|
1 /* |
|
2 * DO NOT ALTER OR REMOVE COPYRIGHT NOTICES OR THIS HEADER. |
|
3 * |
|
4 * Copyright 2021 Mike Becker, Olaf Wintermann All rights reserved. |
|
5 * |
|
6 * Redistribution and use in source and binary forms, with or without |
|
7 * modification, are permitted provided that the following conditions are met: |
|
8 * |
|
9 * 1. Redistributions of source code must retain the above copyright |
|
10 * notice, this list of conditions and the following disclaimer. |
|
11 * |
|
12 * 2. Redistributions in binary form must reproduce the above copyright |
|
13 * notice, this list of conditions and the following disclaimer in the |
|
14 * documentation and/or other materials provided with the distribution. |
|
15 * |
|
16 * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" |
|
17 * AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE |
|
18 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE |
|
19 * ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE |
|
20 * LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR |
|
21 * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF |
|
22 * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS |
|
23 * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN |
|
24 * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) |
|
25 * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE |
|
26 * POSSIBILITY OF SUCH DAMAGE. |
|
27 */ |
|
28 |
|
29 #include "cx/hash_map.h" |
|
30 #include "cx/utils.h" |
|
31 |
|
32 #include <string.h> |
|
33 #include <assert.h> |
|
34 |
|
35 struct cx_hash_map_element_s { |
|
36 /** A pointer to the next element in the current bucket. */ |
|
37 struct cx_hash_map_element_s *next; |
|
38 |
|
39 /** The corresponding key. */ |
|
40 CxHashKey key; |
|
41 |
|
42 /** The value data. */ |
|
43 char data[]; |
|
44 }; |
|
45 |
|
46 static void cx_hash_map_clear(struct cx_map_s *map) { |
|
47 struct cx_hash_map_s *hash_map = (struct cx_hash_map_s *) map; |
|
48 cx_for_n(i, hash_map->bucket_count) { |
|
49 struct cx_hash_map_element_s *elem = hash_map->buckets[i]; |
|
50 if (elem != NULL) { |
|
51 do { |
|
52 struct cx_hash_map_element_s *next = elem->next; |
|
53 // invoke the destructor |
|
54 cx_invoke_destructor(map, elem->data); |
|
55 // free the key data |
|
56 cxFree(map->allocator, (void *) elem->key.data); |
|
57 // free the node |
|
58 cxFree(map->allocator, elem); |
|
59 // proceed |
|
60 elem = next; |
|
61 } while (elem != NULL); |
|
62 |
|
63 // do not leave a dangling pointer |
|
64 hash_map->buckets[i] = NULL; |
|
65 } |
|
66 } |
|
67 map->size = 0; |
|
68 } |
|
69 |
|
70 static void cx_hash_map_destructor(struct cx_map_s *map) { |
|
71 struct cx_hash_map_s *hash_map = (struct cx_hash_map_s *) map; |
|
72 |
|
73 // free the buckets |
|
74 cx_hash_map_clear(map); |
|
75 cxFree(map->allocator, hash_map->buckets); |
|
76 |
|
77 // free the map structure |
|
78 cxFree(map->allocator, map); |
|
79 } |
|
80 |
|
81 static int cx_hash_map_put( |
|
82 CxMap *map, |
|
83 CxHashKey key, |
|
84 void *value |
|
85 ) { |
|
86 struct cx_hash_map_s *hash_map = (struct cx_hash_map_s *) map; |
|
87 CxAllocator const *allocator = map->allocator; |
|
88 |
|
89 unsigned hash = key.hash; |
|
90 if (hash == 0) { |
|
91 cx_hash_murmur(&key); |
|
92 hash = key.hash; |
|
93 } |
|
94 |
|
95 size_t slot = hash % hash_map->bucket_count; |
|
96 struct cx_hash_map_element_s *elm = hash_map->buckets[slot]; |
|
97 struct cx_hash_map_element_s *prev = NULL; |
|
98 |
|
99 while (elm != NULL && elm->key.hash < hash) { |
|
100 prev = elm; |
|
101 elm = elm->next; |
|
102 } |
|
103 |
|
104 if (elm != NULL && elm->key.hash == hash && elm->key.len == key.len && |
|
105 memcmp(elm->key.data, key.data, key.len) == 0) { |
|
106 // overwrite existing element |
|
107 if (map->store_pointer) { |
|
108 memcpy(elm->data, &value, sizeof(void *)); |
|
109 } else { |
|
110 memcpy(elm->data, value, map->item_size); |
|
111 } |
|
112 } else { |
|
113 // allocate new element |
|
114 struct cx_hash_map_element_s *e = cxMalloc( |
|
115 allocator, |
|
116 sizeof(struct cx_hash_map_element_s) + map->item_size |
|
117 ); |
|
118 if (e == NULL) { |
|
119 return -1; |
|
120 } |
|
121 |
|
122 // write the value |
|
123 if (map->store_pointer) { |
|
124 memcpy(e->data, &value, sizeof(void *)); |
|
125 } else { |
|
126 memcpy(e->data, value, map->item_size); |
|
127 } |
|
128 |
|
129 // copy the key |
|
130 void *kd = cxMalloc(allocator, key.len); |
|
131 if (kd == NULL) { |
|
132 return -1; |
|
133 } |
|
134 memcpy(kd, key.data, key.len); |
|
135 e->key.data = kd; |
|
136 e->key.len = key.len; |
|
137 e->key.hash = hash; |
|
138 |
|
139 // insert the element into the linked list |
|
140 if (prev == NULL) { |
|
141 hash_map->buckets[slot] = e; |
|
142 } else { |
|
143 prev->next = e; |
|
144 } |
|
145 e->next = elm; |
|
146 |
|
147 // increase the size |
|
148 map->size++; |
|
149 } |
|
150 |
|
151 return 0; |
|
152 } |
|
153 |
|
154 static void cx_hash_map_unlink( |
|
155 struct cx_hash_map_s *hash_map, |
|
156 size_t slot, |
|
157 struct cx_hash_map_element_s *prev, |
|
158 struct cx_hash_map_element_s *elm |
|
159 ) { |
|
160 // unlink |
|
161 if (prev == NULL) { |
|
162 hash_map->buckets[slot] = elm->next; |
|
163 } else { |
|
164 prev->next = elm->next; |
|
165 } |
|
166 // free element |
|
167 cxFree(hash_map->base.allocator, (void *) elm->key.data); |
|
168 cxFree(hash_map->base.allocator, elm); |
|
169 // decrease size |
|
170 hash_map->base.size--; |
|
171 } |
|
172 |
|
173 /** |
|
174 * Helper function to avoid code duplication. |
|
175 * |
|
176 * @param map the map |
|
177 * @param key the key to look up |
|
178 * @param remove flag indicating whether the looked up entry shall be removed |
|
179 * @param destroy flag indicating whether the destructor shall be invoked |
|
180 * @return a pointer to the value corresponding to the key or \c NULL |
|
181 */ |
|
182 static void *cx_hash_map_get_remove( |
|
183 CxMap *map, |
|
184 CxHashKey key, |
|
185 bool remove, |
|
186 bool destroy |
|
187 ) { |
|
188 struct cx_hash_map_s *hash_map = (struct cx_hash_map_s *) map; |
|
189 |
|
190 unsigned hash = key.hash; |
|
191 if (hash == 0) { |
|
192 cx_hash_murmur(&key); |
|
193 hash = key.hash; |
|
194 } |
|
195 |
|
196 size_t slot = hash % hash_map->bucket_count; |
|
197 struct cx_hash_map_element_s *elm = hash_map->buckets[slot]; |
|
198 struct cx_hash_map_element_s *prev = NULL; |
|
199 while (elm && elm->key.hash <= hash) { |
|
200 if (elm->key.hash == hash && elm->key.len == key.len) { |
|
201 if (memcmp(elm->key.data, key.data, key.len) == 0) { |
|
202 void *data = NULL; |
|
203 if (destroy) { |
|
204 cx_invoke_destructor(map, elm->data); |
|
205 } else { |
|
206 if (map->store_pointer) { |
|
207 data = *(void **) elm->data; |
|
208 } else { |
|
209 data = elm->data; |
|
210 } |
|
211 } |
|
212 if (remove) { |
|
213 cx_hash_map_unlink(hash_map, slot, prev, elm); |
|
214 } |
|
215 return data; |
|
216 } |
|
217 } |
|
218 prev = elm; |
|
219 elm = prev->next; |
|
220 } |
|
221 |
|
222 return NULL; |
|
223 } |
|
224 |
|
225 static void *cx_hash_map_get( |
|
226 CxMap const *map, |
|
227 CxHashKey key |
|
228 ) { |
|
229 // we can safely cast, because we know the map stays untouched |
|
230 return cx_hash_map_get_remove((CxMap *) map, key, false, false); |
|
231 } |
|
232 |
|
233 static void *cx_hash_map_remove( |
|
234 CxMap *map, |
|
235 CxHashKey key, |
|
236 bool destroy |
|
237 ) { |
|
238 return cx_hash_map_get_remove(map, key, true, destroy); |
|
239 } |
|
240 |
|
241 static void *cx_hash_map_iter_current_entry(void const *it) { |
|
242 struct cx_iterator_s const *iter = it; |
|
243 // struct has to have a compatible signature |
|
244 return (struct cx_map_entry_s *) &(iter->kv_data); |
|
245 } |
|
246 |
|
247 static void *cx_hash_map_iter_current_key(void const *it) { |
|
248 struct cx_iterator_s const *iter = it; |
|
249 struct cx_hash_map_element_s *elm = iter->elem_handle; |
|
250 return &elm->key; |
|
251 } |
|
252 |
|
253 static void *cx_hash_map_iter_current_value(void const *it) { |
|
254 struct cx_iterator_s const *iter = it; |
|
255 struct cx_hash_map_s const *map = iter->src_handle; |
|
256 struct cx_hash_map_element_s *elm = iter->elem_handle; |
|
257 if (map->base.store_pointer) { |
|
258 return *(void **) elm->data; |
|
259 } else { |
|
260 return elm->data; |
|
261 } |
|
262 } |
|
263 |
|
264 static bool cx_hash_map_iter_valid(void const *it) { |
|
265 struct cx_iterator_s const *iter = it; |
|
266 return iter->elem_handle != NULL; |
|
267 } |
|
268 |
|
269 static void cx_hash_map_iter_next(void *it) { |
|
270 struct cx_iterator_s *iter = it; |
|
271 struct cx_hash_map_element_s *elm = iter->elem_handle; |
|
272 |
|
273 // remove current element, if asked |
|
274 if (iter->base.remove) { |
|
275 // obtain mutable pointer to the map |
|
276 struct cx_mut_iterator_s *miter = it; |
|
277 struct cx_hash_map_s *map = miter->src_handle; |
|
278 |
|
279 // clear the flag |
|
280 iter->base.remove = false; |
|
281 |
|
282 // determine the next element |
|
283 struct cx_hash_map_element_s *next = elm->next; |
|
284 |
|
285 // search the previous element |
|
286 struct cx_hash_map_element_s *prev = NULL; |
|
287 if (map->buckets[iter->slot] != elm) { |
|
288 prev = map->buckets[iter->slot]; |
|
289 while (prev->next != elm) { |
|
290 prev = prev->next; |
|
291 } |
|
292 } |
|
293 |
|
294 // destroy |
|
295 cx_invoke_destructor((struct cx_map_s *) map, elm->data); |
|
296 |
|
297 // unlink |
|
298 cx_hash_map_unlink(map, iter->slot, prev, elm); |
|
299 |
|
300 // advance |
|
301 elm = next; |
|
302 } else { |
|
303 // just advance |
|
304 elm = elm->next; |
|
305 iter->index++; |
|
306 } |
|
307 |
|
308 // search the next bucket, if required |
|
309 struct cx_hash_map_s const *map = iter->src_handle; |
|
310 while (elm == NULL && ++iter->slot < map->bucket_count) { |
|
311 elm = map->buckets[iter->slot]; |
|
312 } |
|
313 |
|
314 // fill the struct with the next element |
|
315 iter->elem_handle = elm; |
|
316 if (elm == NULL) { |
|
317 iter->kv_data.key = NULL; |
|
318 iter->kv_data.value = NULL; |
|
319 } else { |
|
320 iter->kv_data.key = &elm->key; |
|
321 if (map->base.store_pointer) { |
|
322 iter->kv_data.value = *(void **) elm->data; |
|
323 } else { |
|
324 iter->kv_data.value = elm->data; |
|
325 } |
|
326 } |
|
327 } |
|
328 |
|
329 static bool cx_hash_map_iter_flag_rm(void *it) { |
|
330 struct cx_iterator_base_s *iter = it; |
|
331 if (iter->mutating) { |
|
332 iter->remove = true; |
|
333 return true; |
|
334 } else { |
|
335 return false; |
|
336 } |
|
337 } |
|
338 |
|
339 static CxIterator cx_hash_map_iterator( |
|
340 CxMap const *map, |
|
341 enum cx_map_iterator_type type |
|
342 ) { |
|
343 CxIterator iter; |
|
344 |
|
345 iter.src_handle = map; |
|
346 iter.base.valid = cx_hash_map_iter_valid; |
|
347 iter.base.next = cx_hash_map_iter_next; |
|
348 |
|
349 switch (type) { |
|
350 case CX_MAP_ITERATOR_PAIRS: |
|
351 iter.base.current = cx_hash_map_iter_current_entry; |
|
352 break; |
|
353 case CX_MAP_ITERATOR_KEYS: |
|
354 iter.base.current = cx_hash_map_iter_current_key; |
|
355 break; |
|
356 case CX_MAP_ITERATOR_VALUES: |
|
357 iter.base.current = cx_hash_map_iter_current_value; |
|
358 break; |
|
359 default: |
|
360 assert(false); |
|
361 } |
|
362 |
|
363 iter.base.flag_removal = cx_hash_map_iter_flag_rm; |
|
364 iter.base.remove = false; |
|
365 iter.base.mutating = false; |
|
366 |
|
367 iter.slot = 0; |
|
368 iter.index = 0; |
|
369 |
|
370 if (map->size > 0) { |
|
371 struct cx_hash_map_s *hash_map = (struct cx_hash_map_s *) map; |
|
372 struct cx_hash_map_element_s *elm = hash_map->buckets[0]; |
|
373 while (elm == NULL) { |
|
374 elm = hash_map->buckets[++iter.slot]; |
|
375 } |
|
376 iter.elem_handle = elm; |
|
377 iter.kv_data.key = &elm->key; |
|
378 if (map->store_pointer) { |
|
379 iter.kv_data.value = *(void **) elm->data; |
|
380 } else { |
|
381 iter.kv_data.value = elm->data; |
|
382 } |
|
383 } else { |
|
384 iter.elem_handle = NULL; |
|
385 iter.kv_data.key = NULL; |
|
386 iter.kv_data.value = NULL; |
|
387 } |
|
388 |
|
389 return iter; |
|
390 } |
|
391 |
|
392 static cx_map_class cx_hash_map_class = { |
|
393 cx_hash_map_destructor, |
|
394 cx_hash_map_clear, |
|
395 cx_hash_map_put, |
|
396 cx_hash_map_get, |
|
397 cx_hash_map_remove, |
|
398 cx_hash_map_iterator, |
|
399 }; |
|
400 |
|
401 CxMap *cxHashMapCreate( |
|
402 CxAllocator const *allocator, |
|
403 size_t itemsize, |
|
404 size_t buckets |
|
405 ) { |
|
406 if (buckets == 0) { |
|
407 // implementation defined default |
|
408 buckets = 16; |
|
409 } |
|
410 |
|
411 struct cx_hash_map_s *map = cxCalloc(allocator, 1, |
|
412 sizeof(struct cx_hash_map_s)); |
|
413 if (map == NULL) return NULL; |
|
414 |
|
415 // initialize hash map members |
|
416 map->bucket_count = buckets; |
|
417 map->buckets = cxCalloc(allocator, buckets, |
|
418 sizeof(struct cx_hash_map_element_s *)); |
|
419 if (map->buckets == NULL) { |
|
420 cxFree(allocator, map); |
|
421 return NULL; |
|
422 } |
|
423 |
|
424 // initialize base members |
|
425 map->base.cl = &cx_hash_map_class; |
|
426 map->base.allocator = allocator; |
|
427 |
|
428 if (itemsize > 0) { |
|
429 map->base.store_pointer = false; |
|
430 map->base.item_size = itemsize; |
|
431 } else { |
|
432 map->base.store_pointer = true; |
|
433 map->base.item_size = sizeof(void *); |
|
434 } |
|
435 |
|
436 return (CxMap *) map; |
|
437 } |
|
438 |
|
439 int cxMapRehash(CxMap *map) { |
|
440 struct cx_hash_map_s *hash_map = (struct cx_hash_map_s *) map; |
|
441 if (map->size > ((hash_map->bucket_count * 3) >> 2)) { |
|
442 |
|
443 size_t new_bucket_count = (map->size * 5) >> 1; |
|
444 struct cx_hash_map_element_s **new_buckets = cxCalloc( |
|
445 map->allocator, |
|
446 new_bucket_count, sizeof(struct cx_hash_map_element_s *) |
|
447 ); |
|
448 |
|
449 if (new_buckets == NULL) { |
|
450 return 1; |
|
451 } |
|
452 |
|
453 // iterate through the elements and assign them to their new slots |
|
454 cx_for_n(slot, hash_map->bucket_count) { |
|
455 struct cx_hash_map_element_s *elm = hash_map->buckets[slot]; |
|
456 while (elm != NULL) { |
|
457 struct cx_hash_map_element_s *next = elm->next; |
|
458 size_t new_slot = elm->key.hash % new_bucket_count; |
|
459 |
|
460 // find position where to insert |
|
461 struct cx_hash_map_element_s *bucket_next = new_buckets[new_slot]; |
|
462 struct cx_hash_map_element_s *bucket_prev = NULL; |
|
463 while (bucket_next != NULL && |
|
464 bucket_next->key.hash < elm->key.hash) { |
|
465 bucket_prev = bucket_next; |
|
466 bucket_next = bucket_next->next; |
|
467 } |
|
468 |
|
469 // insert |
|
470 if (bucket_prev == NULL) { |
|
471 elm->next = new_buckets[new_slot]; |
|
472 new_buckets[new_slot] = elm; |
|
473 } else { |
|
474 bucket_prev->next = elm; |
|
475 elm->next = bucket_next; |
|
476 } |
|
477 |
|
478 // advance |
|
479 elm = next; |
|
480 } |
|
481 } |
|
482 |
|
483 // assign result to the map |
|
484 hash_map->bucket_count = new_bucket_count; |
|
485 cxFree(map->allocator, hash_map->buckets); |
|
486 hash_map->buckets = new_buckets; |
|
487 } |
|
488 return 0; |
|
489 } |