xref: /netbsd-src/usr.bin/make/hash.c (revision 5e4c038a45edbc7d63b7c2daa76e29f88b64a4e3)
1 /*	$NetBSD: hash.c,v 1.9 2000/06/11 07:54:32 mycroft Exp $	*/
2 
3 /*
4  * Copyright (c) 1988, 1989, 1990 The Regents of the University of California.
5  * Copyright (c) 1988, 1989 by Adam de Boor
6  * Copyright (c) 1989 by Berkeley Softworks
7  * All rights reserved.
8  *
9  * This code is derived from software contributed to Berkeley by
10  * Adam de Boor.
11  *
12  * Redistribution and use in source and binary forms, with or without
13  * modification, are permitted provided that the following conditions
14  * are met:
15  * 1. Redistributions of source code must retain the above copyright
16  *    notice, this list of conditions and the following disclaimer.
17  * 2. Redistributions in binary form must reproduce the above copyright
18  *    notice, this list of conditions and the following disclaimer in the
19  *    documentation and/or other materials provided with the distribution.
20  * 3. All advertising materials mentioning features or use of this software
21  *    must display the following acknowledgement:
22  *	This product includes software developed by the University of
23  *	California, Berkeley and its contributors.
24  * 4. Neither the name of the University nor the names of its contributors
25  *    may be used to endorse or promote products derived from this software
26  *    without specific prior written permission.
27  *
28  * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
29  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
30  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
31  * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
32  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
33  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
34  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
35  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
36  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
37  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
38  * SUCH DAMAGE.
39  */
40 
41 #ifdef MAKE_BOOTSTRAP
42 static char rcsid[] = "$NetBSD: hash.c,v 1.9 2000/06/11 07:54:32 mycroft Exp $";
43 #else
44 #include <sys/cdefs.h>
45 #ifndef lint
46 #if 0
47 static char sccsid[] = "@(#)hash.c	8.1 (Berkeley) 6/6/93";
48 #else
49 __RCSID("$NetBSD: hash.c,v 1.9 2000/06/11 07:54:32 mycroft Exp $");
50 #endif
51 #endif /* not lint */
52 #endif
53 
54 /* hash.c --
55  *
56  * 	This module contains routines to manipulate a hash table.
57  * 	See hash.h for a definition of the structure of the hash
58  * 	table.  Hash tables grow automatically as the amount of
59  * 	information increases.
60  */
61 #include "sprite.h"
62 #include "make.h"
63 #include "hash.h"
64 
65 /*
66  * Forward references to local procedures that are used before they're
67  * defined:
68  */
69 
70 static void RebuildTable __P((Hash_Table *));
71 
72 /*
73  * The following defines the ratio of # entries to # buckets
74  * at which we rebuild the table to make it larger.
75  */
76 
77 #define rebuildLimit 3
78 
79 /*
80  *---------------------------------------------------------
81  *
82  * Hash_InitTable --
83  *
84  *	This routine just sets up the hash table.
85  *
86  * Results:
87  *	None.
88  *
89  * Side Effects:
90  *	Memory is allocated for the initial bucket area.
91  *
92  *---------------------------------------------------------
93  */
94 
95 void
96 Hash_InitTable(t, numBuckets)
97 	register Hash_Table *t;	/* Structure to use to hold table. */
98 	int numBuckets;		/* How many buckets to create for starters.
99 				 * This number is rounded up to a power of
100 				 * two.   If <= 0, a reasonable default is
101 				 * chosen. The table will grow in size later
102 				 * as needed. */
103 {
104 	register int i;
105 	register struct Hash_Entry **hp;
106 
107 	/*
108 	 * Round up the size to a power of two.
109 	 */
110 	if (numBuckets <= 0)
111 		i = 16;
112 	else {
113 		for (i = 2; i < numBuckets; i <<= 1)
114 			 continue;
115 	}
116 	t->numEntries = 0;
117 	t->size = i;
118 	t->mask = i - 1;
119 	t->bucketPtr = hp = (struct Hash_Entry **)emalloc(sizeof(*hp) * i);
120 	while (--i >= 0)
121 		*hp++ = NULL;
122 }
123 
124 /*
125  *---------------------------------------------------------
126  *
127  * Hash_DeleteTable --
128  *
129  *	This routine removes everything from a hash table
130  *	and frees up the memory space it occupied (except for
131  *	the space in the Hash_Table structure).
132  *
133  * Results:
134  *	None.
135  *
136  * Side Effects:
137  *	Lots of memory is freed up.
138  *
139  *---------------------------------------------------------
140  */
141 
142 void
143 Hash_DeleteTable(t)
144 	Hash_Table *t;
145 {
146 	register struct Hash_Entry **hp, *h, *nexth = NULL;
147 	register int i;
148 
149 	for (hp = t->bucketPtr, i = t->size; --i >= 0;) {
150 		for (h = *hp++; h != NULL; h = nexth) {
151 			nexth = h->next;
152 			free((char *)h);
153 		}
154 	}
155 	free((char *)t->bucketPtr);
156 
157 	/*
158 	 * Set up the hash table to cause memory faults on any future access
159 	 * attempts until re-initialization.
160 	 */
161 	t->bucketPtr = NULL;
162 }
163 
164 /*
165  *---------------------------------------------------------
166  *
167  * Hash_FindEntry --
168  *
169  * 	Searches a hash table for an entry corresponding to key.
170  *
171  * Results:
172  *	The return value is a pointer to the entry for key,
173  *	if key was present in the table.  If key was not
174  *	present, NULL is returned.
175  *
176  * Side Effects:
177  *	None.
178  *
179  *---------------------------------------------------------
180  */
181 
182 Hash_Entry *
183 Hash_FindEntry(t, key)
184 	Hash_Table *t;		/* Hash table to search. */
185 	char *key;		/* A hash key. */
186 {
187 	register Hash_Entry *e;
188 	register unsigned h;
189 	register char *p;
190 
191 	for (h = 0, p = key; *p;)
192 		h = (h << 5) - h + *p++;
193 	p = key;
194 	for (e = t->bucketPtr[h & t->mask]; e != NULL; e = e->next)
195 		if (e->namehash == h && strcmp(e->name, p) == 0)
196 			return (e);
197 	return (NULL);
198 }
199 
200 /*
201  *---------------------------------------------------------
202  *
203  * Hash_CreateEntry --
204  *
205  *	Searches a hash table for an entry corresponding to
206  *	key.  If no entry is found, then one is created.
207  *
208  * Results:
209  *	The return value is a pointer to the entry.  If *newPtr
210  *	isn't NULL, then *newPtr is filled in with TRUE if a
211  *	new entry was created, and FALSE if an entry already existed
212  *	with the given key.
213  *
214  * Side Effects:
215  *	Memory may be allocated, and the hash buckets may be modified.
216  *---------------------------------------------------------
217  */
218 
219 Hash_Entry *
220 Hash_CreateEntry(t, key, newPtr)
221 	register Hash_Table *t;	/* Hash table to search. */
222 	char *key;		/* A hash key. */
223 	Boolean *newPtr;	/* Filled in with TRUE if new entry created,
224 				 * FALSE otherwise. */
225 {
226 	register Hash_Entry *e;
227 	register unsigned h;
228 	register char *p;
229 	int keylen;
230 	struct Hash_Entry **hp;
231 
232 	/*
233 	 * Hash the key.  As a side effect, save the length (strlen) of the
234 	 * key in case we need to create the entry.
235 	 */
236 	for (h = 0, p = key; *p;)
237 		h = (h << 5) - h + *p++;
238 	keylen = p - key;
239 	p = key;
240 	for (e = t->bucketPtr[h & t->mask]; e != NULL; e = e->next) {
241 		if (e->namehash == h && strcmp(e->name, p) == 0) {
242 			if (newPtr != NULL)
243 				*newPtr = FALSE;
244 			return (e);
245 		}
246 	}
247 
248 	/*
249 	 * The desired entry isn't there.  Before allocating a new entry,
250 	 * expand the table if necessary (and this changes the resulting
251 	 * bucket chain).
252 	 */
253 	if (t->numEntries >= rebuildLimit * t->size)
254 		RebuildTable(t);
255 	e = (Hash_Entry *) emalloc(sizeof(*e) + keylen);
256 	hp = &t->bucketPtr[h & t->mask];
257 	e->next = *hp;
258 	*hp = e;
259 	e->clientData = NULL;
260 	e->namehash = h;
261 	(void) strcpy(e->name, p);
262 	t->numEntries++;
263 
264 	if (newPtr != NULL)
265 		*newPtr = TRUE;
266 	return (e);
267 }
268 
269 /*
270  *---------------------------------------------------------
271  *
272  * Hash_DeleteEntry --
273  *
274  * 	Delete the given hash table entry and free memory associated with
275  *	it.
276  *
277  * Results:
278  *	None.
279  *
280  * Side Effects:
281  *	Hash chain that entry lives in is modified and memory is freed.
282  *
283  *---------------------------------------------------------
284  */
285 
286 void
287 Hash_DeleteEntry(t, e)
288 	Hash_Table *t;
289 	Hash_Entry *e;
290 {
291 	register Hash_Entry **hp, *p;
292 
293 	if (e == NULL)
294 		return;
295 	for (hp = &t->bucketPtr[e->namehash & t->mask];
296 	     (p = *hp) != NULL; hp = &p->next) {
297 		if (p == e) {
298 			*hp = p->next;
299 			free((char *)p);
300 			t->numEntries--;
301 			return;
302 		}
303 	}
304 	(void) write(2, "bad call to Hash_DeleteEntry\n", 29);
305 	abort();
306 }
307 
308 /*
309  *---------------------------------------------------------
310  *
311  * Hash_EnumFirst --
312  *	This procedure sets things up for a complete search
313  *	of all entries recorded in the hash table.
314  *
315  * Results:
316  *	The return value is the address of the first entry in
317  *	the hash table, or NULL if the table is empty.
318  *
319  * Side Effects:
320  *	The information in searchPtr is initialized so that successive
321  *	calls to Hash_Next will return successive HashEntry's
322  *	from the table.
323  *
324  *---------------------------------------------------------
325  */
326 
327 Hash_Entry *
328 Hash_EnumFirst(t, searchPtr)
329 	Hash_Table *t;			/* Table to be searched. */
330 	register Hash_Search *searchPtr;/* Area in which to keep state
331 					 * about search.*/
332 {
333 	searchPtr->tablePtr = t;
334 	searchPtr->nextIndex = 0;
335 	searchPtr->hashEntryPtr = NULL;
336 	return Hash_EnumNext(searchPtr);
337 }
338 
339 /*
340  *---------------------------------------------------------
341  *
342  * Hash_EnumNext --
343  *    This procedure returns successive entries in the hash table.
344  *
345  * Results:
346  *    The return value is a pointer to the next HashEntry
347  *    in the table, or NULL when the end of the table is
348  *    reached.
349  *
350  * Side Effects:
351  *    The information in searchPtr is modified to advance to the
352  *    next entry.
353  *
354  *---------------------------------------------------------
355  */
356 
357 Hash_Entry *
358 Hash_EnumNext(searchPtr)
359 	register Hash_Search *searchPtr; /* Area used to keep state about
360 					    search. */
361 {
362 	register Hash_Entry *e;
363 	Hash_Table *t = searchPtr->tablePtr;
364 
365 	/*
366 	 * The hashEntryPtr field points to the most recently returned
367 	 * entry, or is nil if we are starting up.  If not nil, we have
368 	 * to start at the next one in the chain.
369 	 */
370 	e = searchPtr->hashEntryPtr;
371 	if (e != NULL)
372 		e = e->next;
373 	/*
374 	 * If the chain ran out, or if we are starting up, we need to
375 	 * find the next nonempty chain.
376 	 */
377 	while (e == NULL) {
378 		if (searchPtr->nextIndex >= t->size)
379 			return (NULL);
380 		e = t->bucketPtr[searchPtr->nextIndex++];
381 	}
382 	searchPtr->hashEntryPtr = e;
383 	return (e);
384 }
385 
386 /*
387  *---------------------------------------------------------
388  *
389  * RebuildTable --
390  *	This local routine makes a new hash table that
391  *	is larger than the old one.
392  *
393  * Results:
394  * 	None.
395  *
396  * Side Effects:
397  *	The entire hash table is moved, so any bucket numbers
398  *	from the old table are invalid.
399  *
400  *---------------------------------------------------------
401  */
402 
403 static void
404 RebuildTable(t)
405 	register Hash_Table *t;
406 {
407 	register Hash_Entry *e, *next = NULL, **hp, **xp;
408 	register int i, mask;
409         register Hash_Entry **oldhp;
410 	int oldsize;
411 
412 	oldhp = t->bucketPtr;
413 	oldsize = i = t->size;
414 	i <<= 1;
415 	t->size = i;
416 	t->mask = mask = i - 1;
417 	t->bucketPtr = hp = (struct Hash_Entry **) emalloc(sizeof(*hp) * i);
418 	while (--i >= 0)
419 		*hp++ = NULL;
420 	for (hp = oldhp, i = oldsize; --i >= 0;) {
421 		for (e = *hp++; e != NULL; e = next) {
422 			next = e->next;
423 			xp = &t->bucketPtr[e->namehash & mask];
424 			e->next = *xp;
425 			*xp = e;
426 		}
427 	}
428 	free((char *)oldhp);
429 }
430