1 /*
2 * Copyright (c) 2004, Stefan Walter
3 * All rights reserved.
4 *
5 * Redistribution and use in source and binary forms, with or without
6 * modification, are permitted provided that the following conditions
7 * are met:
8 *
9 * * Redistributions of source code must retain the above
10 * copyright notice, this list of conditions and the
11 * following disclaimer.
12 * * Redistributions in binary form must reproduce the
13 * above copyright notice, this list of conditions and
14 * the following disclaimer in the documentation and/or
15 * other materials provided with the distribution.
16 * * The names of contributors to this software may not be
17 * used to endorse or promote products derived from this
18 * software without specific prior written permission.
19 *
20 * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
21 * "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
22 * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS
23 * FOR A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE
24 * COPYRIGHT OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT,
25 * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING,
26 * BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS
27 * OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED
28 * AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY,
29 * OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF
30 * THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH
31 * DAMAGE.
32 */
33
34 /*
35 * Originally from apache 2.0
36 * Modifications for general use by <stef@memberwebs.com>
37 */
38
39 /* Copyright 2000-2004 The Apache Software Foundation
40 *
41 * Licensed under the Apache License, Version 2.0 (the "License");
42 * you may not use this file except in compliance with the License.
43 * You may obtain a copy of the License at
44 *
45 * http://www.apache.org/licenses/LICENSE-2.0
46 *
47 * Unless required by applicable law or agreed to in writing, software
48 * distributed under the License is distributed on an "AS IS" BASIS,
49 * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
50 * See the License for the specific language governing permissions and
51 * limitations under the License.
52 */
53
54 #include <sys/types.h>
55 #include <stdlib.h>
56 #include <string.h>
57 #include "hash.h"
58
59 #define KEY_DATA(he) ((he)->key)
60
61 /*
62 * The internal form of a hash table.
63 *
64 * The table is an array indexed by the hash of the key; collisions
65 * are resolved by hanging a linked list of hash entries off each
66 * element of the array. Although this is a really simple design it
67 * isn't too bad given that pools have a low allocation overhead.
68 */
69
70 typedef struct hsh_entry_t hsh_entry_t;
71
72 struct hsh_entry_t
73 {
74 hsh_entry_t* next;
75 unsigned int hash;
76 const void* key;
77 size_t klen;
78 const void* val;
79 };
80
81 /*
82 * Data structure for iterating through a hash table.
83 *
84 * We keep a pointer to the next hash entry here to allow the current
85 * hash entry to be freed or otherwise mangled between calls to
86 * hsh_next().
87 */
88 struct hsh_index_t
89 {
90 hsh_t* ht;
91 hsh_entry_t* ths;
92 hsh_entry_t* next;
93 unsigned int index;
94 };
95
96 /*
97 * The size of the array is always a power of two. We use the maximum
98 * index rather than the size so that we can use bitwise-AND for
99 * modular arithmetic.
100 * The count of hash entries may be greater depending on the chosen
101 * collision rate.
102 */
103 struct hsh_t
104 {
105 hsh_entry_t** array;
106 hsh_index_t iterator; /* For hsh_first(...) */
107 unsigned int count;
108 unsigned int max;
109 };
110
111
112 #define INITIAL_MAX 15 /* tunable == 2^n - 1 */
113 #define int_malloc malloc
114 #define int_calloc calloc
115 #define int_free free
116
117 /*
118 * Hash creation functions.
119 */
120
alloc_array(hsh_t * ht,unsigned int max)121 static hsh_entry_t** alloc_array(hsh_t* ht, unsigned int max)
122 {
123 return (hsh_entry_t**)int_calloc(sizeof(*(ht->array)), (max + 1));
124 }
125
hsh_create()126 hsh_t* hsh_create()
127 {
128 hsh_t* ht = int_malloc(sizeof(hsh_t));
129 if(ht)
130 {
131 ht->count = 0;
132 ht->max = INITIAL_MAX;
133 ht->array = alloc_array(ht, ht->max);
134 if(!ht->array)
135 {
136 int_free(ht);
137 return NULL;
138 }
139 }
140 return ht;
141 }
142
hsh_free(hsh_t * ht)143 void hsh_free(hsh_t* ht)
144 {
145 hsh_index_t* hi;
146
147 for(hi = hsh_first(ht); hi; hi = hsh_next(hi))
148 int_free(hi->ths);
149
150 if(ht->array)
151 int_free(ht->array);
152
153 int_free(ht);
154 }
155
156 /*
157 * Hash iteration functions.
158 */
159
hsh_next(hsh_index_t * hi)160 hsh_index_t* hsh_next(hsh_index_t* hi)
161 {
162 hi->ths = hi->next;
163 while(!hi->ths)
164 {
165 if(hi->index > hi->ht->max)
166 return NULL;
167
168 hi->ths = hi->ht->array[hi->index++];
169 }
170 hi->next = hi->ths->next;
171 return hi;
172 }
173
hsh_first(hsh_t * ht)174 hsh_index_t* hsh_first(hsh_t* ht)
175 {
176 hsh_index_t* hi = &ht->iterator;
177
178 hi->ht = ht;
179 hi->index = 0;
180 hi->ths = NULL;
181 hi->next = NULL;
182 return hsh_next(hi);
183 }
184
hsh_this(hsh_index_t * hi,const void ** key,size_t * klen)185 void* hsh_this(hsh_index_t* hi, const void** key, size_t* klen)
186 {
187 if(key)
188 *key = KEY_DATA(hi->ths);
189 if(klen)
190 *klen = hi->ths->klen;
191 return (void*)hi->ths->val;
192 }
193
194
195 /*
196 * Expanding a hash table
197 */
198
expand_array(hsh_t * ht)199 static int expand_array(hsh_t* ht)
200 {
201 hsh_index_t* hi;
202 hsh_entry_t** new_array;
203 unsigned int new_max;
204
205 new_max = ht->max * 2 + 1;
206 new_array = alloc_array(ht, new_max);
207
208 if(!new_array)
209 return 0;
210
211 for(hi = hsh_first(ht); hi; hi = hsh_next(hi))
212 {
213 unsigned int i = hi->ths->hash & new_max;
214 hi->ths->next = new_array[i];
215 new_array[i] = hi->ths;
216 }
217
218 if(ht->array)
219 free(ht->array);
220
221 ht->array = new_array;
222 ht->max = new_max;
223 return 1;
224 }
225
226 /*
227 * This is where we keep the details of the hash function and control
228 * the maximum collision rate.
229 *
230 * If val is non-NULL it creates and initializes a new hash entry if
231 * there isn't already one there; it returns an updatable pointer so
232 * that hash entries can be removed.
233 */
234
find_entry(hsh_t * ht,const void * key,size_t klen,const void * val)235 static hsh_entry_t** find_entry(hsh_t* ht, const void* key, size_t klen, const void* val)
236 {
237 hsh_entry_t** hep;
238 hsh_entry_t* he;
239 const unsigned char* p;
240 unsigned int hash;
241 size_t i;
242
243 /*
244 * This is the popular `times 33' hash algorithm which is used by
245 * perl and also appears in Berkeley DB. This is one of the best
246 * known hash functions for strings because it is both computed
247 * very fast and distributes very well.
248 *
249 * The originator may be Dan Bernstein but the code in Berkeley DB
250 * cites Chris Torek as the source. The best citation I have found
251 * is "Chris Torek, Hash function for text in C, Usenet message
252 * <27038@mimsy.umd.edu> in comp.lang.c , October, 1990." in Rich
253 * Salz's USENIX 1992 paper about INN which can be found at
254 * <http://citeseer.nj.nec.com/salz92internetnews.html>.
255 *
256 * The magic of number 33, i.e. why it works better than many other
257 * constants, prime or not, has never been adequately explained by
258 * anyone. So I try an explanation: if one experimentally tests all
259 * multipliers between 1 and 256 (as I did while writing a low-level
260 * data structure library some time ago) one detects that even
261 * numbers are not useable at all. The remaining 128 odd numbers
262 * (except for the number 1) work more or less all equally well.
263 * They all distribute in an acceptable way and this way fill a hash
264 * table with an average percent of approx. 86%.
265 *
266 * If one compares the chi^2 values of the variants (see
267 * Bob Jenkins ``Hashing Frequently Asked Questions'' at
268 * http://burtleburtle.net/bob/hash/hashfaq.html for a description
269 * of chi^2), the number 33 not even has the best value. But the
270 * number 33 and a few other equally good numbers like 17, 31, 63,
271 * 127 and 129 have nevertheless a great advantage to the remaining
272 * numbers in the large set of possible multipliers: their multiply
273 * operation can be replaced by a faster operation based on just one
274 * shift plus either a single addition or subtraction operation. And
275 * because a hash function has to both distribute good _and_ has to
276 * be very fast to compute, those few numbers should be preferred.
277 *
278 * -- Ralf S. Engelschall <rse@engelschall.com>
279 */
280 hash = 0;
281
282 if(klen == HSH_KEY_STRING)
283 {
284 for(p = key; *p; p++)
285 hash = hash * 33 + *p;
286
287 klen = p - (const unsigned char *)key;
288 }
289 else
290 {
291 for(p = key, i = klen; i; i--, p++)
292 hash = hash * 33 + *p;
293 }
294
295 /* scan linked list */
296 for(hep = &ht->array[hash & ht->max], he = *hep;
297 he; hep = &he->next, he = *hep)
298 {
299 if(he->hash == hash &&
300 he->klen == klen &&
301 memcmp(KEY_DATA(he), key, klen) == 0)
302 break;
303 }
304
305 if(he || !val)
306 return hep;
307
308 /* add a new entry for non-NULL val */
309 he = int_malloc(sizeof(*he));
310
311 if(he)
312 {
313 /* Key points to external data */
314 he->key = key;
315 he->klen = klen;
316
317 he->next = NULL;
318 he->hash = hash;
319 he->val = val;
320
321 *hep = he;
322 ht->count++;
323 }
324
325 return hep;
326 }
327
hsh_get(hsh_t * ht,const void * key,size_t klen)328 void* hsh_get(hsh_t* ht, const void *key, size_t klen)
329 {
330 hsh_entry_t** he = find_entry(ht, key, klen, NULL);
331
332 if(he && *he)
333 return (void*)((*he)->val);
334 else
335 return NULL;
336 }
337
hsh_set(hsh_t * ht,const void * key,size_t klen,void * val)338 int hsh_set(hsh_t* ht, const void* key, size_t klen, void* val)
339 {
340 hsh_entry_t** hep = find_entry(ht, key, klen, val);
341
342 if(hep && *hep)
343 {
344 /* replace entry */
345 (*hep)->val = val;
346
347 /* check that the collision rate isn't too high */
348 if(ht->count > ht->max)
349 {
350 if(!expand_array(ht))
351 return 0;
352 }
353
354 return 1;
355 }
356
357 return 0;
358 }
359
hsh_rem(hsh_t * ht,const void * key,size_t klen)360 void* hsh_rem(hsh_t* ht, const void* key, size_t klen)
361 {
362 hsh_entry_t** hep = find_entry(ht, key, klen, NULL);
363 void* val = NULL;
364
365 if(hep && *hep)
366 {
367 hsh_entry_t* old = *hep;
368 *hep = (*hep)->next;
369 --ht->count;
370 val = (void*)old->val;
371 free(old);
372 }
373
374 return val;
375 }
376
hsh_clear(hsh_t * ht)377 void hsh_clear(hsh_t* ht)
378 {
379 hsh_entry_t *he, *next;
380 int i;
381
382 /* Free all entries in the array */
383 for (i = 0; i < ht->max; ++i) {
384 he = ht->array[i];
385 while (he) {
386 next = he->next;
387 free (he);
388 he = next;
389 }
390 }
391
392 memset (ht->array, 0, ht->max * sizeof (hsh_entry_t*));
393 ht->count = 0;
394 }
395
hsh_count(hsh_t * ht)396 unsigned int hsh_count(hsh_t* ht)
397 {
398 return ht->count;
399 }
400
401