]> git.saurik.com Git - bison.git/blob - src/parse-gram.y
Check for unrecognized %define variables similar to checking for
[bison.git] / src / parse-gram.y
1 %{/* Bison Grammar Parser -*- C -*-
2
3 Copyright (C) 2002, 2003, 2004, 2005, 2006, 2007 Free Software Foundation,
4 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 2 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, write to the Free Software
20 Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA
21 02110-1301 USA
22 */
23
24 #include <config.h>
25 #include "system.h"
26
27 #include <strverscmp.h>
28
29 #include "complain.h"
30 #include "conflicts.h"
31 #include "files.h"
32 #include "getargs.h"
33 #include "gram.h"
34 #include "muscle_tab.h"
35 #include "quotearg.h"
36 #include "reader.h"
37 #include "symlist.h"
38 #include "scan-gram.h"
39 #include "scan-code.h"
40
41 #define YYLLOC_DEFAULT(Current, Rhs, N) (Current) = lloc_default (Rhs, N)
42 static YYLTYPE lloc_default (YYLTYPE const *, int);
43
44 #define YY_LOCATION_PRINT(File, Loc) \
45 location_print (File, Loc)
46
47 static void version_check (location const *loc, char const *version);
48
49 /* Request detailed syntax error messages, and pass them to GRAM_ERROR.
50 FIXME: depends on the undocumented availability of YYLLOC. */
51 #undef yyerror
52 #define yyerror(Msg) \
53 gram_error (&yylloc, Msg)
54 static void gram_error (location const *, char const *);
55
56 static char const *char_name (char);
57
58 /** Add a lex-param or a parse-param.
59 *
60 * \param type \a lex_param or \a parse_param
61 * \param decl the formal argument
62 * \param loc the location in the source.
63 */
64 static void add_param (char const *type, char *decl, location loc);
65
66
67 static symbol_class current_class = unknown_sym;
68 static uniqstr current_type = NULL;
69 static symbol *current_lhs;
70 static location current_lhs_location;
71 static int current_prec = 0;
72
73 #define YYTYPE_INT16 int_fast16_t
74 #define YYTYPE_INT8 int_fast8_t
75 #define YYTYPE_UINT16 uint_fast16_t
76 #define YYTYPE_UINT8 uint_fast8_t
77 %}
78
79 %debug
80 %verbose
81 %defines
82 %locations
83 %pure-parser
84 %error-verbose
85 %name-prefix="gram_"
86 %expect 0
87
88 %initial-action
89 {
90 /* Bison's grammar can initial empty locations, hence a default
91 location is needed. */
92 boundary_set (&@$.start, current_file, 1, 1);
93 boundary_set (&@$.end, current_file, 1, 1);
94 }
95
96 %union
97 {
98 symbol *symbol;
99 symbol_list *list;
100 int integer;
101 char const *chars;
102 char *code;
103 assoc assoc;
104 uniqstr uniqstr;
105 unsigned char character;
106 };
107
108 /* Define the tokens together with their human representation. */
109 %token GRAM_EOF 0 "end of file"
110 %token STRING "string"
111 %token INT "integer"
112
113 %token PERCENT_TOKEN "%token"
114 %token PERCENT_NTERM "%nterm"
115
116 %token PERCENT_TYPE "%type"
117 %token PERCENT_DESTRUCTOR "%destructor"
118 %token PERCENT_PRINTER "%printer"
119
120 %token PERCENT_LEFT "%left"
121 %token PERCENT_RIGHT "%right"
122 %token PERCENT_NONASSOC "%nonassoc"
123
124 %token PERCENT_PREC "%prec"
125 %token PERCENT_DPREC "%dprec"
126 %token PERCENT_MERGE "%merge"
127
128
129 /*----------------------.
130 | Global Declarations. |
131 `----------------------*/
132
133 %token
134 PERCENT_CODE "%code"
135 PERCENT_DEBUG "%debug"
136 PERCENT_DEFAULT_PREC "%default-prec"
137 PERCENT_DEFINE "%define"
138 PERCENT_DEFINES "%defines"
139 PERCENT_ERROR_VERBOSE "%error-verbose"
140 PERCENT_EXPECT "%expect"
141 PERCENT_EXPECT_RR "%expect-rr"
142 PERCENT_FILE_PREFIX "%file-prefix"
143 PERCENT_GLR_PARSER "%glr-parser"
144 PERCENT_INITIAL_ACTION "%initial-action"
145 PERCENT_LANGUAGE "%language"
146 PERCENT_LEX_PARAM "%lex-param"
147 PERCENT_LOCATIONS "%locations"
148 PERCENT_NAME_PREFIX "%name-prefix"
149 PERCENT_NO_DEFAULT_PREC "%no-default-prec"
150 PERCENT_NO_LINES "%no-lines"
151 PERCENT_NONDETERMINISTIC_PARSER
152 "%nondeterministic-parser"
153 PERCENT_OUTPUT "%output"
154 PERCENT_PARSE_PARAM "%parse-param"
155 PERCENT_PURE_PARSER "%pure-parser"
156 PERCENT_PUSH_PARSER "%push-parser"
157 PERCENT_PUSH_PULL_PARSER
158 "%push-pull-parser"
159 PERCENT_REQUIRE "%require"
160 PERCENT_SKELETON "%skeleton"
161 PERCENT_START "%start"
162 PERCENT_TOKEN_TABLE "%token-table"
163 PERCENT_VERBOSE "%verbose"
164 PERCENT_YACC "%yacc"
165 ;
166
167 %token BRACED_CODE "{...}"
168 %token CHAR "char"
169 %token EPILOGUE "epilogue"
170 %token EQUAL "="
171 %token ID "identifier"
172 %token ID_COLON "identifier:"
173 %token PERCENT_PERCENT "%%"
174 %token PIPE "|"
175 %token PROLOGUE "%{...%}"
176 %token SEMICOLON ";"
177 %token TYPE "type"
178 %token TYPE_TAG_ANY "<*>"
179 %token TYPE_TAG_NONE "<>"
180
181 %type <character> CHAR
182 %printer { fputs (char_name ($$), stderr); } CHAR
183
184 /* braceless is not to be used for rule or symbol actions, as it
185 calls code_props_plain_init. */
186 %type <chars> STRING "%{...%}" EPILOGUE braceless content content.opt
187 %type <code> "{...}"
188 %printer { fputs (quotearg_style (c_quoting_style, $$), stderr); }
189 STRING
190 %printer { fprintf (stderr, "{\n%s\n}", $$); }
191 braceless content content.opt "{...}" "%{...%}" EPILOGUE
192
193 %type <uniqstr> TYPE ID ID_COLON
194 %printer { fprintf (stderr, "<%s>", $$); } TYPE
195 %printer { fputs ($$, stderr); } ID
196 %printer { fprintf (stderr, "%s:", $$); } ID_COLON
197
198 %type <integer> INT
199 %printer { fprintf (stderr, "%d", $$); } INT
200
201 %type <symbol> id id_colon symbol string_as_id
202 %printer { fprintf (stderr, "%s", $$->tag); } id symbol string_as_id
203 %printer { fprintf (stderr, "%s:", $$->tag); } id_colon
204
205 %type <assoc> precedence_declarator
206 %type <list> symbols.1 generic_symlist generic_symlist_item
207 %%
208
209 input:
210 prologue_declarations "%%" grammar epilogue.opt
211 ;
212
213
214 /*------------------------------------.
215 | Declarations: before the first %%. |
216 `------------------------------------*/
217
218 prologue_declarations:
219 /* Nothing */
220 | prologue_declarations prologue_declaration
221 ;
222
223 prologue_declaration:
224 grammar_declaration
225 | "%{...%}"
226 {
227 code_props plain_code;
228 code_props_plain_init (&plain_code, $1, @1);
229 code_props_translate_code (&plain_code);
230 gram_scanner_last_string_free ();
231 prologue_augment (plain_code.code, @1, union_seen);
232 code_scanner_last_string_free ();
233 }
234 | "%debug" { debug_flag = true; }
235 | "%define" STRING content.opt
236 {
237 /* FIXME: Special characters in $2 may break %define.
238 For example: `['. */
239 if (muscle_find_const ($2))
240 warn_at (@2, _("%s: `%s' redefined"), "%define", $2);
241 muscle_insert ($2, $3);
242 muscle_grow_used_name_list ("used_percent_define_variables", $2, @2);
243 }
244 | "%defines" { defines_flag = true; }
245 | "%defines" STRING
246 {
247 defines_flag = true;
248 spec_defines_file = xstrdup ($2);
249 }
250 | "%error-verbose" { error_verbose = true; }
251 | "%expect" INT { expected_sr_conflicts = $2; }
252 | "%expect-rr" INT { expected_rr_conflicts = $2; }
253 | "%file-prefix" STRING { spec_file_prefix = $2; }
254 | "%file-prefix" "=" STRING { spec_file_prefix = $3; } /* deprecated */
255 | "%glr-parser"
256 {
257 nondeterministic_parser = true;
258 glr_parser = true;
259 }
260 | "%initial-action" "{...}"
261 {
262 code_props action;
263 code_props_symbol_action_init (&action, $2, @2);
264 code_props_translate_code (&action);
265 gram_scanner_last_string_free ();
266 muscle_code_grow ("initial_action", action.code, @2);
267 code_scanner_last_string_free ();
268 }
269 | "%language" STRING { language_argmatch ($2, 1, &@1); }
270 | "%lex-param" "{...}" { add_param ("lex_param", $2, @2); }
271 | "%locations" { locations_flag = true; }
272 | "%name-prefix" STRING { spec_name_prefix = $2; }
273 | "%name-prefix" "=" STRING { spec_name_prefix = $3; } /* deprecated */
274 | "%no-lines" { no_lines_flag = true; }
275 | "%nondeterministic-parser" { nondeterministic_parser = true; }
276 | "%output" STRING { spec_outfile = $2; }
277 | "%output" "=" STRING { spec_outfile = $3; } /* deprecated */
278 | "%parse-param" "{...}" { add_param ("parse_param", $2, @2); }
279 | "%pure-parser" { pure_parser = true; }
280 | "%push-parser" { push_parser = true; pull_parser = false; }
281 | "%push-pull-parser" { push_parser = true; pull_parser = true; }
282 | "%require" STRING { version_check (&@2, $2); }
283 | "%skeleton" STRING { skeleton_arg ($2, 1, &@1); }
284 | "%token-table" { token_table_flag = true; }
285 | "%verbose" { report_flag = report_states; }
286 | "%yacc" { yacc_flag = true; }
287 | /*FIXME: Err? What is this horror doing here? */ ";"
288 ;
289
290 grammar_declaration:
291 precedence_declaration
292 | symbol_declaration
293 | "%start" symbol
294 {
295 grammar_start_symbol_set ($2, @2);
296 }
297 | "%destructor" "{...}" generic_symlist
298 {
299 symbol_list *list;
300 for (list = $3; list; list = list->next)
301 symbol_list_destructor_set (list, $2, @2);
302 symbol_list_free ($3);
303 }
304 | "%printer" "{...}" generic_symlist
305 {
306 symbol_list *list;
307 for (list = $3; list; list = list->next)
308 symbol_list_printer_set (list, $2, @2);
309 symbol_list_free ($3);
310 }
311 | "%default-prec"
312 {
313 default_prec = true;
314 }
315 | "%no-default-prec"
316 {
317 default_prec = false;
318 }
319 | "%code" braceless
320 {
321 muscle_code_grow ("percent_code", $2, @2);
322 code_scanner_last_string_free ();
323 }
324 | "%code" STRING braceless
325 {
326 /* FIXME: Special characters in $2 may break %code.
327 For example: `['. */
328 char const name_prefix[] = "percent_code_";
329 char *name = xmalloc (sizeof name_prefix + strlen ($2));
330 strcpy (name, name_prefix);
331 strcpy (name + sizeof name_prefix - 1, $2);
332 muscle_code_grow (uniqstr_new (name), $3, @3);
333 free (name);
334 code_scanner_last_string_free ();
335 muscle_grow_used_name_list ("used_percent_code_qualifiers", $2, @2);
336 }
337 ;
338
339
340 /*----------*
341 | %union. |
342 *----------*/
343
344 %token PERCENT_UNION "%union";
345
346 union_name:
347 /* Nothing. */ {}
348 | ID { muscle_code_grow ("union_name", $1, @1); }
349 ;
350
351 grammar_declaration:
352 "%union" union_name "{...}"
353 {
354 char const *body = $3;
355
356 /* Concatenate the %union bodies. If this is the first %union, make sure
357 the synchronization line appears after the opening '{' so as not to
358 confuse Doxygen. Otherwise, turn the previous %union's trailing '}'
359 into '\n', and omit the new %union's leading '{'. */
360 if (!union_seen)
361 {
362 muscle_grow ("stype", "{", "");
363 }
364 else
365 {
366 char *code = muscle_find ("stype");
367 code[strlen (code) - 1] = '\n';
368 }
369 body++;
370
371 union_seen = true;
372 muscle_code_grow ("stype", body, @3);
373 gram_scanner_last_string_free ();
374 }
375 ;
376
377
378
379
380 symbol_declaration:
381 "%nterm" { current_class = nterm_sym; } symbol_defs.1
382 {
383 current_class = unknown_sym;
384 current_type = NULL;
385 }
386 | "%token" { current_class = token_sym; } symbol_defs.1
387 {
388 current_class = unknown_sym;
389 current_type = NULL;
390 }
391 | "%type" TYPE symbols.1
392 {
393 symbol_list *list;
394 tag_seen = true;
395 for (list = $3; list; list = list->next)
396 symbol_type_set (list->content.sym, $2, @2);
397 symbol_list_free ($3);
398 }
399 ;
400
401 precedence_declaration:
402 precedence_declarator type.opt symbols.1
403 {
404 symbol_list *list;
405 ++current_prec;
406 for (list = $3; list; list = list->next)
407 {
408 symbol_type_set (list->content.sym, current_type, @2);
409 symbol_precedence_set (list->content.sym, current_prec, $1, @1);
410 }
411 symbol_list_free ($3);
412 current_type = NULL;
413 }
414 ;
415
416 precedence_declarator:
417 "%left" { $$ = left_assoc; }
418 | "%right" { $$ = right_assoc; }
419 | "%nonassoc" { $$ = non_assoc; }
420 ;
421
422 type.opt:
423 /* Nothing. */ { current_type = NULL; }
424 | TYPE { current_type = $1; tag_seen = true; }
425 ;
426
427 /* One or more symbols to be %typed. */
428 symbols.1:
429 symbol
430 { $$ = symbol_list_sym_new ($1, @1); }
431 | symbols.1 symbol
432 { $$ = symbol_list_prepend ($1, symbol_list_sym_new ($2, @2)); }
433 ;
434
435 generic_symlist:
436 generic_symlist_item { $$ = $1; }
437 | generic_symlist generic_symlist_item { $$ = symbol_list_prepend ($1, $2); }
438 ;
439
440 generic_symlist_item:
441 symbol { $$ = symbol_list_sym_new ($1, @1); }
442 | TYPE { $$ = symbol_list_type_new ($1, @1); }
443 | "<*>" { $$ = symbol_list_default_tagged_new (@1); }
444 | "<>" { $$ = symbol_list_default_tagless_new (@1); }
445 ;
446
447 /* One token definition. */
448 symbol_def:
449 TYPE
450 {
451 current_type = $1;
452 tag_seen = true;
453 }
454 | id
455 {
456 symbol_class_set ($1, current_class, @1, true);
457 symbol_type_set ($1, current_type, @1);
458 }
459 | id INT
460 {
461 symbol_class_set ($1, current_class, @1, true);
462 symbol_type_set ($1, current_type, @1);
463 symbol_user_token_number_set ($1, $2, @2);
464 }
465 | id string_as_id
466 {
467 symbol_class_set ($1, current_class, @1, true);
468 symbol_type_set ($1, current_type, @1);
469 symbol_make_alias ($1, $2, @$);
470 }
471 | id INT string_as_id
472 {
473 symbol_class_set ($1, current_class, @1, true);
474 symbol_type_set ($1, current_type, @1);
475 symbol_user_token_number_set ($1, $2, @2);
476 symbol_make_alias ($1, $3, @$);
477 }
478 ;
479
480 /* One or more symbol definitions. */
481 symbol_defs.1:
482 symbol_def
483 | symbol_defs.1 symbol_def
484 ;
485
486
487 /*------------------------------------------.
488 | The grammar section: between the two %%. |
489 `------------------------------------------*/
490
491 grammar:
492 rules_or_grammar_declaration
493 | grammar rules_or_grammar_declaration
494 ;
495
496 /* As a Bison extension, one can use the grammar declarations in the
497 body of the grammar. */
498 rules_or_grammar_declaration:
499 rules
500 | grammar_declaration ";"
501 | error ";"
502 {
503 yyerrok;
504 }
505 ;
506
507 rules:
508 id_colon { current_lhs = $1; current_lhs_location = @1; } rhses.1
509 ;
510
511 rhses.1:
512 rhs { grammar_current_rule_end (@1); }
513 | rhses.1 "|" rhs { grammar_current_rule_end (@3); }
514 | rhses.1 ";"
515 ;
516
517 rhs:
518 /* Nothing. */
519 { grammar_current_rule_begin (current_lhs, current_lhs_location); }
520 | rhs symbol
521 { grammar_current_rule_symbol_append ($2, @2); }
522 | rhs "{...}"
523 { grammar_current_rule_action_append ($2, @2); }
524 | rhs "%prec" symbol
525 { grammar_current_rule_prec_set ($3, @3); }
526 | rhs "%dprec" INT
527 { grammar_current_rule_dprec_set ($3, @3); }
528 | rhs "%merge" TYPE
529 { grammar_current_rule_merge_set ($3, @3); }
530 ;
531
532
533 /*-----------*
534 | content. |
535 *-----------*/
536
537 content:
538 STRING
539 | braceless
540 ;
541
542 /* Some content or "1" by default. */
543 content.opt:
544 /* Nothing. */
545 {
546 static char one[] = "1";
547 $$ = one;
548 }
549 | content
550 ;
551
552
553 braceless:
554 "{...}"
555 {
556 code_props plain_code;
557 $1[strlen ($1) - 1] = '\n';
558 code_props_plain_init (&plain_code, $1+1, @1);
559 code_props_translate_code (&plain_code);
560 gram_scanner_last_string_free ();
561 $$ = plain_code.code;
562 }
563 ;
564
565
566 /*---------------*
567 | Identifiers. |
568 *---------------*/
569
570 /* Identifiers are returned as uniqstr values by the scanner.
571 Depending on their use, we may need to make them genuine symbols. */
572
573 id:
574 ID
575 { $$ = symbol_from_uniqstr ($1, @1); }
576 | CHAR
577 {
578 $$ = symbol_get (char_name ($1), @1);
579 symbol_class_set ($$, token_sym, @1, false);
580 symbol_user_token_number_set ($$, $1, @1);
581 }
582 ;
583
584 id_colon:
585 ID_COLON { $$ = symbol_from_uniqstr ($1, @1); }
586 ;
587
588
589 symbol:
590 id
591 | string_as_id
592 ;
593
594 /* A string used as an ID: quote it. */
595 string_as_id:
596 STRING
597 {
598 $$ = symbol_get (quotearg_style (c_quoting_style, $1), @1);
599 symbol_class_set ($$, token_sym, @1, false);
600 }
601 ;
602
603 epilogue.opt:
604 /* Nothing. */
605 | "%%" EPILOGUE
606 {
607 code_props plain_code;
608 code_props_plain_init (&plain_code, $2, @2);
609 code_props_translate_code (&plain_code);
610 gram_scanner_last_string_free ();
611 muscle_code_grow ("epilogue", plain_code.code, @2);
612 code_scanner_last_string_free ();
613 }
614 ;
615
616 %%
617
618
619 /* Return the location of the left-hand side of a rule whose
620 right-hand side is RHS[1] ... RHS[N]. Ignore empty nonterminals in
621 the right-hand side, and return an empty location equal to the end
622 boundary of RHS[0] if the right-hand side is empty. */
623
624 static YYLTYPE
625 lloc_default (YYLTYPE const *rhs, int n)
626 {
627 int i;
628 YYLTYPE loc;
629
630 /* SGI MIPSpro 7.4.1m miscompiles "loc.start = loc.end = rhs[n].end;".
631 The bug is fixed in 7.4.2m, but play it safe for now. */
632 loc.start = rhs[n].end;
633 loc.end = rhs[n].end;
634
635 /* Ignore empty nonterminals the start of the the right-hand side.
636 Do not bother to ignore them at the end of the right-hand side,
637 since empty nonterminals have the same end as their predecessors. */
638 for (i = 1; i <= n; i++)
639 if (! equal_boundaries (rhs[i].start, rhs[i].end))
640 {
641 loc.start = rhs[i].start;
642 break;
643 }
644
645 return loc;
646 }
647
648
649 /* Add a lex-param or a parse-param (depending on TYPE) with
650 declaration DECL and location LOC. */
651
652 static void
653 add_param (char const *type, char *decl, location loc)
654 {
655 static char const alphanum[26 + 26 + 1 + 10] =
656 "abcdefghijklmnopqrstuvwxyz"
657 "ABCDEFGHIJKLMNOPQRSTUVWXYZ"
658 "_"
659 "0123456789";
660 char const *name_start = NULL;
661 char *p;
662
663 /* Stop on last actual character. */
664 for (p = decl; p[1]; p++)
665 if ((p == decl
666 || ! memchr (alphanum, p[-1], sizeof alphanum))
667 && memchr (alphanum, p[0], sizeof alphanum - 10))
668 name_start = p;
669
670 /* Strip the surrounding '{' and '}', and any blanks just inside
671 the braces. */
672 while (*--p == ' ' || *p == '\t')
673 continue;
674 p[1] = '\0';
675 while (*++decl == ' ' || *decl == '\t')
676 continue;
677
678 if (! name_start)
679 complain_at (loc, _("missing identifier in parameter declaration"));
680 else
681 {
682 char *name;
683 size_t name_len;
684
685 for (name_len = 1;
686 memchr (alphanum, name_start[name_len], sizeof alphanum);
687 name_len++)
688 continue;
689
690 name = xmalloc (name_len + 1);
691 memcpy (name, name_start, name_len);
692 name[name_len] = '\0';
693 muscle_pair_list_grow (type, decl, name);
694 free (name);
695 }
696
697 gram_scanner_last_string_free ();
698 }
699
700
701 static void
702 version_check (location const *loc, char const *version)
703 {
704 if (strverscmp (version, PACKAGE_VERSION) > 0)
705 {
706 complain_at (*loc, "require bison %s, but have %s",
707 version, PACKAGE_VERSION);
708 exit (63);
709 }
710 }
711
712 static void
713 gram_error (location const *loc, char const *msg)
714 {
715 complain_at (*loc, "%s", msg);
716 }
717
718 char const *
719 token_name (int type)
720 {
721 return yytname[YYTRANSLATE (type)];
722 }
723
724 static char const *
725 char_name (char c)
726 {
727 if (c == '\'')
728 return "'\\''";
729 else
730 {
731 char buf[4];
732 buf[0] = '\''; buf[1] = c; buf[2] = '\''; buf[3] = '\0';
733 return quotearg_style (escape_quoting_style, buf);
734 }
735 }