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