hashmap.c

   1 /*
   2  * Generic implementation of hash-based key value mappings.
   3  */
   4 #include "git-compat-util.h"
   5 #include "hashmap.h"
   6
   7 #define FNV32_BASE ((unsigned int) 0x811c9dc5)
   8 #define FNV32_PRIME ((unsigned int) 0x01000193)
   9
  10 unsigned int strhash(const char *str)
  11 {
  12         unsigned int c, hash = FNV32_BASE;
  13         while ((c = (unsigned char) *str++))
  14                 hash = (hash * FNV32_PRIME) ^ c;
  15         return hash;
  16 }
  17
  18 unsigned int strihash(const char *str)
  19 {
  20         unsigned int c, hash = FNV32_BASE;
  21         while ((c = (unsigned char) *str++)) {
  22                 if (c >= 'a' && c <= 'z')
  23                         c -= 'a' - 'A';
  24                 hash = (hash * FNV32_PRIME) ^ c;
  25         }
  26         return hash;
  27 }
  28
  29 unsigned int memhash(const void *buf, size_t len)
  30 {
  31         unsigned int hash = FNV32_BASE;
  32         unsigned char *ucbuf = (unsigned char *) buf;
  33         while (len--) {
  34                 unsigned int c = *ucbuf++;
  35                 hash = (hash * FNV32_PRIME) ^ c;
  36         }
  37         return hash;
  38 }
  39
  40 unsigned int memihash(const void *buf, size_t len)
  41 {
  42         unsigned int hash = FNV32_BASE;
  43         unsigned char *ucbuf = (unsigned char *) buf;
  44         while (len--) {
  45                 unsigned int c = *ucbuf++;
  46                 if (c >= 'a' && c <= 'z')
  47                         c -= 'a' - 'A';
  48                 hash = (hash * FNV32_PRIME) ^ c;
  49         }
  50         return hash;
  51 }
  52
  53 /*
  54  * Incorporate another chunk of data into a memihash
  55  * computation.
  56  */
  57 unsigned int memihash_cont(unsigned int hash_seed, const void *buf, size_t len)
  58 {
  59         unsigned int hash = hash_seed;
  60         unsigned char *ucbuf = (unsigned char *) buf;
  61         while (len--) {
  62                 unsigned int c = *ucbuf++;
  63                 if (c >= 'a' && c <= 'z')
  64                         c -= 'a' - 'A';
  65                 hash = (hash * FNV32_PRIME) ^ c;
  66         }
  67         return hash;
  68 }
  69
  70 #define HASHMAP_INITIAL_SIZE 64
  71 /* grow / shrink by 2^2 */
  72 #define HASHMAP_RESIZE_BITS 2
  73 /* load factor in percent */
  74 #define HASHMAP_LOAD_FACTOR 80
  75
  76 static void alloc_table(struct hashmap *map, unsigned int size)
  77 {
  78         map->tablesize = size;
  79         CALLOC_ARRAY(map->table, size);
  80
  81         /* calculate resize thresholds for new size */
  82         map->grow_at = (unsigned int) ((uint64_t) size * HASHMAP_LOAD_FACTOR / 100);
  83         if (size <= HASHMAP_INITIAL_SIZE)
  84                 map->shrink_at = 0;
  85         else
  86                 /*
  87                  * The shrink-threshold must be slightly smaller than
  88                  * (grow-threshold / resize-factor) to prevent erratic resizing,
  89                  * thus we divide by (resize-factor + 1).
  90                  */
  91                 map->shrink_at = map->grow_at / ((1 << HASHMAP_RESIZE_BITS) + 1);
  92 }
  93
  94 static inline int entry_equals(const struct hashmap *map,
  95                                const struct hashmap_entry *e1,
  96                                const struct hashmap_entry *e2,
  97                                const void *keydata)
  98 {
  99         return (e1 == e2) ||
 100                (e1->hash == e2->hash &&
 101                 !map->cmpfn(map->cmpfn_data, e1, e2, keydata));
 102 }
 103
 104 static inline unsigned int bucket(const struct hashmap *map,
 105                                   const struct hashmap_entry *key)
 106 {
 107         return key->hash & (map->tablesize - 1);
 108 }
 109
 110 int hashmap_bucket(const struct hashmap *map, unsigned int hash)
 111 {
 112         return hash & (map->tablesize - 1);
 113 }
 114
 115 static void rehash(struct hashmap *map, unsigned int newsize)
 116 {
 117         /* map->table MUST NOT be NULL when this function is called */
 118         unsigned int i, oldsize = map->tablesize;
 119         struct hashmap_entry **oldtable = map->table;
 120
 121         alloc_table(map, newsize);
 122         for (i = 0; i < oldsize; i++) {
 123                 struct hashmap_entry *e = oldtable[i];
 124                 while (e) {
 125                         struct hashmap_entry *next = e->next;
 126                         unsigned int b = bucket(map, e);
 127                         e->next = map->table[b];
 128                         map->table[b] = e;
 129                         e = next;
 130                 }
 131         }
 132         free(oldtable);
 133 }
 134
 135 static inline struct hashmap_entry **find_entry_ptr(const struct hashmap *map,
 136                 const struct hashmap_entry *key, const void *keydata)
 137 {
 138         /* map->table MUST NOT be NULL when this function is called */
 139         struct hashmap_entry **e = &map->table[bucket(map, key)];
 140         while (*e && !entry_equals(map, *e, key, keydata))
 141                 e = &(*e)->next;
 142         return e;
 143 }
 144
 145 static int always_equal(const void *cmp_data UNUSED,
 146                         const struct hashmap_entry *entry1 UNUSED,
 147                         const struct hashmap_entry *entry2 UNUSED,
 148                         const void *keydata UNUSED)
 149 {
 150         return 0;
 151 }
 152
 153 void hashmap_init(struct hashmap *map, hashmap_cmp_fn equals_function,
 154                   const void *cmpfn_data, size_t initial_size)
 155 {
 156         unsigned int size = HASHMAP_INITIAL_SIZE;
 157
 158         memset(map, 0, sizeof(*map));
 159
 160         map->cmpfn = equals_function ? equals_function : always_equal;
 161         map->cmpfn_data = cmpfn_data;
 162
 163         /* calculate initial table size and allocate the table */
 164         initial_size = (unsigned int) ((uint64_t) initial_size * 100
 165                         / HASHMAP_LOAD_FACTOR);
 166         while (initial_size > size)
 167                 size <<= HASHMAP_RESIZE_BITS;
 168         alloc_table(map, size);
 169
 170         /*
 171          * Keep track of the number of items in the map and
 172          * allow the map to automatically grow as necessary.
 173          */
 174         map->do_count_items = 1;
 175 }
 176
 177 static void free_individual_entries(struct hashmap *map, ssize_t entry_offset)
 178 {
 179         struct hashmap_iter iter;
 180         struct hashmap_entry *e;
 181
 182         hashmap_iter_init(map, &iter);
 183         while ((e = hashmap_iter_next(&iter)))
 184                 /*
 185                  * like container_of, but using caller-calculated
 186                  * offset (caller being hashmap_clear_and_free)
 187                  */
 188                 free((char *)e - entry_offset);
 189 }
 190
 191 void hashmap_partial_clear_(struct hashmap *map, ssize_t entry_offset)
 192 {
 193         if (!map || !map->table)
 194                 return;
 195         if (entry_offset >= 0)  /* called by hashmap_clear_entries */
 196                 free_individual_entries(map, entry_offset);
 197         memset(map->table, 0, map->tablesize * sizeof(struct hashmap_entry *));
 198         map->shrink_at = 0;
 199         map->private_size = 0;
 200 }
 201
 202 void hashmap_clear_(struct hashmap *map, ssize_t entry_offset)
 203 {
 204         if (!map || !map->table)
 205                 return;
 206         if (entry_offset >= 0)  /* called by hashmap_clear_and_free */
 207                 free_individual_entries(map, entry_offset);
 208         free(map->table);
 209         memset(map, 0, sizeof(*map));
 210 }
 211
 212 struct hashmap_entry *hashmap_get(const struct hashmap *map,
 213                                 const struct hashmap_entry *key,
 214                                 const void *keydata)
 215 {
 216         if (!map->table)
 217                 return NULL;
 218         return *find_entry_ptr(map, key, keydata);
 219 }
 220
 221 struct hashmap_entry *hashmap_get_next(const struct hashmap *map,
 222                                        const struct hashmap_entry *entry)
 223 {
 224         struct hashmap_entry *e = entry->next;
 225         for (; e; e = e->next)
 226                 if (entry_equals(map, entry, e, NULL))
 227                         return e;
 228         return NULL;
 229 }
 230
 231 void hashmap_add(struct hashmap *map, struct hashmap_entry *entry)
 232 {
 233         unsigned int b;
 234
 235         if (!map->table)
 236                 alloc_table(map, HASHMAP_INITIAL_SIZE);
 237
 238         b = bucket(map, entry);
 239         /* add entry */
 240         entry->next = map->table[b];
 241         map->table[b] = entry;
 242
 243         /* fix size and rehash if appropriate */
 244         if (map->do_count_items) {
 245                 map->private_size++;
 246                 if (map->private_size > map->grow_at)
 247                         rehash(map, map->tablesize << HASHMAP_RESIZE_BITS);
 248         }
 249 }
 250
 251 struct hashmap_entry *hashmap_remove(struct hashmap *map,
 252                                      const struct hashmap_entry *key,
 253                                      const void *keydata)
 254 {
 255         struct hashmap_entry *old;
 256         struct hashmap_entry **e;
 257
 258         if (!map->table)
 259                 return NULL;
 260         e = find_entry_ptr(map, key, keydata);
 261         if (!*e)
 262                 return NULL;
 263
 264         /* remove existing entry */
 265         old = *e;
 266         *e = old->next;
 267         old->next = NULL;
 268
 269         /* fix size and rehash if appropriate */
 270         if (map->do_count_items) {
 271                 map->private_size--;
 272                 if (map->private_size < map->shrink_at)
 273                         rehash(map, map->tablesize >> HASHMAP_RESIZE_BITS);
 274         }
 275
 276         return old;
 277 }
 278
 279 struct hashmap_entry *hashmap_put(struct hashmap *map,
 280                                   struct hashmap_entry *entry)
 281 {
 282         struct hashmap_entry *old = hashmap_remove(map, entry, NULL);
 283         hashmap_add(map, entry);
 284         return old;
 285 }
 286
 287 void hashmap_iter_init(struct hashmap *map, struct hashmap_iter *iter)
 288 {
 289         iter->map = map;
 290         iter->tablepos = 0;
 291         iter->next = NULL;
 292 }
 293
 294 struct hashmap_entry *hashmap_iter_next(struct hashmap_iter *iter)
 295 {
 296         struct hashmap_entry *current = iter->next;
 297         for (;;) {
 298                 if (current) {
 299                         iter->next = current->next;
 300                         return current;
 301                 }
 302
 303                 if (iter->tablepos >= iter->map->tablesize)
 304                         return NULL;
 305
 306                 current = iter->map->table[iter->tablepos++];
 307         }
 308 }
 309
 310 struct pool_entry {
 311         struct hashmap_entry ent;
 312         size_t len;
 313         unsigned char data[FLEX_ARRAY];
 314 };
 315
 316 static int pool_entry_cmp(const void *cmp_data UNUSED,
 317                           const struct hashmap_entry *eptr,
 318                           const struct hashmap_entry *entry_or_key,
 319                           const void *keydata)
 320 {
 321         const struct pool_entry *e1, *e2;
 322
 323         e1 = container_of(eptr, const struct pool_entry, ent);
 324         e2 = container_of(entry_or_key, const struct pool_entry, ent);
 325
 326         return e1->data != keydata &&
 327                (e1->len != e2->len || memcmp(e1->data, keydata, e1->len));
 328 }
 329
 330 const void *memintern(const void *data, size_t len)
 331 {
 332         static struct hashmap map;
 333         struct pool_entry key, *e;
 334
 335         /* initialize string pool hashmap */
 336         if (!map.tablesize)
 337                 hashmap_init(&map, pool_entry_cmp, NULL, 0);
 338
 339         /* lookup interned string in pool */
 340         hashmap_entry_init(&key.ent, memhash(data, len));
 341         key.len = len;
 342         e = hashmap_get_entry(&map, &key, ent, data);
 343         if (!e) {
 344                 /* not found: create it */
 345                 FLEX_ALLOC_MEM(e, data, data, len);
 346                 hashmap_entry_init(&e->ent, key.ent.hash);
 347                 e->len = len;
 348                 hashmap_add(&map, &e->ent);
 349         }
 350         return e->data;
 351 }