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