]>
git.proxmox.com Git - libgit2.git/blob - src/khash.h
40e2d1848b77cfc10a10f84a66551e1d7e4e3d5d
3 Copyright (c) 2008, 2009, 2011 by Attractive Chaos <attractor@live.co.uk>
5 Permission is hereby granted, free of charge, to any person obtaining
6 a copy of this software and associated documentation files (the
7 "Software"), to deal in the Software without restriction, including
8 without limitation the rights to use, copy, modify, merge, publish,
9 distribute, sublicense, and/or sell copies of the Software, and to
10 permit persons to whom the Software is furnished to do so, subject to
11 the following conditions:
13 The above copyright notice and this permission notice shall be
14 included in all copies or substantial portions of the Software.
16 THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
17 EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
18 MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
19 NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
20 BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
21 ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
22 CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
30 KHASH_MAP_INIT_INT(32, char)
34 khash_t(32) *h = kh_init(32);
35 k = kh_put(32, h, 5, &ret);
37 k = kh_get(32, h, 10);
38 is_missing = (k == kh_end(h));
41 for (k = kh_begin(h); k != kh_end(h); ++k)
42 if (kh_exist(h, k)) kh_value(h, k) = 1;
51 * Use quadratic probing. When the capacity is power of 2, stepping function
52 i*(i+1)/2 guarantees to traverse each bucket. It is better than double
53 hashing on cache performance and is more robust than linear probing.
55 In theory, double hashing should be more robust than quadratic probing.
56 However, my implementation is probably not for large hash tables, because
57 the second hash function is closely tied to the first hash function,
58 which reduce the effectiveness of double hashing.
60 Reference: http://research.cs.vt.edu/AVresearch/hashing/quadratic.php
64 * Minor code clean up; no actual effect.
68 * The capacity is a power of 2. This seems to dramatically improve the
69 speed for simple keys. Thank Zilong Tan for the suggestion. Reference:
71 - http://code.google.com/p/ulib/
72 - http://nothings.org/computer/judy/
74 * Allow to optionally use linear probing which usually has better
75 performance for random input. Double hashing is still the default as it
76 is more robust to certain non-random input.
78 * Added Wang's integer hash function (not used by default). This hash
79 function is more robust to certain non-random input.
83 * Allow to declare global functions.
91 * Corrected the example
96 * Improved speed a little in kh_put()
101 * Fixed a compiling error
105 * Changed to token concatenation which increases flexibility.
109 * Fixed a bug in kh_get(), which has not been tested previously.
123 Generic hash table library.
126 #define AC_VERSION_KHASH_H "0.2.8"
132 /* compiler specific configuration */
134 #if UINT_MAX == 0xffffffffu
135 typedef unsigned int khint32_t
;
136 #elif ULONG_MAX == 0xffffffffu
137 typedef unsigned long khint32_t
;
140 #if ULONG_MAX == ULLONG_MAX
141 typedef unsigned long khint64_t
;
143 typedef unsigned long long khint64_t
;
148 #define kh_inline __inline
149 #elif defined(__GNUC__)
150 #define kh_inline __inline__
154 #endif /* kh_inline */
156 typedef khint32_t khint_t
;
157 typedef khint_t khiter_t
;
159 #define __ac_isempty(flag, i) ((flag[i>>4]>>((i&0xfU)<<1))&2)
160 #define __ac_isdel(flag, i) ((flag[i>>4]>>((i&0xfU)<<1))&1)
161 #define __ac_iseither(flag, i) ((flag[i>>4]>>((i&0xfU)<<1))&3)
162 #define __ac_set_isdel_false(flag, i) (flag[i>>4]&=~(1ul<<((i&0xfU)<<1)))
163 #define __ac_set_isempty_false(flag, i) (flag[i>>4]&=~(2ul<<((i&0xfU)<<1)))
164 #define __ac_set_isboth_false(flag, i) (flag[i>>4]&=~(3ul<<((i&0xfU)<<1)))
165 #define __ac_set_isdel_true(flag, i) (flag[i>>4]|=1ul<<((i&0xfU)<<1))
167 #define __ac_fsize(m) ((m) < 16? 1 : (m)>>4)
170 #define kroundup32(x) (--(x), (x)|=(x)>>1, (x)|=(x)>>2, (x)|=(x)>>4, (x)|=(x)>>8, (x)|=(x)>>16, ++(x))
174 #define kcalloc(N,Z) calloc(N,Z)
177 #define kmalloc(Z) malloc(Z)
180 #define krealloc(P,Z) realloc(P,Z)
182 #ifndef kreallocarray
183 #define kreallocarray(P,N,Z) ((SIZE_MAX - N < Z) ? NULL : krealloc(P, (N*Z)))
186 #define kfree(P) free(P)
189 static const double __ac_HASH_UPPER
= 0.77;
191 #define __KHASH_TYPE(name, khkey_t, khval_t) \
192 typedef struct kh_##name##_s { \
193 khint_t n_buckets, size, n_occupied, upper_bound; \
199 #define __KHASH_PROTOTYPES(name, khkey_t, khval_t) \
200 extern kh_##name##_t *kh_init_##name(void); \
201 extern void kh_destroy_##name(kh_##name##_t *h); \
202 extern void kh_clear_##name(kh_##name##_t *h); \
203 extern khint_t kh_get_##name(const kh_##name##_t *h, khkey_t key); \
204 extern int kh_resize_##name(kh_##name##_t *h, khint_t new_n_buckets); \
205 extern khint_t kh_put_##name(kh_##name##_t *h, khkey_t key, int *ret); \
206 extern void kh_del_##name(kh_##name##_t *h, khint_t x);
208 #define __KHASH_IMPL(name, SCOPE, khkey_t, khval_t, kh_is_map, __hash_func, __hash_equal) \
209 SCOPE kh_##name##_t *kh_init_##name(void) { \
210 return (kh_##name##_t*)kcalloc(1, sizeof(kh_##name##_t)); \
212 SCOPE void kh_destroy_##name(kh_##name##_t *h) \
215 kfree((void *)h->keys); kfree(h->flags); \
216 kfree((void *)h->vals); \
220 SCOPE void kh_clear_##name(kh_##name##_t *h) \
222 if (h && h->flags) { \
223 memset(h->flags, 0xaa, __ac_fsize(h->n_buckets) * sizeof(khint32_t)); \
224 h->size = h->n_occupied = 0; \
227 SCOPE khint_t kh_get_##name(const kh_##name##_t *h, khkey_t key) \
229 if (h->n_buckets) { \
230 khint_t k, i, last, mask, step = 0; \
231 mask = h->n_buckets - 1; \
232 k = __hash_func(key); i = k & mask; \
234 while (!__ac_isempty(h->flags, i) && (__ac_isdel(h->flags, i) || !__hash_equal(h->keys[i], key))) { \
235 i = (i + (++step)) & mask; \
236 if (i == last) return h->n_buckets; \
238 return __ac_iseither(h->flags, i)? h->n_buckets : i; \
241 SCOPE int kh_resize_##name(kh_##name##_t *h, khint_t new_n_buckets) \
242 { /* This function uses 0.25*n_buckets bytes of working space instead of [sizeof(key_t+val_t)+.25]*n_buckets. */ \
243 khint32_t *new_flags = 0; \
246 kroundup32(new_n_buckets); \
247 if (new_n_buckets < 4) new_n_buckets = 4; \
248 if (h->size >= (khint_t)(new_n_buckets * __ac_HASH_UPPER + 0.5)) j = 0; /* requested size is too small */ \
249 else { /* hash table size to be changed (shrink or expand); rehash */ \
250 new_flags = (khint32_t*)kreallocarray(NULL, __ac_fsize(new_n_buckets), sizeof(khint32_t)); \
251 if (!new_flags) return -1; \
252 memset(new_flags, 0xaa, __ac_fsize(new_n_buckets) * sizeof(khint32_t)); \
253 if (h->n_buckets < new_n_buckets) { /* expand */ \
254 khkey_t *new_keys = (khkey_t*)kreallocarray((void *)h->keys, new_n_buckets, sizeof(khkey_t)); \
255 if (!new_keys) { kfree(new_flags); return -1; } \
256 h->keys = new_keys; \
258 khval_t *new_vals = (khval_t*)kreallocarray((void *)h->vals, new_n_buckets, sizeof(khval_t)); \
259 if (!new_vals) { kfree(new_flags); return -1; } \
260 h->vals = new_vals; \
262 } /* otherwise shrink */ \
265 if (j) { /* rehashing is needed */ \
266 for (j = 0; j != h->n_buckets; ++j) { \
267 if (__ac_iseither(h->flags, j) == 0) { \
268 khkey_t key = h->keys[j]; \
271 new_mask = new_n_buckets - 1; \
272 if (kh_is_map) val = h->vals[j]; \
273 __ac_set_isdel_true(h->flags, j); \
274 while (1) { /* kick-out process; sort of like in Cuckoo hashing */ \
275 khint_t k, i, step = 0; \
276 k = __hash_func(key); \
278 while (!__ac_isempty(new_flags, i)) i = (i + (++step)) & new_mask; \
279 __ac_set_isempty_false(new_flags, i); \
280 if (i < h->n_buckets && __ac_iseither(h->flags, i) == 0) { /* kick out the existing element */ \
281 { khkey_t tmp = h->keys[i]; h->keys[i] = key; key = tmp; } \
282 if (kh_is_map) { khval_t tmp = h->vals[i]; h->vals[i] = val; val = tmp; } \
283 __ac_set_isdel_true(h->flags, i); /* mark it as deleted in the old hash table */ \
284 } else { /* write the element and jump out of the loop */ \
286 if (kh_is_map) h->vals[i] = val; \
292 if (h->n_buckets > new_n_buckets) { /* shrink the hash table */ \
293 h->keys = (khkey_t*)kreallocarray((void *)h->keys, new_n_buckets, sizeof(khkey_t)); \
294 if (kh_is_map) h->vals = (khval_t*)kreallocarray((void *)h->vals, new_n_buckets, sizeof(khval_t)); \
296 kfree(h->flags); /* free the working space */ \
297 h->flags = new_flags; \
298 h->n_buckets = new_n_buckets; \
299 h->n_occupied = h->size; \
300 h->upper_bound = (khint_t)(h->n_buckets * __ac_HASH_UPPER + 0.5); \
304 SCOPE khint_t kh_put_##name(kh_##name##_t *h, khkey_t key, int *ret) \
307 if (h->n_occupied >= h->upper_bound) { /* update the hash table */ \
308 if (h->n_buckets > (h->size<<1)) { \
309 if (kh_resize_##name(h, h->n_buckets - 1) < 0) { /* clear "deleted" elements */ \
310 *ret = -1; return h->n_buckets; \
312 } else if (kh_resize_##name(h, h->n_buckets + 1) < 0) { /* expand the hash table */ \
313 *ret = -1; return h->n_buckets; \
315 } /* TODO: to implement automatically shrinking; resize() already support shrinking */ \
317 khint_t k, i, site, last, mask = h->n_buckets - 1, step = 0; \
318 x = site = h->n_buckets; k = __hash_func(key); i = k & mask; \
319 if (__ac_isempty(h->flags, i)) x = i; /* for speed up */ \
322 while (!__ac_isempty(h->flags, i) && (__ac_isdel(h->flags, i) || !__hash_equal(h->keys[i], key))) { \
323 if (__ac_isdel(h->flags, i)) site = i; \
324 i = (i + (++step)) & mask; \
325 if (i == last) { x = site; break; } \
327 if (x == h->n_buckets) { \
328 if (__ac_isempty(h->flags, i) && site != h->n_buckets) x = site; \
333 if (__ac_isempty(h->flags, x)) { /* not present at all */ \
335 __ac_set_isboth_false(h->flags, x); \
336 ++h->size; ++h->n_occupied; \
338 } else if (__ac_isdel(h->flags, x)) { /* deleted */ \
340 __ac_set_isboth_false(h->flags, x); \
343 } else *ret = 0; /* Don't touch h->keys[x] if present and not deleted */ \
346 SCOPE void kh_del_##name(kh_##name##_t *h, khint_t x) \
348 if (x != h->n_buckets && !__ac_iseither(h->flags, x)) { \
349 __ac_set_isdel_true(h->flags, x); \
354 #define KHASH_DECLARE(name, khkey_t, khval_t) \
355 __KHASH_TYPE(name, khkey_t, khval_t) \
356 __KHASH_PROTOTYPES(name, khkey_t, khval_t)
358 #define KHASH_INIT2(name, SCOPE, khkey_t, khval_t, kh_is_map, __hash_func, __hash_equal) \
359 __KHASH_TYPE(name, khkey_t, khval_t) \
360 __KHASH_IMPL(name, SCOPE, khkey_t, khval_t, kh_is_map, __hash_func, __hash_equal)
362 #define KHASH_INIT(name, khkey_t, khval_t, kh_is_map, __hash_func, __hash_equal) \
363 KHASH_INIT2(name, static kh_inline, khkey_t, khval_t, kh_is_map, __hash_func, __hash_equal)
365 /* --- BEGIN OF HASH FUNCTIONS --- */
368 @abstract Integer hash function
369 @param key The integer [khint32_t]
370 @return The hash value [khint_t]
372 #define kh_int_hash_func(key) (khint32_t)(key)
374 @abstract Integer comparison function
376 #define kh_int_hash_equal(a, b) ((a) == (b))
378 @abstract 64-bit integer hash function
379 @param key The integer [khint64_t]
380 @return The hash value [khint_t]
382 #define kh_int64_hash_func(key) (khint32_t)((key)>>33^(key)^(key)<<11)
384 @abstract 64-bit integer comparison function
386 #define kh_int64_hash_equal(a, b) ((a) == (b))
388 @abstract const char* hash function
389 @param s Pointer to a null terminated string
390 @return The hash value
392 static kh_inline khint_t
__ac_X31_hash_string(const char *s
)
394 khint_t h
= (khint_t
)*s
;
395 if (h
) for (++s
; *s
; ++s
) h
= (h
<< 5) - h
+ (khint_t
)*s
;
399 @abstract Another interface to const char* hash function
400 @param key Pointer to a null terminated string [const char*]
401 @return The hash value [khint_t]
403 #define kh_str_hash_func(key) __ac_X31_hash_string(key)
405 @abstract Const char* comparison function
407 #define kh_str_hash_equal(a, b) (strcmp(a, b) == 0)
409 static kh_inline khint_t
__ac_Wang_hash(khint_t key
)
419 #define kh_int_hash_func2(k) __ac_Wang_hash((khint_t)key)
421 /* --- END OF HASH FUNCTIONS --- */
423 /* Other convenient macros... */
426 @abstract Type of the hash table.
427 @param name Name of the hash table [symbol]
429 #define khash_t(name) kh_##name##_t
432 @abstract Initiate a hash table.
433 @param name Name of the hash table [symbol]
434 @return Pointer to the hash table [khash_t(name)*]
436 #define kh_init(name) kh_init_##name()
439 @abstract Destroy a hash table.
440 @param name Name of the hash table [symbol]
441 @param h Pointer to the hash table [khash_t(name)*]
443 #define kh_destroy(name, h) kh_destroy_##name(h)
446 @abstract Reset a hash table without deallocating memory.
447 @param name Name of the hash table [symbol]
448 @param h Pointer to the hash table [khash_t(name)*]
450 #define kh_clear(name, h) kh_clear_##name(h)
453 @abstract Resize a hash table.
454 @param name Name of the hash table [symbol]
455 @param h Pointer to the hash table [khash_t(name)*]
456 @param s New size [khint_t]
458 #define kh_resize(name, h, s) kh_resize_##name(h, s)
461 @abstract Insert a key to the hash table.
462 @param name Name of the hash table [symbol]
463 @param h Pointer to the hash table [khash_t(name)*]
464 @param k Key [type of keys]
465 @param r Extra return code: -1 if the operation failed;
466 0 if the key is present in the hash table;
467 1 if the bucket is empty (never used); 2 if the element in
468 the bucket has been deleted [int*]
469 @return Iterator to the inserted element [khint_t]
471 #define kh_put(name, h, k, r) kh_put_##name(h, k, r)
474 @abstract Retrieve a key from the hash table.
475 @param name Name of the hash table [symbol]
476 @param h Pointer to the hash table [khash_t(name)*]
477 @param k Key [type of keys]
478 @return Iterator to the found element, or kh_end(h) if the element is absent [khint_t]
480 #define kh_get(name, h, k) kh_get_##name(h, k)
483 @abstract Remove a key from the hash table.
484 @param name Name of the hash table [symbol]
485 @param h Pointer to the hash table [khash_t(name)*]
486 @param k Iterator to the element to be deleted [khint_t]
488 #define kh_del(name, h, k) kh_del_##name(h, k)
491 @abstract Test whether a bucket contains data.
492 @param h Pointer to the hash table [khash_t(name)*]
493 @param x Iterator to the bucket [khint_t]
494 @return 1 if containing data; 0 otherwise [int]
496 #define kh_exist(h, x) (!__ac_iseither((h)->flags, (x)))
499 @abstract Get key given an iterator
500 @param h Pointer to the hash table [khash_t(name)*]
501 @param x Iterator to the bucket [khint_t]
502 @return Key [type of keys]
504 #define kh_key(h, x) ((h)->keys[x])
507 @abstract Get value given an iterator
508 @param h Pointer to the hash table [khash_t(name)*]
509 @param x Iterator to the bucket [khint_t]
510 @return Value [type of values]
511 @discussion For hash sets, calling this results in segfault.
513 #define kh_val(h, x) ((h)->vals[x])
516 @abstract Alias of kh_val()
518 #define kh_value(h, x) ((h)->vals[x])
521 @abstract Get the start iterator
522 @param h Pointer to the hash table [khash_t(name)*]
523 @return The start iterator [khint_t]
525 #define kh_begin(h) (khint_t)(0)
528 @abstract Get the end iterator
529 @param h Pointer to the hash table [khash_t(name)*]
530 @return The end iterator [khint_t]
532 #define kh_end(h) ((h)->n_buckets)
535 @abstract Get the number of elements in the hash table
536 @param h Pointer to the hash table [khash_t(name)*]
537 @return Number of elements in the hash table [khint_t]
539 #define kh_size(h) ((h)->size)
542 @abstract Get the number of buckets in the hash table
543 @param h Pointer to the hash table [khash_t(name)*]
544 @return Number of buckets in the hash table [khint_t]
546 #define kh_n_buckets(h) ((h)->n_buckets)
549 @abstract Iterate over the entries in the hash table
550 @param h Pointer to the hash table [khash_t(name)*]
551 @param kvar Variable to which key will be assigned
552 @param vvar Variable to which value will be assigned
553 @param code Block of code to execute
555 #define kh_foreach(h, kvar, vvar, code) { khint_t __i; \
556 for (__i = kh_begin(h); __i != kh_end(h); ++__i) { \
557 if (!kh_exist(h,__i)) continue; \
558 (kvar) = kh_key(h,__i); \
559 (vvar) = kh_val(h,__i); \
564 @abstract Iterate over the values in the hash table
565 @param h Pointer to the hash table [khash_t(name)*]
566 @param vvar Variable to which value will be assigned
567 @param code Block of code to execute
569 #define kh_foreach_value(h, vvar, code) { khint_t __i; \
570 for (__i = kh_begin(h); __i != kh_end(h); ++__i) { \
571 if (!kh_exist(h,__i)) continue; \
572 (vvar) = kh_val(h,__i); \
576 /* More conenient interfaces */
579 @abstract Instantiate a hash set containing integer keys
580 @param name Name of the hash table [symbol]
582 #define KHASH_SET_INIT_INT(name) \
583 KHASH_INIT(name, khint32_t, char, 0, kh_int_hash_func, kh_int_hash_equal)
586 @abstract Instantiate a hash map containing integer keys
587 @param name Name of the hash table [symbol]
588 @param khval_t Type of values [type]
590 #define KHASH_MAP_INIT_INT(name, khval_t) \
591 KHASH_INIT(name, khint32_t, khval_t, 1, kh_int_hash_func, kh_int_hash_equal)
594 @abstract Instantiate a hash map containing 64-bit integer keys
595 @param name Name of the hash table [symbol]
597 #define KHASH_SET_INIT_INT64(name) \
598 KHASH_INIT(name, khint64_t, char, 0, kh_int64_hash_func, kh_int64_hash_equal)
601 @abstract Instantiate a hash map containing 64-bit integer keys
602 @param name Name of the hash table [symbol]
603 @param khval_t Type of values [type]
605 #define KHASH_MAP_INIT_INT64(name, khval_t) \
606 KHASH_INIT(name, khint64_t, khval_t, 1, kh_int64_hash_func, kh_int64_hash_equal)
608 typedef const char *kh_cstr_t
;
610 @abstract Instantiate a hash map containing const char* keys
611 @param name Name of the hash table [symbol]
613 #define KHASH_SET_INIT_STR(name) \
614 KHASH_INIT(name, kh_cstr_t, char, 0, kh_str_hash_func, kh_str_hash_equal)
617 @abstract Instantiate a hash map containing const char* keys
618 @param name Name of the hash table [symbol]
619 @param khval_t Type of values [type]
621 #define KHASH_MAP_INIT_STR(name, khval_t) \
622 KHASH_INIT(name, kh_cstr_t, khval_t, 1, kh_str_hash_func, kh_str_hash_equal)
624 #endif /* __AC_KHASH_H */