]>
Commit | Line | Data |
---|---|---|
0fb1efaf PE |
1 | /* Symbol table manager for Bison. |
2 | ||
34136e65 | 3 | Copyright (C) 1984, 1989, 2000-2002, 2004-2012 Free Software |
575619af | 4 | Foundation, Inc. |
40675e7c | 5 | |
95e36146 | 6 | This file is part of Bison, the GNU Compiler Compiler. |
40675e7c | 7 | |
f16b0819 | 8 | This program is free software: you can redistribute it and/or modify |
95e36146 | 9 | it under the terms of the GNU General Public License as published by |
f16b0819 PE |
10 | the Free Software Foundation, either version 3 of the License, or |
11 | (at your option) any later version. | |
40675e7c | 12 | |
f16b0819 | 13 | This program is distributed in the hope that it will be useful, |
95e36146 AD |
14 | but WITHOUT ANY WARRANTY; without even the implied warranty of |
15 | MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the | |
16 | GNU General Public License for more details. | |
40675e7c | 17 | |
95e36146 | 18 | You should have received a copy of the GNU General Public License |
f16b0819 | 19 | along with this program. If not, see <http://www.gnu.org/licenses/>. */ |
40675e7c | 20 | |
2cec9080 | 21 | #include <config.h> |
40675e7c | 22 | #include "system.h" |
17ee7397 PE |
23 | |
24 | #include <hash.h> | |
17ee7397 | 25 | |
2f1afb73 | 26 | #include "complain.h" |
40675e7c | 27 | #include "gram.h" |
17ee7397 | 28 | #include "symtab.h" |
40675e7c | 29 | |
83b60c97 JD |
30 | /*-------------------------------------------------------------------. |
31 | | Symbols sorted by tag. Allocated by the first invocation of | | |
32 | | symbols_do, after which no more symbols should be created. | | |
33 | `-------------------------------------------------------------------*/ | |
34 | ||
35 | static symbol **symbols_sorted = NULL; | |
36 | ||
2f1afb73 AD |
37 | /*------------------------. |
38 | | Distinguished symbols. | | |
39 | `------------------------*/ | |
40 | ||
17ee7397 PE |
41 | symbol *errtoken = NULL; |
42 | symbol *undeftoken = NULL; | |
43 | symbol *endtoken = NULL; | |
44 | symbol *accept = NULL; | |
45 | symbol *startsymbol = NULL; | |
46 | location startsymbol_location; | |
2f1afb73 | 47 | |
12e35840 JD |
48 | /*---------------------------------------. |
49 | | Default %destructor's and %printer's. | | |
50 | `---------------------------------------*/ | |
51 | ||
95021767 JD |
52 | static code_props default_tagged_destructor = CODE_PROPS_NONE_INIT; |
53 | static code_props default_tagless_destructor = CODE_PROPS_NONE_INIT; | |
54 | static code_props default_tagged_printer = CODE_PROPS_NONE_INIT; | |
55 | static code_props default_tagless_printer = CODE_PROPS_NONE_INIT; | |
ec5479ce | 56 | |
72a23c97 AD |
57 | /*---------------------------------. |
58 | | Create a new symbol, named TAG. | | |
59 | `---------------------------------*/ | |
1e9798d5 | 60 | |
17ee7397 PE |
61 | static symbol * |
62 | symbol_new (uniqstr tag, location loc) | |
1e9798d5 | 63 | { |
da2a7671 | 64 | symbol *res = xmalloc (sizeof *res); |
1e9798d5 | 65 | |
17ee7397 | 66 | uniqstr_assert (tag); |
4f646c37 AD |
67 | |
68 | /* If the tag is not a string (starts with a double quote), check | |
69 | that it is valid for Yacc. */ | |
84526bf3 | 70 | if (tag[0] != '\"' && tag[0] != '\'' && strchr (tag, '-')) |
4f646c37 AD |
71 | yacc_at (loc, _("POSIX Yacc forbids dashes in symbol names: %s"), |
72 | tag); | |
73 | ||
95612cfa | 74 | res->tag = tag; |
17ee7397 | 75 | res->location = loc; |
24c0aad7 | 76 | |
1e9798d5 | 77 | res->type_name = NULL; |
95021767 JD |
78 | code_props_none_init (&res->destructor); |
79 | code_props_none_init (&res->printer); | |
24c0aad7 | 80 | |
5fbb0954 | 81 | res->number = NUMBER_UNDEFINED; |
1e9798d5 | 82 | res->prec = 0; |
a945ec39 | 83 | res->assoc = undef_assoc; |
b87f8b21 | 84 | res->user_token_number = USER_NUMBER_UNDEFINED; |
260008e5 | 85 | |
1e9798d5 AD |
86 | res->alias = NULL; |
87 | res->class = unknown_sym; | |
073f9288 | 88 | res->declared = false; |
1e9798d5 | 89 | |
3239db74 PE |
90 | if (nsyms == SYMBOL_NUMBER_MAXIMUM) |
91 | fatal (_("too many symbols in input grammar (limit is %d)"), | |
e9690142 | 92 | SYMBOL_NUMBER_MAXIMUM); |
6d0ef4ec | 93 | nsyms++; |
1e9798d5 AD |
94 | return res; |
95 | } | |
96 | ||
b2a0b7ca JD |
97 | /*----------------------------------------. |
98 | | Create a new semantic type, named TAG. | | |
99 | `----------------------------------------*/ | |
100 | ||
101 | static semantic_type * | |
102 | semantic_type_new (uniqstr tag) | |
103 | { | |
104 | semantic_type *res = xmalloc (sizeof *res); | |
105 | ||
106 | uniqstr_assert (tag); | |
107 | res->tag = tag; | |
95021767 JD |
108 | code_props_none_init (&res->destructor); |
109 | code_props_none_init (&res->printer); | |
b2a0b7ca JD |
110 | |
111 | return res; | |
112 | } | |
113 | ||
40675e7c | 114 | |
867a3e00 AD |
115 | /*-----------------. |
116 | | Print a symbol. | | |
117 | `-----------------*/ | |
118 | ||
e9690142 JD |
119 | #define SYMBOL_ATTR_PRINT(Attr) \ |
120 | if (s->Attr) \ | |
ab703f2c | 121 | fprintf (f, " %s { %s }", #Attr, s->Attr) |
867a3e00 | 122 | |
95021767 JD |
123 | #define SYMBOL_CODE_PRINT(Attr) \ |
124 | if (s->Attr.code) \ | |
125 | fprintf (f, " %s { %s }", #Attr, s->Attr.code) | |
126 | ||
867a3e00 | 127 | void |
22dda0f0 | 128 | symbol_print (symbol *s, FILE *f) |
867a3e00 | 129 | { |
affac613 AD |
130 | if (s) |
131 | { | |
132 | fprintf (f, "\"%s\"", s->tag); | |
133 | SYMBOL_ATTR_PRINT (type_name); | |
95021767 JD |
134 | SYMBOL_CODE_PRINT (destructor); |
135 | SYMBOL_CODE_PRINT (printer); | |
affac613 AD |
136 | } |
137 | else | |
138 | fprintf (f, "<NULL>"); | |
867a3e00 AD |
139 | } |
140 | ||
141 | #undef SYMBOL_ATTR_PRINT | |
95021767 | 142 | #undef SYMBOL_CODE_PRINT |
867a3e00 | 143 | |
aea10ef4 AD |
144 | |
145 | /*----------------------------------. | |
146 | | Whether S is a valid identifier. | | |
147 | `----------------------------------*/ | |
148 | ||
149 | static bool | |
150 | is_identifier (uniqstr s) | |
151 | { | |
152 | static char const alphanum[26 + 26 + 1 + 10] = | |
153 | "abcdefghijklmnopqrstuvwxyz" | |
154 | "ABCDEFGHIJKLMNOPQRSTUVWXYZ" | |
155 | "_" | |
156 | "0123456789"; | |
157 | if (!s || ! memchr (alphanum, *s, sizeof alphanum - 10)) | |
158 | return false; | |
159 | for (++s; *s; ++s) | |
160 | if (! memchr (alphanum, *s, sizeof alphanum)) | |
161 | return false; | |
162 | return true; | |
163 | } | |
164 | ||
165 | ||
166 | /*-----------------------------------------------. | |
167 | | Get the identifier associated to this symbol. | | |
168 | `-----------------------------------------------*/ | |
169 | uniqstr | |
170 | symbol_id_get (symbol const *sym) | |
171 | { | |
dfaa4860 | 172 | aver (sym->user_token_number != USER_NUMBER_HAS_STRING_ALIAS); |
aea10ef4 AD |
173 | if (sym->alias) |
174 | sym = sym->alias; | |
175 | return is_identifier (sym->tag) ? sym->tag : 0; | |
176 | } | |
177 | ||
178 | ||
df09ef2e AD |
179 | /*------------------------------------------------------------------. |
180 | | Complain that S's WHAT is redeclared at SECOND, and was first set | | |
181 | | at FIRST. | | |
182 | `------------------------------------------------------------------*/ | |
183 | ||
184 | static void | |
b2a0b7ca JD |
185 | symbol_redeclaration (symbol *s, const char *what, location first, |
186 | location second) | |
df09ef2e AD |
187 | { |
188 | complain_at (second, _("%s redeclaration for %s"), what, s->tag); | |
dd60572a | 189 | complain_at (first, _("previous declaration")); |
df09ef2e AD |
190 | } |
191 | ||
b2a0b7ca JD |
192 | static void |
193 | semantic_type_redeclaration (semantic_type *s, const char *what, location first, | |
194 | location second) | |
195 | { | |
196 | complain_at (second, _("%s redeclaration for <%s>"), what, s->tag); | |
197 | complain_at (first, _("previous declaration")); | |
198 | } | |
199 | ||
df09ef2e | 200 | |
12cf133f | 201 | |
17ee7397 PE |
202 | /*-----------------------------------------------------------------. |
203 | | Set the TYPE_NAME associated with SYM. Does nothing if passed 0 | | |
204 | | as TYPE_NAME. | | |
205 | `-----------------------------------------------------------------*/ | |
3ae2b51f AD |
206 | |
207 | void | |
17ee7397 | 208 | symbol_type_set (symbol *sym, uniqstr type_name, location loc) |
3ae2b51f | 209 | { |
e9955c83 AD |
210 | if (type_name) |
211 | { | |
17ee7397 | 212 | if (sym->type_name) |
e9690142 | 213 | symbol_redeclaration (sym, "%type", sym->type_location, loc); |
17ee7397 PE |
214 | uniqstr_assert (type_name); |
215 | sym->type_name = type_name; | |
df09ef2e | 216 | sym->type_location = loc; |
e9955c83 | 217 | } |
3ae2b51f AD |
218 | } |
219 | ||
95021767 JD |
220 | /*-----------------------------------------. |
221 | | Set the DESTRUCTOR associated with SYM. | | |
222 | `-----------------------------------------*/ | |
9280d3ef AD |
223 | |
224 | void | |
95021767 | 225 | symbol_destructor_set (symbol *sym, code_props const *destructor) |
9280d3ef | 226 | { |
95021767 JD |
227 | if (sym->destructor.code) |
228 | symbol_redeclaration (sym, "%destructor", sym->destructor.location, | |
229 | destructor->location); | |
230 | sym->destructor = *destructor; | |
9280d3ef AD |
231 | } |
232 | ||
95021767 JD |
233 | /*------------------------------------------. |
234 | | Set the DESTRUCTOR associated with TYPE. | | |
235 | `------------------------------------------*/ | |
b2a0b7ca JD |
236 | |
237 | void | |
95021767 JD |
238 | semantic_type_destructor_set (semantic_type *type, |
239 | code_props const *destructor) | |
b2a0b7ca | 240 | { |
95021767 JD |
241 | if (type->destructor.code) |
242 | semantic_type_redeclaration (type, "%destructor", | |
243 | type->destructor.location, | |
244 | destructor->location); | |
245 | type->destructor = *destructor; | |
b2a0b7ca JD |
246 | } |
247 | ||
ec5479ce JD |
248 | /*---------------------------------------. |
249 | | Get the computed %destructor for SYM. | | |
250 | `---------------------------------------*/ | |
251 | ||
95021767 JD |
252 | code_props const * |
253 | symbol_destructor_get (symbol const *sym) | |
ec5479ce | 254 | { |
9350499c | 255 | /* Per-symbol %destructor. */ |
95021767 JD |
256 | if (sym->destructor.code) |
257 | return &sym->destructor; | |
9350499c | 258 | |
b2a0b7ca JD |
259 | /* Per-type %destructor. */ |
260 | if (sym->type_name) | |
261 | { | |
95021767 JD |
262 | code_props const *destructor = |
263 | &semantic_type_get (sym->type_name)->destructor; | |
264 | if (destructor->code) | |
265 | return destructor; | |
b2a0b7ca JD |
266 | } |
267 | ||
12e35840 | 268 | /* Apply default %destructor's only to user-defined symbols. */ |
3508ce36 | 269 | if (sym->tag[0] == '$' || sym == errtoken) |
95021767 | 270 | return &code_props_none; |
12e35840 JD |
271 | |
272 | if (sym->type_name) | |
95021767 JD |
273 | return &default_tagged_destructor; |
274 | return &default_tagless_destructor; | |
ec5479ce JD |
275 | } |
276 | ||
95021767 JD |
277 | /*--------------------------------------. |
278 | | Set the PRINTER associated with SYM. | | |
279 | `--------------------------------------*/ | |
366eea36 AD |
280 | |
281 | void | |
95021767 | 282 | symbol_printer_set (symbol *sym, code_props const *printer) |
366eea36 | 283 | { |
95021767 JD |
284 | if (sym->printer.code) |
285 | symbol_redeclaration (sym, "%printer", | |
286 | sym->printer.location, printer->location); | |
287 | sym->printer = *printer; | |
366eea36 AD |
288 | } |
289 | ||
95021767 JD |
290 | /*---------------------------------------. |
291 | | Set the PRINTER associated with TYPE. | | |
292 | `---------------------------------------*/ | |
b2a0b7ca JD |
293 | |
294 | void | |
95021767 | 295 | semantic_type_printer_set (semantic_type *type, code_props const *printer) |
b2a0b7ca | 296 | { |
95021767 JD |
297 | if (type->printer.code) |
298 | semantic_type_redeclaration (type, "%printer", | |
299 | type->printer.location, printer->location); | |
300 | type->printer = *printer; | |
b2a0b7ca JD |
301 | } |
302 | ||
ec5479ce JD |
303 | /*------------------------------------. |
304 | | Get the computed %printer for SYM. | | |
305 | `------------------------------------*/ | |
306 | ||
95021767 JD |
307 | code_props const * |
308 | symbol_printer_get (symbol const *sym) | |
ec5479ce | 309 | { |
9350499c | 310 | /* Per-symbol %printer. */ |
95021767 JD |
311 | if (sym->printer.code) |
312 | return &sym->printer; | |
9350499c | 313 | |
b2a0b7ca JD |
314 | /* Per-type %printer. */ |
315 | if (sym->type_name) | |
316 | { | |
95021767 JD |
317 | code_props const *printer = &semantic_type_get (sym->type_name)->printer; |
318 | if (printer->code) | |
319 | return printer; | |
b2a0b7ca JD |
320 | } |
321 | ||
9350499c | 322 | /* Apply the default %printer only to user-defined symbols. */ |
3508ce36 | 323 | if (sym->tag[0] == '$' || sym == errtoken) |
95021767 | 324 | return &code_props_none; |
12e35840 JD |
325 | |
326 | if (sym->type_name) | |
95021767 JD |
327 | return &default_tagged_printer; |
328 | return &default_tagless_printer; | |
ec5479ce JD |
329 | } |
330 | ||
17ee7397 PE |
331 | /*-----------------------------------------------------------------. |
332 | | Set the PRECEDENCE associated with SYM. Does nothing if invoked | | |
333 | | with UNDEF_ASSOC as ASSOC. | | |
334 | `-----------------------------------------------------------------*/ | |
3ae2b51f AD |
335 | |
336 | void | |
17ee7397 | 337 | symbol_precedence_set (symbol *sym, int prec, assoc a, location loc) |
3ae2b51f | 338 | { |
17ee7397 | 339 | if (a != undef_assoc) |
e9955c83 | 340 | { |
17ee7397 | 341 | if (sym->prec != 0) |
e9690142 | 342 | symbol_redeclaration (sym, assoc_to_string (a), sym->prec_location, |
b2a0b7ca | 343 | loc); |
17ee7397 PE |
344 | sym->prec = prec; |
345 | sym->assoc = a; | |
df09ef2e | 346 | sym->prec_location = loc; |
e9955c83 AD |
347 | } |
348 | ||
349 | /* Only terminals have a precedence. */ | |
073f9288 | 350 | symbol_class_set (sym, token_sym, loc, false); |
3ae2b51f AD |
351 | } |
352 | ||
353 | ||
17ee7397 PE |
354 | /*------------------------------------. |
355 | | Set the CLASS associated with SYM. | | |
356 | `------------------------------------*/ | |
44536b35 AD |
357 | |
358 | void | |
073f9288 | 359 | symbol_class_set (symbol *sym, symbol_class class, location loc, bool declaring) |
44536b35 | 360 | { |
17ee7397 | 361 | if (sym->class != unknown_sym && sym->class != class) |
073f9288 PE |
362 | { |
363 | complain_at (loc, _("symbol %s redefined"), sym->tag); | |
364 | sym->declared = false; | |
365 | } | |
44536b35 | 366 | |
17ee7397 PE |
367 | if (class == nterm_sym && sym->class != nterm_sym) |
368 | sym->number = nvars++; | |
6d0ef4ec | 369 | else if (class == token_sym && sym->number == NUMBER_UNDEFINED) |
17ee7397 | 370 | sym->number = ntokens++; |
44536b35 | 371 | |
17ee7397 | 372 | sym->class = class; |
073f9288 PE |
373 | |
374 | if (declaring) | |
375 | { | |
376 | if (sym->declared) | |
e9690142 | 377 | warn_at (loc, _("symbol %s redeclared"), sym->tag); |
073f9288 PE |
378 | sym->declared = true; |
379 | } | |
44536b35 AD |
380 | } |
381 | ||
382 | ||
17ee7397 PE |
383 | /*------------------------------------------------. |
384 | | Set the USER_TOKEN_NUMBER associated with SYM. | | |
385 | `------------------------------------------------*/ | |
44536b35 AD |
386 | |
387 | void | |
17ee7397 | 388 | symbol_user_token_number_set (symbol *sym, int user_token_number, location loc) |
44536b35 | 389 | { |
1f6b3679 JD |
390 | int *user_token_numberp; |
391 | ||
dfaa4860 | 392 | if (sym->user_token_number != USER_NUMBER_HAS_STRING_ALIAS) |
1f6b3679 JD |
393 | user_token_numberp = &sym->user_token_number; |
394 | else | |
395 | user_token_numberp = &sym->alias->user_token_number; | |
396 | if (*user_token_numberp != USER_NUMBER_UNDEFINED | |
397 | && *user_token_numberp != user_token_number) | |
17ee7397 | 398 | complain_at (loc, _("redefining user token number of %s"), sym->tag); |
44536b35 | 399 | |
1f6b3679 | 400 | *user_token_numberp = user_token_number; |
88bce5a2 | 401 | /* User defined $end token? */ |
44536b35 AD |
402 | if (user_token_number == 0) |
403 | { | |
17ee7397 | 404 | endtoken = sym; |
44536b35 | 405 | /* It is always mapped to 0, so it was already counted in |
e9690142 | 406 | NTOKENS. */ |
1f36f544 JD |
407 | if (endtoken->number != NUMBER_UNDEFINED) |
408 | --ntokens; | |
409 | endtoken->number = 0; | |
44536b35 AD |
410 | } |
411 | } | |
412 | ||
413 | ||
17ee7397 PE |
414 | /*----------------------------------------------------------. |
415 | | If SYM is not defined, report an error, and consider it a | | |
416 | | nonterminal. | | |
417 | `----------------------------------------------------------*/ | |
2f1afb73 | 418 | |
0fb1efaf | 419 | static inline bool |
17ee7397 | 420 | symbol_check_defined (symbol *sym) |
2f1afb73 | 421 | { |
17ee7397 | 422 | if (sym->class == unknown_sym) |
2f1afb73 | 423 | { |
ee000ba4 | 424 | complain_at |
e9690142 JD |
425 | (sym->location, |
426 | _("symbol %s is used, but is not defined as a token and has no rules"), | |
427 | sym->tag); | |
17ee7397 PE |
428 | sym->class = nterm_sym; |
429 | sym->number = nvars++; | |
2f1afb73 AD |
430 | } |
431 | ||
a3714bce | 432 | return true; |
2f1afb73 AD |
433 | } |
434 | ||
0fb1efaf PE |
435 | static bool |
436 | symbol_check_defined_processor (void *sym, void *null ATTRIBUTE_UNUSED) | |
437 | { | |
438 | return symbol_check_defined (sym); | |
439 | } | |
440 | ||
2f1afb73 | 441 | |
2f1afb73 | 442 | void |
dfaa4860 | 443 | symbol_make_alias (symbol *sym, symbol *str, location loc) |
2f1afb73 | 444 | { |
dfaa4860 | 445 | if (str->alias) |
4a9cd8f2 | 446 | warn_at (loc, _("symbol %s used more than once as a literal string"), |
e9690142 | 447 | str->tag); |
17ee7397 | 448 | else if (sym->alias) |
4a9cd8f2 | 449 | warn_at (loc, _("symbol %s given more than one literal string"), |
e9690142 | 450 | sym->tag); |
2f1afb73 AD |
451 | else |
452 | { | |
dfaa4860 JD |
453 | str->class = token_sym; |
454 | str->user_token_number = sym->user_token_number; | |
455 | sym->user_token_number = USER_NUMBER_HAS_STRING_ALIAS; | |
456 | str->alias = sym; | |
457 | sym->alias = str; | |
458 | str->number = sym->number; | |
459 | symbol_type_set (str, sym->type_name, loc); | |
2f1afb73 AD |
460 | } |
461 | } | |
462 | ||
463 | ||
464 | /*---------------------------------------------------------. | |
465 | | Check that THIS, and its alias, have same precedence and | | |
466 | | associativity. | | |
467 | `---------------------------------------------------------*/ | |
468 | ||
df09ef2e | 469 | static inline void |
0fb1efaf | 470 | symbol_check_alias_consistency (symbol *this) |
2f1afb73 | 471 | { |
dfaa4860 | 472 | symbol *sym = this; |
b143f404 | 473 | symbol *str = this->alias; |
df09ef2e | 474 | |
dfaa4860 JD |
475 | /* Check only the symbol in the symbol-string pair. */ |
476 | if (!(this->alias | |
477 | && this->user_token_number == USER_NUMBER_HAS_STRING_ALIAS)) | |
df09ef2e AD |
478 | return; |
479 | ||
dfaa4860 | 480 | if (str->type_name != sym->type_name) |
2f1afb73 | 481 | { |
dfaa4860 | 482 | if (str->type_name) |
e9690142 | 483 | symbol_type_set (sym, str->type_name, str->type_location); |
df09ef2e | 484 | else |
e9690142 | 485 | symbol_type_set (str, sym->type_name, sym->type_location); |
df09ef2e | 486 | } |
2f1afb73 | 487 | |
df09ef2e | 488 | |
dfaa4860 | 489 | if (str->destructor.code || sym->destructor.code) |
df09ef2e | 490 | { |
dfaa4860 | 491 | if (str->destructor.code) |
e9690142 | 492 | symbol_destructor_set (sym, &str->destructor); |
df09ef2e | 493 | else |
e9690142 | 494 | symbol_destructor_set (str, &sym->destructor); |
df09ef2e AD |
495 | } |
496 | ||
dfaa4860 | 497 | if (str->printer.code || sym->printer.code) |
df09ef2e | 498 | { |
dfaa4860 | 499 | if (str->printer.code) |
e9690142 | 500 | symbol_printer_set (sym, &str->printer); |
df09ef2e | 501 | else |
e9690142 | 502 | symbol_printer_set (str, &sym->printer); |
df09ef2e AD |
503 | } |
504 | ||
dfaa4860 | 505 | if (sym->prec || str->prec) |
df09ef2e | 506 | { |
dfaa4860 | 507 | if (str->prec) |
e9690142 JD |
508 | symbol_precedence_set (sym, str->prec, str->assoc, |
509 | str->prec_location); | |
df09ef2e | 510 | else |
e9690142 JD |
511 | symbol_precedence_set (str, sym->prec, sym->assoc, |
512 | sym->prec_location); | |
2f1afb73 | 513 | } |
2f1afb73 AD |
514 | } |
515 | ||
0fb1efaf PE |
516 | static bool |
517 | symbol_check_alias_consistency_processor (void *this, | |
e9690142 | 518 | void *null ATTRIBUTE_UNUSED) |
0fb1efaf | 519 | { |
df09ef2e AD |
520 | symbol_check_alias_consistency (this); |
521 | return true; | |
0fb1efaf PE |
522 | } |
523 | ||
2f1afb73 AD |
524 | |
525 | /*-------------------------------------------------------------------. | |
526 | | Assign a symbol number, and write the definition of the token name | | |
527 | | into FDEFINES. Put in SYMBOLS. | | |
528 | `-------------------------------------------------------------------*/ | |
529 | ||
0fb1efaf | 530 | static inline bool |
17ee7397 | 531 | symbol_pack (symbol *this) |
2f1afb73 | 532 | { |
f74d6d25 | 533 | aver (this->number != NUMBER_UNDEFINED); |
2f1afb73 | 534 | if (this->class == nterm_sym) |
f74d6d25 JD |
535 | this->number += ntokens; |
536 | else if (this->user_token_number == USER_NUMBER_HAS_STRING_ALIAS) | |
537 | return true; | |
2f1afb73 AD |
538 | |
539 | symbols[this->number] = this; | |
a3714bce | 540 | return true; |
2f1afb73 AD |
541 | } |
542 | ||
0fb1efaf PE |
543 | static bool |
544 | symbol_pack_processor (void *this, void *null ATTRIBUTE_UNUSED) | |
545 | { | |
546 | return symbol_pack (this); | |
547 | } | |
548 | ||
2f1afb73 | 549 | |
12cf133f AD |
550 | static void |
551 | user_token_number_redeclaration (int num, symbol *first, symbol *second) | |
552 | { | |
553 | /* User token numbers are not assigned during the parsing, but in a | |
83b60c97 | 554 | second step, via a traversal of the symbol table sorted on tag. |
2f1afb73 | 555 | |
83b60c97 JD |
556 | However, error messages make more sense if we keep the first |
557 | declaration first. */ | |
12cf133f AD |
558 | if (location_cmp (first->location, second->location) > 0) |
559 | { | |
560 | symbol* tmp = first; | |
561 | first = second; | |
562 | second = tmp; | |
563 | } | |
564 | complain_at (second->location, | |
565 | _("user token number %d redeclaration for %s"), | |
566 | num, second->tag); | |
567 | complain_at (first->location, _("previous declaration for %s"), | |
568 | first->tag); | |
569 | } | |
2f1afb73 AD |
570 | |
571 | /*--------------------------------------------------. | |
572 | | Put THIS in TOKEN_TRANSLATIONS if it is a token. | | |
573 | `--------------------------------------------------*/ | |
574 | ||
0fb1efaf | 575 | static inline bool |
17ee7397 | 576 | symbol_translation (symbol *this) |
2f1afb73 AD |
577 | { |
578 | /* Non-terminal? */ | |
579 | if (this->class == token_sym | |
dfaa4860 | 580 | && this->user_token_number != USER_NUMBER_HAS_STRING_ALIAS) |
2f1afb73 AD |
581 | { |
582 | /* A token which translation has already been set? */ | |
583 | if (token_translations[this->user_token_number] != undeftoken->number) | |
e9690142 | 584 | user_token_number_redeclaration |
12cf133f AD |
585 | (this->user_token_number, |
586 | symbols[token_translations[this->user_token_number]], | |
587 | this); | |
2f1afb73 AD |
588 | |
589 | token_translations[this->user_token_number] = this->number; | |
590 | } | |
591 | ||
a3714bce | 592 | return true; |
2f1afb73 AD |
593 | } |
594 | ||
0fb1efaf PE |
595 | static bool |
596 | symbol_translation_processor (void *this, void *null ATTRIBUTE_UNUSED) | |
597 | { | |
598 | return symbol_translation (this); | |
599 | } | |
600 | ||
72a23c97 | 601 | |
b2a0b7ca JD |
602 | /*---------------------------------------. |
603 | | Symbol and semantic type hash tables. | | |
604 | `---------------------------------------*/ | |
40675e7c | 605 | |
b2a0b7ca | 606 | /* Initial capacity of symbol and semantic type hash table. */ |
72a23c97 AD |
607 | #define HT_INITIAL_CAPACITY 257 |
608 | ||
db8837cb | 609 | static struct hash_table *symbol_table = NULL; |
b2a0b7ca | 610 | static struct hash_table *semantic_type_table = NULL; |
72a23c97 | 611 | |
0fb1efaf | 612 | static inline bool |
17ee7397 | 613 | hash_compare_symbol (const symbol *m1, const symbol *m2) |
72a23c97 | 614 | { |
95612cfa | 615 | /* Since tags are unique, we can compare the pointers themselves. */ |
17ee7397 | 616 | return UNIQSTR_EQ (m1->tag, m2->tag); |
72a23c97 AD |
617 | } |
618 | ||
b2a0b7ca JD |
619 | static inline bool |
620 | hash_compare_semantic_type (const semantic_type *m1, const semantic_type *m2) | |
621 | { | |
622 | /* Since names are unique, we can compare the pointers themselves. */ | |
623 | return UNIQSTR_EQ (m1->tag, m2->tag); | |
624 | } | |
625 | ||
0fb1efaf PE |
626 | static bool |
627 | hash_symbol_comparator (void const *m1, void const *m2) | |
628 | { | |
629 | return hash_compare_symbol (m1, m2); | |
630 | } | |
631 | ||
b2a0b7ca JD |
632 | static bool |
633 | hash_semantic_type_comparator (void const *m1, void const *m2) | |
634 | { | |
635 | return hash_compare_semantic_type (m1, m2); | |
636 | } | |
637 | ||
233a88ad PE |
638 | static inline size_t |
639 | hash_symbol (const symbol *m, size_t tablesize) | |
72a23c97 | 640 | { |
95612cfa | 641 | /* Since tags are unique, we can hash the pointer itself. */ |
0fb1efaf PE |
642 | return ((uintptr_t) m->tag) % tablesize; |
643 | } | |
644 | ||
b2a0b7ca JD |
645 | static inline size_t |
646 | hash_semantic_type (const semantic_type *m, size_t tablesize) | |
647 | { | |
648 | /* Since names are unique, we can hash the pointer itself. */ | |
649 | return ((uintptr_t) m->tag) % tablesize; | |
650 | } | |
651 | ||
233a88ad PE |
652 | static size_t |
653 | hash_symbol_hasher (void const *m, size_t tablesize) | |
0fb1efaf PE |
654 | { |
655 | return hash_symbol (m, tablesize); | |
72a23c97 AD |
656 | } |
657 | ||
b2a0b7ca JD |
658 | static size_t |
659 | hash_semantic_type_hasher (void const *m, size_t tablesize) | |
660 | { | |
661 | return hash_semantic_type (m, tablesize); | |
662 | } | |
72a23c97 AD |
663 | |
664 | /*-------------------------------. | |
2f1afb73 | 665 | | Create the symbol hash table. | |
72a23c97 AD |
666 | `-------------------------------*/ |
667 | ||
668 | void | |
db8837cb | 669 | symbols_new (void) |
72a23c97 | 670 | { |
db8837cb | 671 | symbol_table = hash_initialize (HT_INITIAL_CAPACITY, |
e9690142 JD |
672 | NULL, |
673 | hash_symbol_hasher, | |
674 | hash_symbol_comparator, | |
675 | free); | |
b2a0b7ca | 676 | semantic_type_table = hash_initialize (HT_INITIAL_CAPACITY, |
e9690142 JD |
677 | NULL, |
678 | hash_semantic_type_hasher, | |
679 | hash_semantic_type_comparator, | |
680 | free); | |
40675e7c DM |
681 | } |
682 | ||
683 | ||
1e9798d5 AD |
684 | /*----------------------------------------------------------------. |
685 | | Find the symbol named KEY, and return it. If it does not exist | | |
686 | | yet, create it. | | |
687 | `----------------------------------------------------------------*/ | |
688 | ||
17ee7397 | 689 | symbol * |
203b9274 | 690 | symbol_from_uniqstr (const uniqstr key, location loc) |
40675e7c | 691 | { |
17ee7397 PE |
692 | symbol probe; |
693 | symbol *entry; | |
40675e7c | 694 | |
0fb1efaf | 695 | probe.tag = key; |
db8837cb | 696 | entry = hash_lookup (symbol_table, &probe); |
40675e7c | 697 | |
72a23c97 | 698 | if (!entry) |
40675e7c | 699 | { |
72a23c97 | 700 | /* First insertion in the hash. */ |
83b60c97 | 701 | aver (!symbols_sorted); |
17ee7397 | 702 | entry = symbol_new (key, loc); |
92d7a23a AD |
703 | if (!hash_insert (symbol_table, entry)) |
704 | xalloc_die (); | |
40675e7c | 705 | } |
72a23c97 AD |
706 | return entry; |
707 | } | |
40675e7c | 708 | |
40675e7c | 709 | |
b2a0b7ca JD |
710 | /*-----------------------------------------------------------------------. |
711 | | Find the semantic type named KEY, and return it. If it does not exist | | |
712 | | yet, create it. | | |
713 | `-----------------------------------------------------------------------*/ | |
714 | ||
715 | semantic_type * | |
716 | semantic_type_from_uniqstr (const uniqstr key) | |
717 | { | |
718 | semantic_type probe; | |
719 | semantic_type *entry; | |
720 | ||
721 | probe.tag = key; | |
722 | entry = hash_lookup (semantic_type_table, &probe); | |
723 | ||
724 | if (!entry) | |
725 | { | |
726 | /* First insertion in the hash. */ | |
727 | entry = semantic_type_new (key); | |
92d7a23a AD |
728 | if (!hash_insert (semantic_type_table, entry)) |
729 | xalloc_die (); | |
b2a0b7ca JD |
730 | } |
731 | return entry; | |
732 | } | |
733 | ||
734 | ||
203b9274 AD |
735 | /*----------------------------------------------------------------. |
736 | | Find the symbol named KEY, and return it. If it does not exist | | |
737 | | yet, create it. | | |
738 | `----------------------------------------------------------------*/ | |
739 | ||
740 | symbol * | |
741 | symbol_get (const char *key, location loc) | |
742 | { | |
743 | return symbol_from_uniqstr (uniqstr_new (key), loc); | |
744 | } | |
745 | ||
746 | ||
b2a0b7ca JD |
747 | /*-----------------------------------------------------------------------. |
748 | | Find the semantic type named KEY, and return it. If it does not exist | | |
749 | | yet, create it. | | |
750 | `-----------------------------------------------------------------------*/ | |
751 | ||
752 | semantic_type * | |
753 | semantic_type_get (const char *key) | |
754 | { | |
755 | return semantic_type_from_uniqstr (uniqstr_new (key)); | |
756 | } | |
757 | ||
758 | ||
39f41916 AD |
759 | /*------------------------------------------------------------------. |
760 | | Generate a dummy nonterminal, whose name cannot conflict with the | | |
761 | | user's names. | | |
762 | `------------------------------------------------------------------*/ | |
763 | ||
17ee7397 PE |
764 | symbol * |
765 | dummy_symbol_get (location loc) | |
39f41916 AD |
766 | { |
767 | /* Incremented for each generated symbol. */ | |
768 | static int dummy_count = 0; | |
769 | static char buf[256]; | |
770 | ||
17ee7397 | 771 | symbol *sym; |
39f41916 | 772 | |
f91b1629 | 773 | sprintf (buf, "$@%d", ++dummy_count); |
17ee7397 | 774 | sym = symbol_get (buf, loc); |
39f41916 AD |
775 | sym->class = nterm_sym; |
776 | sym->number = nvars++; | |
777 | return sym; | |
778 | } | |
779 | ||
4d7370cb JD |
780 | bool |
781 | symbol_is_dummy (const symbol *sym) | |
782 | { | |
f91b1629 | 783 | return sym->tag[0] == '@' || (sym->tag[0] == '$' && sym->tag[1] == '@'); |
4d7370cb | 784 | } |
39f41916 | 785 | |
72a23c97 | 786 | /*-------------------. |
db8837cb | 787 | | Free the symbols. | |
72a23c97 AD |
788 | `-------------------*/ |
789 | ||
790 | void | |
db8837cb | 791 | symbols_free (void) |
72a23c97 | 792 | { |
db8837cb | 793 | hash_free (symbol_table); |
b2a0b7ca | 794 | hash_free (semantic_type_table); |
536545f3 | 795 | free (symbols); |
83b60c97 | 796 | free (symbols_sorted); |
40675e7c DM |
797 | } |
798 | ||
799 | ||
72a23c97 | 800 | /*---------------------------------------------------------------. |
db8837cb | 801 | | Look for undefined symbols, report an error, and consider them | |
72a23c97 AD |
802 | | terminals. | |
803 | `---------------------------------------------------------------*/ | |
804 | ||
83b60c97 JD |
805 | static int |
806 | symbols_cmp (symbol const *a, symbol const *b) | |
807 | { | |
808 | return strcmp (a->tag, b->tag); | |
809 | } | |
810 | ||
811 | static int | |
812 | symbols_cmp_qsort (void const *a, void const *b) | |
813 | { | |
814 | return symbols_cmp (*(symbol * const *)a, *(symbol * const *)b); | |
815 | } | |
816 | ||
0fb1efaf PE |
817 | static void |
818 | symbols_do (Hash_processor processor, void *processor_data) | |
40675e7c | 819 | { |
83b60c97 JD |
820 | size_t count = hash_get_n_entries (symbol_table); |
821 | if (!symbols_sorted) | |
822 | { | |
823 | symbols_sorted = xnmalloc (count, sizeof *symbols_sorted); | |
824 | hash_get_entries (symbol_table, (void**)symbols_sorted, count); | |
825 | qsort (symbols_sorted, count, sizeof *symbols_sorted, | |
826 | symbols_cmp_qsort); | |
827 | } | |
828 | { | |
829 | size_t i; | |
830 | for (i = 0; i < count; ++i) | |
831 | processor (symbols_sorted[i], processor_data); | |
832 | } | |
40675e7c | 833 | } |
2f1afb73 | 834 | |
2f1afb73 AD |
835 | /*--------------------------------------------------------------. |
836 | | Check that all the symbols are defined. Report any undefined | | |
837 | | symbols and consider them nonterminals. | | |
838 | `--------------------------------------------------------------*/ | |
839 | ||
840 | void | |
841 | symbols_check_defined (void) | |
842 | { | |
0fb1efaf | 843 | symbols_do (symbol_check_defined_processor, NULL); |
2f1afb73 AD |
844 | } |
845 | ||
846 | /*------------------------------------------------------------------. | |
847 | | Set TOKEN_TRANSLATIONS. Check that no two symbols share the same | | |
848 | | number. | | |
849 | `------------------------------------------------------------------*/ | |
850 | ||
851 | static void | |
852 | symbols_token_translations_init (void) | |
853 | { | |
a3714bce | 854 | bool num_256_available_p = true; |
2f1afb73 AD |
855 | int i; |
856 | ||
857 | /* Find the highest user token number, and whether 256, the POSIX | |
858 | preferred user token number for the error token, is used. */ | |
859 | max_user_token_number = 0; | |
860 | for (i = 0; i < ntokens; ++i) | |
861 | { | |
17ee7397 | 862 | symbol *this = symbols[i]; |
2f1afb73 | 863 | if (this->user_token_number != USER_NUMBER_UNDEFINED) |
e9690142 JD |
864 | { |
865 | if (this->user_token_number > max_user_token_number) | |
866 | max_user_token_number = this->user_token_number; | |
867 | if (this->user_token_number == 256) | |
868 | num_256_available_p = false; | |
869 | } | |
2f1afb73 AD |
870 | } |
871 | ||
872 | /* If 256 is not used, assign it to error, to follow POSIX. */ | |
873 | if (num_256_available_p | |
874 | && errtoken->user_token_number == USER_NUMBER_UNDEFINED) | |
875 | errtoken->user_token_number = 256; | |
876 | ||
877 | /* Set the missing user numbers. */ | |
878 | if (max_user_token_number < 256) | |
879 | max_user_token_number = 256; | |
880 | ||
881 | for (i = 0; i < ntokens; ++i) | |
882 | { | |
17ee7397 | 883 | symbol *this = symbols[i]; |
2f1afb73 | 884 | if (this->user_token_number == USER_NUMBER_UNDEFINED) |
e9690142 | 885 | this->user_token_number = ++max_user_token_number; |
2f1afb73 | 886 | if (this->user_token_number > max_user_token_number) |
e9690142 | 887 | max_user_token_number = this->user_token_number; |
2f1afb73 AD |
888 | } |
889 | ||
da2a7671 | 890 | token_translations = xnmalloc (max_user_token_number + 1, |
e9690142 | 891 | sizeof *token_translations); |
2f1afb73 | 892 | |
b3a28fd4 AD |
893 | /* Initialize all entries for literal tokens to the internal token |
894 | number for $undefined, which represents all invalid inputs. */ | |
2f1afb73 AD |
895 | for (i = 0; i < max_user_token_number + 1; i++) |
896 | token_translations[i] = undeftoken->number; | |
0fb1efaf | 897 | symbols_do (symbol_translation_processor, NULL); |
2f1afb73 AD |
898 | } |
899 | ||
900 | ||
901 | /*----------------------------------------------------------------. | |
902 | | Assign symbol numbers, and write definition of token names into | | |
903 | | FDEFINES. Set up vectors SYMBOL_TABLE, TAGS of symbols. | | |
904 | `----------------------------------------------------------------*/ | |
905 | ||
906 | void | |
907 | symbols_pack (void) | |
908 | { | |
0fb1efaf | 909 | symbols_do (symbol_check_alias_consistency_processor, NULL); |
6d0ef4ec JD |
910 | |
911 | symbols = xcalloc (nsyms, sizeof *symbols); | |
0fb1efaf | 912 | symbols_do (symbol_pack_processor, NULL); |
2f1afb73 | 913 | |
6d0ef4ec JD |
914 | /* Aliases leave empty slots in symbols, so remove them. */ |
915 | { | |
916 | int writei; | |
917 | int readi; | |
918 | int nsyms_old = nsyms; | |
919 | for (writei = 0, readi = 0; readi < nsyms_old; readi += 1) | |
920 | { | |
921 | if (symbols[readi] == NULL) | |
922 | { | |
923 | nsyms -= 1; | |
924 | ntokens -= 1; | |
925 | } | |
926 | else | |
927 | { | |
928 | symbols[writei] = symbols[readi]; | |
929 | symbols[writei]->number = writei; | |
930 | if (symbols[writei]->alias) | |
931 | symbols[writei]->alias->number = writei; | |
932 | writei += 1; | |
933 | } | |
934 | } | |
935 | } | |
936 | symbols = xnrealloc (symbols, nsyms, sizeof *symbols); | |
937 | ||
2f1afb73 AD |
938 | symbols_token_translations_init (); |
939 | ||
940 | if (startsymbol->class == unknown_sym) | |
ee000ba4 | 941 | fatal_at (startsymbol_location, |
e9690142 JD |
942 | _("the start symbol %s is undefined"), |
943 | startsymbol->tag); | |
2f1afb73 | 944 | else if (startsymbol->class == token_sym) |
ee000ba4 | 945 | fatal_at (startsymbol_location, |
e9690142 JD |
946 | _("the start symbol %s is a token"), |
947 | startsymbol->tag); | |
2f1afb73 | 948 | } |
ec5479ce JD |
949 | |
950 | ||
12e35840 JD |
951 | /*--------------------------------------------------. |
952 | | Set default tagged/tagless %destructor/%printer. | | |
953 | `--------------------------------------------------*/ | |
954 | ||
955 | void | |
95021767 | 956 | default_tagged_destructor_set (code_props const *destructor) |
12e35840 | 957 | { |
95021767 | 958 | if (default_tagged_destructor.code) |
12e35840 | 959 | { |
95021767 JD |
960 | complain_at (destructor->location, |
961 | _("redeclaration for default tagged %%destructor")); | |
962 | complain_at (default_tagged_destructor.location, | |
e9690142 | 963 | _("previous declaration")); |
12e35840 | 964 | } |
95021767 | 965 | default_tagged_destructor = *destructor; |
12e35840 JD |
966 | } |
967 | ||
968 | void | |
95021767 | 969 | default_tagless_destructor_set (code_props const *destructor) |
12e35840 | 970 | { |
95021767 | 971 | if (default_tagless_destructor.code) |
12e35840 | 972 | { |
95021767 JD |
973 | complain_at (destructor->location, |
974 | _("redeclaration for default tagless %%destructor")); | |
975 | complain_at (default_tagless_destructor.location, | |
e9690142 | 976 | _("previous declaration")); |
12e35840 | 977 | } |
95021767 | 978 | default_tagless_destructor = *destructor; |
12e35840 | 979 | } |
ec5479ce JD |
980 | |
981 | void | |
95021767 | 982 | default_tagged_printer_set (code_props const *printer) |
ec5479ce | 983 | { |
95021767 | 984 | if (default_tagged_printer.code) |
ec5479ce | 985 | { |
95021767 JD |
986 | complain_at (printer->location, |
987 | _("redeclaration for default tagged %%printer")); | |
988 | complain_at (default_tagged_printer.location, | |
e9690142 | 989 | _("previous declaration")); |
ec5479ce | 990 | } |
95021767 | 991 | default_tagged_printer = *printer; |
ec5479ce JD |
992 | } |
993 | ||
994 | void | |
95021767 | 995 | default_tagless_printer_set (code_props const *printer) |
ec5479ce | 996 | { |
95021767 | 997 | if (default_tagless_printer.code) |
ec5479ce | 998 | { |
95021767 JD |
999 | complain_at (printer->location, |
1000 | _("redeclaration for default tagless %%printer")); | |
1001 | complain_at (default_tagless_printer.location, | |
e9690142 | 1002 | _("previous declaration")); |
ec5479ce | 1003 | } |
95021767 | 1004 | default_tagless_printer = *printer; |
ec5479ce | 1005 | } |