| 1 |
sysadm |
1.13 |
/* SPDX-License-Identifier: GPL-3.0-or-later */
|
| 2 |
|
|
/*
|
| 3 |
|
|
* trie_dict
|
| 4 |
|
|
* - trie-tree based dict feature
|
| 5 |
|
|
*
|
| 6 |
sysadm |
1.27 |
* Copyright (C) 2004-2026 Leaflet <leaflet@leafok.com>
|
| 7 |
sysadm |
1.13 |
*/
|
| 8 |
sysadm |
1.1 |
|
| 9 |
sysadm |
1.16 |
#ifdef HAVE_CONFIG_H
|
| 10 |
|
|
#include "config.h"
|
| 11 |
|
|
#endif
|
| 12 |
|
|
|
| 13 |
sysadm |
1.20 |
#include "common.h"
|
| 14 |
sysadm |
1.11 |
#include "log.h"
|
| 15 |
sysadm |
1.1 |
#include "trie_dict.h"
|
| 16 |
sysadm |
1.7 |
#include <errno.h>
|
| 17 |
sysadm |
1.20 |
#include <fcntl.h>
|
| 18 |
sysadm |
1.10 |
#include <stdio.h>
|
| 19 |
sysadm |
1.1 |
#include <stdlib.h>
|
| 20 |
sysadm |
1.10 |
#include <string.h>
|
| 21 |
sysadm |
1.7 |
#include <time.h>
|
| 22 |
|
|
#include <unistd.h>
|
| 23 |
sysadm |
1.20 |
#include <sys/mman.h>
|
| 24 |
|
|
#include <sys/stat.h>
|
| 25 |
sysadm |
1.7 |
|
| 26 |
|
|
struct trie_node_pool_t
|
| 27 |
|
|
{
|
| 28 |
sysadm |
1.20 |
size_t shm_size;
|
| 29 |
sysadm |
1.8 |
int node_count_limit;
|
| 30 |
|
|
int node_count;
|
| 31 |
sysadm |
1.7 |
TRIE_NODE *p_node_free_list;
|
| 32 |
|
|
};
|
| 33 |
|
|
typedef struct trie_node_pool_t TRIE_NODE_POOL;
|
| 34 |
|
|
|
| 35 |
sysadm |
1.23 |
static char trie_node_shm_name[FILE_NAME_LEN];
|
| 36 |
sysadm |
1.7 |
static TRIE_NODE_POOL *p_trie_node_pool;
|
| 37 |
|
|
|
| 38 |
sysadm |
1.8 |
int trie_dict_init(const char *filename, int node_count_limit)
|
| 39 |
sysadm |
1.7 |
{
|
| 40 |
sysadm |
1.20 |
char filepath[FILE_PATH_LEN];
|
| 41 |
|
|
int fd;
|
| 42 |
sysadm |
1.7 |
size_t size;
|
| 43 |
|
|
void *p_shm;
|
| 44 |
sysadm |
1.8 |
TRIE_NODE *p_trie_nodes;
|
| 45 |
sysadm |
1.7 |
int i;
|
| 46 |
|
|
|
| 47 |
sysadm |
1.20 |
if (filename == NULL)
|
| 48 |
|
|
{
|
| 49 |
sysadm |
1.25 |
log_error("NULL pointer error");
|
| 50 |
sysadm |
1.20 |
return -1;
|
| 51 |
|
|
}
|
| 52 |
|
|
|
| 53 |
sysadm |
1.7 |
if (p_trie_node_pool != NULL)
|
| 54 |
|
|
{
|
| 55 |
sysadm |
1.25 |
log_error("trie_dict_pool already initialized");
|
| 56 |
sysadm |
1.7 |
return -1;
|
| 57 |
|
|
}
|
| 58 |
|
|
|
| 59 |
sysadm |
1.8 |
if (node_count_limit <= 0 || node_count_limit > TRIE_NODE_PER_POOL)
|
| 60 |
|
|
{
|
| 61 |
sysadm |
1.25 |
log_error("trie_dict_init(%d) error: invalid node_count_limit", node_count_limit);
|
| 62 |
sysadm |
1.8 |
return -1;
|
| 63 |
|
|
}
|
| 64 |
|
|
|
| 65 |
sysadm |
1.7 |
// Allocate shared memory
|
| 66 |
sysadm |
1.20 |
size = sizeof(TRIE_NODE_POOL) + sizeof(TRIE_NODE) * (size_t)node_count_limit;
|
| 67 |
|
|
|
| 68 |
|
|
strncpy(filepath, filename, sizeof(filepath) - 1);
|
| 69 |
|
|
filepath[sizeof(filepath) - 1] = '\0';
|
| 70 |
|
|
snprintf(trie_node_shm_name, sizeof(trie_node_shm_name), "/TRIE_DICT_SHM_%s", basename(filepath));
|
| 71 |
|
|
|
| 72 |
|
|
if (shm_unlink(trie_node_shm_name) == -1 && errno != ENOENT)
|
| 73 |
sysadm |
1.7 |
{
|
| 74 |
sysadm |
1.25 |
log_error("shm_unlink(%s) error (%d)", trie_node_shm_name, errno);
|
| 75 |
sysadm |
1.7 |
return -2;
|
| 76 |
|
|
}
|
| 77 |
|
|
|
| 78 |
sysadm |
1.20 |
if ((fd = shm_open(trie_node_shm_name, O_CREAT | O_EXCL | O_RDWR, 0600)) == -1)
|
| 79 |
|
|
{
|
| 80 |
sysadm |
1.25 |
log_error("shm_open(%s) error (%d)", trie_node_shm_name, errno);
|
| 81 |
sysadm |
1.20 |
return -2;
|
| 82 |
|
|
}
|
| 83 |
|
|
if (ftruncate(fd, (off_t)size) == -1)
|
| 84 |
|
|
{
|
| 85 |
sysadm |
1.25 |
log_error("ftruncate(size=%d) error (%d)", size, errno);
|
| 86 |
sysadm |
1.20 |
close(fd);
|
| 87 |
|
|
return -2;
|
| 88 |
|
|
}
|
| 89 |
|
|
|
| 90 |
|
|
p_shm = mmap(NULL, size, PROT_READ | PROT_WRITE, MAP_SHARED, fd, 0L);
|
| 91 |
|
|
if (p_shm == MAP_FAILED)
|
| 92 |
sysadm |
1.7 |
{
|
| 93 |
sysadm |
1.25 |
log_error("mmap() error (%d)", errno);
|
| 94 |
sysadm |
1.20 |
close(fd);
|
| 95 |
|
|
return -2;
|
| 96 |
sysadm |
1.7 |
}
|
| 97 |
sysadm |
1.20 |
|
| 98 |
|
|
if (close(fd) < 0)
|
| 99 |
sysadm |
1.7 |
{
|
| 100 |
sysadm |
1.25 |
log_error("close(fd) error (%d)", errno);
|
| 101 |
sysadm |
1.20 |
return -1;
|
| 102 |
sysadm |
1.7 |
}
|
| 103 |
|
|
|
| 104 |
|
|
p_trie_node_pool = p_shm;
|
| 105 |
sysadm |
1.20 |
p_trie_node_pool->shm_size = size;
|
| 106 |
sysadm |
1.8 |
p_trie_node_pool->node_count_limit = node_count_limit;
|
| 107 |
sysadm |
1.7 |
p_trie_node_pool->node_count = 0;
|
| 108 |
|
|
|
| 109 |
sysadm |
1.12 |
p_trie_nodes = (TRIE_NODE *)((char *)p_shm + sizeof(TRIE_NODE_POOL));
|
| 110 |
sysadm |
1.8 |
p_trie_node_pool->p_node_free_list = &(p_trie_nodes[0]);
|
| 111 |
|
|
for (i = 0; i < node_count_limit - 1; i++)
|
| 112 |
sysadm |
1.7 |
{
|
| 113 |
sysadm |
1.8 |
p_trie_nodes[i].p_nodes[0] = &(p_trie_nodes[i + 1]);
|
| 114 |
sysadm |
1.7 |
}
|
| 115 |
sysadm |
1.8 |
p_trie_nodes[node_count_limit - 1].p_nodes[0] = NULL;
|
| 116 |
sysadm |
1.7 |
|
| 117 |
|
|
return 0;
|
| 118 |
|
|
}
|
| 119 |
|
|
|
| 120 |
sysadm |
1.23 |
void trie_dict_cleanup(void)
|
| 121 |
sysadm |
1.7 |
{
|
| 122 |
|
|
if (p_trie_node_pool == NULL)
|
| 123 |
|
|
{
|
| 124 |
sysadm |
1.23 |
return;
|
| 125 |
sysadm |
1.7 |
}
|
| 126 |
|
|
|
| 127 |
|
|
detach_trie_dict_shm();
|
| 128 |
|
|
|
| 129 |
sysadm |
1.20 |
if (shm_unlink(trie_node_shm_name) == -1 && errno != ENOENT)
|
| 130 |
sysadm |
1.7 |
{
|
| 131 |
sysadm |
1.25 |
log_error("shm_unlink(%s) error (%d)", trie_node_shm_name, errno);
|
| 132 |
sysadm |
1.7 |
}
|
| 133 |
sysadm |
1.20 |
|
| 134 |
|
|
trie_node_shm_name[0] = '\0';
|
| 135 |
sysadm |
1.7 |
}
|
| 136 |
|
|
|
| 137 |
sysadm |
1.20 |
int set_trie_dict_shm_readonly(void)
|
| 138 |
|
|
{
|
| 139 |
sysadm |
1.21 |
if (p_trie_node_pool != NULL && mprotect(p_trie_node_pool, p_trie_node_pool->shm_size, PROT_READ) < 0)
|
| 140 |
sysadm |
1.20 |
{
|
| 141 |
sysadm |
1.25 |
log_error("mprotect() error (%d)", errno);
|
| 142 |
sysadm |
1.21 |
return -1;
|
| 143 |
sysadm |
1.20 |
}
|
| 144 |
sysadm |
1.7 |
|
| 145 |
|
|
return 0;
|
| 146 |
|
|
}
|
| 147 |
|
|
|
| 148 |
|
|
int detach_trie_dict_shm(void)
|
| 149 |
|
|
{
|
| 150 |
sysadm |
1.20 |
if (p_trie_node_pool != NULL && munmap(p_trie_node_pool, p_trie_node_pool->shm_size) < 0)
|
| 151 |
sysadm |
1.7 |
{
|
| 152 |
sysadm |
1.25 |
log_error("munmap() error (%d)", errno);
|
| 153 |
sysadm |
1.7 |
return -1;
|
| 154 |
|
|
}
|
| 155 |
sysadm |
1.8 |
|
| 156 |
sysadm |
1.7 |
p_trie_node_pool = NULL;
|
| 157 |
|
|
|
| 158 |
|
|
return 0;
|
| 159 |
|
|
}
|
| 160 |
sysadm |
1.1 |
|
| 161 |
|
|
TRIE_NODE *trie_dict_create(void)
|
| 162 |
|
|
{
|
| 163 |
sysadm |
1.7 |
TRIE_NODE *p_dict = NULL;
|
| 164 |
|
|
|
| 165 |
|
|
if (p_trie_node_pool != NULL && p_trie_node_pool->p_node_free_list != NULL)
|
| 166 |
|
|
{
|
| 167 |
|
|
p_dict = p_trie_node_pool->p_node_free_list;
|
| 168 |
|
|
p_trie_node_pool->p_node_free_list = p_dict->p_nodes[0];
|
| 169 |
sysadm |
1.1 |
|
| 170 |
sysadm |
1.10 |
memset(p_dict, 0, sizeof(*p_dict));
|
| 171 |
sysadm |
1.8 |
|
| 172 |
|
|
p_trie_node_pool->node_count++;
|
| 173 |
|
|
}
|
| 174 |
|
|
else if (p_trie_node_pool != NULL)
|
| 175 |
|
|
{
|
| 176 |
sysadm |
1.25 |
log_error("trie_dict_create() error: node depleted %d >= %d", p_trie_node_pool->node_count, p_trie_node_pool->node_count_limit);
|
| 177 |
sysadm |
1.7 |
}
|
| 178 |
sysadm |
1.1 |
|
| 179 |
|
|
return p_dict;
|
| 180 |
|
|
}
|
| 181 |
|
|
|
| 182 |
|
|
void trie_dict_destroy(TRIE_NODE *p_dict)
|
| 183 |
|
|
{
|
| 184 |
sysadm |
1.7 |
if (p_trie_node_pool == NULL || p_dict == NULL)
|
| 185 |
sysadm |
1.1 |
{
|
| 186 |
|
|
return;
|
| 187 |
|
|
}
|
| 188 |
|
|
|
| 189 |
|
|
for (int i = 0; i < TRIE_CHILDREN; i++)
|
| 190 |
|
|
{
|
| 191 |
|
|
if (p_dict->p_nodes[i] != NULL)
|
| 192 |
|
|
{
|
| 193 |
|
|
trie_dict_destroy(p_dict->p_nodes[i]);
|
| 194 |
|
|
}
|
| 195 |
sysadm |
1.7 |
}
|
| 196 |
sysadm |
1.3 |
|
| 197 |
sysadm |
1.10 |
memset(p_dict, 0, sizeof(*p_dict));
|
| 198 |
sysadm |
1.1 |
|
| 199 |
sysadm |
1.7 |
p_dict->p_nodes[0] = p_trie_node_pool->p_node_free_list;
|
| 200 |
|
|
p_trie_node_pool->p_node_free_list = p_dict;
|
| 201 |
sysadm |
1.8 |
|
| 202 |
|
|
p_trie_node_pool->node_count--;
|
| 203 |
sysadm |
1.1 |
}
|
| 204 |
|
|
|
| 205 |
|
|
int trie_dict_set(TRIE_NODE *p_dict, const char *key, int64_t value)
|
| 206 |
|
|
{
|
| 207 |
|
|
int offset;
|
| 208 |
|
|
|
| 209 |
sysadm |
1.5 |
if (p_dict == NULL)
|
| 210 |
|
|
{
|
| 211 |
|
|
return -1;
|
| 212 |
|
|
}
|
| 213 |
|
|
|
| 214 |
sysadm |
1.1 |
while (key != NULL && *key != '\0')
|
| 215 |
|
|
{
|
| 216 |
sysadm |
1.26 |
offset = (unsigned char)(*key);
|
| 217 |
sysadm |
1.3 |
if (offset < 0 || offset >= TRIE_CHILDREN) // incorrect key character
|
| 218 |
|
|
{
|
| 219 |
|
|
return -1;
|
| 220 |
|
|
}
|
| 221 |
sysadm |
1.1 |
|
| 222 |
|
|
if (*(key + 1) == '\0')
|
| 223 |
|
|
{
|
| 224 |
|
|
if (p_dict->flags[offset] == 0 || p_dict->values[offset] != value)
|
| 225 |
|
|
{
|
| 226 |
|
|
p_dict->values[offset] = value;
|
| 227 |
|
|
p_dict->flags[offset] = 1;
|
| 228 |
|
|
return 1; // Set to new value
|
| 229 |
|
|
}
|
| 230 |
|
|
return 0; // Unchanged
|
| 231 |
|
|
}
|
| 232 |
|
|
else
|
| 233 |
|
|
{
|
| 234 |
|
|
if (p_dict->p_nodes[offset] == NULL)
|
| 235 |
|
|
{
|
| 236 |
|
|
if ((p_dict->p_nodes[offset] = trie_dict_create()) == NULL)
|
| 237 |
|
|
{
|
| 238 |
|
|
return -2; // OOM
|
| 239 |
|
|
}
|
| 240 |
|
|
}
|
| 241 |
|
|
p_dict = p_dict->p_nodes[offset];
|
| 242 |
|
|
key++;
|
| 243 |
|
|
}
|
| 244 |
|
|
}
|
| 245 |
|
|
|
| 246 |
|
|
return -1; // NULL key
|
| 247 |
|
|
}
|
| 248 |
|
|
|
| 249 |
|
|
int trie_dict_get(TRIE_NODE *p_dict, const char *key, int64_t *p_value)
|
| 250 |
|
|
{
|
| 251 |
|
|
int offset;
|
| 252 |
|
|
|
| 253 |
sysadm |
1.5 |
if (p_dict == NULL)
|
| 254 |
|
|
{
|
| 255 |
|
|
return -1;
|
| 256 |
|
|
}
|
| 257 |
|
|
|
| 258 |
sysadm |
1.1 |
while (key != NULL && *key != '\0')
|
| 259 |
|
|
{
|
| 260 |
sysadm |
1.26 |
offset = (unsigned char)(*key);
|
| 261 |
sysadm |
1.3 |
if (offset < 0 || offset >= TRIE_CHILDREN) // incorrect key character
|
| 262 |
|
|
{
|
| 263 |
|
|
return -1;
|
| 264 |
|
|
}
|
| 265 |
sysadm |
1.1 |
|
| 266 |
|
|
if (*(key + 1) == '\0')
|
| 267 |
|
|
{
|
| 268 |
|
|
if (p_dict->flags[offset] != 0)
|
| 269 |
|
|
{
|
| 270 |
|
|
*p_value = p_dict->values[offset];
|
| 271 |
|
|
return 1; // Found
|
| 272 |
|
|
}
|
| 273 |
|
|
else
|
| 274 |
|
|
{
|
| 275 |
|
|
return 0; // Not exist
|
| 276 |
|
|
}
|
| 277 |
|
|
}
|
| 278 |
|
|
else if (p_dict->p_nodes[offset] == NULL)
|
| 279 |
|
|
{
|
| 280 |
|
|
return 0; // Not exist
|
| 281 |
|
|
}
|
| 282 |
|
|
else
|
| 283 |
|
|
{
|
| 284 |
|
|
p_dict = p_dict->p_nodes[offset];
|
| 285 |
|
|
key++;
|
| 286 |
|
|
}
|
| 287 |
|
|
}
|
| 288 |
|
|
|
| 289 |
|
|
return -1; // NULL key
|
| 290 |
|
|
}
|
| 291 |
|
|
|
| 292 |
|
|
int trie_dict_del(TRIE_NODE *p_dict, const char *key)
|
| 293 |
|
|
{
|
| 294 |
|
|
int offset;
|
| 295 |
|
|
|
| 296 |
sysadm |
1.5 |
if (p_dict == NULL)
|
| 297 |
|
|
{
|
| 298 |
|
|
return -1;
|
| 299 |
|
|
}
|
| 300 |
|
|
|
| 301 |
sysadm |
1.1 |
while (key != NULL && *key != '\0')
|
| 302 |
|
|
{
|
| 303 |
sysadm |
1.26 |
offset = (unsigned char)(*key);
|
| 304 |
sysadm |
1.3 |
if (offset < 0 || offset >= TRIE_CHILDREN) // incorrect key character
|
| 305 |
|
|
{
|
| 306 |
|
|
return -1;
|
| 307 |
|
|
}
|
| 308 |
sysadm |
1.1 |
|
| 309 |
|
|
if (*(key + 1) == '\0')
|
| 310 |
|
|
{
|
| 311 |
|
|
if (p_dict->flags[offset] != 0)
|
| 312 |
|
|
{
|
| 313 |
|
|
p_dict->flags[offset] = 0;
|
| 314 |
|
|
p_dict->values[offset] = 0;
|
| 315 |
|
|
return 1; // Done
|
| 316 |
|
|
}
|
| 317 |
|
|
else
|
| 318 |
|
|
{
|
| 319 |
|
|
return 0; // Not exist
|
| 320 |
|
|
}
|
| 321 |
|
|
}
|
| 322 |
|
|
else if (p_dict->p_nodes[offset] == NULL)
|
| 323 |
|
|
{
|
| 324 |
|
|
return 0; // Not exist
|
| 325 |
|
|
}
|
| 326 |
|
|
else
|
| 327 |
|
|
{
|
| 328 |
|
|
p_dict = p_dict->p_nodes[offset];
|
| 329 |
|
|
key++;
|
| 330 |
|
|
}
|
| 331 |
|
|
}
|
| 332 |
|
|
|
| 333 |
|
|
return -1; // NULL key
|
| 334 |
|
|
}
|
| 335 |
sysadm |
1.2 |
|
| 336 |
|
|
static void _trie_dict_traverse(TRIE_NODE *p_dict, trie_dict_traverse_cb cb, char *key, int depth)
|
| 337 |
|
|
{
|
| 338 |
|
|
if (p_dict == NULL || depth >= TRIE_MAX_KEY_LEN)
|
| 339 |
|
|
{
|
| 340 |
|
|
return;
|
| 341 |
|
|
}
|
| 342 |
|
|
|
| 343 |
|
|
for (int i = 0; i < TRIE_CHILDREN; i++)
|
| 344 |
|
|
{
|
| 345 |
|
|
if (p_dict->flags[i] != 0)
|
| 346 |
|
|
{
|
| 347 |
sysadm |
1.5 |
key[depth] = (char)i;
|
| 348 |
sysadm |
1.3 |
key[depth + 1] = '\0';
|
| 349 |
sysadm |
1.2 |
(*cb)(key, p_dict->values[i]);
|
| 350 |
|
|
}
|
| 351 |
|
|
|
| 352 |
|
|
if (p_dict->p_nodes[i] != NULL && depth + 1 < TRIE_MAX_KEY_LEN)
|
| 353 |
|
|
{
|
| 354 |
sysadm |
1.5 |
key[depth] = (char)i;
|
| 355 |
sysadm |
1.2 |
_trie_dict_traverse(p_dict->p_nodes[i], cb, key, depth + 1);
|
| 356 |
|
|
}
|
| 357 |
sysadm |
1.5 |
}
|
| 358 |
sysadm |
1.2 |
}
|
| 359 |
|
|
|
| 360 |
|
|
void trie_dict_traverse(TRIE_NODE *p_dict, trie_dict_traverse_cb cb)
|
| 361 |
|
|
{
|
| 362 |
|
|
char key[TRIE_MAX_KEY_LEN + 1];
|
| 363 |
sysadm |
1.5 |
|
| 364 |
|
|
if (p_dict == NULL)
|
| 365 |
|
|
{
|
| 366 |
|
|
return;
|
| 367 |
|
|
}
|
| 368 |
sysadm |
1.2 |
|
| 369 |
|
|
_trie_dict_traverse(p_dict, cb, key, 0);
|
| 370 |
|
|
}
|
| 371 |
sysadm |
1.8 |
|
| 372 |
|
|
int trie_dict_used_nodes(void)
|
| 373 |
|
|
{
|
| 374 |
|
|
return (p_trie_node_pool == NULL ? -1 : p_trie_node_pool->node_count);
|
| 375 |
|
|
}
|