]>
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] != '\'' && strchr (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 | /*----------------------------------. | |
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 | { | |
172 | aver (sym->user_token_number != USER_NUMBER_HAS_STRING_ALIAS); | |
173 | if (sym->alias) | |
174 | sym = sym->alias; | |
175 | return is_identifier (sym->tag) ? sym->tag : 0; | |
176 | } | |
177 | ||
178 | ||
179 | /*------------------------------------------------------------------. | |
180 | | Complain that S's WHAT is redeclared at SECOND, and was first set | | |
181 | | at FIRST. | | |
182 | `------------------------------------------------------------------*/ | |
183 | ||
184 | static void | |
185 | symbol_redeclaration (symbol *s, const char *what, location first, | |
186 | location second) | |
187 | { | |
188 | complain_at (second, _("%s redeclaration for %s"), what, s->tag); | |
189 | complain_at (first, _("previous declaration")); | |
190 | } | |
191 | ||
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 | ||
200 | ||
201 | ||
202 | /*-----------------------------------------------------------------. | |
203 | | Set the TYPE_NAME associated with SYM. Does nothing if passed 0 | | |
204 | | as TYPE_NAME. | | |
205 | `-----------------------------------------------------------------*/ | |
206 | ||
207 | void | |
208 | symbol_type_set (symbol *sym, uniqstr type_name, location loc) | |
209 | { | |
210 | if (type_name) | |
211 | { | |
212 | if (sym->type_name) | |
213 | symbol_redeclaration (sym, "%type", sym->type_location, loc); | |
214 | uniqstr_assert (type_name); | |
215 | sym->type_name = type_name; | |
216 | sym->type_location = loc; | |
217 | } | |
218 | } | |
219 | ||
220 | /*-----------------------------------------. | |
221 | | Set the DESTRUCTOR associated with SYM. | | |
222 | `-----------------------------------------*/ | |
223 | ||
224 | void | |
225 | symbol_destructor_set (symbol *sym, code_props const *destructor) | |
226 | { | |
227 | if (sym->destructor.code) | |
228 | symbol_redeclaration (sym, "%destructor", sym->destructor.location, | |
229 | destructor->location); | |
230 | sym->destructor = *destructor; | |
231 | } | |
232 | ||
233 | /*------------------------------------------. | |
234 | | Set the DESTRUCTOR associated with TYPE. | | |
235 | `------------------------------------------*/ | |
236 | ||
237 | void | |
238 | semantic_type_destructor_set (semantic_type *type, | |
239 | code_props const *destructor) | |
240 | { | |
241 | if (type->destructor.code) | |
242 | semantic_type_redeclaration (type, "%destructor", | |
243 | type->destructor.location, | |
244 | destructor->location); | |
245 | type->destructor = *destructor; | |
246 | } | |
247 | ||
248 | /*---------------------------------------. | |
249 | | Get the computed %destructor for SYM. | | |
250 | `---------------------------------------*/ | |
251 | ||
252 | code_props const * | |
253 | symbol_destructor_get (symbol const *sym) | |
254 | { | |
255 | /* Per-symbol %destructor. */ | |
256 | if (sym->destructor.code) | |
257 | return &sym->destructor; | |
258 | ||
259 | /* Per-type %destructor. */ | |
260 | if (sym->type_name) | |
261 | { | |
262 | code_props const *destructor = | |
263 | &semantic_type_get (sym->type_name)->destructor; | |
264 | if (destructor->code) | |
265 | return destructor; | |
266 | } | |
267 | ||
268 | /* Apply default %destructor's only to user-defined symbols. */ | |
269 | if (sym->tag[0] == '$' || sym == errtoken) | |
270 | return &code_props_none; | |
271 | ||
272 | if (sym->type_name) | |
273 | return &default_tagged_destructor; | |
274 | return &default_tagless_destructor; | |
275 | } | |
276 | ||
277 | /*--------------------------------------. | |
278 | | Set the PRINTER associated with SYM. | | |
279 | `--------------------------------------*/ | |
280 | ||
281 | void | |
282 | symbol_printer_set (symbol *sym, code_props const *printer) | |
283 | { | |
284 | if (sym->printer.code) | |
285 | symbol_redeclaration (sym, "%printer", | |
286 | sym->printer.location, printer->location); | |
287 | sym->printer = *printer; | |
288 | } | |
289 | ||
290 | /*---------------------------------------. | |
291 | | Set the PRINTER associated with TYPE. | | |
292 | `---------------------------------------*/ | |
293 | ||
294 | void | |
295 | semantic_type_printer_set (semantic_type *type, code_props const *printer) | |
296 | { | |
297 | if (type->printer.code) | |
298 | semantic_type_redeclaration (type, "%printer", | |
299 | type->printer.location, printer->location); | |
300 | type->printer = *printer; | |
301 | } | |
302 | ||
303 | /*------------------------------------. | |
304 | | Get the computed %printer for SYM. | | |
305 | `------------------------------------*/ | |
306 | ||
307 | code_props const * | |
308 | symbol_printer_get (symbol const *sym) | |
309 | { | |
310 | /* Per-symbol %printer. */ | |
311 | if (sym->printer.code) | |
312 | return &sym->printer; | |
313 | ||
314 | /* Per-type %printer. */ | |
315 | if (sym->type_name) | |
316 | { | |
317 | code_props const *printer = &semantic_type_get (sym->type_name)->printer; | |
318 | if (printer->code) | |
319 | return printer; | |
320 | } | |
321 | ||
322 | /* Apply the default %printer only to user-defined symbols. */ | |
323 | if (sym->tag[0] == '$' || sym == errtoken) | |
324 | return &code_props_none; | |
325 | ||
326 | if (sym->type_name) | |
327 | return &default_tagged_printer; | |
328 | return &default_tagless_printer; | |
329 | } | |
330 | ||
331 | /*-----------------------------------------------------------------. | |
332 | | Set the PRECEDENCE associated with SYM. Does nothing if invoked | | |
333 | | with UNDEF_ASSOC as ASSOC. | | |
334 | `-----------------------------------------------------------------*/ | |
335 | ||
336 | void | |
337 | symbol_precedence_set (symbol *sym, int prec, assoc a, location loc) | |
338 | { | |
339 | if (a != undef_assoc) | |
340 | { | |
341 | if (sym->prec != 0) | |
342 | symbol_redeclaration (sym, assoc_to_string (a), sym->prec_location, | |
343 | loc); | |
344 | sym->prec = prec; | |
345 | sym->assoc = a; | |
346 | sym->prec_location = loc; | |
347 | } | |
348 | ||
349 | /* Only terminals have a precedence. */ | |
350 | symbol_class_set (sym, token_sym, loc, false); | |
351 | } | |
352 | ||
353 | ||
354 | /*------------------------------------. | |
355 | | Set the CLASS associated with SYM. | | |
356 | `------------------------------------*/ | |
357 | ||
358 | void | |
359 | symbol_class_set (symbol *sym, symbol_class class, location loc, bool declaring) | |
360 | { | |
361 | if (sym->class != unknown_sym && sym->class != class) | |
362 | { | |
363 | complain_at (loc, _("symbol %s redefined"), sym->tag); | |
364 | sym->declared = false; | |
365 | } | |
366 | ||
367 | if (class == nterm_sym && sym->class != nterm_sym) | |
368 | sym->number = nvars++; | |
369 | else if (class == token_sym && sym->number == NUMBER_UNDEFINED) | |
370 | sym->number = ntokens++; | |
371 | ||
372 | sym->class = class; | |
373 | ||
374 | if (declaring) | |
375 | { | |
376 | if (sym->declared) | |
377 | warn_at (loc, _("symbol %s redeclared"), sym->tag); | |
378 | sym->declared = true; | |
379 | } | |
380 | } | |
381 | ||
382 | ||
383 | /*------------------------------------------------. | |
384 | | Set the USER_TOKEN_NUMBER associated with SYM. | | |
385 | `------------------------------------------------*/ | |
386 | ||
387 | void | |
388 | symbol_user_token_number_set (symbol *sym, int user_token_number, location loc) | |
389 | { | |
390 | int *user_token_numberp; | |
391 | ||
392 | if (sym->user_token_number != USER_NUMBER_HAS_STRING_ALIAS) | |
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) | |
398 | complain_at (loc, _("redefining user token number of %s"), sym->tag); | |
399 | ||
400 | *user_token_numberp = user_token_number; | |
401 | /* User defined $end token? */ | |
402 | if (user_token_number == 0) | |
403 | { | |
404 | endtoken = sym; | |
405 | /* It is always mapped to 0, so it was already counted in | |
406 | NTOKENS. */ | |
407 | if (endtoken->number != NUMBER_UNDEFINED) | |
408 | --ntokens; | |
409 | endtoken->number = 0; | |
410 | } | |
411 | } | |
412 | ||
413 | ||
414 | /*----------------------------------------------------------. | |
415 | | If SYM is not defined, report an error, and consider it a | | |
416 | | nonterminal. | | |
417 | `----------------------------------------------------------*/ | |
418 | ||
419 | static inline bool | |
420 | symbol_check_defined (symbol *sym) | |
421 | { | |
422 | if (sym->class == unknown_sym) | |
423 | { | |
424 | complain_at | |
425 | (sym->location, | |
426 | _("symbol %s is used, but is not defined as a token and has no rules"), | |
427 | sym->tag); | |
428 | sym->class = nterm_sym; | |
429 | sym->number = nvars++; | |
430 | } | |
431 | ||
432 | return true; | |
433 | } | |
434 | ||
435 | static bool | |
436 | symbol_check_defined_processor (void *sym, void *null ATTRIBUTE_UNUSED) | |
437 | { | |
438 | return symbol_check_defined (sym); | |
439 | } | |
440 | ||
441 | ||
442 | void | |
443 | symbol_make_alias (symbol *sym, symbol *str, location loc) | |
444 | { | |
445 | if (str->alias) | |
446 | warn_at (loc, _("symbol %s used more than once as a literal string"), | |
447 | str->tag); | |
448 | else if (sym->alias) | |
449 | warn_at (loc, _("symbol %s given more than one literal string"), | |
450 | sym->tag); | |
451 | else | |
452 | { | |
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); | |
460 | } | |
461 | } | |
462 | ||
463 | ||
464 | /*---------------------------------------------------------. | |
465 | | Check that THIS, and its alias, have same precedence and | | |
466 | | associativity. | | |
467 | `---------------------------------------------------------*/ | |
468 | ||
469 | static inline void | |
470 | symbol_check_alias_consistency (symbol *this) | |
471 | { | |
472 | symbol *sym = this; | |
473 | symbol *str = this->alias; | |
474 | ||
475 | /* Check only the symbol in the symbol-string pair. */ | |
476 | if (!(this->alias | |
477 | && this->user_token_number == USER_NUMBER_HAS_STRING_ALIAS)) | |
478 | return; | |
479 | ||
480 | if (str->type_name != sym->type_name) | |
481 | { | |
482 | if (str->type_name) | |
483 | symbol_type_set (sym, str->type_name, str->type_location); | |
484 | else | |
485 | symbol_type_set (str, sym->type_name, sym->type_location); | |
486 | } | |
487 | ||
488 | ||
489 | if (str->destructor.code || sym->destructor.code) | |
490 | { | |
491 | if (str->destructor.code) | |
492 | symbol_destructor_set (sym, &str->destructor); | |
493 | else | |
494 | symbol_destructor_set (str, &sym->destructor); | |
495 | } | |
496 | ||
497 | if (str->printer.code || sym->printer.code) | |
498 | { | |
499 | if (str->printer.code) | |
500 | symbol_printer_set (sym, &str->printer); | |
501 | else | |
502 | symbol_printer_set (str, &sym->printer); | |
503 | } | |
504 | ||
505 | if (sym->prec || str->prec) | |
506 | { | |
507 | if (str->prec) | |
508 | symbol_precedence_set (sym, str->prec, str->assoc, | |
509 | str->prec_location); | |
510 | else | |
511 | symbol_precedence_set (str, sym->prec, sym->assoc, | |
512 | sym->prec_location); | |
513 | } | |
514 | } | |
515 | ||
516 | static bool | |
517 | symbol_check_alias_consistency_processor (void *this, | |
518 | void *null ATTRIBUTE_UNUSED) | |
519 | { | |
520 | symbol_check_alias_consistency (this); | |
521 | return true; | |
522 | } | |
523 | ||
524 | ||
525 | /*-------------------------------------------------------------------. | |
526 | | Assign a symbol number, and write the definition of the token name | | |
527 | | into FDEFINES. Put in SYMBOLS. | | |
528 | `-------------------------------------------------------------------*/ | |
529 | ||
530 | static inline bool | |
531 | symbol_pack (symbol *this) | |
532 | { | |
533 | aver (this->number != NUMBER_UNDEFINED); | |
534 | if (this->class == nterm_sym) | |
535 | this->number += ntokens; | |
536 | else if (this->user_token_number == USER_NUMBER_HAS_STRING_ALIAS) | |
537 | return true; | |
538 | ||
539 | symbols[this->number] = this; | |
540 | return true; | |
541 | } | |
542 | ||
543 | static bool | |
544 | symbol_pack_processor (void *this, void *null ATTRIBUTE_UNUSED) | |
545 | { | |
546 | return symbol_pack (this); | |
547 | } | |
548 | ||
549 | ||
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 | |
554 | second step, via a traversal of the symbol table sorted on tag. | |
555 | ||
556 | However, error messages make more sense if we keep the first | |
557 | declaration first. */ | |
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 | } | |
570 | ||
571 | /*--------------------------------------------------. | |
572 | | Put THIS in TOKEN_TRANSLATIONS if it is a token. | | |
573 | `--------------------------------------------------*/ | |
574 | ||
575 | static inline bool | |
576 | symbol_translation (symbol *this) | |
577 | { | |
578 | /* Non-terminal? */ | |
579 | if (this->class == token_sym | |
580 | && this->user_token_number != USER_NUMBER_HAS_STRING_ALIAS) | |
581 | { | |
582 | /* A token which translation has already been set? */ | |
583 | if (token_translations[this->user_token_number] != undeftoken->number) | |
584 | user_token_number_redeclaration | |
585 | (this->user_token_number, | |
586 | symbols[token_translations[this->user_token_number]], | |
587 | this); | |
588 | ||
589 | token_translations[this->user_token_number] = this->number; | |
590 | } | |
591 | ||
592 | return true; | |
593 | } | |
594 | ||
595 | static bool | |
596 | symbol_translation_processor (void *this, void *null ATTRIBUTE_UNUSED) | |
597 | { | |
598 | return symbol_translation (this); | |
599 | } | |
600 | ||
601 | ||
602 | /*---------------------------------------. | |
603 | | Symbol and semantic type hash tables. | | |
604 | `---------------------------------------*/ | |
605 | ||
606 | /* Initial capacity of symbol and semantic type hash table. */ | |
607 | #define HT_INITIAL_CAPACITY 257 | |
608 | ||
609 | static struct hash_table *symbol_table = NULL; | |
610 | static struct hash_table *semantic_type_table = NULL; | |
611 | ||
612 | static inline bool | |
613 | hash_compare_symbol (const symbol *m1, const symbol *m2) | |
614 | { | |
615 | /* Since tags are unique, we can compare the pointers themselves. */ | |
616 | return UNIQSTR_EQ (m1->tag, m2->tag); | |
617 | } | |
618 | ||
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 | ||
626 | static bool | |
627 | hash_symbol_comparator (void const *m1, void const *m2) | |
628 | { | |
629 | return hash_compare_symbol (m1, m2); | |
630 | } | |
631 | ||
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 | ||
638 | static inline size_t | |
639 | hash_symbol (const symbol *m, size_t tablesize) | |
640 | { | |
641 | /* Since tags are unique, we can hash the pointer itself. */ | |
642 | return ((uintptr_t) m->tag) % tablesize; | |
643 | } | |
644 | ||
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 | ||
652 | static size_t | |
653 | hash_symbol_hasher (void const *m, size_t tablesize) | |
654 | { | |
655 | return hash_symbol (m, tablesize); | |
656 | } | |
657 | ||
658 | static size_t | |
659 | hash_semantic_type_hasher (void const *m, size_t tablesize) | |
660 | { | |
661 | return hash_semantic_type (m, tablesize); | |
662 | } | |
663 | ||
664 | /*-------------------------------. | |
665 | | Create the symbol hash table. | | |
666 | `-------------------------------*/ | |
667 | ||
668 | void | |
669 | symbols_new (void) | |
670 | { | |
671 | symbol_table = hash_initialize (HT_INITIAL_CAPACITY, | |
672 | NULL, | |
673 | hash_symbol_hasher, | |
674 | hash_symbol_comparator, | |
675 | free); | |
676 | semantic_type_table = hash_initialize (HT_INITIAL_CAPACITY, | |
677 | NULL, | |
678 | hash_semantic_type_hasher, | |
679 | hash_semantic_type_comparator, | |
680 | free); | |
681 | } | |
682 | ||
683 | ||
684 | /*----------------------------------------------------------------. | |
685 | | Find the symbol named KEY, and return it. If it does not exist | | |
686 | | yet, create it. | | |
687 | `----------------------------------------------------------------*/ | |
688 | ||
689 | symbol * | |
690 | symbol_from_uniqstr (const uniqstr key, location loc) | |
691 | { | |
692 | symbol probe; | |
693 | symbol *entry; | |
694 | ||
695 | probe.tag = key; | |
696 | entry = hash_lookup (symbol_table, &probe); | |
697 | ||
698 | if (!entry) | |
699 | { | |
700 | /* First insertion in the hash. */ | |
701 | aver (!symbols_sorted); | |
702 | entry = symbol_new (key, loc); | |
703 | if (!hash_insert (symbol_table, entry)) | |
704 | xalloc_die (); | |
705 | } | |
706 | return entry; | |
707 | } | |
708 | ||
709 | ||
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); | |
728 | if (!hash_insert (semantic_type_table, entry)) | |
729 | xalloc_die (); | |
730 | } | |
731 | return entry; | |
732 | } | |
733 | ||
734 | ||
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 | ||
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 | ||
759 | /*------------------------------------------------------------------. | |
760 | | Generate a dummy nonterminal, whose name cannot conflict with the | | |
761 | | user's names. | | |
762 | `------------------------------------------------------------------*/ | |
763 | ||
764 | symbol * | |
765 | dummy_symbol_get (location loc) | |
766 | { | |
767 | /* Incremented for each generated symbol. */ | |
768 | static int dummy_count = 0; | |
769 | static char buf[256]; | |
770 | ||
771 | symbol *sym; | |
772 | ||
773 | sprintf (buf, "$@%d", ++dummy_count); | |
774 | sym = symbol_get (buf, loc); | |
775 | sym->class = nterm_sym; | |
776 | sym->number = nvars++; | |
777 | return sym; | |
778 | } | |
779 | ||
780 | bool | |
781 | symbol_is_dummy (const symbol *sym) | |
782 | { | |
783 | return sym->tag[0] == '@' || (sym->tag[0] == '$' && sym->tag[1] == '@'); | |
784 | } | |
785 | ||
786 | /*-------------------. | |
787 | | Free the symbols. | | |
788 | `-------------------*/ | |
789 | ||
790 | void | |
791 | symbols_free (void) | |
792 | { | |
793 | hash_free (symbol_table); | |
794 | hash_free (semantic_type_table); | |
795 | free (symbols); | |
796 | free (symbols_sorted); | |
797 | } | |
798 | ||
799 | ||
800 | /*---------------------------------------------------------------. | |
801 | | Look for undefined symbols, report an error, and consider them | | |
802 | | terminals. | | |
803 | `---------------------------------------------------------------*/ | |
804 | ||
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 | ||
817 | static void | |
818 | symbols_do (Hash_processor processor, void *processor_data) | |
819 | { | |
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 | } | |
833 | } | |
834 | ||
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 | { | |
843 | symbols_do (symbol_check_defined_processor, NULL); | |
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 | { | |
854 | bool num_256_available_p = true; | |
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 | { | |
862 | symbol *this = symbols[i]; | |
863 | if (this->user_token_number != USER_NUMBER_UNDEFINED) | |
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 | } | |
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 | { | |
883 | symbol *this = symbols[i]; | |
884 | if (this->user_token_number == USER_NUMBER_UNDEFINED) | |
885 | this->user_token_number = ++max_user_token_number; | |
886 | if (this->user_token_number > max_user_token_number) | |
887 | max_user_token_number = this->user_token_number; | |
888 | } | |
889 | ||
890 | token_translations = xnmalloc (max_user_token_number + 1, | |
891 | sizeof *token_translations); | |
892 | ||
893 | /* Initialize all entries for literal tokens to the internal token | |
894 | number for $undefined, which represents all invalid inputs. */ | |
895 | for (i = 0; i < max_user_token_number + 1; i++) | |
896 | token_translations[i] = undeftoken->number; | |
897 | symbols_do (symbol_translation_processor, NULL); | |
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 | { | |
909 | symbols_do (symbol_check_alias_consistency_processor, NULL); | |
910 | ||
911 | symbols = xcalloc (nsyms, sizeof *symbols); | |
912 | symbols_do (symbol_pack_processor, NULL); | |
913 | ||
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 | ||
938 | symbols_token_translations_init (); | |
939 | ||
940 | if (startsymbol->class == unknown_sym) | |
941 | fatal_at (startsymbol_location, | |
942 | _("the start symbol %s is undefined"), | |
943 | startsymbol->tag); | |
944 | else if (startsymbol->class == token_sym) | |
945 | fatal_at (startsymbol_location, | |
946 | _("the start symbol %s is a token"), | |
947 | startsymbol->tag); | |
948 | } | |
949 | ||
950 | ||
951 | /*--------------------------------------------------. | |
952 | | Set default tagged/tagless %destructor/%printer. | | |
953 | `--------------------------------------------------*/ | |
954 | ||
955 | void | |
956 | default_tagged_destructor_set (code_props const *destructor) | |
957 | { | |
958 | if (default_tagged_destructor.code) | |
959 | { | |
960 | complain_at (destructor->location, | |
961 | _("redeclaration for default tagged %%destructor")); | |
962 | complain_at (default_tagged_destructor.location, | |
963 | _("previous declaration")); | |
964 | } | |
965 | default_tagged_destructor = *destructor; | |
966 | } | |
967 | ||
968 | void | |
969 | default_tagless_destructor_set (code_props const *destructor) | |
970 | { | |
971 | if (default_tagless_destructor.code) | |
972 | { | |
973 | complain_at (destructor->location, | |
974 | _("redeclaration for default tagless %%destructor")); | |
975 | complain_at (default_tagless_destructor.location, | |
976 | _("previous declaration")); | |
977 | } | |
978 | default_tagless_destructor = *destructor; | |
979 | } | |
980 | ||
981 | void | |
982 | default_tagged_printer_set (code_props const *printer) | |
983 | { | |
984 | if (default_tagged_printer.code) | |
985 | { | |
986 | complain_at (printer->location, | |
987 | _("redeclaration for default tagged %%printer")); | |
988 | complain_at (default_tagged_printer.location, | |
989 | _("previous declaration")); | |
990 | } | |
991 | default_tagged_printer = *printer; | |
992 | } | |
993 | ||
994 | void | |
995 | default_tagless_printer_set (code_props const *printer) | |
996 | { | |
997 | if (default_tagless_printer.code) | |
998 | { | |
999 | complain_at (printer->location, | |
1000 | _("redeclaration for default tagless %%printer")); | |
1001 | complain_at (default_tagless_printer.location, | |
1002 | _("previous declaration")); | |
1003 | } | |
1004 | default_tagless_printer = *printer; | |
1005 | } |