1 /* Symbol table manager for Bison.
3 Copyright (C) 1984, 1989, 2000, 2001, 2002, 2004, 2005, 2006 Free
4 Software Foundation, Inc.
6 This file is part of Bison, the GNU Compiler Compiler.
8 Bison 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 2, or (at your option)
13 Bison 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.
18 You should have received a copy of the GNU General Public License
19 along with Bison; see the file COPYING. If not, write to
20 the Free Software Foundation, Inc., 51 Franklin Street, Fifth Floor,
21 Boston, MA 02110-1301, USA. */
33 /*------------------------.
34 | Distinguished symbols. |
35 `------------------------*/
37 symbol
*errtoken
= NULL
;
38 symbol
*undeftoken
= NULL
;
39 symbol
*endtoken
= NULL
;
40 symbol
*accept
= NULL
;
41 symbol
*startsymbol
= NULL
;
42 location startsymbol_location
;
44 /*-----------------------------------.
45 | Default %destructor and %printer. |
46 `-----------------------------------*/
48 static const char *default_destructor
= NULL
;
49 static location default_destructor_location
;
50 static const char *default_printer
= NULL
;
51 static location default_printer_location
;
53 /*---------------------------------.
54 | Create a new symbol, named TAG. |
55 `---------------------------------*/
58 symbol_new (uniqstr tag
, location loc
)
60 symbol
*res
= xmalloc (sizeof *res
);
66 res
->type_name
= NULL
;
67 res
->destructor
= NULL
;
70 res
->number
= NUMBER_UNDEFINED
;
72 res
->assoc
= undef_assoc
;
73 res
->user_token_number
= USER_NUMBER_UNDEFINED
;
76 res
->class = unknown_sym
;
77 res
->declared
= false;
79 if (nsyms
== SYMBOL_NUMBER_MAXIMUM
)
80 fatal (_("too many symbols in input grammar (limit is %d)"),
81 SYMBOL_NUMBER_MAXIMUM
);
86 /*----------------------------------------.
87 | Create a new semantic type, named TAG. |
88 `----------------------------------------*/
90 static semantic_type
*
91 semantic_type_new (uniqstr tag
)
93 semantic_type
*res
= xmalloc (sizeof *res
);
97 res
->destructor
= NULL
;
108 #define SYMBOL_ATTR_PRINT(Attr) \
110 fprintf (f, " %s { %s }", #Attr, s->Attr)
113 symbol_print (symbol
*s
, FILE *f
)
117 fprintf (f
, "\"%s\"", s
->tag
);
118 SYMBOL_ATTR_PRINT (type_name
);
119 SYMBOL_ATTR_PRINT (destructor
);
120 SYMBOL_ATTR_PRINT (printer
);
123 fprintf (f
, "<NULL>");
126 #undef SYMBOL_ATTR_PRINT
128 /*------------------------------------------------------------------.
129 | Complain that S's WHAT is redeclared at SECOND, and was first set |
131 `------------------------------------------------------------------*/
134 symbol_redeclaration (symbol
*s
, const char *what
, location first
,
137 complain_at (second
, _("%s redeclaration for %s"), what
, s
->tag
);
138 complain_at (first
, _("previous declaration"));
142 semantic_type_redeclaration (semantic_type
*s
, const char *what
, location first
,
145 complain_at (second
, _("%s redeclaration for <%s>"), what
, s
->tag
);
146 complain_at (first
, _("previous declaration"));
150 /*-----------------------------------------------------------------.
151 | Set the TYPE_NAME associated with SYM. Does nothing if passed 0 |
153 `-----------------------------------------------------------------*/
156 symbol_type_set (symbol
*sym
, uniqstr type_name
, location loc
)
161 symbol_redeclaration (sym
, "%type", sym
->type_location
, loc
);
162 uniqstr_assert (type_name
);
163 sym
->type_name
= type_name
;
164 sym
->type_location
= loc
;
169 /*------------------------------------------------------------------.
170 | Set the DESTRUCTOR associated with SYM. Do nothing if passed 0. |
171 `------------------------------------------------------------------*/
174 symbol_destructor_set (symbol
*sym
, const char *destructor
, location loc
)
179 symbol_redeclaration (sym
, "%destructor", sym
->destructor_location
,
181 sym
->destructor
= destructor
;
182 sym
->destructor_location
= loc
;
186 /*-------------------------------------------------------------------.
187 | Set the DESTRUCTOR associated with TYPE. Do nothing if passed 0. |
188 `-------------------------------------------------------------------*/
191 semantic_type_destructor_set (semantic_type
*type
, const char *destructor
,
196 if (type
->destructor
)
197 semantic_type_redeclaration (type
, "%destructor",
198 type
->destructor_location
, loc
);
199 type
->destructor
= destructor
;
200 type
->destructor_location
= loc
;
204 /*---------------------------------------.
205 | Get the computed %destructor for SYM. |
206 `---------------------------------------*/
209 symbol_destructor_get (symbol
*sym
)
211 /* Per-symbol %destructor. */
212 if (sym
->destructor
!= NULL
)
213 return sym
->destructor
;
215 /* Per-type %destructor. */
218 semantic_type
*type
= semantic_type_get (sym
->type_name
);
219 if (type
->destructor
)
220 return type
->destructor
;
223 /* Apply the default %destructor only to user-defined symbols. */
224 if (sym
->tag
[0] == '$' || sym
== errtoken
)
226 return default_destructor
;
229 /*---------------------------------------------------------------.
230 | Get the grammar location of the %destructor computed for SYM. |
231 `---------------------------------------------------------------*/
234 symbol_destructor_location_get (symbol
*sym
)
236 if (sym
->destructor
!= NULL
)
237 return sym
->destructor_location
;
240 semantic_type
*type
= semantic_type_get (sym
->type_name
);
241 if (type
->destructor
)
242 return type
->destructor_location
;
244 return default_destructor_location
;
247 /*---------------------------------------------------------------.
248 | Set the PRINTER associated with SYM. Do nothing if passed 0. |
249 `---------------------------------------------------------------*/
252 symbol_printer_set (symbol
*sym
, const char *printer
, location loc
)
257 symbol_redeclaration (sym
, "%printer", sym
->printer_location
, loc
);
258 sym
->printer
= printer
;
259 sym
->printer_location
= loc
;
263 /*----------------------------------------------------------------.
264 | Set the PRINTER associated with TYPE. Do nothing if passed 0. |
265 `----------------------------------------------------------------*/
268 semantic_type_printer_set (semantic_type
*type
, const char *printer
,
274 semantic_type_redeclaration (type
, "%printer", type
->printer_location
,
276 type
->printer
= printer
;
277 type
->printer_location
= loc
;
281 /*------------------------------------.
282 | Get the computed %printer for SYM. |
283 `------------------------------------*/
286 symbol_printer_get (symbol
*sym
)
288 /* Per-symbol %printer. */
289 if (sym
->printer
!= NULL
)
292 /* Per-type %printer. */
295 semantic_type
*type
= semantic_type_get (sym
->type_name
);
297 return type
->printer
;
300 /* Apply the default %printer only to user-defined symbols. */
301 if (sym
->tag
[0] == '$' || sym
== errtoken
)
303 return default_printer
;
306 /*------------------------------------------------------------.
307 | Get the grammar location of the %printer computed for SYM. |
308 `------------------------------------------------------------*/
311 symbol_printer_location_get (symbol
*sym
)
313 if (sym
->printer
!= NULL
)
314 return sym
->printer_location
;
317 semantic_type
*type
= semantic_type_get (sym
->type_name
);
319 return type
->printer_location
;
321 return default_printer_location
;
325 /*-----------------------------------------------------------------.
326 | Set the PRECEDENCE associated with SYM. Does nothing if invoked |
327 | with UNDEF_ASSOC as ASSOC. |
328 `-----------------------------------------------------------------*/
331 symbol_precedence_set (symbol
*sym
, int prec
, assoc a
, location loc
)
333 if (a
!= undef_assoc
)
336 symbol_redeclaration (sym
, assoc_to_string (a
), sym
->prec_location
,
340 sym
->prec_location
= loc
;
343 /* Only terminals have a precedence. */
344 symbol_class_set (sym
, token_sym
, loc
, false);
348 /*------------------------------------.
349 | Set the CLASS associated with SYM. |
350 `------------------------------------*/
353 symbol_class_set (symbol
*sym
, symbol_class
class, location loc
, bool declaring
)
355 if (sym
->class != unknown_sym
&& sym
->class != class)
357 complain_at (loc
, _("symbol %s redefined"), sym
->tag
);
358 sym
->declared
= false;
361 if (class == nterm_sym
&& sym
->class != nterm_sym
)
362 sym
->number
= nvars
++;
363 else if (class == token_sym
&& sym
->number
== NUMBER_UNDEFINED
)
364 sym
->number
= ntokens
++;
371 warn_at (loc
, _("symbol %s redeclared"), sym
->tag
);
372 sym
->declared
= true;
377 /*------------------------------------------------.
378 | Set the USER_TOKEN_NUMBER associated with SYM. |
379 `------------------------------------------------*/
382 symbol_user_token_number_set (symbol
*sym
, int user_token_number
, location loc
)
384 int *user_token_numberp
;
386 assert (sym
->class == token_sym
);
388 if (sym
->user_token_number
!= USER_NUMBER_ALIAS
)
389 user_token_numberp
= &sym
->user_token_number
;
391 user_token_numberp
= &sym
->alias
->user_token_number
;
392 if (*user_token_numberp
!= USER_NUMBER_UNDEFINED
393 && *user_token_numberp
!= user_token_number
)
394 complain_at (loc
, _("redefining user token number of %s"), sym
->tag
);
396 *user_token_numberp
= user_token_number
;
397 /* User defined $end token? */
398 if (user_token_number
== 0)
401 endtoken
->number
= 0;
402 /* It is always mapped to 0, so it was already counted in
409 /*----------------------------------------------------------.
410 | If SYM is not defined, report an error, and consider it a |
412 `----------------------------------------------------------*/
415 symbol_check_defined (symbol
*sym
)
417 if (sym
->class == unknown_sym
)
421 _("symbol %s is used, but is not defined as a token and has no rules"),
423 sym
->class = nterm_sym
;
424 sym
->number
= nvars
++;
431 symbol_check_defined_processor (void *sym
, void *null ATTRIBUTE_UNUSED
)
433 return symbol_check_defined (sym
);
437 /*------------------------------------------------------------------.
438 | Declare the new symbol SYM. Make it an alias of SYMVAL, and type |
439 | SYMVAL with SYM's type. |
440 `------------------------------------------------------------------*/
443 symbol_make_alias (symbol
*sym
, symbol
*symval
, location loc
)
446 warn_at (loc
, _("symbol `%s' used more than once as a literal string"),
449 warn_at (loc
, _("symbol `%s' given more than one literal string"),
453 symval
->class = token_sym
;
454 symval
->user_token_number
= sym
->user_token_number
;
455 sym
->user_token_number
= USER_NUMBER_ALIAS
;
458 symval
->number
= sym
->number
;
459 symbol_type_set (symval
, sym
->type_name
, loc
);
464 /*---------------------------------------------------------.
465 | Check that THIS, and its alias, have same precedence and |
467 `---------------------------------------------------------*/
470 symbol_check_alias_consistency (symbol
*this)
472 symbol
*alias
= this;
473 symbol
*orig
= this->alias
;
475 /* Check only those that _are_ the aliases. */
476 if (!(this->alias
&& this->user_token_number
== USER_NUMBER_ALIAS
))
479 if (orig
->type_name
!= alias
->type_name
)
482 symbol_type_set (alias
, orig
->type_name
, orig
->type_location
);
484 symbol_type_set (orig
, alias
->type_name
, alias
->type_location
);
488 if (orig
->destructor
|| alias
->destructor
)
490 if (orig
->destructor
)
491 symbol_destructor_set (alias
, orig
->destructor
,
492 orig
->destructor_location
);
494 symbol_destructor_set (orig
, alias
->destructor
,
495 alias
->destructor_location
);
498 if (orig
->printer
|| alias
->printer
)
501 symbol_printer_set (alias
, orig
->printer
, orig
->printer_location
);
503 symbol_printer_set (orig
, alias
->printer
, alias
->printer_location
);
506 if (alias
->prec
|| orig
->prec
)
509 symbol_precedence_set (alias
, orig
->prec
, orig
->assoc
,
510 orig
->prec_location
);
512 symbol_precedence_set (orig
, alias
->prec
, alias
->assoc
,
513 alias
->prec_location
);
518 symbol_check_alias_consistency_processor (void *this,
519 void *null ATTRIBUTE_UNUSED
)
521 symbol_check_alias_consistency (this);
526 /*-------------------------------------------------------------------.
527 | Assign a symbol number, and write the definition of the token name |
528 | into FDEFINES. Put in SYMBOLS. |
529 `-------------------------------------------------------------------*/
532 symbol_pack (symbol
*this)
534 if (this->class == nterm_sym
)
536 this->number
+= ntokens
;
538 else if (this->alias
)
540 /* This symbol and its alias are a single token defn.
541 Allocate a tokno, and assign to both check agreement of
542 prec and assoc fields and make both the same */
543 if (this->number
== NUMBER_UNDEFINED
)
545 if (this == endtoken
|| this->alias
== endtoken
)
546 this->number
= this->alias
->number
= 0;
549 assert (this->alias
->number
!= NUMBER_UNDEFINED
);
550 this->number
= this->alias
->number
;
553 /* Do not do processing below for USER_NUMBER_ALIASes. */
554 if (this->user_token_number
== USER_NUMBER_ALIAS
)
557 else /* this->class == token_sym */
558 assert (this->number
!= NUMBER_UNDEFINED
);
560 symbols
[this->number
] = this;
565 symbol_pack_processor (void *this, void *null ATTRIBUTE_UNUSED
)
567 return symbol_pack (this);
573 /*--------------------------------------------------.
574 | Put THIS in TOKEN_TRANSLATIONS if it is a token. |
575 `--------------------------------------------------*/
578 symbol_translation (symbol
*this)
581 if (this->class == token_sym
582 && this->user_token_number
!= USER_NUMBER_ALIAS
)
584 /* A token which translation has already been set? */
585 if (token_translations
[this->user_token_number
] != undeftoken
->number
)
586 complain_at (this->location
,
587 _("tokens %s and %s both assigned number %d"),
588 symbols
[token_translations
[this->user_token_number
]]->tag
,
589 this->tag
, this->user_token_number
);
591 token_translations
[this->user_token_number
] = this->number
;
598 symbol_translation_processor (void *this, void *null ATTRIBUTE_UNUSED
)
600 return symbol_translation (this);
604 /*---------------------------------------.
605 | Symbol and semantic type hash tables. |
606 `---------------------------------------*/
608 /* Initial capacity of symbol and semantic type hash table. */
609 #define HT_INITIAL_CAPACITY 257
611 static struct hash_table
*symbol_table
= NULL
;
612 static struct hash_table
*semantic_type_table
= NULL
;
615 hash_compare_symbol (const symbol
*m1
, const symbol
*m2
)
617 /* Since tags are unique, we can compare the pointers themselves. */
618 return UNIQSTR_EQ (m1
->tag
, m2
->tag
);
622 hash_compare_semantic_type (const semantic_type
*m1
, const semantic_type
*m2
)
624 /* Since names are unique, we can compare the pointers themselves. */
625 return UNIQSTR_EQ (m1
->tag
, m2
->tag
);
629 hash_symbol_comparator (void const *m1
, void const *m2
)
631 return hash_compare_symbol (m1
, m2
);
635 hash_semantic_type_comparator (void const *m1
, void const *m2
)
637 return hash_compare_semantic_type (m1
, m2
);
641 hash_symbol (const symbol
*m
, size_t tablesize
)
643 /* Since tags are unique, we can hash the pointer itself. */
644 return ((uintptr_t) m
->tag
) % tablesize
;
648 hash_semantic_type (const semantic_type
*m
, size_t tablesize
)
650 /* Since names are unique, we can hash the pointer itself. */
651 return ((uintptr_t) m
->tag
) % tablesize
;
655 hash_symbol_hasher (void const *m
, size_t tablesize
)
657 return hash_symbol (m
, tablesize
);
661 hash_semantic_type_hasher (void const *m
, size_t tablesize
)
663 return hash_semantic_type (m
, tablesize
);
666 /*-------------------------------.
667 | Create the symbol hash table. |
668 `-------------------------------*/
673 symbol_table
= hash_initialize (HT_INITIAL_CAPACITY
,
676 hash_symbol_comparator
,
678 semantic_type_table
= hash_initialize (HT_INITIAL_CAPACITY
,
680 hash_semantic_type_hasher
,
681 hash_semantic_type_comparator
,
686 /*----------------------------------------------------------------.
687 | Find the symbol named KEY, and return it. If it does not exist |
689 `----------------------------------------------------------------*/
692 symbol_from_uniqstr (const uniqstr key
, location loc
)
698 entry
= hash_lookup (symbol_table
, &probe
);
702 /* First insertion in the hash. */
703 entry
= symbol_new (key
, loc
);
704 hash_insert (symbol_table
, entry
);
710 /*-----------------------------------------------------------------------.
711 | Find the semantic type named KEY, and return it. If it does not exist |
713 `-----------------------------------------------------------------------*/
716 semantic_type_from_uniqstr (const uniqstr key
)
719 semantic_type
*entry
;
722 entry
= hash_lookup (semantic_type_table
, &probe
);
726 /* First insertion in the hash. */
727 entry
= semantic_type_new (key
);
728 hash_insert (semantic_type_table
, entry
);
734 /*----------------------------------------------------------------.
735 | Find the symbol named KEY, and return it. If it does not exist |
737 `----------------------------------------------------------------*/
740 symbol_get (const char *key
, location loc
)
742 return symbol_from_uniqstr (uniqstr_new (key
), loc
);
746 /*-----------------------------------------------------------------------.
747 | Find the semantic type named KEY, and return it. If it does not exist |
749 `-----------------------------------------------------------------------*/
752 semantic_type_get (const char *key
)
754 return semantic_type_from_uniqstr (uniqstr_new (key
));
758 /*------------------------------------------------------------------.
759 | Generate a dummy nonterminal, whose name cannot conflict with the |
761 `------------------------------------------------------------------*/
764 dummy_symbol_get (location loc
)
766 /* Incremented for each generated symbol. */
767 static int dummy_count
= 0;
768 static char buf
[256];
772 sprintf (buf
, "@%d", ++dummy_count
);
773 sym
= symbol_get (buf
, loc
);
774 sym
->class = nterm_sym
;
775 sym
->number
= nvars
++;
780 symbol_is_dummy (const symbol
*sym
)
782 return sym
->tag
[0] == '@';
785 /*-------------------.
786 | Free the symbols. |
787 `-------------------*/
792 hash_free (symbol_table
);
793 hash_free (semantic_type_table
);
798 /*---------------------------------------------------------------.
799 | Look for undefined symbols, report an error, and consider them |
801 `---------------------------------------------------------------*/
804 symbols_do (Hash_processor processor
, void *processor_data
)
806 hash_do_for_each (symbol_table
, processor
, processor_data
);
810 /*--------------------------------------------------------------.
811 | Check that all the symbols are defined. Report any undefined |
812 | symbols and consider them nonterminals. |
813 `--------------------------------------------------------------*/
816 symbols_check_defined (void)
818 symbols_do (symbol_check_defined_processor
, NULL
);
821 /*------------------------------------------------------------------.
822 | Set TOKEN_TRANSLATIONS. Check that no two symbols share the same |
824 `------------------------------------------------------------------*/
827 symbols_token_translations_init (void)
829 bool num_256_available_p
= true;
832 /* Find the highest user token number, and whether 256, the POSIX
833 preferred user token number for the error token, is used. */
834 max_user_token_number
= 0;
835 for (i
= 0; i
< ntokens
; ++i
)
837 symbol
*this = symbols
[i
];
838 if (this->user_token_number
!= USER_NUMBER_UNDEFINED
)
840 if (this->user_token_number
> max_user_token_number
)
841 max_user_token_number
= this->user_token_number
;
842 if (this->user_token_number
== 256)
843 num_256_available_p
= false;
847 /* If 256 is not used, assign it to error, to follow POSIX. */
848 if (num_256_available_p
849 && errtoken
->user_token_number
== USER_NUMBER_UNDEFINED
)
850 errtoken
->user_token_number
= 256;
852 /* Set the missing user numbers. */
853 if (max_user_token_number
< 256)
854 max_user_token_number
= 256;
856 for (i
= 0; i
< ntokens
; ++i
)
858 symbol
*this = symbols
[i
];
859 if (this->user_token_number
== USER_NUMBER_UNDEFINED
)
860 this->user_token_number
= ++max_user_token_number
;
861 if (this->user_token_number
> max_user_token_number
)
862 max_user_token_number
= this->user_token_number
;
865 token_translations
= xnmalloc (max_user_token_number
+ 1,
866 sizeof *token_translations
);
868 /* Initialize all entries for literal tokens to 2, the internal
869 token number for $undefined, which represents all invalid inputs.
871 for (i
= 0; i
< max_user_token_number
+ 1; i
++)
872 token_translations
[i
] = undeftoken
->number
;
873 symbols_do (symbol_translation_processor
, NULL
);
877 /*----------------------------------------------------------------.
878 | Assign symbol numbers, and write definition of token names into |
879 | FDEFINES. Set up vectors SYMBOL_TABLE, TAGS of symbols. |
880 `----------------------------------------------------------------*/
885 symbols_do (symbol_check_alias_consistency_processor
, NULL
);
887 symbols
= xcalloc (nsyms
, sizeof *symbols
);
888 symbols_do (symbol_pack_processor
, NULL
);
890 /* Aliases leave empty slots in symbols, so remove them. */
894 int nsyms_old
= nsyms
;
895 for (writei
= 0, readi
= 0; readi
< nsyms_old
; readi
+= 1)
897 if (symbols
[readi
] == NULL
)
904 symbols
[writei
] = symbols
[readi
];
905 symbols
[writei
]->number
= writei
;
906 if (symbols
[writei
]->alias
)
907 symbols
[writei
]->alias
->number
= writei
;
912 symbols
= xnrealloc (symbols
, nsyms
, sizeof *symbols
);
914 symbols_token_translations_init ();
916 if (startsymbol
->class == unknown_sym
)
917 fatal_at (startsymbol_location
,
918 _("the start symbol %s is undefined"),
920 else if (startsymbol
->class == token_sym
)
921 fatal_at (startsymbol_location
,
922 _("the start symbol %s is a token"),
927 /*-----------------------------------.
928 | Set default %destructor/%printer. |
929 `-----------------------------------*/
932 default_destructor_set (const char *destructor
, location loc
)
934 if (default_destructor
!= NULL
)
936 complain_at (loc
, _("redeclaration for default %%destructor"));
937 complain_at (default_destructor_location
, _("previous declaration"));
939 default_destructor
= destructor
;
940 default_destructor_location
= loc
;
944 default_printer_set (const char *printer
, location loc
)
946 if (default_printer
!= NULL
)
948 complain_at (loc
, _("redeclaration for default %%printer"));
949 complain_at (default_printer_location
, _("previous declaration"));
951 default_printer
= printer
;
952 default_printer_location
= loc
;