1 /* Input parser for Bison
3 Copyright (C) 1984, 1986, 1989, 1992, 1998, 2000-2003, 2005-2007,
4 2009-2012 Free Software Foundation, Inc.
6 This file is part of Bison, the GNU Compiler Compiler.
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.
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.
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/>. */
27 #include "conflicts.h"
31 #include "muscle-tab.h"
35 #include "scan-gram.h"
36 #include "scan-code.h"
38 static void prepare_percent_define_front_end_variables (void);
39 static void check_and_convert_grammar (void);
41 static symbol_list
*grammar
= NULL
;
42 static bool start_flag
= false;
43 merger_list
*merge_functions
;
45 /* Was %union seen? */
46 bool union_seen
= false;
49 bool tag_seen
= false;
51 /* Should rules have a default precedence? */
52 bool default_prec
= true;
54 /*-----------------------.
55 | Set the start symbol. |
56 `-----------------------*/
59 grammar_start_symbol_set (symbol
*sym
, location loc
)
62 complain_at (loc
, _("multiple %s declarations"), "%start");
67 startsymbol_location
= loc
;
73 /*------------------------------------------------------------------------.
74 | Return the merger index for a merging function named NAME. Records the |
75 | function, if new, in MERGER_LIST. |
76 `------------------------------------------------------------------------*/
79 get_merge_function (uniqstr name
)
88 head
.next
= merge_functions
;
89 for (syms
= &head
, n
= 1; syms
->next
; syms
= syms
->next
, n
+= 1)
90 if (UNIQSTR_EQ (name
, syms
->next
->name
))
92 if (syms
->next
== NULL
)
94 syms
->next
= xmalloc (sizeof syms
->next
[0]);
95 syms
->next
->name
= uniqstr_new (name
);
96 /* After all symbol type declarations have been parsed, packgram invokes
97 record_merge_function_type to set the type. */
98 syms
->next
->type
= NULL
;
99 syms
->next
->next
= NULL
;
100 merge_functions
= head
.next
;
105 /*-------------------------------------------------------------------------.
106 | For the existing merging function with index MERGER, record the result |
107 | type as TYPE as required by the lhs of the rule whose %merge declaration |
108 | is at DECLARATION_LOC. |
109 `-------------------------------------------------------------------------*/
112 record_merge_function_type (int merger
, uniqstr type
, location declaration_loc
)
115 merger_list
*merge_function
;
121 type
= uniqstr_new ("");
124 for (merge_function
= merge_functions
;
125 merge_function
!= NULL
&& merger_find
!= merger
;
126 merge_function
= merge_function
->next
)
128 aver (merge_function
!= NULL
&& merger_find
== merger
);
129 if (merge_function
->type
!= NULL
&& !UNIQSTR_EQ (merge_function
->type
, type
))
132 complain_at_indent (declaration_loc
, &indent
,
133 _("result type clash on merge function %s: "
135 quote (merge_function
->name
), type
,
136 merge_function
->type
);
137 indent
+= SUB_INDENT
;
138 complain_at_indent (merge_function
->type_declaration_location
, &indent
,
139 _("previous declaration"));
141 merge_function
->type
= uniqstr_new (type
);
142 merge_function
->type_declaration_location
= declaration_loc
;
145 /*--------------------------------------.
146 | Free all merge-function definitions. |
147 `--------------------------------------*/
150 free_merger_functions (void)
152 merger_list
*L0
= merge_functions
;
155 merger_list
*L1
= L0
->next
;
162 /*-------------------------------------------------------------------.
163 | Parse the input grammar into a one symbol_list structure. Each |
164 | rule is represented by a sequence of symbols: the left hand side |
165 | followed by the contents of the right hand side, followed by a |
166 | null pointer instead of a symbol to terminate the rule. The next |
167 | symbol is the lhs of the following rule. |
169 | All actions are copied out, labelled by the rule number they apply |
171 `-------------------------------------------------------------------*/
173 /* The (currently) last symbol of GRAMMAR. */
174 static symbol_list
*grammar_end
= NULL
;
176 /* Append SYM to the grammar. */
178 grammar_symbol_append (symbol
*sym
, location loc
)
180 symbol_list
*p
= symbol_list_sym_new (sym
, loc
);
183 grammar_end
->next
= p
;
189 /* A null SYM stands for an end of rule; it is not an actual
198 assign_named_ref (symbol_list
*p
, named_ref
*name
)
200 symbol
*sym
= p
->content
.sym
;
202 if (name
->id
== sym
->tag
)
205 _("duplicated symbol name for %s ignored"),
207 named_ref_free (name
);
214 /* The rule currently being defined, and the previous rule.
215 CURRENT_RULE points to the first LHS of the current rule, while
216 PREVIOUS_RULE_END points to the *end* of the previous rule (NULL). */
217 static symbol_list
*current_rule
= NULL
;
218 static symbol_list
*previous_rule_end
= NULL
;
221 /*----------------------------------------------.
222 | Create a new rule for LHS in to the GRAMMAR. |
223 `----------------------------------------------*/
226 grammar_current_rule_begin (symbol
*lhs
, location loc
,
231 /* Start a new rule and record its lhs. */
233 previous_rule_end
= grammar_end
;
235 p
= grammar_symbol_append (lhs
, loc
);
237 assign_named_ref (p
, named_ref_copy (lhs_name
));
239 current_rule
= grammar_end
;
241 /* Mark the rule's lhs as a nonterminal if not already so. */
242 if (lhs
->class == unknown_sym
)
244 lhs
->class = nterm_sym
;
248 else if (lhs
->class == token_sym
)
249 complain_at (loc
, _("rule given for %s, which is a token"), lhs
->tag
);
253 /*----------------------------------------------------------------------.
254 | A symbol should be used if either: |
255 | 1. It has a destructor. |
256 | 2. The symbol is a mid-rule symbol (i.e., the generated LHS |
257 | replacing a mid-rule action) that was assigned to or used, as in |
258 | "exp: { $$ = 1; } { $$ = $1; }". |
259 `----------------------------------------------------------------------*/
262 symbol_should_be_used (symbol_list
const *s
, bool *midrule_warning
)
264 if (symbol_destructor_get (s
->content
.sym
)->code
)
266 if ((s
->midrule
&& s
->midrule
->action_props
.is_value_used
)
267 || (s
->midrule_parent_rule
268 && symbol_list_n_get (s
->midrule_parent_rule
,
269 s
->midrule_parent_rhs_index
)
270 ->action_props
.is_value_used
))
272 *midrule_warning
= true;
278 /*----------------------------------------------------------------.
279 | Check that the rule R is properly defined. For instance, there |
280 | should be no type clash on the default action. |
281 `----------------------------------------------------------------*/
284 grammar_rule_check (const symbol_list
*r
)
288 If there is an action, then there is nothing we can do: the user
289 is allowed to shoot herself in the foot.
291 Don't worry about the default action if $$ is untyped, since $$'s
292 value can't be used. */
293 if (!r
->action_props
.code
&& r
->content
.sym
->type_name
)
295 symbol
*first_rhs
= r
->next
->content
.sym
;
296 /* If $$ is being set in default way, report if any type mismatch. */
299 char const *lhs_type
= r
->content
.sym
->type_name
;
300 const char *rhs_type
=
301 first_rhs
->type_name
? first_rhs
->type_name
: "";
302 if (!UNIQSTR_EQ (lhs_type
, rhs_type
))
303 warn_at (r
->location
,
304 _("type clash on default action: <%s> != <%s>"),
307 /* Warn if there is no default for $$ but we need one. */
309 warn_at (r
->location
,
310 _("empty rule for typed nonterminal, and no action"));
313 /* Check that symbol values that should be used are in fact used. */
315 symbol_list
const *l
= r
;
317 for (; l
&& l
->content
.sym
; l
= l
->next
, ++n
)
319 bool midrule_warning
= false;
320 if (!l
->action_props
.is_value_used
321 && symbol_should_be_used (l
, &midrule_warning
)
322 /* The default action, $$ = $1, `uses' both. */
323 && (r
->action_props
.code
|| (n
!= 0 && n
!= 1)))
325 void (*warn_at_ptr
)(location
, char const*, ...) =
326 midrule_warning
? midrule_value_at
: warn_at
;
328 warn_at_ptr (r
->location
, _("unused value: $%d"), n
);
330 warn_at_ptr (r
->location
, _("unset value: $$"));
335 /* See comments in grammar_current_rule_prec_set for how POSIX
336 mandates this complaint. It's only for identifiers, so skip
337 it for char literals and strings, which are always tokens. */
339 && r
->ruleprec
->tag
[0] != '\'' && r
->ruleprec
->tag
[0] != '"'
340 && !r
->ruleprec
->declared
&& !r
->ruleprec
->prec
)
341 warn_at (r
->location
, _("token for %%prec is not defined: %s"),
346 /*-------------------------------------.
347 | End the currently being grown rule. |
348 `-------------------------------------*/
351 grammar_current_rule_end (location loc
)
353 /* Put an empty link in the list to mark the end of this rule */
354 grammar_symbol_append (NULL
, grammar_end
->location
);
355 current_rule
->location
= loc
;
359 /*-------------------------------------------------------------------.
360 | The previous action turns out the be a mid-rule action. Attach it |
361 | to the current rule, i.e., create a dummy symbol, attach it this |
362 | mid-rule action, and append this dummy nonterminal to the current |
364 `-------------------------------------------------------------------*/
367 grammar_midrule_action (void)
369 /* Since the action was written out with this rule's number, we must
370 give the new rule this number by inserting the new rule before
373 /* Make a DUMMY nonterminal, whose location is that of the midrule
374 action. Create the MIDRULE. */
375 location dummy_location
= current_rule
->action_props
.location
;
376 symbol
*dummy
= dummy_symbol_get (dummy_location
);
377 symbol_list
*midrule
= symbol_list_sym_new (dummy
, dummy_location
);
379 /* Remember named_ref of previous action. */
380 named_ref
*action_name
= current_rule
->action_props
.named_ref
;
382 /* Make a new rule, whose body is empty, before the current one, so
383 that the action just read can belong to it. */
386 /* Attach its location and actions to that of the DUMMY. */
387 midrule
->location
= dummy_location
;
388 code_props_rule_action_init (&midrule
->action_props
,
389 current_rule
->action_props
.code
,
390 current_rule
->action_props
.location
,
392 code_props_none_init (¤t_rule
->action_props
);
394 if (previous_rule_end
)
395 previous_rule_end
->next
= midrule
;
399 /* End the dummy's rule. */
400 midrule
->next
= symbol_list_sym_new (NULL
, dummy_location
);
401 midrule
->next
->next
= current_rule
;
403 previous_rule_end
= midrule
->next
;
405 /* Insert the dummy nonterminal replacing the midrule action into
406 the current rule. Bind it to its dedicated rule. */
407 grammar_current_rule_symbol_append (dummy
, dummy_location
,
409 grammar_end
->midrule
= midrule
;
410 midrule
->midrule_parent_rule
= current_rule
;
411 midrule
->midrule_parent_rhs_index
= symbol_list_length (current_rule
->next
);
414 /* Set the precedence symbol of the current rule to PRECSYM. */
417 grammar_current_rule_prec_set (symbol
*precsym
, location loc
)
419 /* POSIX says that any identifier is a nonterminal if it does not
420 appear on the LHS of a grammar rule and is not defined by %token
421 or by one of the directives that assigns precedence to a token. We
422 ignore this here because the only kind of identifier that POSIX
423 allows to follow a %prec is a token and because assuming it's a
424 token now can produce more logical error messages. Nevertheless,
425 grammar_rule_check does obey what we believe is the real intent of
426 POSIX here: that an error be reported for any identifier that
427 appears after %prec but that is not defined separately as a
429 symbol_class_set (precsym
, token_sym
, loc
, false);
430 if (current_rule
->ruleprec
)
431 complain_at (loc
, _("only one %s allowed per rule"), "%prec");
432 current_rule
->ruleprec
= precsym
;
435 /* Attach dynamic precedence DPREC to the current rule. */
438 grammar_current_rule_dprec_set (int dprec
, location loc
)
441 warn_at (loc
, _("%s affects only GLR parsers"), "%dprec");
443 complain_at (loc
, _("%s must be followed by positive number"), "%dprec");
444 else if (current_rule
->dprec
!= 0)
445 complain_at (loc
, _("only one %s allowed per rule"), "%dprec");
446 current_rule
->dprec
= dprec
;
449 /* Attach a merge function NAME with argument type TYPE to current
453 grammar_current_rule_merge_set (uniqstr name
, location loc
)
456 warn_at (loc
, _("%s affects only GLR parsers"), "%merge");
457 if (current_rule
->merger
!= 0)
458 complain_at (loc
, _("only one %s allowed per rule"), "%merge");
459 current_rule
->merger
= get_merge_function (name
);
460 current_rule
->merger_declaration_location
= loc
;
463 /* Attach SYM to the current rule. If needed, move the previous
464 action as a mid-rule action. */
467 grammar_current_rule_symbol_append (symbol
*sym
, location loc
,
471 if (current_rule
->action_props
.code
)
472 grammar_midrule_action ();
473 p
= grammar_symbol_append (sym
, loc
);
475 assign_named_ref(p
, name
);
478 /* Attach an ACTION to the current rule. */
481 grammar_current_rule_action_append (const char *action
, location loc
,
484 if (current_rule
->action_props
.code
)
485 grammar_midrule_action ();
486 /* After all symbol declarations have been parsed, packgram invokes
487 code_props_translate_code. */
488 code_props_rule_action_init (¤t_rule
->action_props
, action
, loc
,
493 /*---------------------------------------------------------------.
494 | Convert the rules into the representation using RRHS, RLHS and |
496 `---------------------------------------------------------------*/
501 unsigned int itemno
= 0;
502 rule_number ruleno
= 0;
503 symbol_list
*p
= grammar
;
505 ritem
= xnmalloc (nritems
+ 1, sizeof *ritem
);
507 /* This sentinel is used by build_relations in gram.c. */
510 rules
= xnmalloc (nrules
, sizeof *rules
);
515 symbol
*ruleprec
= p
->ruleprec
;
516 record_merge_function_type (p
->merger
, p
->content
.sym
->type_name
,
517 p
->merger_declaration_location
);
518 rules
[ruleno
].user_number
= ruleno
;
519 rules
[ruleno
].number
= ruleno
;
520 rules
[ruleno
].lhs
= p
->content
.sym
;
521 rules
[ruleno
].rhs
= ritem
+ itemno
;
522 rules
[ruleno
].prec
= NULL
;
523 rules
[ruleno
].dprec
= p
->dprec
;
524 rules
[ruleno
].merger
= p
->merger
;
525 rules
[ruleno
].precsym
= NULL
;
526 rules
[ruleno
].location
= p
->location
;
527 rules
[ruleno
].useful
= true;
528 rules
[ruleno
].action
= p
->action_props
.code
;
529 rules
[ruleno
].action_location
= p
->action_props
.location
;
531 /* If the midrule's $$ is set or its $n is used, remove the `$' from the
532 symbol name so that it's a user-defined symbol so that the default
533 %destructor and %printer apply. */
534 if (p
->midrule_parent_rule
535 && (p
->action_props
.is_value_used
536 || symbol_list_n_get (p
->midrule_parent_rule
,
537 p
->midrule_parent_rhs_index
)
538 ->action_props
.is_value_used
))
539 p
->content
.sym
->tag
+= 1;
541 /* Don't check the generated rule 0. It has no action, so some rhs
542 symbols may appear unused, but the parsing algorithm ensures that
543 %destructor's are invoked appropriately. */
545 grammar_rule_check (p
);
547 for (p
= p
->next
; p
&& p
->content
.sym
; p
= p
->next
)
551 /* Don't allow rule_length == INT_MAX, since that might
552 cause confusion with strtol if INT_MAX == LONG_MAX. */
553 if (rule_length
== INT_MAX
)
554 fatal_at (rules
[ruleno
].location
, _("rule is too long"));
556 /* item_number = symbol_number.
557 But the former needs to contain more: negative rule numbers. */
559 symbol_number_as_item_number (p
->content
.sym
->number
);
560 /* A rule gets by default the precedence and associativity
561 of its last token. */
562 if (p
->content
.sym
->class == token_sym
&& default_prec
)
563 rules
[ruleno
].prec
= p
->content
.sym
;
566 /* If this rule has a %prec,
567 the specified symbol's precedence replaces the default. */
570 rules
[ruleno
].precsym
= ruleprec
;
571 rules
[ruleno
].prec
= ruleprec
;
573 /* An item ends by the rule number (negated). */
574 ritem
[itemno
++] = rule_number_as_item_number (ruleno
);
575 aver (itemno
< ITEM_NUMBER_MAX
);
577 aver (ruleno
< RULE_NUMBER_MAX
);
583 aver (itemno
== nritems
);
585 if (trace_flag
& trace_sets
)
586 ritem_print (stderr
);
589 /*------------------------------------------------------------------.
590 | Read in the grammar specification and record it in the format |
591 | described in gram.h. All actions are copied into ACTION_OBSTACK, |
592 | in each case forming the body of a C function (YYACTION) which |
593 | contains a switch statement to decide which action to execute. |
594 `------------------------------------------------------------------*/
599 /* Initialize the symbol table. */
602 /* Construct the accept symbol. */
603 accept
= symbol_get ("$accept", empty_location
);
604 accept
->class = nterm_sym
;
605 accept
->number
= nvars
++;
607 /* Construct the error token */
608 errtoken
= symbol_get ("error", empty_location
);
609 errtoken
->class = token_sym
;
610 errtoken
->number
= ntokens
++;
612 /* Construct a token that represents all undefined literal tokens.
613 It is always token number 2. */
614 undeftoken
= symbol_get ("$undefined", empty_location
);
615 undeftoken
->class = token_sym
;
616 undeftoken
->number
= ntokens
++;
618 gram_in
= xfopen (grammar_file
, "r");
620 gram__flex_debug
= trace_flag
& trace_scan
;
621 gram_debug
= trace_flag
& trace_parse
;
622 gram_scanner_initialize ();
624 prepare_percent_define_front_end_variables ();
626 if (! complaint_issued
)
627 check_and_convert_grammar ();
633 prepare_percent_define_front_end_variables (void)
635 /* Set %define front-end variable defaults. */
636 muscle_percent_define_default ("lr.keep-unreachable-states", "false");
639 /* IELR would be a better default, but LALR is historically the
641 muscle_percent_define_default ("lr.type", "lalr");
642 lr_type
= muscle_percent_define_get ("lr.type");
643 if (0 != strcmp (lr_type
, "canonical-lr"))
644 muscle_percent_define_default ("lr.default-reductions", "most");
646 muscle_percent_define_default ("lr.default-reductions", "accepting");
650 /* Check %define front-end variables. */
652 static char const * const values
[] = {
653 "lr.type", "lalr", "ielr", "canonical-lr", NULL
,
654 "lr.default-reductions", "most", "consistent", "accepting", NULL
,
657 muscle_percent_define_check_values (values
);
662 /*-------------------------------------------------------------.
663 | Check the grammar that has just been read, and convert it to |
665 `-------------------------------------------------------------*/
668 check_and_convert_grammar (void)
670 /* Grammar has been read. Do some checking. */
672 fatal (_("no rules in the input grammar"));
674 /* If the user did not define her ENDTOKEN, do it now. */
677 endtoken
= symbol_get ("$end", empty_location
);
678 endtoken
->class = token_sym
;
679 endtoken
->number
= 0;
680 /* Value specified by POSIX. */
681 endtoken
->user_token_number
= 0;
684 /* Report any undefined symbols and consider them nonterminals. */
685 symbols_check_defined ();
687 /* Find the start symbol if no %start. */
692 node
!= NULL
&& symbol_is_dummy (node
->content
.sym
);
695 for (node
= node
->next
;
696 node
!= NULL
&& node
->content
.sym
!= NULL
;
701 grammar_start_symbol_set (node
->content
.sym
,
702 node
->content
.sym
->location
);
705 /* Insert the initial rule, whose line is that of the first rule
706 (not that of the start symbol):
708 accept: %start EOF. */
710 symbol_list
*p
= symbol_list_sym_new (accept
, empty_location
);
711 p
->location
= grammar
->location
;
712 p
->next
= symbol_list_sym_new (startsymbol
, empty_location
);
713 p
->next
->next
= symbol_list_sym_new (endtoken
, empty_location
);
714 p
->next
->next
->next
= symbol_list_sym_new (NULL
, empty_location
);
715 p
->next
->next
->next
->next
= grammar
;
721 aver (nsyms
<= SYMBOL_NUMBER_MAXIMUM
&& nsyms
== ntokens
+ nvars
);
723 /* Assign the symbols their symbol numbers. Write #defines for the
724 token symbols into FDEFINES if requested. */
727 /* Scan rule actions after invoking symbol_check_alias_consistency (in
728 symbols_pack above) so that token types are set correctly before the rule
729 action type checking.
731 Before invoking grammar_rule_check (in packgram below) on any rule, make
732 sure all actions have already been scanned in order to set `used' flags.
733 Otherwise, checking that a midrule's $$ should be set will not always work
734 properly because the check must forward-reference the midrule's parent
735 rule. For the same reason, all the `used' flags must be set before
736 checking whether to remove `$' from any midrule symbol name (also in
740 for (sym
= grammar
; sym
; sym
= sym
->next
)
741 code_props_translate_code (&sym
->action_props
);
744 /* Convert the grammar into the format described in gram.h. */
747 /* The grammar as a symbol_list is no longer needed. */
748 symbol_list_free (grammar
);