Mercurial > vim
annotate src/hashtab.c @ 21834:47742d4ad2a3
Added tag v8.2.1466 for changeset e3f9528bddda314f114f46f2209678d1bf18ad24
| author | Bram Moolenaar <Bram@vim.org> |
|---|---|
| date | Sun, 16 Aug 2020 17:45:23 +0200 |
| parents | 883aa425656a |
| children | 4ed106deb772 |
| rev | line source |
|---|---|
|
10042
4aead6a9b7a9
commit https://github.com/vim/vim/commit/edf3f97ae2af024708ebb4ac614227327033ca47
Christian Brabandt <cb@256bit.org>
parents:
9513
diff
changeset
|
1 /* vi:set ts=8 sts=4 sw=4 noet: |
| 799 | 2 * |
| 3 * VIM - Vi IMproved by Bram Moolenaar | |
| 4 * | |
| 5 * Do ":help uganda" in Vim to read copying and usage conditions. | |
| 6 * Do ":help credits" in Vim to see a list of people who contributed. | |
| 7 * See README.txt for an overview of the Vim source code. | |
| 8 */ | |
| 9 | |
| 10 /* | |
| 11 * hashtab.c: Handling of a hashtable with Vim-specific properties. | |
| 12 * | |
| 13 * Each item in a hashtable has a NUL terminated string key. A key can appear | |
| 14 * only once in the table. | |
| 15 * | |
| 16 * A hash number is computed from the key for quick lookup. When the hashes | |
| 17 * of two different keys point to the same entry an algorithm is used to | |
| 18 * iterate over other entries in the table until the right one is found. | |
| 19 * To make the iteration work removed keys are different from entries where a | |
| 20 * key was never present. | |
| 21 * | |
| 22 * The mechanism has been partly based on how Python Dictionaries are | |
| 23 * implemented. The algorithm is from Knuth Vol. 3, Sec. 6.4. | |
| 24 * | |
| 25 * The hashtable grows to accommodate more entries when needed. At least 1/3 | |
| 26 * of the entries is empty to keep the lookup efficient (at the cost of extra | |
| 27 * memory). | |
| 28 */ | |
| 29 | |
| 30 #include "vim.h" | |
| 31 | |
| 32 #if 0 | |
|
18798
f0f9692d4487
patch 8.1.2387: using old C style comments
Bram Moolenaar <Bram@vim.org>
parents:
17508
diff
changeset
|
33 # define HT_DEBUG // extra checks for table consistency and statistics |
| 799 | 34 |
|
18798
f0f9692d4487
patch 8.1.2387: using old C style comments
Bram Moolenaar <Bram@vim.org>
parents:
17508
diff
changeset
|
35 static long hash_count_lookup = 0; // count number of hashtab lookups |
|
f0f9692d4487
patch 8.1.2387: using old C style comments
Bram Moolenaar <Bram@vim.org>
parents:
17508
diff
changeset
|
36 static long hash_count_perturb = 0; // count number of "misses" |
| 799 | 37 #endif |
| 38 | |
|
18798
f0f9692d4487
patch 8.1.2387: using old C style comments
Bram Moolenaar <Bram@vim.org>
parents:
17508
diff
changeset
|
39 // Magic value for algorithm that walks through the array. |
| 799 | 40 #define PERTURB_SHIFT 5 |
| 41 | |
|
7803
37c929c4a073
commit https://github.com/vim/vim/commit/92b8b2d307e34117f146319872010b0ccc9d2713
Christian Brabandt <cb@256bit.org>
parents:
3970
diff
changeset
|
42 static int hash_may_resize(hashtab_T *ht, int minitems); |
| 799 | 43 |
|
18798
f0f9692d4487
patch 8.1.2387: using old C style comments
Bram Moolenaar <Bram@vim.org>
parents:
17508
diff
changeset
|
44 #if 0 // currently not used |
| 799 | 45 /* |
| 46 * Create an empty hash table. | |
| 47 * Returns NULL when out of memory. | |
| 48 */ | |
| 49 hashtab_T * | |
|
7823
bcef391c101c
commit https://github.com/vim/vim/commit/68c2f638e65d914dc6e84eb7ce2624f08af525c0
Christian Brabandt <cb@256bit.org>
parents:
7803
diff
changeset
|
50 hash_create(void) |
| 799 | 51 { |
| 52 hashtab_T *ht; | |
| 53 | |
|
16825
ce04ebdf26b8
patch 8.1.1414: alloc() returning "char_u *" causes a lot of type casts
Bram Moolenaar <Bram@vim.org>
parents:
16764
diff
changeset
|
54 ht = ALLOC_ONE(hashtab_T); |
| 799 | 55 if (ht != NULL) |
| 56 hash_init(ht); | |
| 57 return ht; | |
| 58 } | |
| 59 #endif | |
| 60 | |
| 61 /* | |
| 62 * Initialize an empty hash table. | |
| 63 */ | |
| 64 void | |
|
7823
bcef391c101c
commit https://github.com/vim/vim/commit/68c2f638e65d914dc6e84eb7ce2624f08af525c0
Christian Brabandt <cb@256bit.org>
parents:
7803
diff
changeset
|
65 hash_init(hashtab_T *ht) |
| 799 | 66 { |
|
18798
f0f9692d4487
patch 8.1.2387: using old C style comments
Bram Moolenaar <Bram@vim.org>
parents:
17508
diff
changeset
|
67 // This zeroes all "ht_" entries and all the "hi_key" in "ht_smallarray". |
|
20007
aadd1cae2ff5
patch 8.2.0559: clearing a struct is verbose
Bram Moolenaar <Bram@vim.org>
parents:
18798
diff
changeset
|
68 CLEAR_POINTER(ht); |
| 799 | 69 ht->ht_array = ht->ht_smallarray; |
| 70 ht->ht_mask = HT_INIT_SIZE - 1; | |
| 71 } | |
| 72 | |
| 73 /* | |
| 74 * Free the array of a hash table. Does not free the items it contains! | |
| 75 * If "ht" is not freed then you should call hash_init() next! | |
| 76 */ | |
| 77 void | |
|
7823
bcef391c101c
commit https://github.com/vim/vim/commit/68c2f638e65d914dc6e84eb7ce2624f08af525c0
Christian Brabandt <cb@256bit.org>
parents:
7803
diff
changeset
|
78 hash_clear(hashtab_T *ht) |
| 799 | 79 { |
| 80 if (ht->ht_array != ht->ht_smallarray) | |
| 81 vim_free(ht->ht_array); | |
| 82 } | |
| 83 | |
|
15555
d89c5b339c2a
patch 8.1.0785: depending on the configuration some functions are unused
Bram Moolenaar <Bram@vim.org>
parents:
15470
diff
changeset
|
84 #if defined(FEAT_SPELL) || defined(PROTO) |
| 799 | 85 /* |
| 86 * Free the array of a hash table and all the keys it contains. The keys must | |
| 87 * have been allocated. "off" is the offset from the start of the allocate | |
| 88 * memory to the location of the key (it's always positive). | |
| 89 */ | |
| 90 void | |
|
7823
bcef391c101c
commit https://github.com/vim/vim/commit/68c2f638e65d914dc6e84eb7ce2624f08af525c0
Christian Brabandt <cb@256bit.org>
parents:
7803
diff
changeset
|
91 hash_clear_all(hashtab_T *ht, int off) |
| 799 | 92 { |
| 835 | 93 long todo; |
| 799 | 94 hashitem_T *hi; |
| 95 | |
| 835 | 96 todo = (long)ht->ht_used; |
| 799 | 97 for (hi = ht->ht_array; todo > 0; ++hi) |
| 98 { | |
| 99 if (!HASHITEM_EMPTY(hi)) | |
| 100 { | |
| 101 vim_free(hi->hi_key - off); | |
| 102 --todo; | |
| 103 } | |
| 104 } | |
| 105 hash_clear(ht); | |
| 106 } | |
|
15555
d89c5b339c2a
patch 8.1.0785: depending on the configuration some functions are unused
Bram Moolenaar <Bram@vim.org>
parents:
15470
diff
changeset
|
107 #endif |
| 799 | 108 |
| 109 /* | |
| 110 * Find "key" in hashtable "ht". "key" must not be NULL. | |
| 111 * Always returns a pointer to a hashitem. If the item was not found then | |
| 112 * HASHITEM_EMPTY() is TRUE. The pointer is then the place where the key | |
| 113 * would be added. | |
| 114 * WARNING: The returned pointer becomes invalid when the hashtable is changed | |
| 115 * (adding, setting or removing an item)! | |
| 116 */ | |
| 117 hashitem_T * | |
|
7823
bcef391c101c
commit https://github.com/vim/vim/commit/68c2f638e65d914dc6e84eb7ce2624f08af525c0
Christian Brabandt <cb@256bit.org>
parents:
7803
diff
changeset
|
118 hash_find(hashtab_T *ht, char_u *key) |
| 799 | 119 { |
| 120 return hash_lookup(ht, key, hash_hash(key)); | |
| 121 } | |
| 122 | |
| 123 /* | |
| 124 * Like hash_find(), but caller computes "hash". | |
| 125 */ | |
| 126 hashitem_T * | |
|
7823
bcef391c101c
commit https://github.com/vim/vim/commit/68c2f638e65d914dc6e84eb7ce2624f08af525c0
Christian Brabandt <cb@256bit.org>
parents:
7803
diff
changeset
|
127 hash_lookup(hashtab_T *ht, char_u *key, hash_T hash) |
| 799 | 128 { |
| 129 hash_T perturb; | |
| 130 hashitem_T *freeitem; | |
| 131 hashitem_T *hi; | |
| 3970 | 132 unsigned idx; |
| 799 | 133 |
| 134 #ifdef HT_DEBUG | |
| 135 ++hash_count_lookup; | |
| 136 #endif | |
| 137 | |
| 138 /* | |
| 139 * Quickly handle the most common situations: | |
| 140 * - return if there is no item at all | |
| 141 * - skip over a removed item | |
| 142 * - return if the item matches | |
| 143 */ | |
| 3970 | 144 idx = (unsigned)(hash & ht->ht_mask); |
| 799 | 145 hi = &ht->ht_array[idx]; |
| 146 | |
| 147 if (hi->hi_key == NULL) | |
| 148 return hi; | |
| 149 if (hi->hi_key == HI_KEY_REMOVED) | |
| 150 freeitem = hi; | |
| 151 else if (hi->hi_hash == hash && STRCMP(hi->hi_key, key) == 0) | |
| 152 return hi; | |
| 153 else | |
| 154 freeitem = NULL; | |
| 155 | |
| 156 /* | |
| 157 * Need to search through the table to find the key. The algorithm | |
| 158 * to step through the table starts with large steps, gradually becoming | |
| 159 * smaller down to (1/4 table size + 1). This means it goes through all | |
| 160 * table entries in the end. | |
| 161 * When we run into a NULL key it's clear that the key isn't there. | |
| 162 * Return the first available slot found (can be a slot of a removed | |
| 163 * item). | |
| 164 */ | |
| 165 for (perturb = hash; ; perturb >>= PERTURB_SHIFT) | |
| 166 { | |
| 167 #ifdef HT_DEBUG | |
|
18798
f0f9692d4487
patch 8.1.2387: using old C style comments
Bram Moolenaar <Bram@vim.org>
parents:
17508
diff
changeset
|
168 ++hash_count_perturb; // count a "miss" for hashtab lookup |
| 799 | 169 #endif |
| 3970 | 170 idx = (unsigned)((idx << 2U) + idx + perturb + 1U); |
| 799 | 171 hi = &ht->ht_array[idx & ht->ht_mask]; |
| 172 if (hi->hi_key == NULL) | |
| 173 return freeitem == NULL ? hi : freeitem; | |
| 174 if (hi->hi_hash == hash | |
| 175 && hi->hi_key != HI_KEY_REMOVED | |
| 176 && STRCMP(hi->hi_key, key) == 0) | |
| 177 return hi; | |
| 178 if (hi->hi_key == HI_KEY_REMOVED && freeitem == NULL) | |
| 179 freeitem = hi; | |
| 180 } | |
| 181 } | |
| 182 | |
|
15555
d89c5b339c2a
patch 8.1.0785: depending on the configuration some functions are unused
Bram Moolenaar <Bram@vim.org>
parents:
15470
diff
changeset
|
183 #if defined(FEAT_EVAL) || defined(FEAT_SYN_HL) || defined(PROTO) |
| 799 | 184 /* |
| 185 * Print the efficiency of hashtable lookups. | |
| 186 * Useful when trying different hash algorithms. | |
| 187 * Called when exiting. | |
| 188 */ | |
| 189 void | |
|
7823
bcef391c101c
commit https://github.com/vim/vim/commit/68c2f638e65d914dc6e84eb7ce2624f08af525c0
Christian Brabandt <cb@256bit.org>
parents:
7803
diff
changeset
|
190 hash_debug_results(void) |
| 799 | 191 { |
| 192 #ifdef HT_DEBUG | |
| 193 fprintf(stderr, "\r\n\r\n\r\n\r\n"); | |
| 194 fprintf(stderr, "Number of hashtable lookups: %ld\r\n", hash_count_lookup); | |
| 195 fprintf(stderr, "Number of perturb loops: %ld\r\n", hash_count_perturb); | |
| 196 fprintf(stderr, "Percentage of perturb loops: %ld%%\r\n", | |
| 197 hash_count_perturb * 100 / hash_count_lookup); | |
| 198 #endif | |
| 199 } | |
|
15555
d89c5b339c2a
patch 8.1.0785: depending on the configuration some functions are unused
Bram Moolenaar <Bram@vim.org>
parents:
15470
diff
changeset
|
200 #endif |
| 799 | 201 |
| 202 /* | |
| 203 * Add item with key "key" to hashtable "ht". | |
| 204 * Returns FAIL when out of memory or the key is already present. | |
| 205 */ | |
| 206 int | |
|
7823
bcef391c101c
commit https://github.com/vim/vim/commit/68c2f638e65d914dc6e84eb7ce2624f08af525c0
Christian Brabandt <cb@256bit.org>
parents:
7803
diff
changeset
|
207 hash_add(hashtab_T *ht, char_u *key) |
| 799 | 208 { |
| 209 hash_T hash = hash_hash(key); | |
| 210 hashitem_T *hi; | |
| 211 | |
| 212 hi = hash_lookup(ht, key, hash); | |
| 213 if (!HASHITEM_EMPTY(hi)) | |
| 214 { | |
|
10359
66f1b5bf3fa6
commit https://github.com/vim/vim/commit/95f096030ed1a8afea028f2ea295d6f6a70f466f
Christian Brabandt <cb@256bit.org>
parents:
10042
diff
changeset
|
215 internal_error("hash_add()"); |
| 799 | 216 return FAIL; |
| 217 } | |
| 218 return hash_add_item(ht, hi, key, hash); | |
| 219 } | |
| 220 | |
| 221 /* | |
| 222 * Add item "hi" with "key" to hashtable "ht". "key" must not be NULL and | |
| 223 * "hi" must have been obtained with hash_lookup() and point to an empty item. | |
| 224 * "hi" is invalid after this! | |
| 225 * Returns OK or FAIL (out of memory). | |
| 226 */ | |
| 227 int | |
|
7823
bcef391c101c
commit https://github.com/vim/vim/commit/68c2f638e65d914dc6e84eb7ce2624f08af525c0
Christian Brabandt <cb@256bit.org>
parents:
7803
diff
changeset
|
228 hash_add_item( |
|
bcef391c101c
commit https://github.com/vim/vim/commit/68c2f638e65d914dc6e84eb7ce2624f08af525c0
Christian Brabandt <cb@256bit.org>
parents:
7803
diff
changeset
|
229 hashtab_T *ht, |
|
bcef391c101c
commit https://github.com/vim/vim/commit/68c2f638e65d914dc6e84eb7ce2624f08af525c0
Christian Brabandt <cb@256bit.org>
parents:
7803
diff
changeset
|
230 hashitem_T *hi, |
|
bcef391c101c
commit https://github.com/vim/vim/commit/68c2f638e65d914dc6e84eb7ce2624f08af525c0
Christian Brabandt <cb@256bit.org>
parents:
7803
diff
changeset
|
231 char_u *key, |
|
bcef391c101c
commit https://github.com/vim/vim/commit/68c2f638e65d914dc6e84eb7ce2624f08af525c0
Christian Brabandt <cb@256bit.org>
parents:
7803
diff
changeset
|
232 hash_T hash) |
| 799 | 233 { |
|
18798
f0f9692d4487
patch 8.1.2387: using old C style comments
Bram Moolenaar <Bram@vim.org>
parents:
17508
diff
changeset
|
234 // If resizing failed before and it fails again we can't add an item. |
| 799 | 235 if (ht->ht_error && hash_may_resize(ht, 0) == FAIL) |
| 236 return FAIL; | |
| 237 | |
| 238 ++ht->ht_used; | |
|
21317
883aa425656a
patch 8.2.1209: Vim9: test failure
Bram Moolenaar <Bram@vim.org>
parents:
20007
diff
changeset
|
239 ++ht->ht_changed; |
| 799 | 240 if (hi->hi_key == NULL) |
| 241 ++ht->ht_filled; | |
| 242 hi->hi_key = key; | |
| 243 hi->hi_hash = hash; | |
| 244 | |
|
18798
f0f9692d4487
patch 8.1.2387: using old C style comments
Bram Moolenaar <Bram@vim.org>
parents:
17508
diff
changeset
|
245 // When the space gets low may resize the array. |
| 799 | 246 return hash_may_resize(ht, 0); |
| 247 } | |
| 248 | |
|
18798
f0f9692d4487
patch 8.1.2387: using old C style comments
Bram Moolenaar <Bram@vim.org>
parents:
17508
diff
changeset
|
249 #if 0 // not used |
| 799 | 250 /* |
| 251 * Overwrite hashtable item "hi" with "key". "hi" must point to the item that | |
| 252 * is to be overwritten. Thus the number of items in the hashtable doesn't | |
| 253 * change. | |
| 254 * Although the key must be identical, the pointer may be different, thus it's | |
| 255 * set anyway (the key is part of an item with that key). | |
| 256 * The caller must take care of freeing the old item. | |
| 257 * "hi" is invalid after this! | |
| 258 */ | |
| 259 void | |
|
7823
bcef391c101c
commit https://github.com/vim/vim/commit/68c2f638e65d914dc6e84eb7ce2624f08af525c0
Christian Brabandt <cb@256bit.org>
parents:
7803
diff
changeset
|
260 hash_set(hashitem_T *hi, char_u *key) |
| 799 | 261 { |
| 262 hi->hi_key = key; | |
| 263 } | |
| 264 #endif | |
| 265 | |
| 266 /* | |
| 267 * Remove item "hi" from hashtable "ht". "hi" must have been obtained with | |
| 268 * hash_lookup(). | |
| 269 * The caller must take care of freeing the item itself. | |
| 270 */ | |
| 271 void | |
|
7823
bcef391c101c
commit https://github.com/vim/vim/commit/68c2f638e65d914dc6e84eb7ce2624f08af525c0
Christian Brabandt <cb@256bit.org>
parents:
7803
diff
changeset
|
272 hash_remove(hashtab_T *ht, hashitem_T *hi) |
| 799 | 273 { |
| 274 --ht->ht_used; | |
|
21317
883aa425656a
patch 8.2.1209: Vim9: test failure
Bram Moolenaar <Bram@vim.org>
parents:
20007
diff
changeset
|
275 ++ht->ht_changed; |
| 799 | 276 hi->hi_key = HI_KEY_REMOVED; |
| 277 hash_may_resize(ht, 0); | |
| 278 } | |
| 279 | |
| 280 /* | |
| 281 * Lock a hashtable: prevent that ht_array changes. | |
| 282 * Don't use this when items are to be added! | |
| 283 * Must call hash_unlock() later. | |
| 284 */ | |
| 285 void | |
|
7823
bcef391c101c
commit https://github.com/vim/vim/commit/68c2f638e65d914dc6e84eb7ce2624f08af525c0
Christian Brabandt <cb@256bit.org>
parents:
7803
diff
changeset
|
286 hash_lock(hashtab_T *ht) |
| 799 | 287 { |
| 288 ++ht->ht_locked; | |
| 289 } | |
| 290 | |
| 291 /* | |
| 292 * Lock a hashtable at the specified number of entries. | |
| 293 * Caller must make sure no more than "size" entries will be added. | |
| 294 * Must call hash_unlock() later. | |
| 295 */ | |
| 296 void | |
|
7823
bcef391c101c
commit https://github.com/vim/vim/commit/68c2f638e65d914dc6e84eb7ce2624f08af525c0
Christian Brabandt <cb@256bit.org>
parents:
7803
diff
changeset
|
297 hash_lock_size(hashtab_T *ht, int size) |
| 799 | 298 { |
| 299 (void)hash_may_resize(ht, size); | |
| 300 ++ht->ht_locked; | |
| 301 } | |
| 302 | |
| 303 /* | |
| 304 * Unlock a hashtable: allow ht_array changes again. | |
| 305 * Table will be resized (shrink) when necessary. | |
| 306 * This must balance a call to hash_lock(). | |
| 307 */ | |
| 308 void | |
|
7823
bcef391c101c
commit https://github.com/vim/vim/commit/68c2f638e65d914dc6e84eb7ce2624f08af525c0
Christian Brabandt <cb@256bit.org>
parents:
7803
diff
changeset
|
309 hash_unlock(hashtab_T *ht) |
| 799 | 310 { |
| 311 --ht->ht_locked; | |
| 312 (void)hash_may_resize(ht, 0); | |
| 313 } | |
| 314 | |
| 315 /* | |
| 316 * Shrink a hashtable when there is too much empty space. | |
| 317 * Grow a hashtable when there is not enough empty space. | |
| 318 * Returns OK or FAIL (out of memory). | |
| 319 */ | |
| 320 static int | |
|
7823
bcef391c101c
commit https://github.com/vim/vim/commit/68c2f638e65d914dc6e84eb7ce2624f08af525c0
Christian Brabandt <cb@256bit.org>
parents:
7803
diff
changeset
|
321 hash_may_resize( |
|
bcef391c101c
commit https://github.com/vim/vim/commit/68c2f638e65d914dc6e84eb7ce2624f08af525c0
Christian Brabandt <cb@256bit.org>
parents:
7803
diff
changeset
|
322 hashtab_T *ht, |
|
18798
f0f9692d4487
patch 8.1.2387: using old C style comments
Bram Moolenaar <Bram@vim.org>
parents:
17508
diff
changeset
|
323 int minitems) // minimal number of items |
| 799 | 324 { |
| 325 hashitem_T temparray[HT_INIT_SIZE]; | |
| 326 hashitem_T *oldarray, *newarray; | |
| 327 hashitem_T *olditem, *newitem; | |
| 3970 | 328 unsigned newi; |
| 799 | 329 int todo; |
| 330 long_u oldsize, newsize; | |
| 331 long_u minsize; | |
| 332 long_u newmask; | |
| 333 hash_T perturb; | |
| 334 | |
|
18798
f0f9692d4487
patch 8.1.2387: using old C style comments
Bram Moolenaar <Bram@vim.org>
parents:
17508
diff
changeset
|
335 // Don't resize a locked table. |
| 799 | 336 if (ht->ht_locked > 0) |
| 337 return OK; | |
| 338 | |
| 339 #ifdef HT_DEBUG | |
| 340 if (ht->ht_used > ht->ht_filled) | |
|
15470
55ccc2d353bd
patch 8.1.0743: giving error messages is not flexible
Bram Moolenaar <Bram@vim.org>
parents:
10605
diff
changeset
|
341 emsg("hash_may_resize(): more used than filled"); |
| 799 | 342 if (ht->ht_filled >= ht->ht_mask + 1) |
|
15470
55ccc2d353bd
patch 8.1.0743: giving error messages is not flexible
Bram Moolenaar <Bram@vim.org>
parents:
10605
diff
changeset
|
343 emsg("hash_may_resize(): table completely filled"); |
| 799 | 344 #endif |
| 345 | |
| 346 if (minitems == 0) | |
| 347 { | |
|
18798
f0f9692d4487
patch 8.1.2387: using old C style comments
Bram Moolenaar <Bram@vim.org>
parents:
17508
diff
changeset
|
348 // Return quickly for small tables with at least two NULL items. NULL |
|
f0f9692d4487
patch 8.1.2387: using old C style comments
Bram Moolenaar <Bram@vim.org>
parents:
17508
diff
changeset
|
349 // items are required for the lookup to decide a key isn't there. |
| 799 | 350 if (ht->ht_filled < HT_INIT_SIZE - 1 |
| 351 && ht->ht_array == ht->ht_smallarray) | |
| 352 return OK; | |
| 353 | |
| 354 /* | |
| 355 * Grow or refill the array when it's more than 2/3 full (including | |
| 356 * removed items, so that they get cleaned up). | |
| 357 * Shrink the array when it's less than 1/5 full. When growing it is | |
| 358 * at least 1/4 full (avoids repeated grow-shrink operations) | |
| 359 */ | |
| 360 oldsize = ht->ht_mask + 1; | |
| 361 if (ht->ht_filled * 3 < oldsize * 2 && ht->ht_used > oldsize / 5) | |
| 362 return OK; | |
| 363 | |
| 364 if (ht->ht_used > 1000) | |
|
18798
f0f9692d4487
patch 8.1.2387: using old C style comments
Bram Moolenaar <Bram@vim.org>
parents:
17508
diff
changeset
|
365 minsize = ht->ht_used * 2; // it's big, don't make too much room |
| 799 | 366 else |
|
18798
f0f9692d4487
patch 8.1.2387: using old C style comments
Bram Moolenaar <Bram@vim.org>
parents:
17508
diff
changeset
|
367 minsize = ht->ht_used * 4; // make plenty of room |
| 799 | 368 } |
| 369 else | |
| 370 { | |
|
17508
34966be2e856
patch 8.1.1752: resizing hashtable is inefficient
Bram Moolenaar <Bram@vim.org>
parents:
16825
diff
changeset
|
371 // Use specified size. |
|
34966be2e856
patch 8.1.1752: resizing hashtable is inefficient
Bram Moolenaar <Bram@vim.org>
parents:
16825
diff
changeset
|
372 if ((long_u)minitems < ht->ht_used) // just in case... |
| 835 | 373 minitems = (int)ht->ht_used; |
|
17508
34966be2e856
patch 8.1.1752: resizing hashtable is inefficient
Bram Moolenaar <Bram@vim.org>
parents:
16825
diff
changeset
|
374 minsize = (minitems * 3 + 1) / 2; // array is up to 2/3 full |
| 799 | 375 } |
| 376 | |
| 377 newsize = HT_INIT_SIZE; | |
| 378 while (newsize < minsize) | |
| 379 { | |
|
18798
f0f9692d4487
patch 8.1.2387: using old C style comments
Bram Moolenaar <Bram@vim.org>
parents:
17508
diff
changeset
|
380 newsize <<= 1; // make sure it's always a power of 2 |
| 799 | 381 if (newsize == 0) |
|
18798
f0f9692d4487
patch 8.1.2387: using old C style comments
Bram Moolenaar <Bram@vim.org>
parents:
17508
diff
changeset
|
382 return FAIL; // overflow |
| 799 | 383 } |
| 384 | |
| 385 if (newsize == HT_INIT_SIZE) | |
| 386 { | |
|
18798
f0f9692d4487
patch 8.1.2387: using old C style comments
Bram Moolenaar <Bram@vim.org>
parents:
17508
diff
changeset
|
387 // Use the small array inside the hashdict structure. |
| 799 | 388 newarray = ht->ht_smallarray; |
| 389 if (ht->ht_array == newarray) | |
| 390 { | |
|
18798
f0f9692d4487
patch 8.1.2387: using old C style comments
Bram Moolenaar <Bram@vim.org>
parents:
17508
diff
changeset
|
391 // Moving from ht_smallarray to ht_smallarray! Happens when there |
|
f0f9692d4487
patch 8.1.2387: using old C style comments
Bram Moolenaar <Bram@vim.org>
parents:
17508
diff
changeset
|
392 // are many removed items. Copy the items to be able to clean up |
|
f0f9692d4487
patch 8.1.2387: using old C style comments
Bram Moolenaar <Bram@vim.org>
parents:
17508
diff
changeset
|
393 // removed items. |
| 799 | 394 mch_memmove(temparray, newarray, sizeof(temparray)); |
| 395 oldarray = temparray; | |
| 396 } | |
| 397 else | |
| 398 oldarray = ht->ht_array; | |
|
20007
aadd1cae2ff5
patch 8.2.0559: clearing a struct is verbose
Bram Moolenaar <Bram@vim.org>
parents:
18798
diff
changeset
|
399 CLEAR_FIELD(ht->ht_smallarray); |
| 799 | 400 } |
| 401 else | |
| 402 { | |
|
18798
f0f9692d4487
patch 8.1.2387: using old C style comments
Bram Moolenaar <Bram@vim.org>
parents:
17508
diff
changeset
|
403 // Allocate an array. |
|
20007
aadd1cae2ff5
patch 8.2.0559: clearing a struct is verbose
Bram Moolenaar <Bram@vim.org>
parents:
18798
diff
changeset
|
404 newarray = ALLOC_CLEAR_MULT(hashitem_T, newsize); |
| 799 | 405 if (newarray == NULL) |
| 406 { | |
|
18798
f0f9692d4487
patch 8.1.2387: using old C style comments
Bram Moolenaar <Bram@vim.org>
parents:
17508
diff
changeset
|
407 // Out of memory. When there are NULL items still return OK. |
|
f0f9692d4487
patch 8.1.2387: using old C style comments
Bram Moolenaar <Bram@vim.org>
parents:
17508
diff
changeset
|
408 // Otherwise set ht_error, because lookup may result in a hang if |
|
f0f9692d4487
patch 8.1.2387: using old C style comments
Bram Moolenaar <Bram@vim.org>
parents:
17508
diff
changeset
|
409 // we add another item. |
| 799 | 410 if (ht->ht_filled < ht->ht_mask) |
| 411 return OK; | |
| 412 ht->ht_error = TRUE; | |
| 413 return FAIL; | |
| 414 } | |
| 415 oldarray = ht->ht_array; | |
| 416 } | |
| 417 | |
| 418 /* | |
| 419 * Move all the items from the old array to the new one, placing them in | |
| 420 * the right spot. The new array won't have any removed items, thus this | |
| 421 * is also a cleanup action. | |
| 422 */ | |
| 423 newmask = newsize - 1; | |
| 835 | 424 todo = (int)ht->ht_used; |
| 799 | 425 for (olditem = oldarray; todo > 0; ++olditem) |
| 426 if (!HASHITEM_EMPTY(olditem)) | |
| 427 { | |
| 428 /* | |
| 429 * The algorithm to find the spot to add the item is identical to | |
| 430 * the algorithm to find an item in hash_lookup(). But we only | |
| 431 * need to search for a NULL key, thus it's simpler. | |
| 432 */ | |
| 3970 | 433 newi = (unsigned)(olditem->hi_hash & newmask); |
| 799 | 434 newitem = &newarray[newi]; |
| 435 | |
| 436 if (newitem->hi_key != NULL) | |
| 437 for (perturb = olditem->hi_hash; ; perturb >>= PERTURB_SHIFT) | |
| 438 { | |
| 3970 | 439 newi = (unsigned)((newi << 2U) + newi + perturb + 1U); |
| 799 | 440 newitem = &newarray[newi & newmask]; |
| 441 if (newitem->hi_key == NULL) | |
| 442 break; | |
| 443 } | |
| 444 *newitem = *olditem; | |
| 445 --todo; | |
| 446 } | |
| 447 | |
| 448 if (ht->ht_array != ht->ht_smallarray) | |
| 449 vim_free(ht->ht_array); | |
| 450 ht->ht_array = newarray; | |
| 451 ht->ht_mask = newmask; | |
| 452 ht->ht_filled = ht->ht_used; | |
|
21317
883aa425656a
patch 8.2.1209: Vim9: test failure
Bram Moolenaar <Bram@vim.org>
parents:
20007
diff
changeset
|
453 ++ht->ht_changed; |
| 799 | 454 ht->ht_error = FALSE; |
| 455 | |
| 456 return OK; | |
| 457 } | |
| 458 | |
| 459 /* | |
| 460 * Get the hash number for a key. | |
| 461 * If you think you know a better hash function: Compile with HT_DEBUG set and | |
| 462 * run a script that uses hashtables a lot. Vim will then print statistics | |
| 463 * when exiting. Try that with the current hash algorithm and yours. The | |
| 464 * lower the percentage the better. | |
| 465 */ | |
| 466 hash_T | |
|
7823
bcef391c101c
commit https://github.com/vim/vim/commit/68c2f638e65d914dc6e84eb7ce2624f08af525c0
Christian Brabandt <cb@256bit.org>
parents:
7803
diff
changeset
|
467 hash_hash(char_u *key) |
| 799 | 468 { |
| 469 hash_T hash; | |
| 470 char_u *p; | |
| 471 | |
| 472 if ((hash = *key) == 0) | |
|
8839
9fa567d13551
commit https://github.com/vim/vim/commit/0921ecff1c5a74541bad6c073e8ade32247403d8
Christian Brabandt <cb@256bit.org>
parents:
7823
diff
changeset
|
473 return (hash_T)0; |
| 799 | 474 p = key + 1; |
| 475 | |
|
18798
f0f9692d4487
patch 8.1.2387: using old C style comments
Bram Moolenaar <Bram@vim.org>
parents:
17508
diff
changeset
|
476 // A simplistic algorithm that appears to do very well. |
|
f0f9692d4487
patch 8.1.2387: using old C style comments
Bram Moolenaar <Bram@vim.org>
parents:
17508
diff
changeset
|
477 // Suggested by George Reilly. |
| 799 | 478 while (*p != NUL) |
| 479 hash = hash * 101 + *p++; | |
| 480 | |
| 481 return hash; | |
| 482 } |