|
428
|
1 /* Hash tables.
|
|
|
2 Copyright (C) 1992, 1993, 1994 Free Software Foundation, Inc.
|
|
1292
|
3 Copyright (C) 2003 Ben Wing.
|
|
428
|
4
|
|
|
5 This file is part of XEmacs.
|
|
|
6
|
|
|
7 XEmacs is free software; you can redistribute it and/or modify it
|
|
|
8 under the terms of the GNU General Public License as published by the
|
|
|
9 Free Software Foundation; either version 2, or (at your option) any
|
|
|
10 later version.
|
|
|
11
|
|
|
12 XEmacs is distributed in the hope that it will be useful, but WITHOUT
|
|
|
13 ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
|
|
|
14 FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License
|
|
|
15 for more details.
|
|
|
16
|
|
|
17 You should have received a copy of the GNU General Public License
|
|
|
18 along with XEmacs; see the file COPYING. If not, write to
|
|
|
19 the Free Software Foundation, Inc., 59 Temple Place - Suite 330,
|
|
|
20 Boston, MA 02111-1307, USA. */
|
|
|
21
|
|
|
22 /* Synched up with: Not in FSF. */
|
|
|
23
|
|
1292
|
24 /* Author: Lost in the mists of history. At least back to Lucid 19.3,
|
|
|
25 circa Sep 1992. */
|
|
|
26
|
|
428
|
27 #include <config.h>
|
|
|
28 #include "lisp.h"
|
|
|
29 #include "hash.h"
|
|
|
30
|
|
1204
|
31 #define NULL_ENTRY ((void *) 0xdeadbeef) /* -559038737 base 10 */
|
|
428
|
32
|
|
|
33 #define COMFORTABLE_SIZE(size) (21 * (size) / 16)
|
|
|
34
|
|
|
35 #define KEYS_DIFFER_P(old, new, testfun) \
|
|
|
36 (((old) != (new)) && (!(testfun) || !(testfun) ((old),(new))))
|
|
|
37
|
|
665
|
38 static void rehash (hentry *harray, struct hash_table *ht, Elemcount size);
|
|
428
|
39
|
|
665
|
40 Hashcode
|
|
|
41 memory_hash (const void *xv, Bytecount size)
|
|
428
|
42 {
|
|
665
|
43 Hashcode h = 0;
|
|
442
|
44 unsigned const char *x = (unsigned const char *) xv;
|
|
428
|
45
|
|
|
46 if (!x) return 0;
|
|
|
47
|
|
|
48 while (size--)
|
|
|
49 {
|
|
665
|
50 Hashcode g;
|
|
428
|
51 h = (h << 4) + *x++;
|
|
|
52 if ((g = h & 0xf0000000) != 0)
|
|
|
53 h = (h ^ (g >> 24)) ^ g;
|
|
|
54 }
|
|
|
55
|
|
|
56 return h;
|
|
|
57 }
|
|
|
58
|
|
665
|
59 Hashcode
|
|
442
|
60 string_hash (const char *xv)
|
|
|
61 {
|
|
665
|
62 Hashcode h = 0;
|
|
442
|
63 unsigned const char *x = (unsigned const char *) xv;
|
|
|
64
|
|
|
65 if (!x) return 0;
|
|
|
66
|
|
|
67 while (*x)
|
|
|
68 {
|
|
665
|
69 Hashcode g;
|
|
442
|
70 h = (h << 4) + *x++;
|
|
|
71 if ((g = h & 0xf0000000) != 0)
|
|
|
72 h = (h ^ (g >> 24)) ^ g;
|
|
|
73 }
|
|
|
74
|
|
|
75 return h;
|
|
|
76 }
|
|
|
77
|
|
428
|
78 /* Return a suitable size for a hash table, with at least SIZE slots. */
|
|
665
|
79 static Elemcount
|
|
|
80 hash_table_size (Elemcount requested_size)
|
|
428
|
81 {
|
|
|
82 /* Return some prime near, but greater than or equal to, SIZE.
|
|
|
83 Decades from the time of writing, someone will have a system large
|
|
|
84 enough that the list below will be too short... */
|
|
665
|
85 static const Elemcount primes [] =
|
|
428
|
86 {
|
|
|
87 19, 29, 41, 59, 79, 107, 149, 197, 263, 347, 457, 599, 787, 1031,
|
|
|
88 1361, 1777, 2333, 3037, 3967, 5167, 6719, 8737, 11369, 14783,
|
|
|
89 19219, 24989, 32491, 42257, 54941, 71429, 92861, 120721, 156941,
|
|
|
90 204047, 265271, 344857, 448321, 582821, 757693, 985003, 1280519,
|
|
|
91 1664681, 2164111, 2813353, 3657361, 4754591, 6180989, 8035301,
|
|
|
92 10445899, 13579681, 17653589, 22949669, 29834603, 38784989,
|
|
|
93 50420551, 65546729, 85210757, 110774011, 144006217, 187208107,
|
|
|
94 243370577, 316381771, 411296309, 534685237, 695090819, 903618083,
|
|
647
|
95 1174703521, 1527114613, 1985248999 /* , 2580823717UL, 3355070839UL */
|
|
428
|
96 };
|
|
|
97 /* We've heard of binary search. */
|
|
|
98 int low, high;
|
|
|
99 for (low = 0, high = countof (primes) - 1; high - low > 1;)
|
|
|
100 {
|
|
|
101 /* Loop Invariant: size < primes [high] */
|
|
|
102 int mid = (low + high) / 2;
|
|
|
103 if (primes [mid] < requested_size)
|
|
|
104 low = mid;
|
|
|
105 else
|
|
|
106 high = mid;
|
|
|
107 }
|
|
|
108 return primes [high];
|
|
|
109 }
|
|
|
110
|
|
442
|
111 const void *
|
|
|
112 gethash (const void *key, struct hash_table *hash_table, const void **ret_value)
|
|
428
|
113 {
|
|
|
114 if (!key)
|
|
|
115 {
|
|
|
116 *ret_value = hash_table->zero_entry;
|
|
|
117 return (void *) hash_table->zero_set;
|
|
|
118 }
|
|
|
119 else
|
|
|
120 {
|
|
|
121 hentry *harray = hash_table->harray;
|
|
|
122 hash_table_test_function test_function = hash_table->test_function;
|
|
665
|
123 Elemcount size = hash_table->size;
|
|
|
124 Hashcode hcode_initial =
|
|
428
|
125 hash_table->hash_function ?
|
|
|
126 hash_table->hash_function (key) :
|
|
665
|
127 (Hashcode) key;
|
|
|
128 Elemcount hcode = (Elemcount) (hcode_initial % size);
|
|
428
|
129 hentry *e = &harray [hcode];
|
|
442
|
130 const void *e_key = e->key;
|
|
428
|
131
|
|
|
132 if (e_key ?
|
|
|
133 KEYS_DIFFER_P (e_key, key, test_function) :
|
|
|
134 e->contents == NULL_ENTRY)
|
|
|
135 {
|
|
665
|
136 Elemcount h2 = size - 2;
|
|
|
137 Elemcount incr = (Elemcount) (1 + (hcode_initial % h2));
|
|
428
|
138 do
|
|
|
139 {
|
|
|
140 hcode += incr; if (hcode >= size) hcode -= size;
|
|
|
141 e = &harray [hcode];
|
|
|
142 e_key = e->key;
|
|
|
143 }
|
|
|
144 while (e_key ?
|
|
|
145 KEYS_DIFFER_P (e_key, key, test_function) :
|
|
|
146 e->contents == NULL_ENTRY);
|
|
|
147 }
|
|
|
148
|
|
|
149 *ret_value = e->contents;
|
|
|
150 return e->key;
|
|
|
151 }
|
|
|
152 }
|
|
|
153
|
|
|
154 void
|
|
|
155 clrhash (struct hash_table *hash_table)
|
|
|
156 {
|
|
|
157 memset (hash_table->harray, 0, sizeof (hentry) * hash_table->size);
|
|
|
158 hash_table->zero_entry = 0;
|
|
|
159 hash_table->zero_set = 0;
|
|
|
160 hash_table->fullness = 0;
|
|
|
161 }
|
|
|
162
|
|
|
163 void
|
|
|
164 free_hash_table (struct hash_table *hash_table)
|
|
|
165 {
|
|
1726
|
166 xfree (hash_table->harray, hentry *);
|
|
|
167 xfree (hash_table, struct hash_table *);
|
|
428
|
168 }
|
|
|
169
|
|
|
170 struct hash_table*
|
|
665
|
171 make_hash_table (Elemcount size)
|
|
428
|
172 {
|
|
|
173 struct hash_table *hash_table = xnew_and_zero (struct hash_table);
|
|
|
174 hash_table->size = hash_table_size (COMFORTABLE_SIZE (size));
|
|
|
175 hash_table->harray = xnew_array (hentry, hash_table->size);
|
|
|
176 clrhash (hash_table);
|
|
|
177 return hash_table;
|
|
|
178 }
|
|
|
179
|
|
|
180 struct hash_table *
|
|
665
|
181 make_general_hash_table (Elemcount size,
|
|
428
|
182 hash_table_hash_function hash_function,
|
|
|
183 hash_table_test_function test_function)
|
|
|
184 {
|
|
|
185 struct hash_table* hash_table = make_hash_table (size);
|
|
|
186 hash_table->hash_function = hash_function;
|
|
|
187 hash_table->test_function = test_function;
|
|
|
188 return hash_table;
|
|
|
189 }
|
|
|
190
|
|
|
191 static void
|
|
665
|
192 grow_hash_table (struct hash_table *hash_table, Elemcount new_size)
|
|
428
|
193 {
|
|
665
|
194 Elemcount old_size = hash_table->size;
|
|
428
|
195 hentry *old_harray = hash_table->harray;
|
|
|
196
|
|
|
197 hash_table->size = hash_table_size (new_size);
|
|
|
198 hash_table->harray = xnew_array (hentry, hash_table->size);
|
|
|
199
|
|
|
200 /* do the rehash on the "grown" table */
|
|
|
201 {
|
|
|
202 long old_zero_set = hash_table->zero_set;
|
|
|
203 void *old_zero_entry = hash_table->zero_entry;
|
|
|
204 clrhash (hash_table);
|
|
|
205 hash_table->zero_set = old_zero_set;
|
|
|
206 hash_table->zero_entry = old_zero_entry;
|
|
|
207 rehash (old_harray, hash_table, old_size);
|
|
|
208 }
|
|
|
209
|
|
1726
|
210 xfree (old_harray, hentry *);
|
|
428
|
211 }
|
|
|
212
|
|
|
213 void
|
|
1292
|
214 pregrow_hash_table_if_necessary (struct hash_table *hash_table,
|
|
|
215 Elemcount breathing_room)
|
|
|
216 {
|
|
|
217 Elemcount comfortable_size = COMFORTABLE_SIZE (hash_table->fullness);
|
|
|
218 if (hash_table->size < comfortable_size - breathing_room)
|
|
|
219 grow_hash_table (hash_table, comfortable_size + 1);
|
|
|
220 }
|
|
|
221
|
|
|
222 void
|
|
442
|
223 puthash (const void *key, void *contents, struct hash_table *hash_table)
|
|
428
|
224 {
|
|
|
225 if (!key)
|
|
|
226 {
|
|
|
227 hash_table->zero_entry = contents;
|
|
|
228 hash_table->zero_set = 1;
|
|
|
229 }
|
|
|
230 else
|
|
|
231 {
|
|
|
232 hash_table_test_function test_function = hash_table->test_function;
|
|
665
|
233 Elemcount size = hash_table->size;
|
|
428
|
234 hentry *harray = hash_table->harray;
|
|
665
|
235 Hashcode hcode_initial =
|
|
428
|
236 hash_table->hash_function ?
|
|
|
237 hash_table->hash_function (key) :
|
|
665
|
238 (Hashcode) key;
|
|
|
239 Elemcount hcode = (Elemcount) (hcode_initial % size);
|
|
|
240 Elemcount h2 = size - 2;
|
|
|
241 Elemcount incr = (Elemcount) (1 + (hcode_initial % h2));
|
|
442
|
242 const void *e_key = harray [hcode].key;
|
|
|
243 const void *oldcontents;
|
|
428
|
244
|
|
|
245 if (e_key && KEYS_DIFFER_P (e_key, key, test_function))
|
|
|
246 {
|
|
|
247 do
|
|
|
248 {
|
|
|
249 hcode += incr; if (hcode >= size) hcode -= size;
|
|
|
250 e_key = harray [hcode].key;
|
|
|
251 }
|
|
|
252 while (e_key && KEYS_DIFFER_P (e_key, key, test_function));
|
|
|
253 }
|
|
|
254 oldcontents = harray [hcode].contents;
|
|
|
255 harray [hcode].key = key;
|
|
|
256 harray [hcode].contents = contents;
|
|
|
257 /* If the entry that we used was a deleted entry,
|
|
|
258 check for a non deleted entry of the same key,
|
|
|
259 then delete it. */
|
|
|
260 if (!e_key && oldcontents == NULL_ENTRY)
|
|
|
261 {
|
|
|
262 hentry *e;
|
|
|
263
|
|
|
264 do
|
|
|
265 {
|
|
|
266 hcode += incr; if (hcode >= size) hcode -= size;
|
|
|
267 e = &harray [hcode];
|
|
|
268 e_key = e->key;
|
|
|
269 }
|
|
|
270 while (e_key ?
|
|
|
271 KEYS_DIFFER_P (e_key, key, test_function):
|
|
|
272 e->contents == NULL_ENTRY);
|
|
|
273
|
|
|
274 if (e_key)
|
|
|
275 {
|
|
|
276 e->key = 0;
|
|
|
277 e->contents = NULL_ENTRY;
|
|
|
278 }
|
|
|
279 }
|
|
|
280
|
|
|
281 /* only increment the fullness when we used up a new hentry */
|
|
|
282 if (!e_key || KEYS_DIFFER_P (e_key, key, test_function))
|
|
|
283 {
|
|
665
|
284 Elemcount comfortable_size = COMFORTABLE_SIZE (++(hash_table->fullness));
|
|
428
|
285 if (hash_table->size < comfortable_size)
|
|
|
286 grow_hash_table (hash_table, comfortable_size + 1);
|
|
|
287 }
|
|
|
288 }
|
|
|
289 }
|
|
|
290
|
|
|
291 static void
|
|
665
|
292 rehash (hentry *harray, struct hash_table *hash_table, Elemcount size)
|
|
428
|
293 {
|
|
|
294 hentry *limit = harray + size;
|
|
|
295 hentry *e;
|
|
|
296 for (e = harray; e < limit; e++)
|
|
|
297 {
|
|
|
298 if (e->key)
|
|
|
299 puthash (e->key, e->contents, hash_table);
|
|
|
300 }
|
|
|
301 }
|
|
|
302
|
|
|
303 void
|
|
442
|
304 remhash (const void *key, struct hash_table *hash_table)
|
|
428
|
305 {
|
|
|
306 if (!key)
|
|
|
307 {
|
|
|
308 hash_table->zero_entry = 0;
|
|
|
309 hash_table->zero_set = 0;
|
|
|
310 }
|
|
|
311 else
|
|
|
312 {
|
|
|
313 hentry *harray = hash_table->harray;
|
|
|
314 hash_table_test_function test_function = hash_table->test_function;
|
|
665
|
315 Elemcount size = hash_table->size;
|
|
|
316 Hashcode hcode_initial =
|
|
428
|
317 (hash_table->hash_function) ?
|
|
|
318 (hash_table->hash_function (key)) :
|
|
665
|
319 ((Hashcode) key);
|
|
|
320 Elemcount hcode = (Elemcount) (hcode_initial % size);
|
|
428
|
321 hentry *e = &harray [hcode];
|
|
442
|
322 const void *e_key = e->key;
|
|
428
|
323
|
|
|
324 if (e_key ?
|
|
|
325 KEYS_DIFFER_P (e_key, key, test_function) :
|
|
|
326 e->contents == NULL_ENTRY)
|
|
|
327 {
|
|
665
|
328 Elemcount h2 = size - 2;
|
|
|
329 Elemcount incr = (Elemcount) (1 + (hcode_initial % h2));
|
|
428
|
330 do
|
|
|
331 {
|
|
|
332 hcode += incr; if (hcode >= size) hcode -= size;
|
|
|
333 e = &harray [hcode];
|
|
|
334 e_key = e->key;
|
|
|
335 }
|
|
|
336 while (e_key?
|
|
|
337 KEYS_DIFFER_P (e_key, key, test_function):
|
|
|
338 e->contents == NULL_ENTRY);
|
|
|
339 }
|
|
|
340 if (e_key)
|
|
|
341 {
|
|
|
342 e->key = 0;
|
|
|
343 e->contents = NULL_ENTRY;
|
|
|
344 /* Note: you can't do fullness-- here, it breaks the world. */
|
|
|
345 }
|
|
|
346 }
|
|
|
347 }
|
|
|
348
|
|
|
349 void
|
|
|
350 maphash (maphash_function mf, struct hash_table *hash_table, void *arg)
|
|
|
351 {
|
|
|
352 hentry *e;
|
|
|
353 hentry *limit;
|
|
|
354
|
|
|
355 if (hash_table->zero_set)
|
|
|
356 {
|
|
|
357 if (mf (0, hash_table->zero_entry, arg))
|
|
|
358 return;
|
|
|
359 }
|
|
|
360
|
|
|
361 for (e = hash_table->harray, limit = e + hash_table->size; e < limit; e++)
|
|
|
362 {
|
|
|
363 if (e->key && mf (e->key, e->contents, arg))
|
|
|
364 return;
|
|
|
365 }
|
|
|
366 }
|
|
|
367
|
|
|
368 void
|
|
|
369 map_remhash (remhash_predicate predicate, struct hash_table *hash_table, void *arg)
|
|
|
370 {
|
|
|
371 hentry *e;
|
|
|
372 hentry *limit;
|
|
|
373
|
|
|
374 if (hash_table->zero_set && predicate (0, hash_table->zero_entry, arg))
|
|
|
375 {
|
|
|
376 hash_table->zero_set = 0;
|
|
|
377 hash_table->zero_entry = 0;
|
|
|
378 }
|
|
|
379
|
|
|
380 for (e = hash_table->harray, limit = e + hash_table->size; e < limit; e++)
|
|
|
381 if (predicate (e->key, e->contents, arg))
|
|
|
382 {
|
|
|
383 e->key = 0;
|
|
|
384 e->contents = NULL_ENTRY;
|
|
|
385 }
|
|
|
386 }
|
