1 /* An expandable hash tables datatype.
2 Copyright (C) 1999, 2000, 2001, 2002 Free Software Foundation, Inc.
3 Contributed by Vladimir Makarov (vmakarov@cygnus.com).
5 This file is part of the libiberty library.
6 Libiberty is free software; you can redistribute it and/or
7 modify it under the terms of the GNU Library General Public
8 License as published by the Free Software Foundation; either
9 version 2 of the License, or (at your option) any later version.
11 Libiberty is distributed in the hope that it will be useful,
12 but WITHOUT ANY WARRANTY; without even the implied warranty of
13 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
14 Library General Public License for more details.
16 You should have received a copy of the GNU Library General Public
17 License along with libiberty; see the file COPYING.LIB. If
18 not, write to the Free Software Foundation, Inc., 59 Temple Place - Suite 330,
19 Boston, MA 02111-1307, USA. */
21 /* This package implements basic hash table functionality. It is possible
22 to search for an entry, create an entry and destroy an entry.
24 Elements in the table are generic pointers.
26 The size of the table is not fixed; if the occupancy of the table
27 grows too high the hash table will be expanded.
29 The abstract data implementation is based on generalized Algorithm D
30 from Knuth's book "The art of computer programming". Hash table is
31 expanded by creation of new hash table and transferring elements from
32 the old table to the new table. */
38 #include <sys/types.h>
50 #include "libiberty.h"
53 /* This macro defines reserved value for empty table entry. */
55 #define EMPTY_ENTRY ((PTR) 0)
57 /* This macro defines reserved value for table entry which contained
60 #define DELETED_ENTRY ((PTR) 1)
62 static unsigned long higher_prime_number PARAMS ((unsigned long));
63 static hashval_t hash_pointer PARAMS ((const void *));
64 static int eq_pointer PARAMS ((const void *, const void *));
65 static int htab_expand PARAMS ((htab_t));
66 static PTR *find_empty_slot_for_expand PARAMS ((htab_t, hashval_t));
68 /* At some point, we could make these be NULL, and modify the
69 hash-table routines to handle NULL specially; that would avoid
70 function-call overhead for the common case of hashing pointers. */
71 htab_hash htab_hash_pointer = hash_pointer;
72 htab_eq htab_eq_pointer = eq_pointer;
74 /* The following function returns a nearest prime number which is
75 greater than N, and near a power of two. */
78 higher_prime_number (n)
81 /* These are primes that are near, but slightly smaller than, a
83 static const unsigned long primes[] = {
95 (unsigned long) 16381,
96 (unsigned long) 32749,
97 (unsigned long) 65521,
98 (unsigned long) 131071,
99 (unsigned long) 262139,
100 (unsigned long) 524287,
101 (unsigned long) 1048573,
102 (unsigned long) 2097143,
103 (unsigned long) 4194301,
104 (unsigned long) 8388593,
105 (unsigned long) 16777213,
106 (unsigned long) 33554393,
107 (unsigned long) 67108859,
108 (unsigned long) 134217689,
109 (unsigned long) 268435399,
110 (unsigned long) 536870909,
111 (unsigned long) 1073741789,
112 (unsigned long) 2147483647,
114 ((unsigned long) 2147483647) + ((unsigned long) 2147483644),
117 const unsigned long *low = &primes[0];
118 const unsigned long *high = &primes[sizeof(primes) / sizeof(primes[0])];
122 const unsigned long *mid = low + (high - low) / 2;
129 /* If we've run out of primes, abort. */
132 fprintf (stderr, "Cannot find prime bigger than %lu\n", n);
139 /* Returns a hash code for P. */
145 return (hashval_t) ((long)p >> 3);
148 /* Returns non-zero if P1 and P2 are equal. */
158 /* This function creates table with length slightly longer than given
159 source length. Created hash table is initiated as empty (all the
160 hash table entries are EMPTY_ENTRY). The function returns the
161 created hash table, or NULL if memory allocation fails. */
164 htab_create_alloc (size, hash_f, eq_f, del_f, alloc_f, free_f)
174 size = higher_prime_number (size);
175 result = (htab_t) (*alloc_f) (1, sizeof (struct htab));
178 result->entries = (PTR *) (*alloc_f) (size, sizeof (PTR));
179 if (result->entries == NULL)
186 result->hash_f = hash_f;
188 result->del_f = del_f;
189 result->alloc_f = alloc_f;
190 result->free_f = free_f;
194 /* This function frees all memory allocated for given hash table.
195 Naturally the hash table must already exist. */
204 for (i = htab->size - 1; i >= 0; i--)
205 if (htab->entries[i] != EMPTY_ENTRY
206 && htab->entries[i] != DELETED_ENTRY)
207 (*htab->del_f) (htab->entries[i]);
209 if (htab->free_f != NULL)
211 (*htab->free_f) (htab->entries);
212 (*htab->free_f) (htab);
216 /* This function clears all entries in the given hash table. */
225 for (i = htab->size - 1; i >= 0; i--)
226 if (htab->entries[i] != EMPTY_ENTRY
227 && htab->entries[i] != DELETED_ENTRY)
228 (*htab->del_f) (htab->entries[i]);
230 memset (htab->entries, 0, htab->size * sizeof (PTR));
233 /* Similar to htab_find_slot, but without several unwanted side effects:
234 - Does not call htab->eq_f when it finds an existing entry.
235 - Does not change the count of elements/searches/collisions in the
237 This function also assumes there are no deleted entries in the table.
238 HASH is the hash value for the element to be inserted. */
241 find_empty_slot_for_expand (htab, hash)
245 size_t size = htab->size;
246 unsigned int index = hash % size;
247 PTR *slot = htab->entries + index;
250 if (*slot == EMPTY_ENTRY)
252 else if (*slot == DELETED_ENTRY)
255 hash2 = 1 + hash % (size - 2);
262 slot = htab->entries + index;
263 if (*slot == EMPTY_ENTRY)
265 else if (*slot == DELETED_ENTRY)
270 /* The following function changes size of memory allocated for the
271 entries and repeatedly inserts the table elements. The occupancy
272 of the table after the call will be about 50%. Naturally the hash
273 table must already exist. Remember also that the place of the
274 table entries is changed. If memory allocation failures are allowed,
275 this function will return zero, indicating that the table could not be
276 expanded. If all goes well, it will return a non-zero value. */
287 oentries = htab->entries;
288 olimit = oentries + htab->size;
290 htab->size = higher_prime_number (htab->size * 2);
292 nentries = (PTR *) (*htab->alloc_f) (htab->size, sizeof (PTR *));
293 if (nentries == NULL)
295 htab->entries = nentries;
297 htab->n_elements -= htab->n_deleted;
305 if (x != EMPTY_ENTRY && x != DELETED_ENTRY)
307 PTR *q = find_empty_slot_for_expand (htab, (*htab->hash_f) (x));
316 if (htab->free_f != NULL)
317 (*htab->free_f) (oentries);
321 /* This function searches for a hash table entry equal to the given
322 element. It cannot be used to insert or delete an element. */
325 htab_find_with_hash (htab, element, hash)
339 entry = htab->entries[index];
340 if (entry == EMPTY_ENTRY
341 || (entry != DELETED_ENTRY && (*htab->eq_f) (entry, element)))
344 hash2 = 1 + hash % (size - 2);
353 entry = htab->entries[index];
354 if (entry == EMPTY_ENTRY
355 || (entry != DELETED_ENTRY && (*htab->eq_f) (entry, element)))
360 /* Like htab_find_slot_with_hash, but compute the hash value from the
364 htab_find (htab, element)
368 return htab_find_with_hash (htab, element, (*htab->hash_f) (element));
371 /* This function searches for a hash table slot containing an entry
372 equal to the given element. To delete an entry, call this with
373 INSERT = 0, then call htab_clear_slot on the slot returned (possibly
374 after doing some checks). To insert an entry, call this with
375 INSERT = 1, then write the value you want into the returned slot.
376 When inserting an entry, NULL may be returned if memory allocation
380 htab_find_slot_with_hash (htab, element, hash, insert)
384 enum insert_option insert;
386 PTR *first_deleted_slot;
392 if (insert == INSERT && htab->size * 3 <= htab->n_elements * 4
393 && htab_expand (htab) == 0)
400 first_deleted_slot = NULL;
402 entry = htab->entries[index];
403 if (entry == EMPTY_ENTRY)
405 else if (entry == DELETED_ENTRY)
406 first_deleted_slot = &htab->entries[index];
407 else if ((*htab->eq_f) (entry, element))
408 return &htab->entries[index];
410 hash2 = 1 + hash % (size - 2);
418 entry = htab->entries[index];
419 if (entry == EMPTY_ENTRY)
421 else if (entry == DELETED_ENTRY)
423 if (!first_deleted_slot)
424 first_deleted_slot = &htab->entries[index];
426 else if ((*htab->eq_f) (entry, element))
427 return &htab->entries[index];
431 if (insert == NO_INSERT)
436 if (first_deleted_slot)
438 *first_deleted_slot = EMPTY_ENTRY;
439 return first_deleted_slot;
442 return &htab->entries[index];
445 /* Like htab_find_slot_with_hash, but compute the hash value from the
449 htab_find_slot (htab, element, insert)
452 enum insert_option insert;
454 return htab_find_slot_with_hash (htab, element, (*htab->hash_f) (element),
458 /* This function deletes an element with the given value from hash
459 table. If there is no matching element in the hash table, this
460 function does nothing. */
463 htab_remove_elt (htab, element)
469 slot = htab_find_slot (htab, element, NO_INSERT);
470 if (*slot == EMPTY_ENTRY)
474 (*htab->del_f) (*slot);
476 *slot = DELETED_ENTRY;
480 /* This function clears a specified slot in a hash table. It is
481 useful when you've already done the lookup and don't want to do it
485 htab_clear_slot (htab, slot)
489 if (slot < htab->entries || slot >= htab->entries + htab->size
490 || *slot == EMPTY_ENTRY || *slot == DELETED_ENTRY)
494 (*htab->del_f) (*slot);
496 *slot = DELETED_ENTRY;
500 /* This function scans over the entire hash table calling
501 CALLBACK for each live entry. If CALLBACK returns false,
502 the iteration stops. INFO is passed as CALLBACK's second
506 htab_traverse (htab, callback, info)
511 PTR *slot = htab->entries;
512 PTR *limit = slot + htab->size;
518 if (x != EMPTY_ENTRY && x != DELETED_ENTRY)
519 if (!(*callback) (slot, info))
522 while (++slot < limit);
525 /* Return the current size of given hash table. */
534 /* Return the current number of elements in given hash table. */
540 return htab->n_elements - htab->n_deleted;
543 /* Return the fraction of fixed collisions during all work with given
547 htab_collisions (htab)
550 if (htab->searches == 0)
553 return (double) htab->collisions / (double) htab->searches;
556 /* Hash P as a null-terminated string.
558 Copied from gcc/hashtable.c. Zack had the following to say with respect
559 to applicability, though note that unlike hashtable.c, this hash table
560 implementation re-hashes rather than chain buckets.
562 http://gcc.gnu.org/ml/gcc-patches/2001-08/msg01021.html
563 From: Zack Weinberg <zackw@panix.com>
564 Date: Fri, 17 Aug 2001 02:15:56 -0400
566 I got it by extracting all the identifiers from all the source code
567 I had lying around in mid-1999, and testing many recurrences of
568 the form "H_n = H_{n-1} * K + c_n * L + M" where K, L, M were either
569 prime numbers or the appropriate identity. This was the best one.
570 I don't remember exactly what constituted "best", except I was
571 looking at bucket-length distributions mostly.
573 So it should be very good at hashing identifiers, but might not be
574 as good at arbitrary strings.
576 I'll add that it thoroughly trounces the hash functions recommended
577 for this use at http://burtleburtle.net/bob/hash/index.html, both
578 on speed and bucket distribution. I haven't tried it against the
579 function they just started using for Perl's hashes. */
585 const unsigned char *str = (const unsigned char *) p;
589 while ((c = *str++) != 0)
590 r = r * 67 + c - 113;