| 1 | /* Symbol table manager for Bison. |
| 2 | |
| 3 | Copyright (C) 1984, 1989, 2000, 2001, 2002, 2004, 2005, 2006, 2007, |
| 4 | 2008, 2009 |
| 5 | Free Software Foundation, Inc. |
| 6 | |
| 7 | This file is part of Bison, the GNU Compiler Compiler. |
| 8 | |
| 9 | This program is free software: you can redistribute it and/or modify |
| 10 | it under the terms of the GNU General Public License as published by |
| 11 | the Free Software Foundation, either version 3 of the License, or |
| 12 | (at your option) any later version. |
| 13 | |
| 14 | This program is distributed in the hope that it will be useful, |
| 15 | but WITHOUT ANY WARRANTY; without even the implied warranty of |
| 16 | MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the |
| 17 | GNU General Public License for more details. |
| 18 | |
| 19 | You should have received a copy of the GNU General Public License |
| 20 | along with this program. If not, see <http://www.gnu.org/licenses/>. */ |
| 21 | |
| 22 | #include <config.h> |
| 23 | #include "system.h" |
| 24 | |
| 25 | #include <hash.h> |
| 26 | #include <quotearg.h> |
| 27 | |
| 28 | #include "complain.h" |
| 29 | #include "gram.h" |
| 30 | #include "symtab.h" |
| 31 | |
| 32 | /*-------------------------------------------------------------------. |
| 33 | | Symbols sorted by tag. Allocated by the first invocation of | |
| 34 | | symbols_do, after which no more symbols should be created. | |
| 35 | `-------------------------------------------------------------------*/ |
| 36 | |
| 37 | static symbol **symbols_sorted = NULL; |
| 38 | |
| 39 | /*------------------------. |
| 40 | | Distinguished symbols. | |
| 41 | `------------------------*/ |
| 42 | |
| 43 | symbol *errtoken = NULL; |
| 44 | symbol *undeftoken = NULL; |
| 45 | symbol *endtoken = NULL; |
| 46 | symbol *accept = NULL; |
| 47 | symbol *startsymbol = NULL; |
| 48 | location startsymbol_location; |
| 49 | |
| 50 | /*---------------------------------------. |
| 51 | | Default %destructor's and %printer's. | |
| 52 | `---------------------------------------*/ |
| 53 | |
| 54 | static code_props default_tagged_destructor = CODE_PROPS_NONE_INIT; |
| 55 | static code_props default_tagless_destructor = CODE_PROPS_NONE_INIT; |
| 56 | static code_props default_tagged_printer = CODE_PROPS_NONE_INIT; |
| 57 | static code_props default_tagless_printer = CODE_PROPS_NONE_INIT; |
| 58 | |
| 59 | /*---------------------------------. |
| 60 | | Create a new symbol, named TAG. | |
| 61 | `---------------------------------*/ |
| 62 | |
| 63 | static symbol * |
| 64 | symbol_new (uniqstr tag, location loc) |
| 65 | { |
| 66 | symbol *res = xmalloc (sizeof *res); |
| 67 | |
| 68 | uniqstr_assert (tag); |
| 69 | |
| 70 | /* If the tag is not a string (starts with a double quote), check |
| 71 | that it is valid for Yacc. */ |
| 72 | if (tag[0] != '\"' && tag[0] != '\'' && strchr (tag, '-')) |
| 73 | yacc_at (loc, _("POSIX Yacc forbids dashes in symbol names: %s"), |
| 74 | tag); |
| 75 | |
| 76 | res->tag = tag; |
| 77 | res->location = loc; |
| 78 | |
| 79 | res->type_name = NULL; |
| 80 | code_props_none_init (&res->destructor); |
| 81 | code_props_none_init (&res->printer); |
| 82 | |
| 83 | res->number = NUMBER_UNDEFINED; |
| 84 | res->prec = 0; |
| 85 | res->assoc = undef_assoc; |
| 86 | res->user_token_number = USER_NUMBER_UNDEFINED; |
| 87 | |
| 88 | res->alias = NULL; |
| 89 | res->class = unknown_sym; |
| 90 | res->declared = false; |
| 91 | |
| 92 | if (nsyms == SYMBOL_NUMBER_MAXIMUM) |
| 93 | fatal (_("too many symbols in input grammar (limit is %d)"), |
| 94 | SYMBOL_NUMBER_MAXIMUM); |
| 95 | nsyms++; |
| 96 | return res; |
| 97 | } |
| 98 | |
| 99 | /*----------------------------------------. |
| 100 | | Create a new semantic type, named TAG. | |
| 101 | `----------------------------------------*/ |
| 102 | |
| 103 | static semantic_type * |
| 104 | semantic_type_new (uniqstr tag) |
| 105 | { |
| 106 | semantic_type *res = xmalloc (sizeof *res); |
| 107 | |
| 108 | uniqstr_assert (tag); |
| 109 | res->tag = tag; |
| 110 | code_props_none_init (&res->destructor); |
| 111 | code_props_none_init (&res->printer); |
| 112 | |
| 113 | return res; |
| 114 | } |
| 115 | |
| 116 | |
| 117 | /*-----------------. |
| 118 | | Print a symbol. | |
| 119 | `-----------------*/ |
| 120 | |
| 121 | #define SYMBOL_ATTR_PRINT(Attr) \ |
| 122 | if (s->Attr) \ |
| 123 | fprintf (f, " %s { %s }", #Attr, s->Attr) |
| 124 | |
| 125 | #define SYMBOL_CODE_PRINT(Attr) \ |
| 126 | if (s->Attr.code) \ |
| 127 | fprintf (f, " %s { %s }", #Attr, s->Attr.code) |
| 128 | |
| 129 | void |
| 130 | symbol_print (symbol *s, FILE *f) |
| 131 | { |
| 132 | if (s) |
| 133 | { |
| 134 | fprintf (f, "\"%s\"", s->tag); |
| 135 | SYMBOL_ATTR_PRINT (type_name); |
| 136 | SYMBOL_CODE_PRINT (destructor); |
| 137 | SYMBOL_CODE_PRINT (printer); |
| 138 | } |
| 139 | else |
| 140 | fprintf (f, "<NULL>"); |
| 141 | } |
| 142 | |
| 143 | #undef SYMBOL_ATTR_PRINT |
| 144 | #undef SYMBOL_CODE_PRINT |
| 145 | |
| 146 | |
| 147 | /*----------------------------------. |
| 148 | | Whether S is a valid identifier. | |
| 149 | `----------------------------------*/ |
| 150 | |
| 151 | static bool |
| 152 | is_identifier (uniqstr s) |
| 153 | { |
| 154 | static char const alphanum[26 + 26 + 1 + 10] = |
| 155 | "abcdefghijklmnopqrstuvwxyz" |
| 156 | "ABCDEFGHIJKLMNOPQRSTUVWXYZ" |
| 157 | "_" |
| 158 | "0123456789"; |
| 159 | if (!s || ! memchr (alphanum, *s, sizeof alphanum - 10)) |
| 160 | return false; |
| 161 | for (++s; *s; ++s) |
| 162 | if (! memchr (alphanum, *s, sizeof alphanum)) |
| 163 | return false; |
| 164 | return true; |
| 165 | } |
| 166 | |
| 167 | |
| 168 | /*-----------------------------------------------. |
| 169 | | Get the identifier associated to this symbol. | |
| 170 | `-----------------------------------------------*/ |
| 171 | uniqstr |
| 172 | symbol_id_get (symbol const *sym) |
| 173 | { |
| 174 | aver (sym->user_token_number != USER_NUMBER_HAS_STRING_ALIAS); |
| 175 | if (sym->alias) |
| 176 | sym = sym->alias; |
| 177 | return is_identifier (sym->tag) ? sym->tag : 0; |
| 178 | } |
| 179 | |
| 180 | |
| 181 | /*------------------------------------------------------------------. |
| 182 | | Complain that S's WHAT is redeclared at SECOND, and was first set | |
| 183 | | at FIRST. | |
| 184 | `------------------------------------------------------------------*/ |
| 185 | |
| 186 | static void |
| 187 | symbol_redeclaration (symbol *s, const char *what, location first, |
| 188 | location second) |
| 189 | { |
| 190 | complain_at (second, _("%s redeclaration for %s"), what, s->tag); |
| 191 | complain_at (first, _("previous declaration")); |
| 192 | } |
| 193 | |
| 194 | static void |
| 195 | semantic_type_redeclaration (semantic_type *s, const char *what, location first, |
| 196 | location second) |
| 197 | { |
| 198 | complain_at (second, _("%s redeclaration for <%s>"), what, s->tag); |
| 199 | complain_at (first, _("previous declaration")); |
| 200 | } |
| 201 | |
| 202 | |
| 203 | |
| 204 | /*-----------------------------------------------------------------. |
| 205 | | Set the TYPE_NAME associated with SYM. Does nothing if passed 0 | |
| 206 | | as TYPE_NAME. | |
| 207 | `-----------------------------------------------------------------*/ |
| 208 | |
| 209 | void |
| 210 | symbol_type_set (symbol *sym, uniqstr type_name, location loc) |
| 211 | { |
| 212 | if (type_name) |
| 213 | { |
| 214 | if (sym->type_name) |
| 215 | symbol_redeclaration (sym, "%type", sym->type_location, loc); |
| 216 | uniqstr_assert (type_name); |
| 217 | sym->type_name = type_name; |
| 218 | sym->type_location = loc; |
| 219 | } |
| 220 | } |
| 221 | |
| 222 | /*-----------------------------------------. |
| 223 | | Set the DESTRUCTOR associated with SYM. | |
| 224 | `-----------------------------------------*/ |
| 225 | |
| 226 | void |
| 227 | symbol_destructor_set (symbol *sym, code_props const *destructor) |
| 228 | { |
| 229 | if (sym->destructor.code) |
| 230 | symbol_redeclaration (sym, "%destructor", sym->destructor.location, |
| 231 | destructor->location); |
| 232 | sym->destructor = *destructor; |
| 233 | } |
| 234 | |
| 235 | /*------------------------------------------. |
| 236 | | Set the DESTRUCTOR associated with TYPE. | |
| 237 | `------------------------------------------*/ |
| 238 | |
| 239 | void |
| 240 | semantic_type_destructor_set (semantic_type *type, |
| 241 | code_props const *destructor) |
| 242 | { |
| 243 | if (type->destructor.code) |
| 244 | semantic_type_redeclaration (type, "%destructor", |
| 245 | type->destructor.location, |
| 246 | destructor->location); |
| 247 | type->destructor = *destructor; |
| 248 | } |
| 249 | |
| 250 | /*---------------------------------------. |
| 251 | | Get the computed %destructor for SYM. | |
| 252 | `---------------------------------------*/ |
| 253 | |
| 254 | code_props const * |
| 255 | symbol_destructor_get (symbol const *sym) |
| 256 | { |
| 257 | /* Per-symbol %destructor. */ |
| 258 | if (sym->destructor.code) |
| 259 | return &sym->destructor; |
| 260 | |
| 261 | /* Per-type %destructor. */ |
| 262 | if (sym->type_name) |
| 263 | { |
| 264 | code_props const *destructor = |
| 265 | &semantic_type_get (sym->type_name)->destructor; |
| 266 | if (destructor->code) |
| 267 | return destructor; |
| 268 | } |
| 269 | |
| 270 | /* Apply default %destructor's only to user-defined symbols. */ |
| 271 | if (sym->tag[0] == '$' || sym == errtoken) |
| 272 | return &code_props_none; |
| 273 | |
| 274 | if (sym->type_name) |
| 275 | return &default_tagged_destructor; |
| 276 | return &default_tagless_destructor; |
| 277 | } |
| 278 | |
| 279 | /*--------------------------------------. |
| 280 | | Set the PRINTER associated with SYM. | |
| 281 | `--------------------------------------*/ |
| 282 | |
| 283 | void |
| 284 | symbol_printer_set (symbol *sym, code_props const *printer) |
| 285 | { |
| 286 | if (sym->printer.code) |
| 287 | symbol_redeclaration (sym, "%printer", |
| 288 | sym->printer.location, printer->location); |
| 289 | sym->printer = *printer; |
| 290 | } |
| 291 | |
| 292 | /*---------------------------------------. |
| 293 | | Set the PRINTER associated with TYPE. | |
| 294 | `---------------------------------------*/ |
| 295 | |
| 296 | void |
| 297 | semantic_type_printer_set (semantic_type *type, code_props const *printer) |
| 298 | { |
| 299 | if (type->printer.code) |
| 300 | semantic_type_redeclaration (type, "%printer", |
| 301 | type->printer.location, printer->location); |
| 302 | type->printer = *printer; |
| 303 | } |
| 304 | |
| 305 | /*------------------------------------. |
| 306 | | Get the computed %printer for SYM. | |
| 307 | `------------------------------------*/ |
| 308 | |
| 309 | code_props const * |
| 310 | symbol_printer_get (symbol const *sym) |
| 311 | { |
| 312 | /* Per-symbol %printer. */ |
| 313 | if (sym->printer.code) |
| 314 | return &sym->printer; |
| 315 | |
| 316 | /* Per-type %printer. */ |
| 317 | if (sym->type_name) |
| 318 | { |
| 319 | code_props const *printer = &semantic_type_get (sym->type_name)->printer; |
| 320 | if (printer->code) |
| 321 | return printer; |
| 322 | } |
| 323 | |
| 324 | /* Apply the default %printer only to user-defined symbols. */ |
| 325 | if (sym->tag[0] == '$' || sym == errtoken) |
| 326 | return &code_props_none; |
| 327 | |
| 328 | if (sym->type_name) |
| 329 | return &default_tagged_printer; |
| 330 | return &default_tagless_printer; |
| 331 | } |
| 332 | |
| 333 | /*-----------------------------------------------------------------. |
| 334 | | Set the PRECEDENCE associated with SYM. Does nothing if invoked | |
| 335 | | with UNDEF_ASSOC as ASSOC. | |
| 336 | `-----------------------------------------------------------------*/ |
| 337 | |
| 338 | void |
| 339 | symbol_precedence_set (symbol *sym, int prec, assoc a, location loc) |
| 340 | { |
| 341 | if (a != undef_assoc) |
| 342 | { |
| 343 | if (sym->prec != 0) |
| 344 | symbol_redeclaration (sym, assoc_to_string (a), sym->prec_location, |
| 345 | loc); |
| 346 | sym->prec = prec; |
| 347 | sym->assoc = a; |
| 348 | sym->prec_location = loc; |
| 349 | } |
| 350 | |
| 351 | /* Only terminals have a precedence. */ |
| 352 | symbol_class_set (sym, token_sym, loc, false); |
| 353 | } |
| 354 | |
| 355 | |
| 356 | /*------------------------------------. |
| 357 | | Set the CLASS associated with SYM. | |
| 358 | `------------------------------------*/ |
| 359 | |
| 360 | void |
| 361 | symbol_class_set (symbol *sym, symbol_class class, location loc, bool declaring) |
| 362 | { |
| 363 | if (sym->class != unknown_sym && sym->class != class) |
| 364 | { |
| 365 | complain_at (loc, _("symbol %s redefined"), sym->tag); |
| 366 | sym->declared = false; |
| 367 | } |
| 368 | |
| 369 | if (class == nterm_sym && sym->class != nterm_sym) |
| 370 | sym->number = nvars++; |
| 371 | else if (class == token_sym && sym->number == NUMBER_UNDEFINED) |
| 372 | sym->number = ntokens++; |
| 373 | |
| 374 | sym->class = class; |
| 375 | |
| 376 | if (declaring) |
| 377 | { |
| 378 | if (sym->declared) |
| 379 | warn_at (loc, _("symbol %s redeclared"), sym->tag); |
| 380 | sym->declared = true; |
| 381 | } |
| 382 | } |
| 383 | |
| 384 | |
| 385 | /*------------------------------------------------. |
| 386 | | Set the USER_TOKEN_NUMBER associated with SYM. | |
| 387 | `------------------------------------------------*/ |
| 388 | |
| 389 | void |
| 390 | symbol_user_token_number_set (symbol *sym, int user_token_number, location loc) |
| 391 | { |
| 392 | int *user_token_numberp; |
| 393 | |
| 394 | if (sym->user_token_number != USER_NUMBER_HAS_STRING_ALIAS) |
| 395 | user_token_numberp = &sym->user_token_number; |
| 396 | else |
| 397 | user_token_numberp = &sym->alias->user_token_number; |
| 398 | if (*user_token_numberp != USER_NUMBER_UNDEFINED |
| 399 | && *user_token_numberp != user_token_number) |
| 400 | complain_at (loc, _("redefining user token number of %s"), sym->tag); |
| 401 | |
| 402 | *user_token_numberp = user_token_number; |
| 403 | /* User defined $end token? */ |
| 404 | if (user_token_number == 0) |
| 405 | { |
| 406 | endtoken = sym; |
| 407 | endtoken->number = 0; |
| 408 | /* It is always mapped to 0, so it was already counted in |
| 409 | NTOKENS. */ |
| 410 | --ntokens; |
| 411 | } |
| 412 | } |
| 413 | |
| 414 | |
| 415 | /*----------------------------------------------------------. |
| 416 | | If SYM is not defined, report an error, and consider it a | |
| 417 | | nonterminal. | |
| 418 | `----------------------------------------------------------*/ |
| 419 | |
| 420 | static inline bool |
| 421 | symbol_check_defined (symbol *sym) |
| 422 | { |
| 423 | if (sym->class == unknown_sym) |
| 424 | { |
| 425 | complain_at |
| 426 | (sym->location, |
| 427 | _("symbol %s is used, but is not defined as a token and has no rules"), |
| 428 | sym->tag); |
| 429 | sym->class = nterm_sym; |
| 430 | sym->number = nvars++; |
| 431 | } |
| 432 | |
| 433 | return true; |
| 434 | } |
| 435 | |
| 436 | static bool |
| 437 | symbol_check_defined_processor (void *sym, void *null ATTRIBUTE_UNUSED) |
| 438 | { |
| 439 | return symbol_check_defined (sym); |
| 440 | } |
| 441 | |
| 442 | |
| 443 | void |
| 444 | symbol_make_alias (symbol *sym, symbol *str, location loc) |
| 445 | { |
| 446 | if (str->alias) |
| 447 | warn_at (loc, _("symbol `%s' used more than once as a literal string"), |
| 448 | str->tag); |
| 449 | else if (sym->alias) |
| 450 | warn_at (loc, _("symbol `%s' given more than one literal string"), |
| 451 | sym->tag); |
| 452 | else |
| 453 | { |
| 454 | str->class = token_sym; |
| 455 | str->user_token_number = sym->user_token_number; |
| 456 | sym->user_token_number = USER_NUMBER_HAS_STRING_ALIAS; |
| 457 | str->alias = sym; |
| 458 | sym->alias = str; |
| 459 | str->number = sym->number; |
| 460 | symbol_type_set (str, sym->type_name, loc); |
| 461 | } |
| 462 | } |
| 463 | |
| 464 | |
| 465 | /*---------------------------------------------------------. |
| 466 | | Check that THIS, and its alias, have same precedence and | |
| 467 | | associativity. | |
| 468 | `---------------------------------------------------------*/ |
| 469 | |
| 470 | static inline void |
| 471 | symbol_check_alias_consistency (symbol *this) |
| 472 | { |
| 473 | symbol *sym = this; |
| 474 | symbol *str = this->alias; |
| 475 | |
| 476 | /* Check only the symbol in the symbol-string pair. */ |
| 477 | if (!(this->alias |
| 478 | && this->user_token_number == USER_NUMBER_HAS_STRING_ALIAS)) |
| 479 | return; |
| 480 | |
| 481 | if (str->type_name != sym->type_name) |
| 482 | { |
| 483 | if (str->type_name) |
| 484 | symbol_type_set (sym, str->type_name, str->type_location); |
| 485 | else |
| 486 | symbol_type_set (str, sym->type_name, sym->type_location); |
| 487 | } |
| 488 | |
| 489 | |
| 490 | if (str->destructor.code || sym->destructor.code) |
| 491 | { |
| 492 | if (str->destructor.code) |
| 493 | symbol_destructor_set (sym, &str->destructor); |
| 494 | else |
| 495 | symbol_destructor_set (str, &sym->destructor); |
| 496 | } |
| 497 | |
| 498 | if (str->printer.code || sym->printer.code) |
| 499 | { |
| 500 | if (str->printer.code) |
| 501 | symbol_printer_set (sym, &str->printer); |
| 502 | else |
| 503 | symbol_printer_set (str, &sym->printer); |
| 504 | } |
| 505 | |
| 506 | if (sym->prec || str->prec) |
| 507 | { |
| 508 | if (str->prec) |
| 509 | symbol_precedence_set (sym, str->prec, str->assoc, |
| 510 | str->prec_location); |
| 511 | else |
| 512 | symbol_precedence_set (str, sym->prec, sym->assoc, |
| 513 | sym->prec_location); |
| 514 | } |
| 515 | } |
| 516 | |
| 517 | static bool |
| 518 | symbol_check_alias_consistency_processor (void *this, |
| 519 | void *null ATTRIBUTE_UNUSED) |
| 520 | { |
| 521 | symbol_check_alias_consistency (this); |
| 522 | return true; |
| 523 | } |
| 524 | |
| 525 | |
| 526 | /*-------------------------------------------------------------------. |
| 527 | | Assign a symbol number, and write the definition of the token name | |
| 528 | | into FDEFINES. Put in SYMBOLS. | |
| 529 | `-------------------------------------------------------------------*/ |
| 530 | |
| 531 | static inline bool |
| 532 | symbol_pack (symbol *this) |
| 533 | { |
| 534 | aver (this->number != NUMBER_UNDEFINED); |
| 535 | if (this->class == nterm_sym) |
| 536 | this->number += ntokens; |
| 537 | else if (this->user_token_number == USER_NUMBER_HAS_STRING_ALIAS) |
| 538 | return true; |
| 539 | |
| 540 | symbols[this->number] = this; |
| 541 | return true; |
| 542 | } |
| 543 | |
| 544 | static bool |
| 545 | symbol_pack_processor (void *this, void *null ATTRIBUTE_UNUSED) |
| 546 | { |
| 547 | return symbol_pack (this); |
| 548 | } |
| 549 | |
| 550 | |
| 551 | static void |
| 552 | user_token_number_redeclaration (int num, symbol *first, symbol *second) |
| 553 | { |
| 554 | /* User token numbers are not assigned during the parsing, but in a |
| 555 | second step, via a traversal of the symbol table sorted on tag. |
| 556 | |
| 557 | However, error messages make more sense if we keep the first |
| 558 | declaration first. */ |
| 559 | if (location_cmp (first->location, second->location) > 0) |
| 560 | { |
| 561 | symbol* tmp = first; |
| 562 | first = second; |
| 563 | second = tmp; |
| 564 | } |
| 565 | complain_at (second->location, |
| 566 | _("user token number %d redeclaration for %s"), |
| 567 | num, second->tag); |
| 568 | complain_at (first->location, _("previous declaration for %s"), |
| 569 | first->tag); |
| 570 | } |
| 571 | |
| 572 | /*--------------------------------------------------. |
| 573 | | Put THIS in TOKEN_TRANSLATIONS if it is a token. | |
| 574 | `--------------------------------------------------*/ |
| 575 | |
| 576 | static inline bool |
| 577 | symbol_translation (symbol *this) |
| 578 | { |
| 579 | /* Non-terminal? */ |
| 580 | if (this->class == token_sym |
| 581 | && this->user_token_number != USER_NUMBER_HAS_STRING_ALIAS) |
| 582 | { |
| 583 | /* A token which translation has already been set? */ |
| 584 | if (token_translations[this->user_token_number] != undeftoken->number) |
| 585 | user_token_number_redeclaration |
| 586 | (this->user_token_number, |
| 587 | symbols[token_translations[this->user_token_number]], |
| 588 | this); |
| 589 | |
| 590 | token_translations[this->user_token_number] = this->number; |
| 591 | } |
| 592 | |
| 593 | return true; |
| 594 | } |
| 595 | |
| 596 | static bool |
| 597 | symbol_translation_processor (void *this, void *null ATTRIBUTE_UNUSED) |
| 598 | { |
| 599 | return symbol_translation (this); |
| 600 | } |
| 601 | |
| 602 | |
| 603 | /*---------------------------------------. |
| 604 | | Symbol and semantic type hash tables. | |
| 605 | `---------------------------------------*/ |
| 606 | |
| 607 | /* Initial capacity of symbol and semantic type hash table. */ |
| 608 | #define HT_INITIAL_CAPACITY 257 |
| 609 | |
| 610 | static struct hash_table *symbol_table = NULL; |
| 611 | static struct hash_table *semantic_type_table = NULL; |
| 612 | |
| 613 | static inline bool |
| 614 | hash_compare_symbol (const symbol *m1, const symbol *m2) |
| 615 | { |
| 616 | /* Since tags are unique, we can compare the pointers themselves. */ |
| 617 | return UNIQSTR_EQ (m1->tag, m2->tag); |
| 618 | } |
| 619 | |
| 620 | static inline bool |
| 621 | hash_compare_semantic_type (const semantic_type *m1, const semantic_type *m2) |
| 622 | { |
| 623 | /* Since names are unique, we can compare the pointers themselves. */ |
| 624 | return UNIQSTR_EQ (m1->tag, m2->tag); |
| 625 | } |
| 626 | |
| 627 | static bool |
| 628 | hash_symbol_comparator (void const *m1, void const *m2) |
| 629 | { |
| 630 | return hash_compare_symbol (m1, m2); |
| 631 | } |
| 632 | |
| 633 | static bool |
| 634 | hash_semantic_type_comparator (void const *m1, void const *m2) |
| 635 | { |
| 636 | return hash_compare_semantic_type (m1, m2); |
| 637 | } |
| 638 | |
| 639 | static inline size_t |
| 640 | hash_symbol (const symbol *m, size_t tablesize) |
| 641 | { |
| 642 | /* Since tags are unique, we can hash the pointer itself. */ |
| 643 | return ((uintptr_t) m->tag) % tablesize; |
| 644 | } |
| 645 | |
| 646 | static inline size_t |
| 647 | hash_semantic_type (const semantic_type *m, size_t tablesize) |
| 648 | { |
| 649 | /* Since names are unique, we can hash the pointer itself. */ |
| 650 | return ((uintptr_t) m->tag) % tablesize; |
| 651 | } |
| 652 | |
| 653 | static size_t |
| 654 | hash_symbol_hasher (void const *m, size_t tablesize) |
| 655 | { |
| 656 | return hash_symbol (m, tablesize); |
| 657 | } |
| 658 | |
| 659 | static size_t |
| 660 | hash_semantic_type_hasher (void const *m, size_t tablesize) |
| 661 | { |
| 662 | return hash_semantic_type (m, tablesize); |
| 663 | } |
| 664 | |
| 665 | /*-------------------------------. |
| 666 | | Create the symbol hash table. | |
| 667 | `-------------------------------*/ |
| 668 | |
| 669 | void |
| 670 | symbols_new (void) |
| 671 | { |
| 672 | symbol_table = hash_initialize (HT_INITIAL_CAPACITY, |
| 673 | NULL, |
| 674 | hash_symbol_hasher, |
| 675 | hash_symbol_comparator, |
| 676 | free); |
| 677 | semantic_type_table = hash_initialize (HT_INITIAL_CAPACITY, |
| 678 | NULL, |
| 679 | hash_semantic_type_hasher, |
| 680 | hash_semantic_type_comparator, |
| 681 | free); |
| 682 | } |
| 683 | |
| 684 | |
| 685 | /*----------------------------------------------------------------. |
| 686 | | Find the symbol named KEY, and return it. If it does not exist | |
| 687 | | yet, create it. | |
| 688 | `----------------------------------------------------------------*/ |
| 689 | |
| 690 | symbol * |
| 691 | symbol_from_uniqstr (const uniqstr key, location loc) |
| 692 | { |
| 693 | symbol probe; |
| 694 | symbol *entry; |
| 695 | |
| 696 | probe.tag = key; |
| 697 | entry = hash_lookup (symbol_table, &probe); |
| 698 | |
| 699 | if (!entry) |
| 700 | { |
| 701 | /* First insertion in the hash. */ |
| 702 | aver (!symbols_sorted); |
| 703 | entry = symbol_new (key, loc); |
| 704 | if (!hash_insert (symbol_table, entry)) |
| 705 | xalloc_die (); |
| 706 | } |
| 707 | return entry; |
| 708 | } |
| 709 | |
| 710 | |
| 711 | /*-----------------------------------------------------------------------. |
| 712 | | Find the semantic type named KEY, and return it. If it does not exist | |
| 713 | | yet, create it. | |
| 714 | `-----------------------------------------------------------------------*/ |
| 715 | |
| 716 | semantic_type * |
| 717 | semantic_type_from_uniqstr (const uniqstr key) |
| 718 | { |
| 719 | semantic_type probe; |
| 720 | semantic_type *entry; |
| 721 | |
| 722 | probe.tag = key; |
| 723 | entry = hash_lookup (semantic_type_table, &probe); |
| 724 | |
| 725 | if (!entry) |
| 726 | { |
| 727 | /* First insertion in the hash. */ |
| 728 | entry = semantic_type_new (key); |
| 729 | if (!hash_insert (semantic_type_table, entry)) |
| 730 | xalloc_die (); |
| 731 | } |
| 732 | return entry; |
| 733 | } |
| 734 | |
| 735 | |
| 736 | /*----------------------------------------------------------------. |
| 737 | | Find the symbol named KEY, and return it. If it does not exist | |
| 738 | | yet, create it. | |
| 739 | `----------------------------------------------------------------*/ |
| 740 | |
| 741 | symbol * |
| 742 | symbol_get (const char *key, location loc) |
| 743 | { |
| 744 | return symbol_from_uniqstr (uniqstr_new (key), loc); |
| 745 | } |
| 746 | |
| 747 | |
| 748 | /*-----------------------------------------------------------------------. |
| 749 | | Find the semantic type named KEY, and return it. If it does not exist | |
| 750 | | yet, create it. | |
| 751 | `-----------------------------------------------------------------------*/ |
| 752 | |
| 753 | semantic_type * |
| 754 | semantic_type_get (const char *key) |
| 755 | { |
| 756 | return semantic_type_from_uniqstr (uniqstr_new (key)); |
| 757 | } |
| 758 | |
| 759 | |
| 760 | /*------------------------------------------------------------------. |
| 761 | | Generate a dummy nonterminal, whose name cannot conflict with the | |
| 762 | | user's names. | |
| 763 | `------------------------------------------------------------------*/ |
| 764 | |
| 765 | symbol * |
| 766 | dummy_symbol_get (location loc) |
| 767 | { |
| 768 | /* Incremented for each generated symbol. */ |
| 769 | static int dummy_count = 0; |
| 770 | static char buf[256]; |
| 771 | |
| 772 | symbol *sym; |
| 773 | |
| 774 | sprintf (buf, "$@%d", ++dummy_count); |
| 775 | sym = symbol_get (buf, loc); |
| 776 | sym->class = nterm_sym; |
| 777 | sym->number = nvars++; |
| 778 | return sym; |
| 779 | } |
| 780 | |
| 781 | bool |
| 782 | symbol_is_dummy (const symbol *sym) |
| 783 | { |
| 784 | return sym->tag[0] == '@' || (sym->tag[0] == '$' && sym->tag[1] == '@'); |
| 785 | } |
| 786 | |
| 787 | /*-------------------. |
| 788 | | Free the symbols. | |
| 789 | `-------------------*/ |
| 790 | |
| 791 | void |
| 792 | symbols_free (void) |
| 793 | { |
| 794 | hash_free (symbol_table); |
| 795 | hash_free (semantic_type_table); |
| 796 | free (symbols); |
| 797 | free (symbols_sorted); |
| 798 | } |
| 799 | |
| 800 | |
| 801 | /*---------------------------------------------------------------. |
| 802 | | Look for undefined symbols, report an error, and consider them | |
| 803 | | terminals. | |
| 804 | `---------------------------------------------------------------*/ |
| 805 | |
| 806 | static int |
| 807 | symbols_cmp (symbol const *a, symbol const *b) |
| 808 | { |
| 809 | return strcmp (a->tag, b->tag); |
| 810 | } |
| 811 | |
| 812 | static int |
| 813 | symbols_cmp_qsort (void const *a, void const *b) |
| 814 | { |
| 815 | return symbols_cmp (*(symbol * const *)a, *(symbol * const *)b); |
| 816 | } |
| 817 | |
| 818 | static void |
| 819 | symbols_do (Hash_processor processor, void *processor_data) |
| 820 | { |
| 821 | size_t count = hash_get_n_entries (symbol_table); |
| 822 | if (!symbols_sorted) |
| 823 | { |
| 824 | symbols_sorted = xnmalloc (count, sizeof *symbols_sorted); |
| 825 | hash_get_entries (symbol_table, (void**)symbols_sorted, count); |
| 826 | qsort (symbols_sorted, count, sizeof *symbols_sorted, |
| 827 | symbols_cmp_qsort); |
| 828 | } |
| 829 | { |
| 830 | size_t i; |
| 831 | for (i = 0; i < count; ++i) |
| 832 | processor (symbols_sorted[i], processor_data); |
| 833 | } |
| 834 | } |
| 835 | |
| 836 | /*--------------------------------------------------------------. |
| 837 | | Check that all the symbols are defined. Report any undefined | |
| 838 | | symbols and consider them nonterminals. | |
| 839 | `--------------------------------------------------------------*/ |
| 840 | |
| 841 | void |
| 842 | symbols_check_defined (void) |
| 843 | { |
| 844 | symbols_do (symbol_check_defined_processor, NULL); |
| 845 | } |
| 846 | |
| 847 | /*------------------------------------------------------------------. |
| 848 | | Set TOKEN_TRANSLATIONS. Check that no two symbols share the same | |
| 849 | | number. | |
| 850 | `------------------------------------------------------------------*/ |
| 851 | |
| 852 | static void |
| 853 | symbols_token_translations_init (void) |
| 854 | { |
| 855 | bool num_256_available_p = true; |
| 856 | int i; |
| 857 | |
| 858 | /* Find the highest user token number, and whether 256, the POSIX |
| 859 | preferred user token number for the error token, is used. */ |
| 860 | max_user_token_number = 0; |
| 861 | for (i = 0; i < ntokens; ++i) |
| 862 | { |
| 863 | symbol *this = symbols[i]; |
| 864 | if (this->user_token_number != USER_NUMBER_UNDEFINED) |
| 865 | { |
| 866 | if (this->user_token_number > max_user_token_number) |
| 867 | max_user_token_number = this->user_token_number; |
| 868 | if (this->user_token_number == 256) |
| 869 | num_256_available_p = false; |
| 870 | } |
| 871 | } |
| 872 | |
| 873 | /* If 256 is not used, assign it to error, to follow POSIX. */ |
| 874 | if (num_256_available_p |
| 875 | && errtoken->user_token_number == USER_NUMBER_UNDEFINED) |
| 876 | errtoken->user_token_number = 256; |
| 877 | |
| 878 | /* Set the missing user numbers. */ |
| 879 | if (max_user_token_number < 256) |
| 880 | max_user_token_number = 256; |
| 881 | |
| 882 | for (i = 0; i < ntokens; ++i) |
| 883 | { |
| 884 | symbol *this = symbols[i]; |
| 885 | if (this->user_token_number == USER_NUMBER_UNDEFINED) |
| 886 | this->user_token_number = ++max_user_token_number; |
| 887 | if (this->user_token_number > max_user_token_number) |
| 888 | max_user_token_number = this->user_token_number; |
| 889 | } |
| 890 | |
| 891 | token_translations = xnmalloc (max_user_token_number + 1, |
| 892 | sizeof *token_translations); |
| 893 | |
| 894 | /* Initialize all entries for literal tokens to the internal token |
| 895 | number for $undefined, which represents all invalid inputs. */ |
| 896 | for (i = 0; i < max_user_token_number + 1; i++) |
| 897 | token_translations[i] = undeftoken->number; |
| 898 | symbols_do (symbol_translation_processor, NULL); |
| 899 | } |
| 900 | |
| 901 | |
| 902 | /*----------------------------------------------------------------. |
| 903 | | Assign symbol numbers, and write definition of token names into | |
| 904 | | FDEFINES. Set up vectors SYMBOL_TABLE, TAGS of symbols. | |
| 905 | `----------------------------------------------------------------*/ |
| 906 | |
| 907 | void |
| 908 | symbols_pack (void) |
| 909 | { |
| 910 | symbols_do (symbol_check_alias_consistency_processor, NULL); |
| 911 | |
| 912 | symbols = xcalloc (nsyms, sizeof *symbols); |
| 913 | symbols_do (symbol_pack_processor, NULL); |
| 914 | |
| 915 | /* Aliases leave empty slots in symbols, so remove them. */ |
| 916 | { |
| 917 | int writei; |
| 918 | int readi; |
| 919 | int nsyms_old = nsyms; |
| 920 | for (writei = 0, readi = 0; readi < nsyms_old; readi += 1) |
| 921 | { |
| 922 | if (symbols[readi] == NULL) |
| 923 | { |
| 924 | nsyms -= 1; |
| 925 | ntokens -= 1; |
| 926 | } |
| 927 | else |
| 928 | { |
| 929 | symbols[writei] = symbols[readi]; |
| 930 | symbols[writei]->number = writei; |
| 931 | if (symbols[writei]->alias) |
| 932 | symbols[writei]->alias->number = writei; |
| 933 | writei += 1; |
| 934 | } |
| 935 | } |
| 936 | } |
| 937 | symbols = xnrealloc (symbols, nsyms, sizeof *symbols); |
| 938 | |
| 939 | symbols_token_translations_init (); |
| 940 | |
| 941 | if (startsymbol->class == unknown_sym) |
| 942 | fatal_at (startsymbol_location, |
| 943 | _("the start symbol %s is undefined"), |
| 944 | startsymbol->tag); |
| 945 | else if (startsymbol->class == token_sym) |
| 946 | fatal_at (startsymbol_location, |
| 947 | _("the start symbol %s is a token"), |
| 948 | startsymbol->tag); |
| 949 | } |
| 950 | |
| 951 | |
| 952 | /*--------------------------------------------------. |
| 953 | | Set default tagged/tagless %destructor/%printer. | |
| 954 | `--------------------------------------------------*/ |
| 955 | |
| 956 | void |
| 957 | default_tagged_destructor_set (code_props const *destructor) |
| 958 | { |
| 959 | if (default_tagged_destructor.code) |
| 960 | { |
| 961 | complain_at (destructor->location, |
| 962 | _("redeclaration for default tagged %%destructor")); |
| 963 | complain_at (default_tagged_destructor.location, |
| 964 | _("previous declaration")); |
| 965 | } |
| 966 | default_tagged_destructor = *destructor; |
| 967 | } |
| 968 | |
| 969 | void |
| 970 | default_tagless_destructor_set (code_props const *destructor) |
| 971 | { |
| 972 | if (default_tagless_destructor.code) |
| 973 | { |
| 974 | complain_at (destructor->location, |
| 975 | _("redeclaration for default tagless %%destructor")); |
| 976 | complain_at (default_tagless_destructor.location, |
| 977 | _("previous declaration")); |
| 978 | } |
| 979 | default_tagless_destructor = *destructor; |
| 980 | } |
| 981 | |
| 982 | void |
| 983 | default_tagged_printer_set (code_props const *printer) |
| 984 | { |
| 985 | if (default_tagged_printer.code) |
| 986 | { |
| 987 | complain_at (printer->location, |
| 988 | _("redeclaration for default tagged %%printer")); |
| 989 | complain_at (default_tagged_printer.location, |
| 990 | _("previous declaration")); |
| 991 | } |
| 992 | default_tagged_printer = *printer; |
| 993 | } |
| 994 | |
| 995 | void |
| 996 | default_tagless_printer_set (code_props const *printer) |
| 997 | { |
| 998 | if (default_tagless_printer.code) |
| 999 | { |
| 1000 | complain_at (printer->location, |
| 1001 | _("redeclaration for default tagless %%printer")); |
| 1002 | complain_at (default_tagless_printer.location, |
| 1003 | _("previous declaration")); |
| 1004 | } |
| 1005 | default_tagless_printer = *printer; |
| 1006 | } |