version2/src/C/hashtable.h

   1 /*      Copyright (c) 2015 Regents of the University of California
   2  *
   3  *      Author: Brian Demsky <bdemsky@uci.edu>
   4  *
   5  *      This program is free software; you can redistribute it and/or
   6  *      modify it under the terms of the GNU General Public License
   7  *      version 2 as published by the Free Software Foundation.
   8  */
   9
  10 /** @file hashtable.h
  11  *  @brief Hashtable.  Standard chained bucket variety.
  12  */
  13
  14 #ifndef HASHTABLE_H__
  15 #define HASHTABLE_H__
  16
  17 #include <stdlib.h>
  18 #include <stdio.h>
  19 #include <string.h>
  20 #include "common.h"
  21 #include "mymemory.h"
  22
  23 /**
  24  * @brief Hashtable node
  25  *
  26  * @tparam _Key    Type name for the key
  27  * @tparam _Val    Type name for the values to be stored
  28  */
  29 template<typename _Key, typename _Val>
  30 struct Hashlistnode {
  31         _Key key;
  32         _Val val;
  33         uint hashcode;
  34 };
  35
  36 template<typename _Key, int _Shift, typename _KeyInt>
  37 inline unsigned int defaultHashFunction(_Key hash) {
  38         return (unsigned int)(((_KeyInt)hash) >> _Shift);
  39 }
  40
  41 template<typename _Key>
  42 inline bool defaultEquals(_Key key1, _Key key2) {
  43         return key1 == key2;
  44 }
  45
  46 /**
  47  * @brief A simple, custom hash table
  48  *
  49  * By default it is snapshotting, but you can pass in your own allocation
  50  * functions. Note that this table does not support the value 0 (NULL) used as
  51  * a key and is designed primarily with pointer-based keys in mind. Other
  52  * primitive key types are supported only for non-zero values.
  53  *
  54  * @tparam _Key    Type name for the key
  55  * @tparam _Val    Type name for the values to be stored
  56  * @tparam _KeyInt Integer type that is at least as large as _Key. Used for key
  57  *                 manipulation and storage.
  58  * @tparam _Shift  Logical shift to apply to all keys. Default 0.
  59  */
  60 template<typename _Key, typename _Val, typename _KeyInt, int _Shift = 0, unsigned int (*hash_function) (_Key) = defaultHashFunction<_Key, _Shift, _KeyInt>, bool (*equals) (_Key, _Key) = defaultEquals<_Key> >
  61 class Hashtable {
  62 public:
  63         /**
  64          * @brief Hash table constructor
  65          * @param initialcapacity Sets the initial capacity of the hash table.
  66          * Default size 1024.
  67          * @param factor Sets the percentage full before the hashtable is
  68          * resized. Default ratio 0.5.
  69          */
  70         Hashtable(unsigned int initialcapacity = 1024, double factor = 0.5) {
  71                 // Allocate space for the hash table
  72                 table = (struct Hashlistnode<_Key, _Val> *)ourcalloc(initialcapacity, sizeof(struct Hashlistnode<_Key, _Val>));
  73                 zero = NULL;
  74                 loadfactor = factor;
  75                 capacity = initialcapacity;
  76                 capacitymask = initialcapacity - 1;
  77
  78                 threshold = (unsigned int)(initialcapacity * loadfactor);
  79                 size = 0;                                                       // Initial number of elements in the hash
  80         }
  81
  82         /** @brief Hash table destructor */
  83         ~Hashtable() {
  84                 ourfree(table);
  85                 if (zero)
  86                         ourfree(zero);
  87         }
  88
  89         /** Override: new operator */
  90         void *operator new(size_t size) {
  91                 return ourmalloc(size);
  92         }
  93
  94         /** Override: delete operator */
  95         void operator delete(void *p, size_t size) {
  96                 ourfree(p);
  97         }
  98
  99         /** Override: new[] operator */
 100         void *operator new[](size_t size) {
 101                 return ourmalloc(size);
 102         }
 103
 104         /** Override: delete[] operator */
 105         void operator delete[](void *p, size_t size) {
 106                 ourfree(p);
 107         }
 108
 109         /** @brief Reset the table to its initial state. */
 110         void reset() {
 111                 memset(table, 0, capacity * sizeof(struct Hashlistnode<_Key, _Val>));
 112                 if (zero) {
 113                         ourfree(zero);
 114                         zero = NULL;
 115                 }
 116                 size = 0;
 117         }
 118
 119         /** Doesn't work with zero value */
 120         _Val getRandomValue() {
 121                 while (true) {
 122                         unsigned int index = random() & capacitymask;
 123                         struct Hashlistnode<_Key, _Val> *bin = &table[index];
 124                         if (bin->key != NULL && bin->val != NULL) {
 125                                 return bin->val;
 126                         }
 127                 }
 128         }
 129
 130         void resetAndDeleteKeys() {
 131                 for (unsigned int i = 0; i < capacity; i++) {
 132                         struct Hashlistnode<_Key, _Val> *bin = &table[i];
 133                         if (bin->key != NULL) {
 134                                 delete bin->key;
 135                                 bin->key = NULL;
 136                                 if (bin->val != NULL) {
 137                                         bin->val = NULL;
 138                                 }
 139                         }
 140                 }
 141                 if (zero) {
 142                         ourfree(zero);
 143                         zero = NULL;
 144                 }
 145                 size = 0;
 146         }
 147
 148         void resetAndDeleteVals() {
 149                 for (unsigned int i = 0; i < capacity; i++) {
 150                         struct Hashlistnode<_Key, _Val> *bin = &table[i];
 151                         if (bin->key != NULL) {
 152                                 bin->key = NULL;
 153                                 if (bin->val != NULL) {
 154                                         delete bin->val;
 155                                         bin->val = NULL;
 156                                 }
 157                         }
 158                 }
 159                 if (zero) {
 160                         if (zero->val != NULL)
 161                                 delete zero->val;
 162                         ourfree(zero);
 163                         zero = NULL;
 164                 }
 165                 size = 0;
 166         }
 167
 168         void resetAndFreeVals() {
 169                 for (unsigned int i = 0; i < capacity; i++) {
 170                         struct Hashlistnode<_Key, _Val> *bin = &table[i];
 171                         if (bin->key != NULL) {
 172                                 bin->key = NULL;
 173                                 if (bin->val != NULL) {
 174                                         ourfree(bin->val);
 175                                         bin->val = NULL;
 176                                 }
 177                         }
 178                 }
 179                 if (zero) {
 180                         if (zero->val != NULL)
 181                                 ourfree(zero->val);
 182                         ourfree(zero);
 183                         zero = NULL;
 184                 }
 185                 size = 0;
 186         }
 187
 188         /**
 189          * @brief Put a key/value pair into the table
 190          * @param key The key for the new value; must not be 0 or NULL
 191          * @param val The value to store in the table
 192          */
 193         void put(_Key key, _Val val) {
 194                 /* Hashtable cannot handle 0 as a key */
 195                 if (!key) {
 196                         if (!zero) {
 197                                 zero = (struct Hashlistnode<_Key, _Val> *)ourmalloc(sizeof(struct Hashlistnode<_Key, _Val>));
 198                                 size++;
 199                         }
 200                         zero->key = key;
 201                         zero->val = val;
 202                         return;
 203                 }
 204
 205                 if (size > threshold)
 206                         resize(capacity << 1);
 207
 208                 struct Hashlistnode<_Key, _Val> *search;
 209
 210                 unsigned int hashcode = hash_function(key);
 211                 unsigned int index = hashcode;
 212                 do {
 213                         index &= capacitymask;
 214                         search = &table[index];
 215                         if (!search->key) {
 216                                 //key is null, probably done
 217                                 break;
 218                         }
 219                         if (search->hashcode == hashcode)
 220                                 if (equals(search->key, key)) {
 221                                         search->val = val;
 222                                         return;
 223                                 }
 224                         index++;
 225                 } while (true);
 226
 227                 search->key = key;
 228                 search->val = val;
 229                 search->hashcode = hashcode;
 230                 size++;
 231         }
 232
 233         /**
 234          * @brief Lookup the corresponding value for the given key
 235          * @param key The key for finding the value; must not be 0 or NULL
 236          * @return The value in the table, if the key is found; otherwise 0
 237          */
 238         _Val get(_Key key) const {
 239                 struct Hashlistnode<_Key, _Val> *search;
 240
 241                 /* Hashtable cannot handle 0 as a key */
 242                 if (!key) {
 243                         if (zero)
 244                                 return zero->val;
 245                         else
 246                                 return (_Val) 0;
 247                 }
 248
 249                 unsigned int hashcode = hash_function(key);
 250                 unsigned int oindex = hashcode & capacitymask;
 251                 unsigned int index = oindex;
 252                 do {
 253                         search = &table[index];
 254                         if (!search->key) {
 255                                 if (!search->val)
 256                                         break;
 257                         } else
 258                         if (hashcode == search->hashcode)
 259                                 if (equals(search->key, key))
 260                                         return search->val;
 261                         index++;
 262                         index &= capacitymask;
 263                         if (index == oindex)
 264                                 break;
 265                 } while (true);
 266                 return (_Val)0;
 267         }
 268
 269         /**
 270          * @brief Remove the given key and return the corresponding value
 271          * @param key The key for finding the value; must not be 0 or NULL
 272          * @return The value in the table, if the key is found; otherwise 0
 273          */
 274         _Val remove(_Key key) {
 275                 struct Hashlistnode<_Key, _Val> *search;
 276
 277                 /* Hashtable cannot handle 0 as a key */
 278                 if (!key) {
 279                         if (!zero) {
 280                                 return (_Val)0;
 281                         } else {
 282                                 _Val v = zero->val;
 283                                 ourfree(zero);
 284                                 zero = NULL;
 285                                 size--;
 286                                 return v;
 287                         }
 288                 }
 289
 290
 291                 unsigned int hashcode = hash_function(key);
 292                 unsigned int index = hashcode;
 293                 do {
 294                         index &= capacitymask;
 295                         search = &table[index];
 296                         if (!search->key) {
 297                                 if (!search->val)
 298                                         break;
 299                         } else
 300                         if (hashcode == search->hashcode)
 301                                 if (equals(search->key, key)) {
 302                                         _Val v = search->val;
 303                                         //empty out this bin
 304                                         search->val = (_Val) 1;
 305                                         search->key = 0;
 306                                         size--;
 307                                         return v;
 308                                 }
 309                         index++;
 310                 } while (true);
 311                 return (_Val)0;
 312         }
 313
 314         unsigned int getSize() const {
 315                 return size;
 316         }
 317
 318
 319         /**
 320          * @brief Check whether the table contains a value for the given key
 321          * @param key The key for finding the value; must not be 0 or NULL
 322          * @return True, if the key is found; false otherwise
 323          */
 324         bool contains(_Key key) const {
 325                 struct Hashlistnode<_Key, _Val> *search;
 326
 327                 /* Hashtable cannot handle 0 as a key */
 328                 if (!key) {
 329                         return zero != NULL;
 330                 }
 331
 332                 unsigned int index = hash_function(key);
 333                 unsigned int hashcode = index;
 334                 do {
 335                         index &= capacitymask;
 336                         search = &table[index];
 337                         if (!search->key) {
 338                                 if (!search->val)
 339                                         break;
 340                         } else
 341                         if (hashcode == search->hashcode)
 342                                 if (equals(search->key, key))
 343                                         return true;
 344                         index++;
 345                 } while (true);
 346                 return false;
 347         }
 348
 349         /**
 350          * @brief Resize the table
 351          * @param newsize The new size of the table
 352          */
 353         void resize(unsigned int newsize) {
 354                 struct Hashlistnode<_Key, _Val> *oldtable = table;
 355                 struct Hashlistnode<_Key, _Val> *newtable;
 356                 unsigned int oldcapacity = capacity;
 357
 358                 if ((newtable = (struct Hashlistnode<_Key, _Val> *)ourcalloc(newsize, sizeof(struct Hashlistnode<_Key, _Val>))) == NULL) {
 359                         model_print("calloc error %s %d\n", __FILE__, __LINE__);
 360                         exit(EXIT_FAILURE);
 361                 }
 362
 363                 table = newtable;                                                                                       // Update the global hashtable upon resize()
 364                 capacity = newsize;
 365                 capacitymask = newsize - 1;
 366
 367                 threshold = (unsigned int)(newsize * loadfactor);
 368
 369                 struct Hashlistnode<_Key, _Val> *bin = &oldtable[0];
 370                 struct Hashlistnode<_Key, _Val> *lastbin = &oldtable[oldcapacity];
 371                 for (; bin < lastbin; bin++) {
 372                         _Key key = bin->key;
 373
 374                         struct Hashlistnode<_Key, _Val> *search;
 375                         if (!key)
 376                                 continue;
 377
 378                         unsigned int hashcode = bin->hashcode;
 379                         unsigned int index = hashcode;
 380                         do {
 381                                 index &= capacitymask;
 382                                 search = &table[index];
 383                                 index++;
 384                         } while (search->key);
 385
 386                         search->hashcode = hashcode;
 387                         search->key = key;
 388                         search->val = bin->val;
 389                 }
 390
 391                 ourfree(oldtable);                                                                                              // Free the memory of the old hash table
 392         }
 393         double getLoadFactor() {return loadfactor;}
 394         unsigned int getCapacity() {return capacity;}
 395         struct Hashlistnode<_Key, _Val> *table;
 396         struct Hashlistnode<_Key, _Val> *zero;
 397         unsigned int capacity;
 398         unsigned int size;
 399 private:
 400         unsigned int capacitymask;
 401         unsigned int threshold;
 402         double loadfactor;
 403 };
 404
 405 #endif/* __HASHTABLE_H__ */