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 /**
 325  * Resize the hashset
 326  * @internal
 327  */
 328 static INLINE
 329 void resize(HashSet *self, size_t new_size)
 330 {
 331         size_t num_buckets = self->num_buckets;
 332         size_t i;
 333         HashSetEntry *old_entries = self->entries;
 334         HashSetEntry *new_entries;
 335
 336         /* allocate a new array with double size */
 337         new_entries = Alloc(new_size);
 338         SetRangeEmpty(new_entries, new_size);
 339
 340         /* use the new array */
 341         self->entries      = new_entries;
 342         self->num_buckets  = new_size;
 343         self->num_elements = 0;
 344         self->num_deleted  = 0;
 345 #ifndef NDEBUG
 346         self->entries_version++;
 347 #endif
 348         reset_thresholds(self);
 349
 350         /* reinsert all elements */
 351         for(i = 0; i < num_buckets; ++i) {
 352                 HashSetEntry *entry = & old_entries[i];
 353                 if(EntryIsEmpty(*entry) || EntryIsDeleted(*entry))
 354                         continue;
 355
 356                 insert_new(self, EntryGetHash(self, *entry), EntryGetValue(*entry));
 357         }
 358
 359         /* now we can free the old array */
 360         Free(old_entries);
 361 }
 362
 363 /**
 364  * grow the hashset if adding 1 more elements would make it too crowded
 365  * @internal
 366  */
 367 static INLINE
 368 void maybe_grow(HashSet *self)
 369 {
 370         size_t resize_to;
 371
 372         if(LIKELY(self->num_elements + 1 <= self->enlarge_threshold))
 373                 return;
 374
 375         /* double table size */
 376         resize_to = self->num_buckets * 2;
 377         resize(self, resize_to);
 378 }
 379
 380 /**
 381  * shrink the hashset if it is only sparsely filled
 382  * @internal
 383  */
 384 static INLINE
 385 void maybe_shrink(HashSet *self)
 386 {
 387         size_t size;
 388         size_t resize_to;
 389
 390         if(!self->consider_shrink)
 391                 return;
 392
 393         self->consider_shrink = 0;
 394         size                  = hashset_size(self);
 395         if(size <= HT_MIN_BUCKETS)
 396                 return;
 397
 398         if(LIKELY(size > self->shrink_threshold))
 399                 return;
 400
 401         resize_to = ceil_po2(size);
 402
 403         if(resize_to < 4)
 404                 resize_to = 4;
 405
 406         resize(self, resize_to);
 407 }
 408
 409 /**
 410  * Insert an element into the hashset. If no element with the given key exists yet,
 411  * then a new one is created and initialized with the InitData function.
 412  * Otherwise the existing element is returned (for hashs where key is equal to
 413  * value, nothing is returned.)
 414  *
 415  * @param self   the hashset
 416  * @param key    the key that identifies the data
 417  * @returns      the existing or newly created data element (or nothing in case of hashs where keys are the while value)
 418  */
 419 InsertReturnValue hashset_insert(HashSet *self, KeyType key)
 420 {
 421 #ifndef NDEBUG
 422         self->entries_version++;
 423 #endif
 424
 425         maybe_shrink(self);
 426         maybe_grow(self);
 427         return insert_nogrow(self, key);
 428 }
 429
 430 /**
 431  * Searches for an element with key @p key.
 432  *
 433  * @param self      the hashset
 434  * @param key       the key to search for
 435  * @returns         the found value or NullValue if nothing was found
 436  */
 437 InsertReturnValue hashset_find(const HashSet *self, ConstKeyType key)
 438 {
 439         size_t   num_probes  = 0;
 440         size_t   num_buckets = self->num_buckets;
 441         size_t   hashmask    = num_buckets - 1;
 442         unsigned hash        = Hash(self, key);
 443         size_t   bucknum     = hash & hashmask;
 444
 445         while(1) {
 446                 HashSetEntry *entry = & self->entries[bucknum];
 447
 448                 if(EntryIsEmpty(*entry)) {
 449                         return NullReturnValue;
 450                 }
 451                 if(EntryIsDeleted(*entry)) {
 452                         // value is deleted
 453                 } else if(EntryGetHash(self, *entry) == hash) {
 454                         if(KeysEqual(self, GetKey(EntryGetValue(*entry)), key)) {
 455                                 // found the value
 456                                 return GetInsertReturnValue(*entry, 1);
 457                         }
 458                 }
 459
 460                 ++num_probes;
 461                 bucknum = (bucknum + JUMP(num_probes)) & hashmask;
 462                 assert(num_probes < num_buckets);
 463         }
 464 }
 465
 466 /**
 467  * Removes an element from a hashset. Does nothing if the set doesn't contain
 468  * the element.
 469  *
 470  * @param self    the hashset
 471  * @param key     key that identifies the data to remove
 472  */
 473 void hashset_remove(HashSet *self, ConstKeyType key)
 474 {
 475         size_t   num_probes  = 0;
 476         size_t   num_buckets = self->num_buckets;
 477         size_t   hashmask    = num_buckets - 1;
 478         unsigned hash        = Hash(self, key);
 479         size_t   bucknum     = hash & hashmask;
 480
 481 #ifndef NDEBUG
 482         self->entries_version++;
 483 #endif
 484
 485         while(1) {
 486                 HashSetEntry *entry = & self->entries[bucknum];
 487
 488                 if(EntryIsEmpty(*entry)) {
 489                         return;
 490                 }
 491                 if(EntryIsDeleted(*entry)) {
 492                         // entry is deleted
 493                 } else if(EntryGetHash(self, *entry) == hash) {
 494                         if(KeysEqual(self, GetKey(EntryGetValue(*entry)), key)) {
 495                                 EntrySetDeleted(*entry);
 496                                 self->num_deleted++;
 497                                 self->consider_shrink = 1;
 498                                 return;
 499                         }
 500                 }
 501
 502                 ++num_probes;
 503                 bucknum = (bucknum + JUMP(num_probes)) & hashmask;
 504                 assert(num_probes < num_buckets);
 505         }
 506 }
 507
 508 /**
 509  * Initializes hashset with a specific size
 510  * @internal
 511  */
 512 static INLINE
 513 void init_size(HashSet *self, size_t initial_size)
 514 {
 515         if(initial_size < 4)
 516                 initial_size = 4;
 517
 518         self->entries         = Alloc(initial_size);
 519         SetRangeEmpty(self->entries, initial_size);
 520         self->num_buckets     = initial_size;
 521         self->consider_shrink = 0;
 522         self->num_elements    = 0;
 523         self->num_deleted     = 0;
 524 #ifndef NDEBUG
 525         self->entries_version = 0;
 526 #endif
 527 #ifdef ADDITIONAL_INIT
 528         ADDITIONAL_INIT
 529 #endif
 530
 531         reset_thresholds(self);
 532 }
 533
 534 /**
 535  * Initializes a hashset with the default size. The memory for the set has to
 536  * already allocated.
 537  */
 538 void hashset_init(HashSet *self)
 539 {
 540         init_size(self, HT_MIN_BUCKETS);
 541 }
 542
 543 /**
 544  * Destroys a hashset, freeing all used memory (except the memory for the
 545  * HashSet struct itself).
 546  */
 547 void hashset_destroy(HashSet *self)
 548 {
 549 #ifdef ADDITIONAL_TERM
 550         ADDITIONAL_TERM
 551 #endif
 552         Free(self->entries);
 553 #ifndef NDEBUG
 554         self->entries = NULL;
 555 #endif
 556 }
 557
 558 /**
 559  * Initializes a hashset expecting expected_element size.
 560  */
 561 void hashset_init_size(HashSet *self, size_t expected_elements)
 562 {
 563         size_t needed_size;
 564         size_t po2size;
 565
 566         if(expected_elements >= UINT_MAX/2) {
 567                 abort();
 568         }
 569
 570         needed_size = expected_elements * HT_1_DIV_OCCUPANCY_FLT;
 571         po2size     = ceil_po2(needed_size);
 572         init_size(self, po2size);
 573 }
 574
 575 #ifndef NO_ITERATOR
 576 /**
 577  * Initializes a hashset iterator. The memory for the allocator has to be
 578  * already allocated.
 579  * @note it is not allowed to remove or insert elements while iterating
 580  */
 581 void hashset_iterator_init(HashSetIterator *self, const HashSet *hashset)
 582 {
 583         self->current_bucket = hashset->entries - 1;
 584         self->end            = hashset->entries + hashset->num_buckets;
 585 #ifndef NDEBUG
 586         self->set             = hashset;
 587         self->entries_version = hashset->entries_version;
 588 #endif
 589 }
 590
 591 /**
 592  * Returns the next value in the iterator or NULL if no value is left
 593  * in the hashset.
 594  * @note it is not allowed to remove or insert elements while iterating
 595  */
 596 ValueType hashset_iterator_next(HashSetIterator *self)
 597 {
 598         HashSetEntry *current_bucket = self->current_bucket;
 599         HashSetEntry *end            = self->end;
 600
 601         /* using hashset_insert or hashset_remove is not allowed while iterating */
 602         assert(self->entries_version == self->set->entries_version);
 603
 604         do {
 605                 current_bucket++;
 606                 if(current_bucket >= end)
 607                         return NullValue;
 608         } while(EntryIsEmpty(*current_bucket) || EntryIsDeleted(*current_bucket));
 609
 610         self->current_bucket = current_bucket;
 611         return EntryGetValue(*current_bucket);
 612 }
 613
 614 /**
 615  * Removes the element the iterator points to. Removing an element a second time
 616  * has no result.
 617  */
 618 void hashset_remove_iterator(HashSet *self, const HashSetIterator *iter)
 619 {
 620         HashSetEntry *entry = iter->current_bucket;
 621
 622         /* iterator_next needs to have been called at least once */
 623         assert(entry >= self->entries);
 624         /* needs to be on a valid element */
 625         assert(entry < self->entries + self->num_buckets);
 626
 627         if(EntryIsDeleted(*entry))
 628                 return;
 629
 630         EntrySetDeleted(*entry);
 631         self->num_deleted++;
 632         self->consider_shrink = 1;
 633 }
 634 #endif /* NO_ITERATOR */
 635
 636 #endif /* HashSet */