| Karsten Blees | 6a364ce | 2013-11-14 19:17:54 | [diff] [blame] | 1 | #ifndef HASHMAP_H |
| 2 | #define HASHMAP_H |
| 3 | |
| Jeff King | d40abc8 | 2019-06-20 07:41:49 | [diff] [blame] | 4 | #include "hash.h" |
| 5 | |
| Karsten Blees | 6a364ce | 2013-11-14 19:17:54 | [diff] [blame] | 6 | /* |
| 7 | * Generic implementation of hash-based key-value mappings. |
| Stefan Beller | 1ecbf31 | 2017-06-30 19:14:07 | [diff] [blame] | 8 | * |
| 9 | * An example that maps long to a string: |
| 10 | * For the sake of the example this allows to lookup exact values, too |
| 11 | * (i.e. it is operated as a set, the value is part of the key) |
| 12 | * ------------------------------------- |
| 13 | * |
| 14 | * struct hashmap map; |
| 15 | * struct long2string { |
| Eric Wong | e2b5038 | 2019-10-06 23:30:43 | [diff] [blame] | 16 | * struct hashmap_entry ent; |
| Stefan Beller | 1ecbf31 | 2017-06-30 19:14:07 | [diff] [blame] | 17 | * long key; |
| 18 | * char value[FLEX_ARRAY]; // be careful with allocating on stack! |
| 19 | * }; |
| 20 | * |
| 21 | * #define COMPARE_VALUE 1 |
| 22 | * |
| Johannes Schindelin | 826c778 | 2017-11-29 23:51:41 | [diff] [blame] | 23 | * static int long2string_cmp(const void *hashmap_cmp_fn_data, |
| Eric Wong | 939af16 | 2019-10-06 23:30:37 | [diff] [blame] | 24 | * const struct hashmap_entry *eptr, |
| 25 | * const struct hashmap_entry *entry_or_key, |
| Johannes Schindelin | 826c778 | 2017-11-29 23:51:41 | [diff] [blame] | 26 | * const void *keydata) |
| Stefan Beller | 1ecbf31 | 2017-06-30 19:14:07 | [diff] [blame] | 27 | * { |
| Johannes Schindelin | 826c778 | 2017-11-29 23:51:41 | [diff] [blame] | 28 | * const char *string = keydata; |
| 29 | * unsigned flags = *(unsigned *)hashmap_cmp_fn_data; |
| Eric Wong | 939af16 | 2019-10-06 23:30:37 | [diff] [blame] | 30 | * const struct long2string *e1, *e2; |
| 31 | * |
| 32 | * e1 = container_of(eptr, const struct long2string, ent); |
| 33 | * e2 = container_of(entry_or_key, const struct long2string, ent); |
| Stefan Beller | 1ecbf31 | 2017-06-30 19:14:07 | [diff] [blame] | 34 | * |
| 35 | * if (flags & COMPARE_VALUE) |
| Johannes Schindelin | 826c778 | 2017-11-29 23:51:41 | [diff] [blame] | 36 | * return e1->key != e2->key || |
| 37 | * strcmp(e1->value, string ? string : e2->value); |
| Stefan Beller | 1ecbf31 | 2017-06-30 19:14:07 | [diff] [blame] | 38 | * else |
| Johannes Schindelin | 826c778 | 2017-11-29 23:51:41 | [diff] [blame] | 39 | * return e1->key != e2->key; |
| Stefan Beller | 1ecbf31 | 2017-06-30 19:14:07 | [diff] [blame] | 40 | * } |
| 41 | * |
| 42 | * int main(int argc, char **argv) |
| 43 | * { |
| 44 | * long key; |
| Johannes Schindelin | 826c778 | 2017-11-29 23:51:41 | [diff] [blame] | 45 | * char value[255], action[32]; |
| 46 | * unsigned flags = 0; |
| Stefan Beller | 1ecbf31 | 2017-06-30 19:14:07 | [diff] [blame] | 47 | * |
| Eric Wong | 939af16 | 2019-10-06 23:30:37 | [diff] [blame] | 48 | * hashmap_init(&map, long2string_cmp, &flags, 0); |
| Stefan Beller | 1ecbf31 | 2017-06-30 19:14:07 | [diff] [blame] | 49 | * |
| Johannes Schindelin | 826c778 | 2017-11-29 23:51:41 | [diff] [blame] | 50 | * while (scanf("%s %ld %s", action, &key, value)) { |
| Stefan Beller | 1ecbf31 | 2017-06-30 19:14:07 | [diff] [blame] | 51 | * |
| 52 | * if (!strcmp("add", action)) { |
| 53 | * struct long2string *e; |
| Johannes Schindelin | 826c778 | 2017-11-29 23:51:41 | [diff] [blame] | 54 | * FLEX_ALLOC_STR(e, value, value); |
| Eric Wong | d22245a | 2019-10-06 23:30:27 | [diff] [blame] | 55 | * hashmap_entry_init(&e->ent, memhash(&key, sizeof(long))); |
| Stefan Beller | 1ecbf31 | 2017-06-30 19:14:07 | [diff] [blame] | 56 | * e->key = key; |
| Eric Wong | b94e5c1 | 2019-10-06 23:30:29 | [diff] [blame] | 57 | * hashmap_add(&map, &e->ent); |
| Stefan Beller | 1ecbf31 | 2017-06-30 19:14:07 | [diff] [blame] | 58 | * } |
| 59 | * |
| 60 | * if (!strcmp("print_all_by_key", action)) { |
| Johannes Schindelin | 826c778 | 2017-11-29 23:51:41 | [diff] [blame] | 61 | * struct long2string k, *e; |
| Eric Wong | d22245a | 2019-10-06 23:30:27 | [diff] [blame] | 62 | * hashmap_entry_init(&k->ent, memhash(&key, sizeof(long))); |
| Stefan Beller | 1ecbf31 | 2017-06-30 19:14:07 | [diff] [blame] | 63 | * k.key = key; |
| 64 | * |
| Johannes Schindelin | 826c778 | 2017-11-29 23:51:41 | [diff] [blame] | 65 | * flags &= ~COMPARE_VALUE; |
| Eric Wong | 404ab78 | 2019-10-06 23:30:42 | [diff] [blame] | 66 | * e = hashmap_get_entry(&map, &k, ent, NULL); |
| Eric Wong | f0e63c4 | 2019-10-06 23:30:35 | [diff] [blame] | 67 | * if (e) { |
| Johannes Schindelin | 826c778 | 2017-11-29 23:51:41 | [diff] [blame] | 68 | * printf("first: %ld %s\n", e->key, e->value); |
| Eric Wong | f0e63c4 | 2019-10-06 23:30:35 | [diff] [blame] | 69 | * while ((e = hashmap_get_next_entry(&map, e, |
| 70 | * struct long2string, ent))) { |
| Johannes Schindelin | 826c778 | 2017-11-29 23:51:41 | [diff] [blame] | 71 | * printf("found more: %ld %s\n", e->key, e->value); |
| Eric Wong | 6bcbdfb | 2019-10-06 23:30:34 | [diff] [blame] | 72 | * } |
| Stefan Beller | 1ecbf31 | 2017-06-30 19:14:07 | [diff] [blame] | 73 | * } |
| 74 | * } |
| 75 | * |
| 76 | * if (!strcmp("has_exact_match", action)) { |
| Stefan Beller | 1ecbf31 | 2017-06-30 19:14:07 | [diff] [blame] | 77 | * struct long2string *e; |
| Johannes Schindelin | 826c778 | 2017-11-29 23:51:41 | [diff] [blame] | 78 | * FLEX_ALLOC_STR(e, value, value); |
| Eric Wong | d22245a | 2019-10-06 23:30:27 | [diff] [blame] | 79 | * hashmap_entry_init(&e->ent, memhash(&key, sizeof(long))); |
| Stefan Beller | 1ecbf31 | 2017-06-30 19:14:07 | [diff] [blame] | 80 | * e->key = key; |
| Stefan Beller | 1ecbf31 | 2017-06-30 19:14:07 | [diff] [blame] | 81 | * |
| Johannes Schindelin | 826c778 | 2017-11-29 23:51:41 | [diff] [blame] | 82 | * flags |= COMPARE_VALUE; |
| Eric Wong | b6c5241 | 2019-10-06 23:30:30 | [diff] [blame] | 83 | * printf("%sfound\n", |
| 84 | * hashmap_get(&map, &e->ent, NULL) ? "" : "not "); |
| Johannes Schindelin | 826c778 | 2017-11-29 23:51:41 | [diff] [blame] | 85 | * free(e); |
| Stefan Beller | 1ecbf31 | 2017-06-30 19:14:07 | [diff] [blame] | 86 | * } |
| 87 | * |
| 88 | * if (!strcmp("has_exact_match_no_heap_alloc", action)) { |
| Johannes Schindelin | 826c778 | 2017-11-29 23:51:41 | [diff] [blame] | 89 | * struct long2string k; |
| Eric Wong | d22245a | 2019-10-06 23:30:27 | [diff] [blame] | 90 | * hashmap_entry_init(&k->ent, memhash(&key, sizeof(long))); |
| Johannes Schindelin | 826c778 | 2017-11-29 23:51:41 | [diff] [blame] | 91 | * k.key = key; |
| 92 | * |
| Stefan Beller | 1ecbf31 | 2017-06-30 19:14:07 | [diff] [blame] | 93 | * flags |= COMPARE_VALUE; |
| Eric Wong | b6c5241 | 2019-10-06 23:30:30 | [diff] [blame] | 94 | * printf("%sfound\n", |
| 95 | * hashmap_get(&map, &k->ent, value) ? "" : "not "); |
| Stefan Beller | 1ecbf31 | 2017-06-30 19:14:07 | [diff] [blame] | 96 | * } |
| 97 | * |
| 98 | * if (!strcmp("end", action)) { |
| Eric Wong | c8e424c | 2019-10-06 23:30:40 | [diff] [blame] | 99 | * hashmap_free_entries(&map, struct long2string, ent); |
| Stefan Beller | 1ecbf31 | 2017-06-30 19:14:07 | [diff] [blame] | 100 | * break; |
| 101 | * } |
| 102 | * } |
| Johannes Schindelin | 826c778 | 2017-11-29 23:51:41 | [diff] [blame] | 103 | * |
| 104 | * return 0; |
| Stefan Beller | 1ecbf31 | 2017-06-30 19:14:07 | [diff] [blame] | 105 | * } |
| Karsten Blees | 6a364ce | 2013-11-14 19:17:54 | [diff] [blame] | 106 | */ |
| 107 | |
| Stefan Beller | 1ecbf31 | 2017-06-30 19:14:07 | [diff] [blame] | 108 | /* |
| 109 | * Ready-to-use hash functions for strings, using the FNV-1 algorithm (see |
| 110 | * http://www.isthe.com/chongo/tech/comp/fnv). |
| 111 | * `strhash` and `strihash` take 0-terminated strings, while `memhash` and |
| 112 | * `memihash` operate on arbitrary-length memory. |
| 113 | * `strihash` and `memihash` are case insensitive versions. |
| 114 | * `memihash_cont` is a variant of `memihash` that allows a computation to be |
| 115 | * continued with another chunk of data. |
| 116 | */ |
| Denton Liu | 5545442 | 2019-04-29 08:28:14 | [diff] [blame] | 117 | unsigned int strhash(const char *buf); |
| 118 | unsigned int strihash(const char *buf); |
| 119 | unsigned int memhash(const void *buf, size_t len); |
| 120 | unsigned int memihash(const void *buf, size_t len); |
| 121 | unsigned int memihash_cont(unsigned int hash_seed, const void *buf, size_t len); |
| Karsten Blees | 6a364ce | 2013-11-14 19:17:54 | [diff] [blame] | 122 | |
| Stefan Beller | 1ecbf31 | 2017-06-30 19:14:07 | [diff] [blame] | 123 | /* |
| 124 | * Converts a cryptographic hash (e.g. SHA-1) into an int-sized hash code |
| 125 | * for use in hash tables. Cryptographic hashes are supposed to have |
| 126 | * uniform distribution, so in contrast to `memhash()`, this just copies |
| 127 | * the first `sizeof(int)` bytes without shuffling any bits. Note that |
| 128 | * the results will be different on big-endian and little-endian |
| 129 | * platforms, so they should not be stored or transferred over the net. |
| 130 | */ |
| Jeff King | d40abc8 | 2019-06-20 07:41:49 | [diff] [blame] | 131 | static inline unsigned int oidhash(const struct object_id *oid) |
| Karsten Blees | 039dc71 | 2014-07-02 22:20:20 | [diff] [blame] | 132 | { |
| 133 | /* |
| Jeff King | d40abc8 | 2019-06-20 07:41:49 | [diff] [blame] | 134 | * Equivalent to 'return *(unsigned int *)oid->hash;', but safe on |
| Karsten Blees | 039dc71 | 2014-07-02 22:20:20 | [diff] [blame] | 135 | * platforms that don't support unaligned reads. |
| 136 | */ |
| 137 | unsigned int hash; |
| Jeff King | d40abc8 | 2019-06-20 07:41:49 | [diff] [blame] | 138 | memcpy(&hash, oid->hash, sizeof(hash)); |
| Karsten Blees | 039dc71 | 2014-07-02 22:20:20 | [diff] [blame] | 139 | return hash; |
| 140 | } |
| 141 | |
| Stefan Beller | 1ecbf31 | 2017-06-30 19:14:07 | [diff] [blame] | 142 | /* |
| 143 | * struct hashmap_entry is an opaque structure representing an entry in the |
| Eric Wong | e2b5038 | 2019-10-06 23:30:43 | [diff] [blame] | 144 | * hash table. |
| Stefan Beller | 1ecbf31 | 2017-06-30 19:14:07 | [diff] [blame] | 145 | * Ideally it should be followed by an int-sized member to prevent unused |
| 146 | * memory on 64-bit systems due to alignment. |
| 147 | */ |
| Karsten Blees | 6a364ce | 2013-11-14 19:17:54 | [diff] [blame] | 148 | struct hashmap_entry { |
| Stefan Beller | 1ecbf31 | 2017-06-30 19:14:07 | [diff] [blame] | 149 | /* |
| 150 | * next points to the next entry in case of collisions (i.e. if |
| 151 | * multiple entries map to the same bucket) |
| 152 | */ |
| Karsten Blees | 6a364ce | 2013-11-14 19:17:54 | [diff] [blame] | 153 | struct hashmap_entry *next; |
| Stefan Beller | 1ecbf31 | 2017-06-30 19:14:07 | [diff] [blame] | 154 | |
| 155 | /* entry's hash code */ |
| Karsten Blees | 6a364ce | 2013-11-14 19:17:54 | [diff] [blame] | 156 | unsigned int hash; |
| 157 | }; |
| 158 | |
| Stefan Beller | 1ecbf31 | 2017-06-30 19:14:07 | [diff] [blame] | 159 | /* |
| 160 | * User-supplied function to test two hashmap entries for equality. Shall |
| 161 | * return 0 if the entries are equal. |
| 162 | * |
| 163 | * This function is always called with non-NULL `entry` and `entry_or_key` |
| 164 | * parameters that have the same hash code. |
| 165 | * |
| 166 | * When looking up an entry, the `key` and `keydata` parameters to hashmap_get |
| 167 | * and hashmap_remove are always passed as second `entry_or_key` and third |
| 168 | * argument `keydata`, respectively. Otherwise, `keydata` is NULL. |
| 169 | * |
| 170 | * When it is too expensive to allocate a user entry (either because it is |
| 171 | * large or varialbe sized, such that it is not on the stack), then the |
| 172 | * relevant data to check for equality should be passed via `keydata`. |
| 173 | * In this case `key` can be a stripped down version of the user key data |
| 174 | * or even just a hashmap_entry having the correct hash. |
| 175 | * |
| 176 | * The `hashmap_cmp_fn_data` entry is the pointer given in the init function. |
| 177 | */ |
| Stefan Beller | 7663cdc | 2017-06-30 19:14:05 | [diff] [blame] | 178 | typedef int (*hashmap_cmp_fn)(const void *hashmap_cmp_fn_data, |
| Eric Wong | 939af16 | 2019-10-06 23:30:37 | [diff] [blame] | 179 | const struct hashmap_entry *entry, |
| 180 | const struct hashmap_entry *entry_or_key, |
| Stefan Beller | 7663cdc | 2017-06-30 19:14:05 | [diff] [blame] | 181 | const void *keydata); |
| Karsten Blees | 6a364ce | 2013-11-14 19:17:54 | [diff] [blame] | 182 | |
| Stefan Beller | 1ecbf31 | 2017-06-30 19:14:07 | [diff] [blame] | 183 | /* |
| 184 | * struct hashmap is the hash table structure. Members can be used as follows, |
| 185 | * but should not be modified directly. |
| 186 | */ |
| Karsten Blees | 6a364ce | 2013-11-14 19:17:54 | [diff] [blame] | 187 | struct hashmap { |
| 188 | struct hashmap_entry **table; |
| Stefan Beller | 1ecbf31 | 2017-06-30 19:14:07 | [diff] [blame] | 189 | |
| 190 | /* Stores the comparison function specified in `hashmap_init()`. */ |
| Karsten Blees | 6a364ce | 2013-11-14 19:17:54 | [diff] [blame] | 191 | hashmap_cmp_fn cmpfn; |
| Stefan Beller | 7663cdc | 2017-06-30 19:14:05 | [diff] [blame] | 192 | const void *cmpfn_data; |
| Karsten Blees | 6a364ce | 2013-11-14 19:17:54 | [diff] [blame] | 193 | |
| Stefan Beller | 1ecbf31 | 2017-06-30 19:14:07 | [diff] [blame] | 194 | /* total number of entries (0 means the hashmap is empty) */ |
| Jeff Hostetler | 8b604d1 | 2017-09-06 15:43:48 | [diff] [blame] | 195 | unsigned int private_size; /* use hashmap_get_size() */ |
| Stefan Beller | 1ecbf31 | 2017-06-30 19:14:07 | [diff] [blame] | 196 | |
| 197 | /* |
| 198 | * tablesize is the allocated size of the hash table. A non-0 value |
| 199 | * indicates that the hashmap is initialized. It may also be useful |
| 200 | * for statistical purposes (i.e. `size / tablesize` is the current |
| 201 | * load factor). |
| 202 | */ |
| 203 | unsigned int tablesize; |
| 204 | |
| 205 | unsigned int grow_at; |
| 206 | unsigned int shrink_at; |
| 207 | |
| Jeff Hostetler | 8b604d1 | 2017-09-06 15:43:48 | [diff] [blame] | 208 | unsigned int do_count_items : 1; |
| Karsten Blees | 6a364ce | 2013-11-14 19:17:54 | [diff] [blame] | 209 | }; |
| 210 | |
| 211 | /* hashmap functions */ |
| 212 | |
| Stefan Beller | 1ecbf31 | 2017-06-30 19:14:07 | [diff] [blame] | 213 | /* |
| 214 | * Initializes a hashmap structure. |
| 215 | * |
| 216 | * `map` is the hashmap to initialize. |
| 217 | * |
| 218 | * The `equals_function` can be specified to compare two entries for equality. |
| 219 | * If NULL, entries are considered equal if their hash codes are equal. |
| 220 | * |
| 221 | * The `equals_function_data` parameter can be used to provide additional data |
| 222 | * (a callback cookie) that will be passed to `equals_function` each time it |
| 223 | * is called. This allows a single `equals_function` to implement multiple |
| 224 | * comparison functions. |
| 225 | * |
| 226 | * If the total number of entries is known in advance, the `initial_size` |
| 227 | * parameter may be used to preallocate a sufficiently large table and thus |
| 228 | * prevent expensive resizing. If 0, the table is dynamically resized. |
| 229 | */ |
| Denton Liu | 5545442 | 2019-04-29 08:28:14 | [diff] [blame] | 230 | void hashmap_init(struct hashmap *map, |
| Stefan Beller | 7663cdc | 2017-06-30 19:14:05 | [diff] [blame] | 231 | hashmap_cmp_fn equals_function, |
| 232 | const void *equals_function_data, |
| 233 | size_t initial_size); |
| Stefan Beller | 1ecbf31 | 2017-06-30 19:14:07 | [diff] [blame] | 234 | |
| Eric Wong | c8e424c | 2019-10-06 23:30:40 | [diff] [blame] | 235 | /* internal function for freeing hashmap */ |
| 236 | void hashmap_free_(struct hashmap *map, ssize_t offset); |
| 237 | |
| Stefan Beller | 1ecbf31 | 2017-06-30 19:14:07 | [diff] [blame] | 238 | /* |
| Eric Wong | c8e424c | 2019-10-06 23:30:40 | [diff] [blame] | 239 | * Frees a hashmap structure and allocated memory, leaves entries undisturbed |
| Stefan Beller | 1ecbf31 | 2017-06-30 19:14:07 | [diff] [blame] | 240 | */ |
| Eric Wong | c8e424c | 2019-10-06 23:30:40 | [diff] [blame] | 241 | #define hashmap_free(map) hashmap_free_(map, -1) |
| 242 | |
| 243 | /* |
| 244 | * Frees @map and all entries. @type is the struct type of the entry |
| 245 | * where @member is the hashmap_entry struct used to associate with @map |
| 246 | */ |
| 247 | #define hashmap_free_entries(map, type, member) \ |
| 248 | hashmap_free_(map, offsetof(type, member)); |
| Karsten Blees | 6a364ce | 2013-11-14 19:17:54 | [diff] [blame] | 249 | |
| 250 | /* hashmap_entry functions */ |
| 251 | |
| Stefan Beller | 1ecbf31 | 2017-06-30 19:14:07 | [diff] [blame] | 252 | /* |
| 253 | * Initializes a hashmap_entry structure. |
| 254 | * |
| 255 | * `entry` points to the entry to initialize. |
| 256 | * `hash` is the hash code of the entry. |
| 257 | * |
| 258 | * The hashmap_entry structure does not hold references to external resources, |
| 259 | * and it is safe to just discard it once you are done with it (i.e. if |
| 260 | * your structure was allocated with xmalloc(), you can just free(3) it, |
| 261 | * and if it is on stack, you can just let it go out of scope). |
| 262 | */ |
| Eric Wong | d22245a | 2019-10-06 23:30:27 | [diff] [blame] | 263 | static inline void hashmap_entry_init(struct hashmap_entry *e, |
| 264 | unsigned int hash) |
| Karsten Blees | 6a364ce | 2013-11-14 19:17:54 | [diff] [blame] | 265 | { |
| Karsten Blees | 6a364ce | 2013-11-14 19:17:54 | [diff] [blame] | 266 | e->hash = hash; |
| 267 | e->next = NULL; |
| 268 | } |
| Karsten Blees | 6a364ce | 2013-11-14 19:17:54 | [diff] [blame] | 269 | |
| Stefan Beller | 1ecbf31 | 2017-06-30 19:14:07 | [diff] [blame] | 270 | /* |
| Jeff Hostetler | 8b604d1 | 2017-09-06 15:43:48 | [diff] [blame] | 271 | * Return the number of items in the map. |
| 272 | */ |
| 273 | static inline unsigned int hashmap_get_size(struct hashmap *map) |
| 274 | { |
| 275 | if (map->do_count_items) |
| 276 | return map->private_size; |
| 277 | |
| 278 | BUG("hashmap_get_size: size not set"); |
| 279 | return 0; |
| 280 | } |
| 281 | |
| 282 | /* |
| Stefan Beller | 1ecbf31 | 2017-06-30 19:14:07 | [diff] [blame] | 283 | * Returns the hashmap entry for the specified key, or NULL if not found. |
| 284 | * |
| 285 | * `map` is the hashmap structure. |
| 286 | * |
| 287 | * `key` is a user data structure that starts with hashmap_entry that has at |
| 288 | * least been initialized with the proper hash code (via `hashmap_entry_init`). |
| 289 | * |
| 290 | * `keydata` is a data structure that holds just enough information to check |
| 291 | * for equality to a given entry. |
| 292 | * |
| 293 | * If the key data is variable-sized (e.g. a FLEX_ARRAY string) or quite large, |
| 294 | * it is undesirable to create a full-fledged entry structure on the heap and |
| 295 | * copy all the key data into the structure. |
| 296 | * |
| 297 | * In this case, the `keydata` parameter can be used to pass |
| 298 | * variable-sized key data directly to the comparison function, and the `key` |
| 299 | * parameter can be a stripped-down, fixed size entry structure allocated on the |
| 300 | * stack. |
| 301 | * |
| 302 | * If an entry with matching hash code is found, `key` and `keydata` are passed |
| 303 | * to `hashmap_cmp_fn` to decide whether the entry matches the key. |
| 304 | */ |
| Eric Wong | f23a465 | 2019-10-06 23:30:36 | [diff] [blame] | 305 | struct hashmap_entry *hashmap_get(const struct hashmap *map, |
| 306 | const struct hashmap_entry *key, |
| 307 | const void *keydata); |
| Stefan Beller | 1ecbf31 | 2017-06-30 19:14:07 | [diff] [blame] | 308 | |
| 309 | /* |
| 310 | * Returns the hashmap entry for the specified hash code and key data, |
| 311 | * or NULL if not found. |
| 312 | * |
| 313 | * `map` is the hashmap structure. |
| 314 | * `hash` is the hash code of the entry to look up. |
| 315 | * |
| 316 | * If an entry with matching hash code is found, `keydata` is passed to |
| 317 | * `hashmap_cmp_fn` to decide whether the entry matches the key. The |
| 318 | * `entry_or_key` parameter of `hashmap_cmp_fn` points to a hashmap_entry |
| 319 | * structure that should not be used in the comparison. |
| 320 | */ |
| Eric Wong | f23a465 | 2019-10-06 23:30:36 | [diff] [blame] | 321 | static inline struct hashmap_entry *hashmap_get_from_hash( |
| 322 | const struct hashmap *map, |
| 323 | unsigned int hash, |
| 324 | const void *keydata) |
| Karsten Blees | ab73a9d | 2014-07-02 22:22:11 | [diff] [blame] | 325 | { |
| 326 | struct hashmap_entry key; |
| 327 | hashmap_entry_init(&key, hash); |
| 328 | return hashmap_get(map, &key, keydata); |
| 329 | } |
| 330 | |
| Stefan Beller | 1ecbf31 | 2017-06-30 19:14:07 | [diff] [blame] | 331 | /* |
| 332 | * Returns the next equal hashmap entry, or NULL if not found. This can be |
| 333 | * used to iterate over duplicate entries (see `hashmap_add`). |
| 334 | * |
| 335 | * `map` is the hashmap structure. |
| 336 | * `entry` is the hashmap_entry to start the search from, obtained via a previous |
| 337 | * call to `hashmap_get` or `hashmap_get_next`. |
| 338 | */ |
| Eric Wong | 6bcbdfb | 2019-10-06 23:30:34 | [diff] [blame] | 339 | struct hashmap_entry *hashmap_get_next(const struct hashmap *map, |
| Eric Wong | f6eb6bd | 2019-10-06 23:30:28 | [diff] [blame] | 340 | const struct hashmap_entry *entry); |
| Stefan Beller | 1ecbf31 | 2017-06-30 19:14:07 | [diff] [blame] | 341 | |
| 342 | /* |
| 343 | * Adds a hashmap entry. This allows to add duplicate entries (i.e. |
| 344 | * separate values with the same key according to hashmap_cmp_fn). |
| 345 | * |
| 346 | * `map` is the hashmap structure. |
| 347 | * `entry` is the entry to add. |
| 348 | */ |
| Eric Wong | b94e5c1 | 2019-10-06 23:30:29 | [diff] [blame] | 349 | void hashmap_add(struct hashmap *map, struct hashmap_entry *entry); |
| Stefan Beller | 1ecbf31 | 2017-06-30 19:14:07 | [diff] [blame] | 350 | |
| 351 | /* |
| 352 | * Adds or replaces a hashmap entry. If the hashmap contains duplicate |
| 353 | * entries equal to the specified entry, only one of them will be replaced. |
| 354 | * |
| 355 | * `map` is the hashmap structure. |
| 356 | * `entry` is the entry to add or replace. |
| 357 | * Returns the replaced entry, or NULL if not found (i.e. the entry was added). |
| 358 | */ |
| Eric Wong | 8a973d0 | 2019-10-06 23:30:39 | [diff] [blame] | 359 | struct hashmap_entry *hashmap_put(struct hashmap *map, |
| 360 | struct hashmap_entry *entry); |
| 361 | |
| Eric Wong | 404ab78 | 2019-10-06 23:30:42 | [diff] [blame] | 362 | /* |
| 363 | * Adds or replaces a hashmap entry contained within @keyvar, |
| 364 | * where @keyvar is a pointer to a struct containing a |
| 365 | * "struct hashmap_entry" @member. |
| 366 | * |
| 367 | * Returns the replaced pointer which is of the same type as @keyvar, |
| 368 | * or NULL if not found. |
| 369 | */ |
| 370 | #define hashmap_put_entry(map, keyvar, member) \ |
| 371 | container_of_or_null_offset(hashmap_put(map, &(keyvar)->member), \ |
| 372 | OFFSETOF_VAR(keyvar, member)) |
| Stefan Beller | 1ecbf31 | 2017-06-30 19:14:07 | [diff] [blame] | 373 | |
| 374 | /* |
| 375 | * Removes a hashmap entry matching the specified key. If the hashmap contains |
| 376 | * duplicate entries equal to the specified key, only one of them will be |
| 377 | * removed. Returns the removed entry, or NULL if not found. |
| 378 | * |
| 379 | * Argument explanation is the same as in `hashmap_get`. |
| 380 | */ |
| Eric Wong | 8a973d0 | 2019-10-06 23:30:39 | [diff] [blame] | 381 | struct hashmap_entry *hashmap_remove(struct hashmap *map, |
| 382 | const struct hashmap_entry *key, |
| 383 | const void *keydata); |
| 384 | |
| Eric Wong | 404ab78 | 2019-10-06 23:30:42 | [diff] [blame] | 385 | /* |
| 386 | * Removes a hashmap entry contained within @keyvar, |
| 387 | * where @keyvar is a pointer to a struct containing a |
| 388 | * "struct hashmap_entry" @member. |
| 389 | * |
| 390 | * See `hashmap_get` for an explanation of @keydata |
| 391 | * |
| 392 | * Returns the replaced pointer which is of the same type as @keyvar, |
| 393 | * or NULL if not found. |
| 394 | */ |
| 395 | #define hashmap_remove_entry(map, keyvar, member, keydata) \ |
| 396 | container_of_or_null_offset( \ |
| 397 | hashmap_remove(map, &(keyvar)->member, keydata), \ |
| 398 | OFFSETOF_VAR(keyvar, member)) |
| Stefan Beller | 1ecbf31 | 2017-06-30 19:14:07 | [diff] [blame] | 399 | |
| 400 | /* |
| 401 | * Returns the `bucket` an entry is stored in. |
| 402 | * Useful for multithreaded read access. |
| 403 | */ |
| Jeff Hostetler | 0607e10 | 2017-03-22 17:14:22 | [diff] [blame] | 404 | int hashmap_bucket(const struct hashmap *map, unsigned int hash); |
| 405 | |
| 406 | /* |
| Stefan Beller | 1ecbf31 | 2017-06-30 19:14:07 | [diff] [blame] | 407 | * Used to iterate over all entries of a hashmap. Note that it is |
| 408 | * not safe to add or remove entries to the hashmap while |
| 409 | * iterating. |
| 410 | */ |
| 411 | struct hashmap_iter { |
| 412 | struct hashmap *map; |
| 413 | struct hashmap_entry *next; |
| 414 | unsigned int tablepos; |
| 415 | }; |
| Karsten Blees | 6a364ce | 2013-11-14 19:17:54 | [diff] [blame] | 416 | |
| Stefan Beller | 1ecbf31 | 2017-06-30 19:14:07 | [diff] [blame] | 417 | /* Initializes a `hashmap_iter` structure. */ |
| Denton Liu | 5545442 | 2019-04-29 08:28:14 | [diff] [blame] | 418 | void hashmap_iter_init(struct hashmap *map, struct hashmap_iter *iter); |
| Stefan Beller | 1ecbf31 | 2017-06-30 19:14:07 | [diff] [blame] | 419 | |
| 420 | /* Returns the next hashmap_entry, or NULL if there are no more entries. */ |
| Eric Wong | 87571c3 | 2019-10-06 23:30:38 | [diff] [blame] | 421 | struct hashmap_entry *hashmap_iter_next(struct hashmap_iter *iter); |
| Stefan Beller | 1ecbf31 | 2017-06-30 19:14:07 | [diff] [blame] | 422 | |
| 423 | /* Initializes the iterator and returns the first entry, if any. */ |
| Eric Wong | 87571c3 | 2019-10-06 23:30:38 | [diff] [blame] | 424 | static inline struct hashmap_entry *hashmap_iter_first(struct hashmap *map, |
| Karsten Blees | 6a364ce | 2013-11-14 19:17:54 | [diff] [blame] | 425 | struct hashmap_iter *iter) |
| 426 | { |
| 427 | hashmap_iter_init(map, iter); |
| 428 | return hashmap_iter_next(iter); |
| 429 | } |
| 430 | |
| Eric Wong | 23dee69 | 2019-10-06 23:30:41 | [diff] [blame] | 431 | /* |
| 432 | * returns the first entry in @map using @iter, where the entry is of |
| 433 | * @type (e.g. "struct foo") and @member is the name of the |
| 434 | * "struct hashmap_entry" in @type |
| 435 | */ |
| Eric Wong | 87571c3 | 2019-10-06 23:30:38 | [diff] [blame] | 436 | #define hashmap_iter_first_entry(map, iter, type, member) \ |
| 437 | container_of_or_null(hashmap_iter_first(map, iter), type, member) |
| 438 | |
| Eric Wong | 23dee69 | 2019-10-06 23:30:41 | [diff] [blame] | 439 | /* internal macro for hashmap_for_each_entry */ |
| 440 | #define hashmap_iter_next_entry_offset(iter, offset) \ |
| 441 | container_of_or_null_offset(hashmap_iter_next(iter), offset) |
| 442 | |
| 443 | /* internal macro for hashmap_for_each_entry */ |
| 444 | #define hashmap_iter_first_entry_offset(map, iter, offset) \ |
| 445 | container_of_or_null_offset(hashmap_iter_first(map, iter), offset) |
| 446 | |
| 447 | /* |
| 448 | * iterate through @map using @iter, @var is a pointer to a type |
| 449 | * containing a @member which is a "struct hashmap_entry" |
| 450 | */ |
| 451 | #define hashmap_for_each_entry(map, iter, var, member) \ |
| 452 | for (var = hashmap_iter_first_entry_offset(map, iter, \ |
| 453 | OFFSETOF_VAR(var, member)); \ |
| Eric Wong | 87571c3 | 2019-10-06 23:30:38 | [diff] [blame] | 454 | var; \ |
| Eric Wong | 23dee69 | 2019-10-06 23:30:41 | [diff] [blame] | 455 | var = hashmap_iter_next_entry_offset(iter, \ |
| 456 | OFFSETOF_VAR(var, member))) |
| Eric Wong | 87571c3 | 2019-10-06 23:30:38 | [diff] [blame] | 457 | |
| Jeff Hostetler | 8b604d1 | 2017-09-06 15:43:48 | [diff] [blame] | 458 | /* |
| Eric Wong | 404ab78 | 2019-10-06 23:30:42 | [diff] [blame] | 459 | * returns a pointer of type matching @keyvar, or NULL if nothing found. |
| 460 | * @keyvar is a pointer to a struct containing a |
| 461 | * "struct hashmap_entry" @member. |
| Eric Wong | f0e63c4 | 2019-10-06 23:30:35 | [diff] [blame] | 462 | */ |
| Eric Wong | 404ab78 | 2019-10-06 23:30:42 | [diff] [blame] | 463 | #define hashmap_get_entry(map, keyvar, member, keydata) \ |
| 464 | container_of_or_null_offset( \ |
| 465 | hashmap_get(map, &(keyvar)->member, keydata), \ |
| 466 | OFFSETOF_VAR(keyvar, member)) |
| Eric Wong | f0e63c4 | 2019-10-06 23:30:35 | [diff] [blame] | 467 | |
| 468 | #define hashmap_get_entry_from_hash(map, hash, keydata, type, member) \ |
| 469 | container_of_or_null(hashmap_get_from_hash(map, hash, keydata), \ |
| 470 | type, member) |
| 471 | /* |
| Eric Wong | 23dee69 | 2019-10-06 23:30:41 | [diff] [blame] | 472 | * returns the next equal pointer to @var, or NULL if not found. |
| 473 | * @var is a pointer of any type containing "struct hashmap_entry" |
| 474 | * @member is the name of the "struct hashmap_entry" field |
| Eric Wong | f0e63c4 | 2019-10-06 23:30:35 | [diff] [blame] | 475 | */ |
| Eric Wong | 23dee69 | 2019-10-06 23:30:41 | [diff] [blame] | 476 | #define hashmap_get_next_entry(map, var, member) \ |
| 477 | container_of_or_null_offset(hashmap_get_next(map, &(var)->member), \ |
| 478 | OFFSETOF_VAR(var, member)) |
| Eric Wong | f0e63c4 | 2019-10-06 23:30:35 | [diff] [blame] | 479 | |
| 480 | /* |
| 481 | * iterate @map starting from @var, where @var is a pointer of @type |
| 482 | * and @member is the name of the "struct hashmap_entry" field in @type |
| 483 | */ |
| Eric Wong | 23dee69 | 2019-10-06 23:30:41 | [diff] [blame] | 484 | #define hashmap_for_each_entry_from(map, var, member) \ |
| Eric Wong | f0e63c4 | 2019-10-06 23:30:35 | [diff] [blame] | 485 | for (; \ |
| 486 | var; \ |
| Eric Wong | 23dee69 | 2019-10-06 23:30:41 | [diff] [blame] | 487 | var = hashmap_get_next_entry(map, var, member)) |
| Eric Wong | f0e63c4 | 2019-10-06 23:30:35 | [diff] [blame] | 488 | |
| 489 | /* |
| Jeff Hostetler | 8b604d1 | 2017-09-06 15:43:48 | [diff] [blame] | 490 | * Disable item counting and automatic rehashing when adding/removing items. |
| 491 | * |
| 492 | * Normally, the hashmap keeps track of the number of items in the map |
| 493 | * and uses it to dynamically resize it. This (both the counting and |
| 494 | * the resizing) can cause problems when the map is being used by |
| 495 | * threaded callers (because the hashmap code does not know about the |
| 496 | * locking strategy used by the threaded callers and therefore, does |
| 497 | * not know how to protect the "private_size" counter). |
| 498 | */ |
| 499 | static inline void hashmap_disable_item_counting(struct hashmap *map) |
| 500 | { |
| 501 | map->do_count_items = 0; |
| 502 | } |
| 503 | |
| 504 | /* |
| 505 | * Re-enable item couting when adding/removing items. |
| 506 | * If counting is currently disabled, it will force count them. |
| 507 | * It WILL NOT automatically rehash them. |
| 508 | */ |
| 509 | static inline void hashmap_enable_item_counting(struct hashmap *map) |
| 510 | { |
| Jeff Hostetler | 8b604d1 | 2017-09-06 15:43:48 | [diff] [blame] | 511 | unsigned int n = 0; |
| 512 | struct hashmap_iter iter; |
| 513 | |
| 514 | if (map->do_count_items) |
| 515 | return; |
| 516 | |
| 517 | hashmap_iter_init(map, &iter); |
| Randall S. Becker | 7d68bb0 | 2018-01-14 18:07:48 | [diff] [blame] | 518 | while (hashmap_iter_next(&iter)) |
| Jeff Hostetler | 8b604d1 | 2017-09-06 15:43:48 | [diff] [blame] | 519 | n++; |
| 520 | |
| 521 | map->do_count_items = 1; |
| 522 | map->private_size = n; |
| 523 | } |
| 524 | |
| Stefan Beller | 1ecbf31 | 2017-06-30 19:14:07 | [diff] [blame] | 525 | /* String interning */ |
| Karsten Blees | 7b64d42 | 2014-07-02 22:22:54 | [diff] [blame] | 526 | |
| Stefan Beller | 1ecbf31 | 2017-06-30 19:14:07 | [diff] [blame] | 527 | /* |
| 528 | * Returns the unique, interned version of the specified string or data, |
| 529 | * similar to the `String.intern` API in Java and .NET, respectively. |
| 530 | * Interned strings remain valid for the entire lifetime of the process. |
| 531 | * |
| 532 | * Can be used as `[x]strdup()` or `xmemdupz` replacement, except that interned |
| 533 | * strings / data must not be modified or freed. |
| 534 | * |
| 535 | * Interned strings are best used for short strings with high probability of |
| 536 | * duplicates. |
| 537 | * |
| 538 | * Uses a hashmap to store the pool of interned strings. |
| 539 | */ |
| Denton Liu | 5545442 | 2019-04-29 08:28:14 | [diff] [blame] | 540 | const void *memintern(const void *data, size_t len); |
| Karsten Blees | 7b64d42 | 2014-07-02 22:22:54 | [diff] [blame] | 541 | static inline const char *strintern(const char *string) |
| 542 | { |
| 543 | return memintern(string, strlen(string)); |
| 544 | } |
| 545 | |
| Karsten Blees | 6a364ce | 2013-11-14 19:17:54 | [diff] [blame] | 546 | #endif |