]>
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 | static symbol **semantic_types_sorted = NULL; | |
37 | ||
38 | /*------------------------. | |
39 | | Distinguished symbols. | | |
40 | `------------------------*/ | |
41 | ||
42 | symbol *errtoken = NULL; | |
43 | symbol *undeftoken = NULL; | |
44 | symbol *endtoken = NULL; | |
45 | symbol *accept = NULL; | |
46 | symbol *startsymbol = NULL; | |
47 | location startsymbol_location; | |
48 | ||
49 | /*---------------------------------------. | |
50 | | Default %destructor's and %printer's. | | |
51 | `---------------------------------------*/ | |
52 | ||
53 | static code_props default_tagged_code_props[CODE_PROPS_SIZE] = | |
54 | { | |
55 | CODE_PROPS_NONE_INIT, | |
56 | CODE_PROPS_NONE_INIT, | |
57 | }; | |
58 | static code_props default_tagless_code_props[CODE_PROPS_SIZE] = | |
59 | { | |
60 | CODE_PROPS_NONE_INIT, | |
61 | CODE_PROPS_NONE_INIT, | |
62 | }; | |
63 | ||
64 | /*---------------------------------. | |
65 | | Create a new symbol, named TAG. | | |
66 | `---------------------------------*/ | |
67 | ||
68 | static symbol * | |
69 | symbol_new (uniqstr tag, location loc) | |
70 | { | |
71 | symbol *res = xmalloc (sizeof *res); | |
72 | ||
73 | uniqstr_assert (tag); | |
74 | ||
75 | /* If the tag is not a string (starts with a double quote), check | |
76 | that it is valid for Yacc. */ | |
77 | if (tag[0] != '\"' && tag[0] != '\'' && strchr (tag, '-')) | |
78 | yacc_at (loc, _("POSIX Yacc forbids dashes in symbol names: %s"), | |
79 | tag); | |
80 | ||
81 | res->tag = tag; | |
82 | res->location = loc; | |
83 | ||
84 | res->type_name = NULL; | |
85 | for (int i = 0; i < CODE_PROPS_SIZE; ++i) | |
86 | code_props_none_init (&res->props[i]); | |
87 | ||
88 | res->number = NUMBER_UNDEFINED; | |
89 | res->prec = 0; | |
90 | res->assoc = undef_assoc; | |
91 | res->user_token_number = USER_NUMBER_UNDEFINED; | |
92 | ||
93 | res->alias = NULL; | |
94 | res->class = unknown_sym; | |
95 | res->status = undeclared; | |
96 | ||
97 | if (nsyms == SYMBOL_NUMBER_MAXIMUM) | |
98 | fatal (_("too many symbols in input grammar (limit is %d)"), | |
99 | SYMBOL_NUMBER_MAXIMUM); | |
100 | nsyms++; | |
101 | return res; | |
102 | } | |
103 | ||
104 | char const * | |
105 | code_props_type_string (code_props_type kind) | |
106 | { | |
107 | switch (kind) | |
108 | { | |
109 | case destructor: | |
110 | return "%destructor"; | |
111 | case printer: | |
112 | return "%printer"; | |
113 | } | |
114 | assert (0); | |
115 | } | |
116 | ||
117 | /*----------------------------------------. | |
118 | | Create a new semantic type, named TAG. | | |
119 | `----------------------------------------*/ | |
120 | ||
121 | static semantic_type * | |
122 | semantic_type_new (uniqstr tag, const location *loc) | |
123 | { | |
124 | semantic_type *res = xmalloc (sizeof *res); | |
125 | ||
126 | uniqstr_assert (tag); | |
127 | res->tag = tag; | |
128 | if (loc) | |
129 | res->location = *loc; | |
130 | for (int i = 0; i < CODE_PROPS_SIZE; ++i) | |
131 | code_props_none_init (&res->props[i]); | |
132 | ||
133 | return res; | |
134 | } | |
135 | ||
136 | ||
137 | /*-----------------. | |
138 | | Print a symbol. | | |
139 | `-----------------*/ | |
140 | ||
141 | #define SYMBOL_ATTR_PRINT(Attr) \ | |
142 | if (s->Attr) \ | |
143 | fprintf (f, " %s { %s }", #Attr, s->Attr) | |
144 | ||
145 | #define SYMBOL_CODE_PRINT(Attr) \ | |
146 | if (s->props[Attr].code) \ | |
147 | fprintf (f, " %s { %s }", #Attr, s->props[Attr].code) | |
148 | ||
149 | void | |
150 | symbol_print (symbol *s, FILE *f) | |
151 | { | |
152 | if (s) | |
153 | { | |
154 | fprintf (f, "\"%s\"", s->tag); | |
155 | SYMBOL_ATTR_PRINT (type_name); | |
156 | SYMBOL_CODE_PRINT (destructor); | |
157 | SYMBOL_CODE_PRINT (printer); | |
158 | } | |
159 | else | |
160 | fprintf (f, "<NULL>"); | |
161 | } | |
162 | ||
163 | #undef SYMBOL_ATTR_PRINT | |
164 | #undef SYMBOL_CODE_PRINT | |
165 | ||
166 | ||
167 | /*----------------------------------. | |
168 | | Whether S is a valid identifier. | | |
169 | `----------------------------------*/ | |
170 | ||
171 | static bool | |
172 | is_identifier (uniqstr s) | |
173 | { | |
174 | static char const alphanum[26 + 26 + 1 + 10] = | |
175 | "abcdefghijklmnopqrstuvwxyz" | |
176 | "ABCDEFGHIJKLMNOPQRSTUVWXYZ" | |
177 | "_" | |
178 | "0123456789"; | |
179 | if (!s || ! memchr (alphanum, *s, sizeof alphanum - 10)) | |
180 | return false; | |
181 | for (++s; *s; ++s) | |
182 | if (! memchr (alphanum, *s, sizeof alphanum)) | |
183 | return false; | |
184 | return true; | |
185 | } | |
186 | ||
187 | ||
188 | /*-----------------------------------------------. | |
189 | | Get the identifier associated to this symbol. | | |
190 | `-----------------------------------------------*/ | |
191 | uniqstr | |
192 | symbol_id_get (symbol const *sym) | |
193 | { | |
194 | aver (sym->user_token_number != USER_NUMBER_HAS_STRING_ALIAS); | |
195 | if (sym->alias) | |
196 | sym = sym->alias; | |
197 | return is_identifier (sym->tag) ? sym->tag : 0; | |
198 | } | |
199 | ||
200 | ||
201 | /*------------------------------------------------------------------. | |
202 | | Complain that S's WHAT is redeclared at SECOND, and was first set | | |
203 | | at FIRST. | | |
204 | `------------------------------------------------------------------*/ | |
205 | ||
206 | static void | |
207 | symbol_redeclaration (symbol *s, const char *what, location first, | |
208 | location second) | |
209 | { | |
210 | complain_at (second, _("%s redeclaration for %s"), what, s->tag); | |
211 | complain_at (first, _("previous declaration")); | |
212 | } | |
213 | ||
214 | static void | |
215 | semantic_type_redeclaration (semantic_type *s, const char *what, location first, | |
216 | location second) | |
217 | { | |
218 | complain_at (second, _("%s redeclaration for <%s>"), what, s->tag); | |
219 | complain_at (first, _("previous declaration")); | |
220 | } | |
221 | ||
222 | ||
223 | ||
224 | /*-----------------------------------------------------------------. | |
225 | | Set the TYPE_NAME associated with SYM. Does nothing if passed 0 | | |
226 | | as TYPE_NAME. | | |
227 | `-----------------------------------------------------------------*/ | |
228 | ||
229 | void | |
230 | symbol_type_set (symbol *sym, uniqstr type_name, location loc) | |
231 | { | |
232 | if (type_name) | |
233 | { | |
234 | if (sym->type_name) | |
235 | symbol_redeclaration (sym, "%type", sym->type_location, loc); | |
236 | uniqstr_assert (type_name); | |
237 | sym->type_name = type_name; | |
238 | sym->type_location = loc; | |
239 | } | |
240 | } | |
241 | ||
242 | /*--------------------------------------------------------. | |
243 | | Set the DESTRUCTOR or PRINTER associated with the SYM. | | |
244 | `--------------------------------------------------------*/ | |
245 | ||
246 | void | |
247 | symbol_code_props_set (symbol *sym, code_props_type kind, | |
248 | code_props const *code) | |
249 | { | |
250 | if (sym->props[kind].code) | |
251 | symbol_redeclaration (sym, code_props_type_string (kind), | |
252 | sym->props[kind].location, | |
253 | code->location); | |
254 | sym->props[kind] = *code; | |
255 | } | |
256 | ||
257 | /*-----------------------------------------------------. | |
258 | | Set the DESTRUCTOR or PRINTER associated with TYPE. | | |
259 | `-----------------------------------------------------*/ | |
260 | ||
261 | void | |
262 | semantic_type_code_props_set (semantic_type *type, | |
263 | code_props_type kind, | |
264 | code_props const *code) | |
265 | { | |
266 | if (type->props[kind].code) | |
267 | semantic_type_redeclaration (type, code_props_type_string (kind), | |
268 | type->props[kind].location, | |
269 | code->location); | |
270 | type->props[kind] = *code; | |
271 | } | |
272 | ||
273 | /*---------------------------------------------------. | |
274 | | Get the computed %destructor or %printer for SYM. | | |
275 | `---------------------------------------------------*/ | |
276 | ||
277 | code_props const * | |
278 | symbol_code_props_get (symbol const *sym, | |
279 | code_props_type kind) | |
280 | { | |
281 | /* Per-symbol code props. */ | |
282 | if (sym->props[kind].code) | |
283 | return &sym->props[kind]; | |
284 | ||
285 | /* Per-type code props. */ | |
286 | if (sym->type_name) | |
287 | { | |
288 | code_props const *code = | |
289 | &semantic_type_get (sym->type_name, NULL)->props[kind]; | |
290 | if (code->code) | |
291 | return code; | |
292 | } | |
293 | ||
294 | /* Apply default code props's only to user-defined symbols. */ | |
295 | if (sym->tag[0] == '$' || sym == errtoken) | |
296 | return &code_props_none; | |
297 | ||
298 | if (sym->type_name) | |
299 | return &default_tagged_code_props[kind]; | |
300 | return &default_tagless_code_props[kind]; | |
301 | } | |
302 | ||
303 | /*-----------------------------------------------------------------. | |
304 | | Set the PRECEDENCE associated with SYM. Does nothing if invoked | | |
305 | | with UNDEF_ASSOC as ASSOC. | | |
306 | `-----------------------------------------------------------------*/ | |
307 | ||
308 | void | |
309 | symbol_precedence_set (symbol *sym, int prec, assoc a, location loc) | |
310 | { | |
311 | if (a != undef_assoc) | |
312 | { | |
313 | if (sym->prec != 0) | |
314 | symbol_redeclaration (sym, assoc_to_string (a), sym->prec_location, | |
315 | loc); | |
316 | sym->prec = prec; | |
317 | sym->assoc = a; | |
318 | sym->prec_location = loc; | |
319 | } | |
320 | ||
321 | /* Only terminals have a precedence. */ | |
322 | symbol_class_set (sym, token_sym, loc, false); | |
323 | } | |
324 | ||
325 | ||
326 | /*------------------------------------. | |
327 | | Set the CLASS associated with SYM. | | |
328 | `------------------------------------*/ | |
329 | ||
330 | void | |
331 | symbol_class_set (symbol *sym, symbol_class class, location loc, bool declaring) | |
332 | { | |
333 | bool warned = false; | |
334 | if (sym->class != unknown_sym && sym->class != class) | |
335 | { | |
336 | complain_at (loc, _("symbol %s redefined"), sym->tag); | |
337 | // Don't report both "redefined" and "redeclared". | |
338 | warned = true; | |
339 | } | |
340 | ||
341 | if (class == nterm_sym && sym->class != nterm_sym) | |
342 | sym->number = nvars++; | |
343 | else if (class == token_sym && sym->number == NUMBER_UNDEFINED) | |
344 | sym->number = ntokens++; | |
345 | ||
346 | sym->class = class; | |
347 | ||
348 | if (declaring) | |
349 | { | |
350 | if (sym->status == declared && !warned) | |
351 | warn_at (loc, _("symbol %s redeclared"), sym->tag); | |
352 | sym->status = declared; | |
353 | } | |
354 | } | |
355 | ||
356 | ||
357 | /*------------------------------------------------. | |
358 | | Set the USER_TOKEN_NUMBER associated with SYM. | | |
359 | `------------------------------------------------*/ | |
360 | ||
361 | void | |
362 | symbol_user_token_number_set (symbol *sym, int user_token_number, location loc) | |
363 | { | |
364 | int *user_token_numberp; | |
365 | ||
366 | if (sym->user_token_number != USER_NUMBER_HAS_STRING_ALIAS) | |
367 | user_token_numberp = &sym->user_token_number; | |
368 | else | |
369 | user_token_numberp = &sym->alias->user_token_number; | |
370 | if (*user_token_numberp != USER_NUMBER_UNDEFINED | |
371 | && *user_token_numberp != user_token_number) | |
372 | complain_at (loc, _("redefining user token number of %s"), sym->tag); | |
373 | ||
374 | *user_token_numberp = user_token_number; | |
375 | /* User defined $end token? */ | |
376 | if (user_token_number == 0) | |
377 | { | |
378 | endtoken = sym; | |
379 | /* It is always mapped to 0, so it was already counted in | |
380 | NTOKENS. */ | |
381 | if (endtoken->number != NUMBER_UNDEFINED) | |
382 | --ntokens; | |
383 | endtoken->number = 0; | |
384 | } | |
385 | } | |
386 | ||
387 | ||
388 | /*----------------------------------------------------------. | |
389 | | If SYM is not defined, report an error, and consider it a | | |
390 | | nonterminal. | | |
391 | `----------------------------------------------------------*/ | |
392 | ||
393 | static inline bool | |
394 | symbol_check_defined (symbol *sym) | |
395 | { | |
396 | if (sym->class == unknown_sym) | |
397 | { | |
398 | switch (sym->status) | |
399 | { | |
400 | case used: | |
401 | warn_at (sym->location, | |
402 | _("symbol %s is used, but is not defined as a token" | |
403 | " and has no rules"), | |
404 | sym->tag); | |
405 | break; | |
406 | case undeclared: | |
407 | case needed: | |
408 | complain_at (sym->location, | |
409 | _("symbol %s is used, but is not defined as a token" | |
410 | " and has no rules"), | |
411 | sym->tag); | |
412 | break; | |
413 | case declared: | |
414 | /* If declared, then sym->class != unknown_sym. */ | |
415 | assert (0); | |
416 | } | |
417 | ||
418 | sym->class = nterm_sym; | |
419 | sym->number = nvars++; | |
420 | } | |
421 | ||
422 | for (int i = 0; i < 2; ++i) | |
423 | if (sym->props[i].kind == CODE_PROPS_NONE && sym->type_name) | |
424 | { | |
425 | semantic_type *sem_type = semantic_type_get (sym->type_name, NULL); | |
426 | if (sem_type | |
427 | && sem_type->props[i].kind != CODE_PROPS_NONE) | |
428 | sem_type->props[i].is_used = true; | |
429 | } | |
430 | ||
431 | /* Set the semantic type status associated to the current symbol to | |
432 | 'declared' so that we could check semantic types unnecessary uses. */ | |
433 | if (sym->type_name) | |
434 | { | |
435 | semantic_type *sem_type = semantic_type_get (sym->type_name, NULL); | |
436 | if (sem_type) | |
437 | sem_type->status = declared; | |
438 | } | |
439 | ||
440 | return true; | |
441 | } | |
442 | ||
443 | static inline bool | |
444 | semantic_type_check_defined (semantic_type *sem_type) | |
445 | { | |
446 | if (sem_type->status == declared) | |
447 | { | |
448 | for (int i = 0; i < 2; ++i) | |
449 | if (sem_type->props[i].kind != CODE_PROPS_NONE | |
450 | && ! sem_type->props[i].is_used) | |
451 | warn_at (sem_type->location, | |
452 | _("useless %s for type <%s>"), | |
453 | code_props_type_string (i), sem_type->tag); | |
454 | } | |
455 | else | |
456 | warn_at (sem_type->location, | |
457 | _("type <%s> is used, but is not associated to any symbol"), | |
458 | sem_type->tag); | |
459 | ||
460 | return true; | |
461 | } | |
462 | ||
463 | static bool | |
464 | symbol_check_defined_processor (void *sym, void *null ATTRIBUTE_UNUSED) | |
465 | { | |
466 | return symbol_check_defined (sym); | |
467 | } | |
468 | ||
469 | static bool | |
470 | semantic_type_check_defined_processor (void *sem_type, | |
471 | void *null ATTRIBUTE_UNUSED) | |
472 | { | |
473 | return semantic_type_check_defined (sem_type); | |
474 | } | |
475 | ||
476 | ||
477 | void | |
478 | symbol_make_alias (symbol *sym, symbol *str, location loc) | |
479 | { | |
480 | if (str->alias) | |
481 | warn_at (loc, _("symbol %s used more than once as a literal string"), | |
482 | str->tag); | |
483 | else if (sym->alias) | |
484 | warn_at (loc, _("symbol %s given more than one literal string"), | |
485 | sym->tag); | |
486 | else | |
487 | { | |
488 | str->class = token_sym; | |
489 | str->user_token_number = sym->user_token_number; | |
490 | sym->user_token_number = USER_NUMBER_HAS_STRING_ALIAS; | |
491 | str->alias = sym; | |
492 | sym->alias = str; | |
493 | str->number = sym->number; | |
494 | symbol_type_set (str, sym->type_name, loc); | |
495 | } | |
496 | } | |
497 | ||
498 | ||
499 | /*---------------------------------------------------------. | |
500 | | Check that THIS, and its alias, have same precedence and | | |
501 | | associativity. | | |
502 | `---------------------------------------------------------*/ | |
503 | ||
504 | static inline void | |
505 | symbol_check_alias_consistency (symbol *this) | |
506 | { | |
507 | symbol *sym = this; | |
508 | symbol *str = this->alias; | |
509 | ||
510 | /* Check only the symbol in the symbol-string pair. */ | |
511 | if (!(this->alias | |
512 | && this->user_token_number == USER_NUMBER_HAS_STRING_ALIAS)) | |
513 | return; | |
514 | ||
515 | if (str->type_name != sym->type_name) | |
516 | { | |
517 | if (str->type_name) | |
518 | symbol_type_set (sym, str->type_name, str->type_location); | |
519 | else | |
520 | symbol_type_set (str, sym->type_name, sym->type_location); | |
521 | } | |
522 | ||
523 | ||
524 | for (int i = 0; i < CODE_PROPS_SIZE; ++i) | |
525 | if (str->props[i].code) | |
526 | symbol_code_props_set (sym, i, &str->props[i]); | |
527 | else if (sym->props[i].code) | |
528 | symbol_code_props_set (str, i, &sym->props[i]); | |
529 | ||
530 | if (sym->prec || str->prec) | |
531 | { | |
532 | if (str->prec) | |
533 | symbol_precedence_set (sym, str->prec, str->assoc, | |
534 | str->prec_location); | |
535 | else | |
536 | symbol_precedence_set (str, sym->prec, sym->assoc, | |
537 | sym->prec_location); | |
538 | } | |
539 | } | |
540 | ||
541 | static bool | |
542 | symbol_check_alias_consistency_processor (void *this, | |
543 | void *null ATTRIBUTE_UNUSED) | |
544 | { | |
545 | symbol_check_alias_consistency (this); | |
546 | return true; | |
547 | } | |
548 | ||
549 | ||
550 | /*-------------------------------------------------------------------. | |
551 | | Assign a symbol number, and write the definition of the token name | | |
552 | | into FDEFINES. Put in SYMBOLS. | | |
553 | `-------------------------------------------------------------------*/ | |
554 | ||
555 | static inline bool | |
556 | symbol_pack (symbol *this) | |
557 | { | |
558 | aver (this->number != NUMBER_UNDEFINED); | |
559 | if (this->class == nterm_sym) | |
560 | this->number += ntokens; | |
561 | else if (this->user_token_number == USER_NUMBER_HAS_STRING_ALIAS) | |
562 | return true; | |
563 | ||
564 | symbols[this->number] = this; | |
565 | return true; | |
566 | } | |
567 | ||
568 | static bool | |
569 | symbol_pack_processor (void *this, void *null ATTRIBUTE_UNUSED) | |
570 | { | |
571 | return symbol_pack (this); | |
572 | } | |
573 | ||
574 | ||
575 | static void | |
576 | user_token_number_redeclaration (int num, symbol *first, symbol *second) | |
577 | { | |
578 | /* User token numbers are not assigned during the parsing, but in a | |
579 | second step, via a traversal of the symbol table sorted on tag. | |
580 | ||
581 | However, error messages make more sense if we keep the first | |
582 | declaration first. */ | |
583 | if (location_cmp (first->location, second->location) > 0) | |
584 | { | |
585 | symbol* tmp = first; | |
586 | first = second; | |
587 | second = tmp; | |
588 | } | |
589 | complain_at (second->location, | |
590 | _("user token number %d redeclaration for %s"), | |
591 | num, second->tag); | |
592 | complain_at (first->location, _("previous declaration for %s"), | |
593 | first->tag); | |
594 | } | |
595 | ||
596 | /*--------------------------------------------------. | |
597 | | Put THIS in TOKEN_TRANSLATIONS if it is a token. | | |
598 | `--------------------------------------------------*/ | |
599 | ||
600 | static inline bool | |
601 | symbol_translation (symbol *this) | |
602 | { | |
603 | /* Non-terminal? */ | |
604 | if (this->class == token_sym | |
605 | && this->user_token_number != USER_NUMBER_HAS_STRING_ALIAS) | |
606 | { | |
607 | /* A token which translation has already been set? */ | |
608 | if (token_translations[this->user_token_number] != undeftoken->number) | |
609 | user_token_number_redeclaration | |
610 | (this->user_token_number, | |
611 | symbols[token_translations[this->user_token_number]], | |
612 | this); | |
613 | ||
614 | token_translations[this->user_token_number] = this->number; | |
615 | } | |
616 | ||
617 | return true; | |
618 | } | |
619 | ||
620 | static bool | |
621 | symbol_translation_processor (void *this, void *null ATTRIBUTE_UNUSED) | |
622 | { | |
623 | return symbol_translation (this); | |
624 | } | |
625 | ||
626 | ||
627 | /*---------------------------------------. | |
628 | | Symbol and semantic type hash tables. | | |
629 | `---------------------------------------*/ | |
630 | ||
631 | /* Initial capacity of symbol and semantic type hash table. */ | |
632 | #define HT_INITIAL_CAPACITY 257 | |
633 | ||
634 | static struct hash_table *symbol_table = NULL; | |
635 | static struct hash_table *semantic_type_table = NULL; | |
636 | ||
637 | static inline bool | |
638 | hash_compare_symbol (const symbol *m1, const symbol *m2) | |
639 | { | |
640 | /* Since tags are unique, we can compare the pointers themselves. */ | |
641 | return UNIQSTR_EQ (m1->tag, m2->tag); | |
642 | } | |
643 | ||
644 | static inline bool | |
645 | hash_compare_semantic_type (const semantic_type *m1, const semantic_type *m2) | |
646 | { | |
647 | /* Since names are unique, we can compare the pointers themselves. */ | |
648 | return UNIQSTR_EQ (m1->tag, m2->tag); | |
649 | } | |
650 | ||
651 | static bool | |
652 | hash_symbol_comparator (void const *m1, void const *m2) | |
653 | { | |
654 | return hash_compare_symbol (m1, m2); | |
655 | } | |
656 | ||
657 | static bool | |
658 | hash_semantic_type_comparator (void const *m1, void const *m2) | |
659 | { | |
660 | return hash_compare_semantic_type (m1, m2); | |
661 | } | |
662 | ||
663 | static inline size_t | |
664 | hash_symbol (const symbol *m, size_t tablesize) | |
665 | { | |
666 | /* Since tags are unique, we can hash the pointer itself. */ | |
667 | return ((uintptr_t) m->tag) % tablesize; | |
668 | } | |
669 | ||
670 | static inline size_t | |
671 | hash_semantic_type (const semantic_type *m, size_t tablesize) | |
672 | { | |
673 | /* Since names are unique, we can hash the pointer itself. */ | |
674 | return ((uintptr_t) m->tag) % tablesize; | |
675 | } | |
676 | ||
677 | static size_t | |
678 | hash_symbol_hasher (void const *m, size_t tablesize) | |
679 | { | |
680 | return hash_symbol (m, tablesize); | |
681 | } | |
682 | ||
683 | static size_t | |
684 | hash_semantic_type_hasher (void const *m, size_t tablesize) | |
685 | { | |
686 | return hash_semantic_type (m, tablesize); | |
687 | } | |
688 | ||
689 | /*-------------------------------. | |
690 | | Create the symbol hash table. | | |
691 | `-------------------------------*/ | |
692 | ||
693 | void | |
694 | symbols_new (void) | |
695 | { | |
696 | symbol_table = hash_initialize (HT_INITIAL_CAPACITY, | |
697 | NULL, | |
698 | hash_symbol_hasher, | |
699 | hash_symbol_comparator, | |
700 | free); | |
701 | semantic_type_table = hash_initialize (HT_INITIAL_CAPACITY, | |
702 | NULL, | |
703 | hash_semantic_type_hasher, | |
704 | hash_semantic_type_comparator, | |
705 | free); | |
706 | } | |
707 | ||
708 | ||
709 | /*----------------------------------------------------------------. | |
710 | | Find the symbol named KEY, and return it. If it does not exist | | |
711 | | yet, create it. | | |
712 | `----------------------------------------------------------------*/ | |
713 | ||
714 | symbol * | |
715 | symbol_from_uniqstr (const uniqstr key, location loc) | |
716 | { | |
717 | symbol probe; | |
718 | symbol *entry; | |
719 | ||
720 | probe.tag = key; | |
721 | entry = hash_lookup (symbol_table, &probe); | |
722 | ||
723 | if (!entry) | |
724 | { | |
725 | /* First insertion in the hash. */ | |
726 | aver (!symbols_sorted); | |
727 | entry = symbol_new (key, loc); | |
728 | if (!hash_insert (symbol_table, entry)) | |
729 | xalloc_die (); | |
730 | } | |
731 | return entry; | |
732 | } | |
733 | ||
734 | ||
735 | /*-----------------------------------------------------------------------. | |
736 | | Find the semantic type named KEY, and return it. If it does not exist | | |
737 | | yet, create it. | | |
738 | `-----------------------------------------------------------------------*/ | |
739 | ||
740 | semantic_type * | |
741 | semantic_type_from_uniqstr (const uniqstr key, const location *loc) | |
742 | { | |
743 | semantic_type probe; | |
744 | semantic_type *entry; | |
745 | ||
746 | probe.tag = key; | |
747 | entry = hash_lookup (semantic_type_table, &probe); | |
748 | ||
749 | if (!entry) | |
750 | { | |
751 | /* First insertion in the hash. */ | |
752 | entry = semantic_type_new (key, loc); | |
753 | if (!hash_insert (semantic_type_table, entry)) | |
754 | xalloc_die (); | |
755 | } | |
756 | return entry; | |
757 | } | |
758 | ||
759 | ||
760 | /*----------------------------------------------------------------. | |
761 | | Find the symbol named KEY, and return it. If it does not exist | | |
762 | | yet, create it. | | |
763 | `----------------------------------------------------------------*/ | |
764 | ||
765 | symbol * | |
766 | symbol_get (const char *key, location loc) | |
767 | { | |
768 | return symbol_from_uniqstr (uniqstr_new (key), loc); | |
769 | } | |
770 | ||
771 | ||
772 | /*-----------------------------------------------------------------------. | |
773 | | Find the semantic type named KEY, and return it. If it does not exist | | |
774 | | yet, create it. | | |
775 | `-----------------------------------------------------------------------*/ | |
776 | ||
777 | semantic_type * | |
778 | semantic_type_get (const char *key, const location *loc) | |
779 | { | |
780 | return semantic_type_from_uniqstr (uniqstr_new (key), loc); | |
781 | } | |
782 | ||
783 | ||
784 | /*------------------------------------------------------------------. | |
785 | | Generate a dummy nonterminal, whose name cannot conflict with the | | |
786 | | user's names. | | |
787 | `------------------------------------------------------------------*/ | |
788 | ||
789 | symbol * | |
790 | dummy_symbol_get (location loc) | |
791 | { | |
792 | /* Incremented for each generated symbol. */ | |
793 | static int dummy_count = 0; | |
794 | static char buf[256]; | |
795 | ||
796 | symbol *sym; | |
797 | ||
798 | sprintf (buf, "$@%d", ++dummy_count); | |
799 | sym = symbol_get (buf, loc); | |
800 | sym->class = nterm_sym; | |
801 | sym->number = nvars++; | |
802 | return sym; | |
803 | } | |
804 | ||
805 | bool | |
806 | symbol_is_dummy (const symbol *sym) | |
807 | { | |
808 | return sym->tag[0] == '@' || (sym->tag[0] == '$' && sym->tag[1] == '@'); | |
809 | } | |
810 | ||
811 | /*-------------------. | |
812 | | Free the symbols. | | |
813 | `-------------------*/ | |
814 | ||
815 | void | |
816 | symbols_free (void) | |
817 | { | |
818 | hash_free (symbol_table); | |
819 | hash_free (semantic_type_table); | |
820 | free (symbols); | |
821 | free (symbols_sorted); | |
822 | } | |
823 | ||
824 | ||
825 | /*---------------------------------------------------------------. | |
826 | | Look for undefined symbols, report an error, and consider them | | |
827 | | terminals. | | |
828 | `---------------------------------------------------------------*/ | |
829 | ||
830 | static int | |
831 | symbols_cmp (symbol const *a, symbol const *b) | |
832 | { | |
833 | return strcmp (a->tag, b->tag); | |
834 | } | |
835 | ||
836 | static int | |
837 | symbols_cmp_qsort (void const *a, void const *b) | |
838 | { | |
839 | return symbols_cmp (*(symbol * const *)a, *(symbol * const *)b); | |
840 | } | |
841 | ||
842 | static void | |
843 | symbols_do (Hash_processor processor, void *processor_data, | |
844 | struct hash_table *table, symbol **sorted) | |
845 | { | |
846 | size_t count = hash_get_n_entries (table); | |
847 | if (!sorted) | |
848 | { | |
849 | sorted = xnmalloc (count, sizeof *sorted); | |
850 | hash_get_entries (table, (void**)sorted, count); | |
851 | qsort (sorted, count, sizeof *sorted, symbols_cmp_qsort); | |
852 | } | |
853 | { | |
854 | size_t i; | |
855 | for (i = 0; i < count; ++i) | |
856 | processor (sorted[i], processor_data); | |
857 | } | |
858 | } | |
859 | ||
860 | /*--------------------------------------------------------------. | |
861 | | Check that all the symbols are defined. Report any undefined | | |
862 | | symbols and consider them nonterminals. | | |
863 | `--------------------------------------------------------------*/ | |
864 | ||
865 | void | |
866 | symbols_check_defined (void) | |
867 | { | |
868 | symbols_do (symbol_check_defined_processor, NULL, | |
869 | symbol_table, symbols_sorted); | |
870 | symbols_do (semantic_type_check_defined_processor, NULL, | |
871 | semantic_type_table, semantic_types_sorted); | |
872 | } | |
873 | ||
874 | /*------------------------------------------------------------------. | |
875 | | Set TOKEN_TRANSLATIONS. Check that no two symbols share the same | | |
876 | | number. | | |
877 | `------------------------------------------------------------------*/ | |
878 | ||
879 | static void | |
880 | symbols_token_translations_init (void) | |
881 | { | |
882 | bool num_256_available_p = true; | |
883 | int i; | |
884 | ||
885 | /* Find the highest user token number, and whether 256, the POSIX | |
886 | preferred user token number for the error token, is used. */ | |
887 | max_user_token_number = 0; | |
888 | for (i = 0; i < ntokens; ++i) | |
889 | { | |
890 | symbol *this = symbols[i]; | |
891 | if (this->user_token_number != USER_NUMBER_UNDEFINED) | |
892 | { | |
893 | if (this->user_token_number > max_user_token_number) | |
894 | max_user_token_number = this->user_token_number; | |
895 | if (this->user_token_number == 256) | |
896 | num_256_available_p = false; | |
897 | } | |
898 | } | |
899 | ||
900 | /* If 256 is not used, assign it to error, to follow POSIX. */ | |
901 | if (num_256_available_p | |
902 | && errtoken->user_token_number == USER_NUMBER_UNDEFINED) | |
903 | errtoken->user_token_number = 256; | |
904 | ||
905 | /* Set the missing user numbers. */ | |
906 | if (max_user_token_number < 256) | |
907 | max_user_token_number = 256; | |
908 | ||
909 | for (i = 0; i < ntokens; ++i) | |
910 | { | |
911 | symbol *this = symbols[i]; | |
912 | if (this->user_token_number == USER_NUMBER_UNDEFINED) | |
913 | this->user_token_number = ++max_user_token_number; | |
914 | if (this->user_token_number > max_user_token_number) | |
915 | max_user_token_number = this->user_token_number; | |
916 | } | |
917 | ||
918 | token_translations = xnmalloc (max_user_token_number + 1, | |
919 | sizeof *token_translations); | |
920 | ||
921 | /* Initialize all entries for literal tokens to the internal token | |
922 | number for $undefined, which represents all invalid inputs. */ | |
923 | for (i = 0; i < max_user_token_number + 1; i++) | |
924 | token_translations[i] = undeftoken->number; | |
925 | symbols_do (symbol_translation_processor, NULL, | |
926 | symbol_table, symbols_sorted); | |
927 | } | |
928 | ||
929 | ||
930 | /*----------------------------------------------------------------. | |
931 | | Assign symbol numbers, and write definition of token names into | | |
932 | | FDEFINES. Set up vectors SYMBOL_TABLE, TAGS of symbols. | | |
933 | `----------------------------------------------------------------*/ | |
934 | ||
935 | void | |
936 | symbols_pack (void) | |
937 | { | |
938 | symbols_do (symbol_check_alias_consistency_processor, NULL, | |
939 | symbol_table, symbols_sorted); | |
940 | ||
941 | symbols = xcalloc (nsyms, sizeof *symbols); | |
942 | symbols_do (symbol_pack_processor, NULL, symbol_table, symbols_sorted); | |
943 | ||
944 | /* Aliases leave empty slots in symbols, so remove them. */ | |
945 | { | |
946 | int writei; | |
947 | int readi; | |
948 | int nsyms_old = nsyms; | |
949 | for (writei = 0, readi = 0; readi < nsyms_old; readi += 1) | |
950 | { | |
951 | if (symbols[readi] == NULL) | |
952 | { | |
953 | nsyms -= 1; | |
954 | ntokens -= 1; | |
955 | } | |
956 | else | |
957 | { | |
958 | symbols[writei] = symbols[readi]; | |
959 | symbols[writei]->number = writei; | |
960 | if (symbols[writei]->alias) | |
961 | symbols[writei]->alias->number = writei; | |
962 | writei += 1; | |
963 | } | |
964 | } | |
965 | } | |
966 | symbols = xnrealloc (symbols, nsyms, sizeof *symbols); | |
967 | ||
968 | symbols_token_translations_init (); | |
969 | ||
970 | if (startsymbol->class == unknown_sym) | |
971 | fatal_at (startsymbol_location, | |
972 | _("the start symbol %s is undefined"), | |
973 | startsymbol->tag); | |
974 | else if (startsymbol->class == token_sym) | |
975 | fatal_at (startsymbol_location, | |
976 | _("the start symbol %s is a token"), | |
977 | startsymbol->tag); | |
978 | } | |
979 | ||
980 | ||
981 | /*--------------------------------------------------. | |
982 | | Set default tagged/tagless %destructor/%printer. | | |
983 | `--------------------------------------------------*/ | |
984 | ||
985 | void | |
986 | default_tagged_code_props_set (code_props_type kind, code_props const *code) | |
987 | { | |
988 | if (default_tagged_code_props[kind].code) | |
989 | { | |
990 | complain_at (code->location, | |
991 | _("redeclaration for default tagged %s"), | |
992 | code_props_type_string (kind)); | |
993 | complain_at (default_tagged_code_props[kind].location, | |
994 | _("previous declaration")); | |
995 | } | |
996 | default_tagged_code_props[kind] = *code; | |
997 | } | |
998 | ||
999 | void | |
1000 | default_tagless_code_props_set (code_props_type kind, code_props const *code) | |
1001 | { | |
1002 | if (default_tagless_code_props[kind].code) | |
1003 | { | |
1004 | complain_at (code->location, | |
1005 | _("redeclaration for default tagless %s"), | |
1006 | code_props_type_string (kind)); | |
1007 | complain_at (default_tagless_code_props[kind].location, | |
1008 | _("previous declaration")); | |
1009 | } | |
1010 | default_tagless_code_props[kind] = *code; | |
1011 | } |