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