| 1 | /* -*-c-*- |
| 2 | * |
| 3 | * $Id: sym.c,v 1.11 2000/06/17 10:37:39 mdw Exp $ |
| 4 | * |
| 5 | * Symbol table management |
| 6 | * |
| 7 | * (c) 1998 Straylight/Edgeware |
| 8 | */ |
| 9 | |
| 10 | /*----- Licensing notice --------------------------------------------------* |
| 11 | * |
| 12 | * This file is part of the mLib utilities library. |
| 13 | * |
| 14 | * mLib is free software; you can redistribute it and/or modify |
| 15 | * it under the terms of the GNU Library General Public License as |
| 16 | * published by the Free Software Foundation; either version 2 of the |
| 17 | * License, or (at your option) any later version. |
| 18 | * |
| 19 | * mLib is distributed in the hope that it will be useful, |
| 20 | * but WITHOUT ANY WARRANTY; without even the implied warranty of |
| 21 | * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the |
| 22 | * GNU Library General Public License for more details. |
| 23 | * |
| 24 | * You should have received a copy of the GNU Library General Public |
| 25 | * License along with mLib; if not, write to the Free |
| 26 | * Software Foundation, Inc., 59 Temple Place - Suite 330, Boston, |
| 27 | * MA 02111-1307, USA. |
| 28 | */ |
| 29 | |
| 30 | /*----- Revision history --------------------------------------------------* |
| 31 | * |
| 32 | * $Log: sym.c,v $ |
| 33 | * Revision 1.11 2000/06/17 10:37:39 mdw |
| 34 | * Add support for arena management. |
| 35 | * |
| 36 | * Revision 1.10 1999/12/10 23:42:04 mdw |
| 37 | * Change header file guard names. |
| 38 | * |
| 39 | * Revision 1.9 1999/10/22 22:36:37 mdw |
| 40 | * New test structure for symbol tables. |
| 41 | * |
| 42 | * Revision 1.8 1999/08/02 14:45:48 mdw |
| 43 | * Break low-level hashtable code out from sym. |
| 44 | * |
| 45 | * Revision 1.7 1999/06/01 09:49:08 mdw |
| 46 | * Allow things to be looked up by just their caller-supplied hashes. This |
| 47 | * actually needs to be thought through better. |
| 48 | * |
| 49 | * Revision 1.6 1999/05/26 21:08:31 mdw |
| 50 | * Rename symbols in line with newer conventions. |
| 51 | * |
| 52 | * Revision 1.5 1999/05/13 22:48:37 mdw |
| 53 | * Twiddle the extension threshold. Change `-ise' to `-ize' throughout. |
| 54 | * |
| 55 | * Revision 1.4 1999/05/06 19:51:35 mdw |
| 56 | * Reformatted the LGPL notice a little bit. |
| 57 | * |
| 58 | * Revision 1.3 1999/05/05 18:50:31 mdw |
| 59 | * Change licensing conditions to LGPL. |
| 60 | * |
| 61 | * Revision 1.2 1998/11/26 19:27:33 mdw |
| 62 | * Move SYM_NAME into the header file. Fix bugs. |
| 63 | * |
| 64 | * Revision 1.1.1.1 1998/06/17 23:44:42 mdw |
| 65 | * Initial version of mLib |
| 66 | * |
| 67 | */ |
| 68 | |
| 69 | /*----- Header files ------------------------------------------------------*/ |
| 70 | |
| 71 | /* --- ANSI headers --- */ |
| 72 | |
| 73 | #include <stdio.h> |
| 74 | #include <stdlib.h> |
| 75 | #include <string.h> |
| 76 | |
| 77 | /* --- Local headers --- */ |
| 78 | |
| 79 | #include "alloc.h" |
| 80 | #include "arena.h" |
| 81 | #include "bits.h" |
| 82 | #include "crc32.h" |
| 83 | #include "exc.h" |
| 84 | #include "hash.h" |
| 85 | #include "sub.h" |
| 86 | #include "sym.h" |
| 87 | |
| 88 | /*----- Tuning parameters -------------------------------------------------*/ |
| 89 | |
| 90 | /* --- Initial hash table size --- * |
| 91 | * |
| 92 | * This is the initial @mask@ value. It must be of the form %$2^n - 1$%, |
| 93 | * so that it can be used to mask of the bottom bits of a hash value. |
| 94 | */ |
| 95 | |
| 96 | #define SYM_INITSZ 64 /* Size of a new hash table */ |
| 97 | |
| 98 | /* --- Maximum load factor --- * |
| 99 | * |
| 100 | * This parameter controls how much the table has to be loaded before the |
| 101 | * table is extended. The number of elements %$n$%, the number of bins %$b$% |
| 102 | * and the limit %$l$% satisfy the relation %$n < bl$%; if a new item is |
| 103 | * added to the table and this relation is found to be false, the table is |
| 104 | * doubled in size. |
| 105 | */ |
| 106 | |
| 107 | #define SYM_LIMIT(n) ((n) * 2) /* Load factor for growing table */ |
| 108 | |
| 109 | /*----- Main code ---------------------------------------------------------*/ |
| 110 | |
| 111 | /* --- @sym_create@ --- * |
| 112 | * |
| 113 | * Arguments: @sym_table *t@ = symbol table to initialize |
| 114 | * |
| 115 | * Returns: --- |
| 116 | * |
| 117 | * Use: Initializes the given symbol table. Raises @EXC_NOMEM@ if |
| 118 | * there isn't enough memory. |
| 119 | */ |
| 120 | |
| 121 | void sym_create(sym_table *t) |
| 122 | { |
| 123 | hash_create(&t->t, SYM_INITSZ); |
| 124 | t->s = &sub_global; |
| 125 | t->load = SYM_LIMIT(SYM_INITSZ); |
| 126 | } |
| 127 | |
| 128 | /* --- @sym_destroy@ --- * |
| 129 | * |
| 130 | * Arguments: @sym_table *t@ = pointer to symbol table in question |
| 131 | * |
| 132 | * Returns: --- |
| 133 | * |
| 134 | * Use: Destroys a symbol table, freeing all the memory it used to |
| 135 | * occupy. |
| 136 | */ |
| 137 | |
| 138 | void sym_destroy(sym_table *t) |
| 139 | { |
| 140 | sym_iter i; |
| 141 | |
| 142 | SYM_MKITER(&i, t); |
| 143 | for (;;) { |
| 144 | sym_base *p; |
| 145 | SYM_NEXT(&i, p); |
| 146 | if (!p) |
| 147 | break; |
| 148 | if (p->len > SYM_BUFSZ) |
| 149 | subarena_free(t->s, p->name.p, p->len); |
| 150 | x_free(t->t.a, p); |
| 151 | } |
| 152 | hash_destroy(&t->t); |
| 153 | } |
| 154 | |
| 155 | /* --- @sym_find@ --- * |
| 156 | * |
| 157 | * Arguments: @sym_table *t@ = pointer to symbol table in question |
| 158 | * @const char *n@ = pointer to symbol table to look up |
| 159 | * @long l@ = length of the name string or negative to measure |
| 160 | * @size_t sz@ = size of desired symbol object, or zero |
| 161 | * @unsigned *f@ = pointer to a flag, or null. |
| 162 | * |
| 163 | * Returns: The address of a @sym_base@ structure, or null if not found |
| 164 | * and @sz@ is zero. |
| 165 | * |
| 166 | * Use: Looks up a symbol in a given symbol table. The name is |
| 167 | * passed by the address of its first character. The length |
| 168 | * may be given, in which case the name may contain arbitrary |
| 169 | * binary data, or it may be given as a negative number, in |
| 170 | * which case the length of the name is calculated as |
| 171 | * @strlen(n) + 1@. |
| 172 | * |
| 173 | * The return value is the address of a pointer to a @sym_base@ |
| 174 | * block (which may have other things on the end, as above). If |
| 175 | * the symbol could be found, the return value points to the |
| 176 | * symbol block. If the symbol wasn't there, then if @sz@ is |
| 177 | * nonzero, a new symbol is created and its address is returned; |
| 178 | * otherwise a null pointer is returned. The exception |
| 179 | * @EXC_NOMEM@ is raised if the block can't be allocated. |
| 180 | * |
| 181 | * The value of @*f@ indicates whether a new symbol entry was |
| 182 | * created: a nonzero value indicates that an old value was |
| 183 | * found. |
| 184 | */ |
| 185 | |
| 186 | void *sym_find(sym_table *t, const char *n, long l, size_t sz, unsigned *f) |
| 187 | { |
| 188 | uint32 hash; |
| 189 | size_t len = 0; |
| 190 | hash_base **bin, **p; |
| 191 | sym_base *q; |
| 192 | |
| 193 | /* --- Find the correct bin --- */ |
| 194 | |
| 195 | len = l < 0 ? strlen(n) + 1 : l; |
| 196 | CRC32(hash, 0, n, len); |
| 197 | bin = HASH_BIN(&t->t, hash); |
| 198 | |
| 199 | /* --- Search the bin list --- */ |
| 200 | |
| 201 | for (p = bin; *p; p = &(*p)->next) { |
| 202 | q = (sym_base *)*p; |
| 203 | if (hash == q->b.hash && len == q->len && !memcmp(n, SYM_NAME(q), len)) { |
| 204 | |
| 205 | /* --- Found a match --- * |
| 206 | * |
| 207 | * As a minor, and probably pointless, tweak, move the item to the |
| 208 | * front of its bin list. |
| 209 | */ |
| 210 | |
| 211 | (*p) = q->b.next; |
| 212 | q->b.next = *bin; |
| 213 | *bin = &q->b; |
| 214 | |
| 215 | /* --- Return the block --- */ |
| 216 | |
| 217 | if (f) *f = 1; |
| 218 | return (q); |
| 219 | } |
| 220 | } |
| 221 | |
| 222 | /* --- Couldn't find the item there --- */ |
| 223 | |
| 224 | if (f) *f = 0; |
| 225 | if (!sz) return (0); |
| 226 | |
| 227 | /* --- Create a new symbol block and initialize it --- */ |
| 228 | |
| 229 | q = x_alloc(t->t.a, sz); |
| 230 | q->b.next = *bin; |
| 231 | q->b.hash = hash; |
| 232 | q->len = len; |
| 233 | if (n) { |
| 234 | if (len <= SYM_BUFSZ) |
| 235 | memcpy(q->name.b, n, len); |
| 236 | else { |
| 237 | q->name.p = subarena_alloc(t->s, len); |
| 238 | memcpy(q->name.p, n, len); |
| 239 | } |
| 240 | } |
| 241 | |
| 242 | *bin = &q->b; |
| 243 | |
| 244 | /* --- Consider growing the array --- */ |
| 245 | |
| 246 | if (t->load) |
| 247 | t->load--; |
| 248 | if (!t->load && hash_extend(&t->t)) |
| 249 | t->load = SYM_LIMIT(t->t.mask / 2 + 1); |
| 250 | |
| 251 | /* --- Finished that, so return the new symbol block --- */ |
| 252 | |
| 253 | return (q); |
| 254 | } |
| 255 | |
| 256 | /* --- @sym_remove@ --- * |
| 257 | * |
| 258 | * Arguments: @sym_table *t@ = pointer to a symbol table object |
| 259 | * @void *p@ = pointer to symbol table entry |
| 260 | * |
| 261 | * Returns: --- |
| 262 | * |
| 263 | * Use: Removes the object from the symbol table. The space occupied |
| 264 | * by the object and its name is freed; anything else attached |
| 265 | * to the entry should already be gone by this point. |
| 266 | */ |
| 267 | |
| 268 | void sym_remove(sym_table *t, void *p) |
| 269 | { |
| 270 | sym_base *q = p; |
| 271 | hash_remove(&t->t, &q->b); |
| 272 | if (q->len > SYM_BUFSZ) |
| 273 | subarena_free(t->s, q->name.p, q->len); |
| 274 | xfree(q); |
| 275 | t->load++; |
| 276 | } |
| 277 | |
| 278 | /* --- @sym_mkiter@ --- * |
| 279 | * |
| 280 | * Arguments: @sym_iter *i@ = pointer to an iterator object |
| 281 | * @sym_table *t@ = pointer to a symbol table object |
| 282 | * |
| 283 | * Returns: --- |
| 284 | * |
| 285 | * Use: Creates a new symbol table iterator which may be used to |
| 286 | * iterate through a symbol table. |
| 287 | */ |
| 288 | |
| 289 | void sym_mkiter(sym_iter *i, sym_table *t) { SYM_MKITER(i, t); } |
| 290 | |
| 291 | /* --- @sym_next@ --- * |
| 292 | * |
| 293 | * Arguments: @sym_iter *i@ = pointer to iterator object |
| 294 | * |
| 295 | * Returns: Pointer to the next symbol found, or null when finished. |
| 296 | * |
| 297 | * Use: Returns the next symbol from the table. Symbols are not |
| 298 | * returned in any particular order. |
| 299 | */ |
| 300 | |
| 301 | void *sym_next(sym_iter *i) |
| 302 | { |
| 303 | void *p; |
| 304 | SYM_NEXT(i, p); |
| 305 | return (p); |
| 306 | } |
| 307 | |
| 308 | /*----- That's all, folks -------------------------------------------------*/ |