]> git.saurik.com Git - bison.git/blob - src/parse-gram.y
Document how `%define "var" "value"' is not M4-friendly.
[bison.git] / src / parse-gram.y
1 %{/* Bison Grammar Parser -*- C -*-
2
3 Copyright (C) 2002, 2003, 2004, 2005, 2006, 2007, 2008, 2009 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 "complain.h"
25 #include "conflicts.h"
26 #include "files.h"
27 #include "getargs.h"
28 #include "gram.h"
29 #include "muscle-tab.h"
30 #include "quotearg.h"
31 #include "reader.h"
32 #include "symlist.h"
33 #include "scan-gram.h"
34 #include "scan-code.h"
35
36 #define YYLLOC_DEFAULT(Current, Rhs, N) (Current) = lloc_default (Rhs, N)
37 static YYLTYPE lloc_default (YYLTYPE const *, int);
38
39 #define YY_LOCATION_PRINT(File, Loc) \
40 location_print (File, Loc)
41
42 static void version_check (location const *loc, char const *version);
43
44 /* Request detailed syntax error messages, and pass them to GRAM_ERROR.
45 FIXME: depends on the undocumented availability of YYLLOC. */
46 #undef yyerror
47 #define yyerror(Msg) \
48 gram_error (&yylloc, Msg)
49 static void gram_error (location const *, char const *);
50
51 static char const *char_name (char);
52
53 /** Add a lex-param or a parse-param.
54 *
55 * \param type \a lex_param or \a parse_param
56 * \param decl the formal argument
57 * \param loc the location in the source.
58 */
59 static void add_param (char const *type, char *decl, location loc);
60
61
62 static symbol_class current_class = unknown_sym;
63 static uniqstr current_type = NULL;
64 static symbol *current_lhs;
65 static location current_lhs_location;
66 static int current_prec = 0;
67
68 #define YYTYPE_INT16 int_fast16_t
69 #define YYTYPE_INT8 int_fast8_t
70 #define YYTYPE_UINT16 uint_fast16_t
71 #define YYTYPE_UINT8 uint_fast8_t
72 %}
73
74 %debug
75 %verbose
76 %defines
77 %locations
78 %pure-parser
79 %error-verbose
80 %name-prefix="gram_"
81 %expect 0
82
83 %initial-action
84 {
85 /* Bison's grammar can initial empty locations, hence a default
86 location is needed. */
87 boundary_set (&@$.start, current_file, 1, 1);
88 boundary_set (&@$.end, current_file, 1, 1);
89 }
90
91 %union
92 {
93 symbol *symbol;
94 symbol_list *list;
95 int integer;
96 char const *chars;
97 char *code;
98 assoc assoc;
99 uniqstr uniqstr;
100 unsigned char character;
101 };
102
103 /* Define the tokens together with their human representation. */
104 %token GRAM_EOF 0 "end of file"
105 %token STRING "string"
106 %token INT "integer"
107
108 %token PERCENT_TOKEN "%token"
109 %token PERCENT_NTERM "%nterm"
110
111 %token PERCENT_TYPE "%type"
112 %token PERCENT_DESTRUCTOR "%destructor"
113 %token PERCENT_PRINTER "%printer"
114
115 %token PERCENT_LEFT "%left"
116 %token PERCENT_RIGHT "%right"
117 %token PERCENT_NONASSOC "%nonassoc"
118 %token PERCENT_PRECEDENCE "%precedence"
119
120 %token PERCENT_PREC "%prec"
121 %token PERCENT_DPREC "%dprec"
122 %token PERCENT_MERGE "%merge"
123
124
125 /*----------------------.
126 | Global Declarations. |
127 `----------------------*/
128
129 %token
130 PERCENT_CODE "%code"
131 PERCENT_DEFAULT_PREC "%default-prec"
132 PERCENT_DEFINE "%define"
133 PERCENT_DEFINES "%defines"
134 PERCENT_EXPECT "%expect"
135 PERCENT_EXPECT_RR "%expect-rr"
136 PERCENT_FLAG "%<flag>"
137 PERCENT_FILE_PREFIX "%file-prefix"
138 PERCENT_GLR_PARSER "%glr-parser"
139 PERCENT_INITIAL_ACTION "%initial-action"
140 PERCENT_LANGUAGE "%language"
141 PERCENT_LEX_PARAM "%lex-param"
142 PERCENT_NAME_PREFIX "%name-prefix"
143 PERCENT_NO_DEFAULT_PREC "%no-default-prec"
144 PERCENT_NO_LINES "%no-lines"
145 PERCENT_NONDETERMINISTIC_PARSER
146 "%nondeterministic-parser"
147 PERCENT_OUTPUT "%output"
148 PERCENT_PARSE_PARAM "%parse-param"
149 PERCENT_REQUIRE "%require"
150 PERCENT_SKELETON "%skeleton"
151 PERCENT_START "%start"
152 PERCENT_TOKEN_TABLE "%token-table"
153 PERCENT_VERBOSE "%verbose"
154 PERCENT_YACC "%yacc"
155 ;
156
157 %token BRACED_CODE "{...}"
158 %token CHAR "char"
159 %token EPILOGUE "epilogue"
160 %token EQUAL "="
161 %token ID "identifier"
162 %token ID_COLON "identifier:"
163 %token PERCENT_PERCENT "%%"
164 %token PIPE "|"
165 %token PROLOGUE "%{...%}"
166 %token SEMICOLON ";"
167 %token TAG "<tag>"
168 %token TAG_ANY "<*>"
169 %token TAG_NONE "<>"
170
171 %type <character> CHAR
172 %printer { fputs (char_name ($$), stderr); } CHAR
173
174 /* braceless is not to be used for rule or symbol actions, as it
175 calls code_props_plain_init. */
176 %type <chars> STRING "%{...%}" EPILOGUE braceless content.opt
177 %type <code> "{...}"
178 %printer { fputs (quotearg_style (c_quoting_style, $$), stderr); }
179 STRING
180 %printer { fprintf (stderr, "{\n%s\n}", $$); }
181 braceless content.opt "{...}" "%{...%}" EPILOGUE
182
183 %type <uniqstr> TAG ID ID_COLON PERCENT_FLAG variable
184 %printer { fputs ($$, stderr); } ID variable
185 %printer { fprintf (stderr, "%s:", $$); } ID_COLON
186 %printer { fprintf (stderr, "%%%s", $$); } PERCENT_FLAG
187 %printer { fprintf (stderr, "<%s>", $$); } TAG
188
189 %type <integer> INT
190 %printer { fprintf (stderr, "%d", $$); } INT
191
192 %type <symbol> id id_colon symbol symbol.prec string_as_id
193 %printer { fprintf (stderr, "%s", $$->tag); } id symbol string_as_id
194 %printer { fprintf (stderr, "%s:", $$->tag); } id_colon
195
196 %type <assoc> precedence_declarator
197 %type <list> symbols.1 symbols.prec generic_symlist generic_symlist_item
198 %%
199
200 input:
201 prologue_declarations "%%" grammar epilogue.opt
202 ;
203
204
205 /*------------------------------------.
206 | Declarations: before the first %%. |
207 `------------------------------------*/
208
209 prologue_declarations:
210 /* Nothing */
211 | prologue_declarations prologue_declaration
212 ;
213
214 prologue_declaration:
215 grammar_declaration
216 | "%{...%}"
217 {
218 code_props plain_code;
219 code_props_plain_init (&plain_code, $1, @1);
220 code_props_translate_code (&plain_code);
221 gram_scanner_last_string_free ();
222 muscle_code_grow (union_seen ? "post_prologue" : "pre_prologue",
223 plain_code.code, @1);
224 code_scanner_last_string_free ();
225 }
226 | "%<flag>"
227 {
228 muscle_percent_define_ensure ($1, @1, true);
229 }
230 | "%define" variable content.opt
231 {
232 muscle_percent_define_insert ($2, @2, $3);
233 }
234 | "%defines" { defines_flag = true; }
235 | "%defines" STRING
236 {
237 defines_flag = true;
238 spec_defines_file = xstrdup ($2);
239 }
240 | "%expect" INT { expected_sr_conflicts = $2; }
241 | "%expect-rr" INT { expected_rr_conflicts = $2; }
242 | "%file-prefix" STRING { spec_file_prefix = $2; }
243 | "%file-prefix" "=" STRING { spec_file_prefix = $3; } /* deprecated */
244 | "%glr-parser"
245 {
246 nondeterministic_parser = true;
247 glr_parser = true;
248 }
249 | "%initial-action" "{...}"
250 {
251 code_props action;
252 code_props_symbol_action_init (&action, $2, @2);
253 code_props_translate_code (&action);
254 gram_scanner_last_string_free ();
255 muscle_code_grow ("initial_action", action.code, @2);
256 code_scanner_last_string_free ();
257 }
258 | "%language" STRING { language_argmatch ($2, grammar_prio, @1); }
259 | "%lex-param" "{...}" { add_param ("lex_param", $2, @2); }
260 | "%name-prefix" STRING { spec_name_prefix = $2; }
261 | "%name-prefix" "=" STRING { spec_name_prefix = $3; } /* deprecated */
262 | "%no-lines" { no_lines_flag = true; }
263 | "%nondeterministic-parser" { nondeterministic_parser = true; }
264 | "%output" STRING { spec_outfile = $2; }
265 | "%output" "=" STRING { spec_outfile = $3; } /* deprecated */
266 | "%parse-param" "{...}" { add_param ("parse_param", $2, @2); }
267 | "%require" STRING { version_check (&@2, $2); }
268 | "%skeleton" STRING
269 {
270 char const *skeleton_user = $2;
271 if (strchr (skeleton_user, '/'))
272 {
273 size_t dir_length = strlen (current_file);
274 char *skeleton_build;
275 while (dir_length && current_file[dir_length - 1] != '/')
276 --dir_length;
277 while (dir_length && current_file[dir_length - 1] == '/')
278 --dir_length;
279 skeleton_build =
280 xmalloc (dir_length + 1 + strlen (skeleton_user) + 1);
281 if (dir_length > 0)
282 {
283 strncpy (skeleton_build, current_file, dir_length);
284 skeleton_build[dir_length++] = '/';
285 }
286 strcpy (skeleton_build + dir_length, skeleton_user);
287 skeleton_user = uniqstr_new (skeleton_build);
288 free (skeleton_build);
289 }
290 skeleton_arg (skeleton_user, grammar_prio, @1);
291 }
292 | "%token-table" { token_table_flag = true; }
293 | "%verbose" { report_flag |= report_states; }
294 | "%yacc" { yacc_flag = true; }
295 | /*FIXME: Err? What is this horror doing here? */ ";"
296 ;
297
298 grammar_declaration:
299 precedence_declaration
300 | symbol_declaration
301 | "%start" symbol
302 {
303 grammar_start_symbol_set ($2, @2);
304 }
305 | "%destructor" "{...}" generic_symlist
306 {
307 symbol_list *list;
308 for (list = $3; list; list = list->next)
309 symbol_list_destructor_set (list, $2, @2);
310 symbol_list_free ($3);
311 }
312 | "%printer" "{...}" generic_symlist
313 {
314 symbol_list *list;
315 for (list = $3; list; list = list->next)
316 symbol_list_printer_set (list, $2, @2);
317 symbol_list_free ($3);
318 }
319 | "%default-prec"
320 {
321 default_prec = true;
322 }
323 | "%no-default-prec"
324 {
325 default_prec = false;
326 }
327 | "%code" braceless
328 {
329 /* Do not invoke muscle_percent_code_grow here since it invokes
330 muscle_user_name_list_grow. */
331 muscle_code_grow ("percent_code()", $2, @2);
332 code_scanner_last_string_free ();
333 }
334 | "%code" ID braceless
335 {
336 muscle_percent_code_grow ($2, @2, $3, @3);
337 code_scanner_last_string_free ();
338 }
339 ;
340
341
342 /*---------.
343 | %union. |
344 `---------*/
345
346 %token PERCENT_UNION "%union";
347
348 union_name:
349 /* Nothing. */ {}
350 | ID { muscle_code_grow ("union_name", $1, @1); }
351 ;
352
353 grammar_declaration:
354 "%union" union_name braceless
355 {
356 union_seen = true;
357 muscle_code_grow ("stype", $3, @3);
358 code_scanner_last_string_free ();
359 }
360 ;
361
362
363
364
365 symbol_declaration:
366 "%nterm" { current_class = nterm_sym; } symbol_defs.1
367 {
368 current_class = unknown_sym;
369 current_type = NULL;
370 }
371 | "%token" { current_class = token_sym; } symbol_defs.1
372 {
373 current_class = unknown_sym;
374 current_type = NULL;
375 }
376 | "%type" TAG symbols.1
377 {
378 symbol_list *list;
379 tag_seen = true;
380 for (list = $3; list; list = list->next)
381 symbol_type_set (list->content.sym, $2, @2);
382 symbol_list_free ($3);
383 }
384 ;
385
386 precedence_declaration:
387 precedence_declarator tag.opt symbols.prec
388 {
389 symbol_list *list;
390 ++current_prec;
391 for (list = $3; list; list = list->next)
392 {
393 symbol_type_set (list->content.sym, current_type, @2);
394 symbol_precedence_set (list->content.sym, current_prec, $1, @1);
395 }
396 symbol_list_free ($3);
397 current_type = NULL;
398 }
399 ;
400
401 precedence_declarator:
402 "%left" { $$ = left_assoc; }
403 | "%right" { $$ = right_assoc; }
404 | "%nonassoc" { $$ = non_assoc; }
405 | "%precedence" { $$ = precedence_assoc; }
406 ;
407
408 tag.opt:
409 /* Nothing. */ { current_type = NULL; }
410 | TAG { current_type = $1; tag_seen = true; }
411 ;
412
413 /* Just like symbols.1 but accept INT for the sake of POSIX. */
414 symbols.prec:
415 symbol.prec
416 { $$ = symbol_list_sym_new ($1, @1); }
417 | symbols.prec symbol.prec
418 { $$ = symbol_list_prepend ($1, symbol_list_sym_new ($2, @2)); }
419 ;
420
421 symbol.prec:
422 symbol { $$ = $1; }
423 | symbol INT { $$ = $1; symbol_user_token_number_set ($1, $2, @2); }
424 ;
425
426 /* One or more symbols to be %typed. */
427 symbols.1:
428 symbol
429 { $$ = symbol_list_sym_new ($1, @1); }
430 | symbols.1 symbol
431 { $$ = symbol_list_prepend ($1, symbol_list_sym_new ($2, @2)); }
432 ;
433
434 generic_symlist:
435 generic_symlist_item { $$ = $1; }
436 | generic_symlist generic_symlist_item { $$ = symbol_list_prepend ($1, $2); }
437 ;
438
439 generic_symlist_item:
440 symbol { $$ = symbol_list_sym_new ($1, @1); }
441 | TAG { $$ = symbol_list_type_new ($1, @1); }
442 | "<*>" { $$ = symbol_list_default_tagged_new (@1); }
443 | "<>" { $$ = symbol_list_default_tagless_new (@1); }
444 ;
445
446 /* One token definition. */
447 symbol_def:
448 TAG
449 {
450 current_type = $1;
451 tag_seen = true;
452 }
453 | id
454 {
455 symbol_class_set ($1, current_class, @1, true);
456 symbol_type_set ($1, current_type, @1);
457 }
458 | id INT
459 {
460 symbol_class_set ($1, current_class, @1, true);
461 symbol_type_set ($1, current_type, @1);
462 symbol_user_token_number_set ($1, $2, @2);
463 }
464 | id string_as_id
465 {
466 symbol_class_set ($1, current_class, @1, true);
467 symbol_type_set ($1, current_type, @1);
468 symbol_make_alias ($1, $2, @$);
469 }
470 | id INT string_as_id
471 {
472 symbol_class_set ($1, current_class, @1, true);
473 symbol_type_set ($1, current_type, @1);
474 symbol_user_token_number_set ($1, $2, @2);
475 symbol_make_alias ($1, $3, @$);
476 }
477 ;
478
479 /* One or more symbol definitions. */
480 symbol_defs.1:
481 symbol_def
482 | symbol_defs.1 symbol_def
483 ;
484
485
486 /*------------------------------------------.
487 | The grammar section: between the two %%. |
488 `------------------------------------------*/
489
490 grammar:
491 rules_or_grammar_declaration
492 | grammar rules_or_grammar_declaration
493 ;
494
495 /* As a Bison extension, one can use the grammar declarations in the
496 body of the grammar. */
497 rules_or_grammar_declaration:
498 rules
499 | grammar_declaration ";"
500 | error ";"
501 {
502 yyerrok;
503 }
504 ;
505
506 rules:
507 id_colon { current_lhs = $1; current_lhs_location = @1; } rhses.1
508 ;
509
510 rhses.1:
511 rhs { grammar_current_rule_end (@1); }
512 | rhses.1 "|" rhs { grammar_current_rule_end (@3); }
513 | rhses.1 ";"
514 ;
515
516 rhs:
517 /* Nothing. */
518 { grammar_current_rule_begin (current_lhs, current_lhs_location); }
519 | rhs symbol
520 { grammar_current_rule_symbol_append ($2, @2); }
521 | rhs "{...}"
522 { grammar_current_rule_action_append ($2, @2); }
523 | rhs "%prec" symbol
524 { grammar_current_rule_prec_set ($3, @3); }
525 | rhs "%dprec" INT
526 { grammar_current_rule_dprec_set ($3, @3); }
527 | rhs "%merge" TAG
528 { grammar_current_rule_merge_set ($3, @3); }
529 ;
530
531
532 /*---------------------------.
533 | variable and content.opt. |
534 `---------------------------*/
535
536 /* The STRING form of variable is deprecated and is not M4-friendly.
537 For example, M4 fails for `%define "[" "value"'. */
538 variable:
539 ID
540 | STRING { $$ = uniqstr_new ($1); }
541 ;
542
543 /* Some content or empty by default. */
544 content.opt:
545 /* Nothing. */ { $$ = ""; }
546 | STRING
547 ;
548
549
550 /*------------.
551 | braceless. |
552 `------------*/
553
554 braceless:
555 "{...}"
556 {
557 code_props plain_code;
558 $1[strlen ($1) - 1] = '\n';
559 code_props_plain_init (&plain_code, $1+1, @1);
560 code_props_translate_code (&plain_code);
561 gram_scanner_last_string_free ();
562 $$ = plain_code.code;
563 }
564 ;
565
566
567 /*--------------.
568 | Identifiers. |
569 `--------------*/
570
571 /* Identifiers are returned as uniqstr values by the scanner.
572 Depending on their use, we may need to make them genuine symbols. */
573
574 id:
575 ID
576 { $$ = symbol_from_uniqstr ($1, @1); }
577 | CHAR
578 {
579 $$ = symbol_get (char_name ($1), @1);
580 symbol_class_set ($$, token_sym, @1, false);
581 symbol_user_token_number_set ($$, $1, @1);
582 }
583 ;
584
585 id_colon:
586 ID_COLON { $$ = symbol_from_uniqstr ($1, @1); }
587 ;
588
589
590 symbol:
591 id
592 | string_as_id
593 ;
594
595 /* A string used as an ID: quote it. */
596 string_as_id:
597 STRING
598 {
599 $$ = symbol_get (quotearg_style (c_quoting_style, $1), @1);
600 symbol_class_set ($$, token_sym, @1, false);
601 }
602 ;
603
604 epilogue.opt:
605 /* Nothing. */
606 | "%%" EPILOGUE
607 {
608 code_props plain_code;
609 code_props_plain_init (&plain_code, $2, @2);
610 code_props_translate_code (&plain_code);
611 gram_scanner_last_string_free ();
612 muscle_code_grow ("epilogue", plain_code.code, @2);
613 code_scanner_last_string_free ();
614 }
615 ;
616
617 %%
618
619
620 /* Return the location of the left-hand side of a rule whose
621 right-hand side is RHS[1] ... RHS[N]. Ignore empty nonterminals in
622 the right-hand side, and return an empty location equal to the end
623 boundary of RHS[0] if the right-hand side is empty. */
624
625 static YYLTYPE
626 lloc_default (YYLTYPE const *rhs, int n)
627 {
628 int i;
629 YYLTYPE loc;
630
631 /* SGI MIPSpro 7.4.1m miscompiles "loc.start = loc.end = rhs[n].end;".
632 The bug is fixed in 7.4.2m, but play it safe for now. */
633 loc.start = rhs[n].end;
634 loc.end = rhs[n].end;
635
636 /* Ignore empty nonterminals the start of the the right-hand side.
637 Do not bother to ignore them at the end of the right-hand side,
638 since empty nonterminals have the same end as their predecessors. */
639 for (i = 1; i <= n; i++)
640 if (! equal_boundaries (rhs[i].start, rhs[i].end))
641 {
642 loc.start = rhs[i].start;
643 break;
644 }
645
646 return loc;
647 }
648
649
650 /* Add a lex-param or a parse-param (depending on TYPE) with
651 declaration DECL and location LOC. */
652
653 static void
654 add_param (char const *type, char *decl, location loc)
655 {
656 static char const alphanum[26 + 26 + 1 + 10] =
657 "abcdefghijklmnopqrstuvwxyz"
658 "ABCDEFGHIJKLMNOPQRSTUVWXYZ"
659 "_"
660 "0123456789";
661 char const *name_start = NULL;
662 char *p;
663
664 /* Stop on last actual character. */
665 for (p = decl; p[1]; p++)
666 if ((p == decl
667 || ! memchr (alphanum, p[-1], sizeof alphanum))
668 && memchr (alphanum, p[0], sizeof alphanum - 10))
669 name_start = p;
670
671 /* Strip the surrounding '{' and '}', and any blanks just inside
672 the braces. */
673 while (*--p == ' ' || *p == '\t')
674 continue;
675 p[1] = '\0';
676 while (*++decl == ' ' || *decl == '\t')
677 continue;
678
679 if (! name_start)
680 complain_at (loc, _("missing identifier in parameter declaration"));
681 else
682 {
683 char *name;
684 size_t name_len;
685
686 for (name_len = 1;
687 memchr (alphanum, name_start[name_len], sizeof alphanum);
688 name_len++)
689 continue;
690
691 name = xmalloc (name_len + 1);
692 memcpy (name, name_start, name_len);
693 name[name_len] = '\0';
694 muscle_pair_list_grow (type, decl, name);
695 free (name);
696 }
697
698 gram_scanner_last_string_free ();
699 }
700
701
702 static void
703 version_check (location const *loc, char const *version)
704 {
705 if (strverscmp (version, PACKAGE_VERSION) > 0)
706 {
707 complain_at (*loc, "require bison %s, but have %s",
708 version, PACKAGE_VERSION);
709 exit (63);
710 }
711 }
712
713 static void
714 gram_error (location const *loc, char const *msg)
715 {
716 complain_at (*loc, "%s", msg);
717 }
718
719 char const *
720 token_name (int type)
721 {
722 return yytname[YYTRANSLATE (type)];
723 }
724
725 static char const *
726 char_name (char c)
727 {
728 if (c == '\'')
729 return "'\\''";
730 else
731 {
732 char buf[4];
733 buf[0] = '\''; buf[1] = c; buf[2] = '\''; buf[3] = '\0';
734 return quotearg_style (escape_quoting_style, buf);
735 }
736 }