;
 639         self->consider_shrink = 1;
 640 }
 641 #endif /* NO_ITERATOR */
 642
 643 #endif /* HashSet */