]> git.saurik.com Git - bison.git/blob - src/parse-gram.y
Fix options documentation.
[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_ERROR_VERBOSE "%error-verbose"
135 PERCENT_EXPECT "%expect"
136 PERCENT_EXPECT_RR "%expect-rr"
137 PERCENT_FLAG "%<flag>"
138 PERCENT_FILE_PREFIX "%file-prefix"
139 PERCENT_GLR_PARSER "%glr-parser"
140 PERCENT_INITIAL_ACTION "%initial-action"
141 PERCENT_LANGUAGE "%language"
142 PERCENT_LEX_PARAM "%lex-param"
143 PERCENT_LOCATIONS "%locations"
144 PERCENT_NAME_PREFIX "%name-prefix"
145 PERCENT_NO_DEFAULT_PREC "%no-default-prec"
146 PERCENT_NO_LINES "%no-lines"
147 PERCENT_NONDETERMINISTIC_PARSER
148 "%nondeterministic-parser"
149 PERCENT_OUTPUT "%output"
150 PERCENT_PARSE_PARAM "%parse-param"
151 PERCENT_PURE_PARSER "%pure-parser"
152 PERCENT_REQUIRE "%require"
153 PERCENT_SKELETON "%skeleton"
154 PERCENT_START "%start"
155 PERCENT_TOKEN_TABLE "%token-table"
156 PERCENT_VERBOSE "%verbose"
157 PERCENT_YACC "%yacc"
158 ;
159
160 %token BRACED_CODE "{...}"
161 %token CHAR "char"
162 %token EPILOGUE "epilogue"
163 %token EQUAL "="
164 %token ID "identifier"
165 %token ID_COLON "identifier:"
166 %token PERCENT_PERCENT "%%"
167 %token PIPE "|"
168 %token PROLOGUE "%{...%}"
169 %token SEMICOLON ";"
170 %token TAG "<tag>"
171 %token TAG_ANY "<*>"
172 %token TAG_NONE "<>"
173
174 %type <character> CHAR
175 %printer { fputs (char_name ($$), stderr); } CHAR
176
177 /* braceless is not to be used for rule or symbol actions, as it
178 calls code_props_plain_init. */
179 %type <chars> STRING "%{...%}" EPILOGUE braceless content.opt
180 %type <code> "{...}"
181 %printer { fputs (quotearg_style (c_quoting_style, $$), stderr); }
182 STRING
183 %printer { fprintf (stderr, "{\n%s\n}", $$); }
184 braceless content.opt "{...}" "%{...%}" EPILOGUE
185
186 %type <uniqstr> TAG ID ID_COLON PERCENT_FLAG variable
187 %printer { fputs ($$, stderr); } ID variable
188 %printer { fprintf (stderr, "%s:", $$); } ID_COLON
189 %printer { fprintf (stderr, "%%%s", $$); } PERCENT_FLAG
190 %printer { fprintf (stderr, "<%s>", $$); } TAG
191
192 %type <integer> INT
193 %printer { fprintf (stderr, "%d", $$); } INT
194
195 %type <symbol> id id_colon symbol symbol.prec string_as_id
196 %printer { fprintf (stderr, "%s", $$->tag); } id symbol string_as_id
197 %printer { fprintf (stderr, "%s:", $$->tag); } id_colon
198
199 %type <assoc> precedence_declarator
200 %type <list> symbols.1 symbols.prec generic_symlist generic_symlist_item
201 %%
202
203 input:
204 prologue_declarations "%%" grammar epilogue.opt
205 ;
206
207
208 /*------------------------------------.
209 | Declarations: before the first %%. |
210 `------------------------------------*/
211
212 prologue_declarations:
213 /* Nothing */
214 | prologue_declarations prologue_declaration
215 ;
216
217 prologue_declaration:
218 grammar_declaration
219 | "%{...%}"
220 {
221 code_props plain_code;
222 code_props_plain_init (&plain_code, $1, @1);
223 code_props_translate_code (&plain_code);
224 gram_scanner_last_string_free ();
225 muscle_code_grow (union_seen ? "post_prologue" : "pre_prologue",
226 plain_code.code, @1);
227 code_scanner_last_string_free ();
228 }
229 | "%<flag>"
230 {
231 muscle_percent_define_insert ($1, @1, "");
232 }
233 | "%define" variable content.opt
234 {
235 muscle_percent_define_insert ($2, @2, $3);
236 }
237 | "%defines" { defines_flag = true; }
238 | "%defines" STRING
239 {
240 defines_flag = true;
241 spec_defines_file = xstrdup ($2);
242 }
243 | "%expect" INT { expected_sr_conflicts = $2; }
244 | "%expect-rr" INT { expected_rr_conflicts = $2; }
245 | "%file-prefix" STRING { spec_file_prefix = $2; }
246 | "%file-prefix" "=" STRING { spec_file_prefix = $3; } /* deprecated */
247 | "%glr-parser"
248 {
249 nondeterministic_parser = true;
250 glr_parser = true;
251 }
252 | "%initial-action" "{...}"
253 {
254 code_props action;
255 code_props_symbol_action_init (&action, $2, @2);
256 code_props_translate_code (&action);
257 gram_scanner_last_string_free ();
258 muscle_code_grow ("initial_action", action.code, @2);
259 code_scanner_last_string_free ();
260 }
261 | "%language" STRING { language_argmatch ($2, grammar_prio, @1); }
262 | "%lex-param" "{...}" { add_param ("lex_param", $2, @2); }
263 | "%locations" { locations_flag = true; }
264 | "%name-prefix" STRING { spec_name_prefix = $2; }
265 | "%name-prefix" "=" STRING { spec_name_prefix = $3; } /* deprecated */
266 | "%no-lines" { no_lines_flag = true; }
267 | "%nondeterministic-parser" { nondeterministic_parser = true; }
268 | "%output" STRING { spec_outfile = $2; }
269 | "%output" "=" STRING { spec_outfile = $3; } /* deprecated */
270 | "%parse-param" "{...}" { add_param ("parse_param", $2, @2); }
271 | "%pure-parser"
272 {
273 /* %pure-parser is deprecated in favor of `%define api.pure', so use
274 `%define api.pure' in a backward-compatible manner here. First, don't
275 complain if %pure-parser is specified multiple times. */
276 if (!muscle_find_const ("percent_define(api.pure)"))
277 muscle_percent_define_insert ("api.pure", @1, "");
278 /* In all cases, use api.pure now so that the backend doesn't complain if
279 the skeleton ignores api.pure, but do warn now if there's a previous
280 conflicting definition from an actual %define. */
281 if (!muscle_percent_define_flag_if ("api.pure"))
282 muscle_percent_define_insert ("api.pure", @1, "");
283 }
284 | "%require" STRING { version_check (&@2, $2); }
285 | "%skeleton" STRING
286 {
287 char const *skeleton_user = $2;
288 if (strchr (skeleton_user, '/'))
289 {
290 size_t dir_length = strlen (current_file);
291 char *skeleton_build;
292 while (dir_length && current_file[dir_length - 1] != '/')
293 --dir_length;
294 while (dir_length && current_file[dir_length - 1] == '/')
295 --dir_length;
296 skeleton_build =
297 xmalloc (dir_length + 1 + strlen (skeleton_user) + 1);
298 if (dir_length > 0)
299 {
300 strncpy (skeleton_build, current_file, dir_length);
301 skeleton_build[dir_length++] = '/';
302 }
303 strcpy (skeleton_build + dir_length, skeleton_user);
304 skeleton_user = uniqstr_new (skeleton_build);
305 free (skeleton_build);
306 }
307 skeleton_arg (skeleton_user, grammar_prio, @1);
308 }
309 | "%token-table" { token_table_flag = true; }
310 | "%verbose" { report_flag |= report_states; }
311 | "%yacc" { yacc_flag = true; }
312 | /*FIXME: Err? What is this horror doing here? */ ";"
313 ;
314
315 grammar_declaration:
316 precedence_declaration
317 | symbol_declaration
318 | "%start" symbol
319 {
320 grammar_start_symbol_set ($2, @2);
321 }
322 | "%destructor" "{...}" generic_symlist
323 {
324 symbol_list *list;
325 for (list = $3; list; list = list->next)
326 symbol_list_destructor_set (list, $2, @2);
327 symbol_list_free ($3);
328 }
329 | "%printer" "{...}" generic_symlist
330 {
331 symbol_list *list;
332 for (list = $3; list; list = list->next)
333 symbol_list_printer_set (list, $2, @2);
334 symbol_list_free ($3);
335 }
336 | "%default-prec"
337 {
338 default_prec = true;
339 }
340 | "%no-default-prec"
341 {
342 default_prec = false;
343 }
344 | "%code" braceless
345 {
346 /* Do not invoke muscle_percent_code_grow here since it invokes
347 muscle_user_name_list_grow. */
348 muscle_code_grow ("percent_code()", $2, @2);
349 code_scanner_last_string_free ();
350 }
351 | "%code" ID braceless
352 {
353 muscle_percent_code_grow ($2, @2, $3, @3);
354 code_scanner_last_string_free ();
355 }
356 ;
357
358
359 /*---------.
360 | %union. |
361 `---------*/
362
363 %token PERCENT_UNION "%union";
364
365 union_name:
366 /* Nothing. */ {}
367 | ID { muscle_code_grow ("union_name", $1, @1); }
368 ;
369
370 grammar_declaration:
371 "%union" union_name braceless
372 {
373 union_seen = true;
374 muscle_code_grow ("stype", $3, @3);
375 code_scanner_last_string_free ();
376 }
377 ;
378
379
380
381
382 symbol_declaration:
383 "%nterm" { current_class = nterm_sym; } symbol_defs.1
384 {
385 current_class = unknown_sym;
386 current_type = NULL;
387 }
388 | "%token" { current_class = token_sym; } symbol_defs.1
389 {
390 current_class = unknown_sym;
391 current_type = NULL;
392 }
393 | "%type" TAG symbols.1
394 {
395 symbol_list *list;
396 tag_seen = true;
397 for (list = $3; list; list = list->next)
398 symbol_type_set (list->content.sym, $2, @2);
399 symbol_list_free ($3);
400 }
401 ;
402
403 precedence_declaration:
404 precedence_declarator tag.opt symbols.prec
405 {
406 symbol_list *list;
407 ++current_prec;
408 for (list = $3; list; list = list->next)
409 {
410 symbol_type_set (list->content.sym, current_type, @2);
411 symbol_precedence_set (list->content.sym, current_prec, $1, @1);
412 }
413 symbol_list_free ($3);
414 current_type = NULL;
415 }
416 ;
417
418 precedence_declarator:
419 "%left" { $$ = left_assoc; }
420 | "%right" { $$ = right_assoc; }
421 | "%nonassoc" { $$ = non_assoc; }
422 | "%precedence" { $$ = precedence_assoc; }
423 ;
424
425 tag.opt:
426 /* Nothing. */ { current_type = NULL; }
427 | TAG { current_type = $1; tag_seen = true; }
428 ;
429
430 /* Just like symbols.1 but accept INT for the sake of POSIX. */
431 symbols.prec:
432 symbol.prec
433 { $$ = symbol_list_sym_new ($1, @1); }
434 | symbols.prec symbol.prec
435 { $$ = symbol_list_prepend ($1, symbol_list_sym_new ($2, @2)); }
436 ;
437
438 symbol.prec:
439 symbol { $$ = $1; }
440 | symbol INT { $$ = $1; symbol_user_token_number_set ($1, $2, @2); }
441 ;
442
443 /* One or more symbols to be %typed. */
444 symbols.1:
445 symbol
446 { $$ = symbol_list_sym_new ($1, @1); }
447 | symbols.1 symbol
448 { $$ = symbol_list_prepend ($1, symbol_list_sym_new ($2, @2)); }
449 ;
450
451 generic_symlist:
452 generic_symlist_item { $$ = $1; }
453 | generic_symlist generic_symlist_item { $$ = symbol_list_prepend ($1, $2); }
454 ;
455
456 generic_symlist_item:
457 symbol { $$ = symbol_list_sym_new ($1, @1); }
458 | TAG { $$ = symbol_list_type_new ($1, @1); }
459 | "<*>" { $$ = symbol_list_default_tagged_new (@1); }
460 | "<>" { $$ = symbol_list_default_tagless_new (@1); }
461 ;
462
463 /* One token definition. */
464 symbol_def:
465 TAG
466 {
467 current_type = $1;
468 tag_seen = true;
469 }
470 | id
471 {
472 symbol_class_set ($1, current_class, @1, true);
473 symbol_type_set ($1, current_type, @1);
474 }
475 | id INT
476 {
477 symbol_class_set ($1, current_class, @1, true);
478 symbol_type_set ($1, current_type, @1);
479 symbol_user_token_number_set ($1, $2, @2);
480 }
481 | id string_as_id
482 {
483 symbol_class_set ($1, current_class, @1, true);
484 symbol_type_set ($1, current_type, @1);
485 symbol_make_alias ($1, $2, @$);
486 }
487 | id INT string_as_id
488 {
489 symbol_class_set ($1, current_class, @1, true);
490 symbol_type_set ($1, current_type, @1);
491 symbol_user_token_number_set ($1, $2, @2);
492 symbol_make_alias ($1, $3, @$);
493 }
494 ;
495
496 /* One or more symbol definitions. */
497 symbol_defs.1:
498 symbol_def
499 | symbol_defs.1 symbol_def
500 ;
501
502
503 /*------------------------------------------.
504 | The grammar section: between the two %%. |
505 `------------------------------------------*/
506
507 grammar:
508 rules_or_grammar_declaration
509 | grammar rules_or_grammar_declaration
510 ;
511
512 /* As a Bison extension, one can use the grammar declarations in the
513 body of the grammar. */
514 rules_or_grammar_declaration:
515 rules
516 | grammar_declaration ";"
517 | error ";"
518 {
519 yyerrok;
520 }
521 ;
522
523 rules:
524 id_colon { current_lhs = $1; current_lhs_location = @1; } rhses.1
525 ;
526
527 rhses.1:
528 rhs { grammar_current_rule_end (@1); }
529 | rhses.1 "|" rhs { grammar_current_rule_end (@3); }
530 | rhses.1 ";"
531 ;
532
533 rhs:
534 /* Nothing. */
535 { grammar_current_rule_begin (current_lhs, current_lhs_location); }
536 | rhs symbol
537 { grammar_current_rule_symbol_append ($2, @2); }
538 | rhs "{...}"
539 { grammar_current_rule_action_append ($2, @2); }
540 | rhs "%prec" symbol
541 { grammar_current_rule_prec_set ($3, @3); }
542 | rhs "%dprec" INT
543 { grammar_current_rule_dprec_set ($3, @3); }
544 | rhs "%merge" TAG
545 { grammar_current_rule_merge_set ($3, @3); }
546 ;
547
548
549 /*---------------------------.
550 | variable and content.opt. |
551 `---------------------------*/
552
553 variable:
554 ID
555 | STRING { $$ = uniqstr_new ($1); } /* deprecated and not M4-friendly */
556 ;
557
558 /* Some content or empty by default. */
559 content.opt:
560 /* Nothing. */
561 {
562 $$ = "";
563 }
564 | STRING
565 ;
566
567
568 /*------------.
569 | braceless. |
570 `------------*/
571
572 braceless:
573 "{...}"
574 {
575 code_props plain_code;
576 $1[strlen ($1) - 1] = '\n';
577 code_props_plain_init (&plain_code, $1+1, @1);
578 code_props_translate_code (&plain_code);
579 gram_scanner_last_string_free ();
580 $$ = plain_code.code;
581 }
582 ;
583
584
585 /*--------------.
586 | Identifiers. |
587 `--------------*/
588
589 /* Identifiers are returned as uniqstr values by the scanner.
590 Depending on their use, we may need to make them genuine symbols. */
591
592 id:
593 ID
594 { $$ = symbol_from_uniqstr ($1, @1); }
595 | CHAR
596 {
597 $$ = symbol_get (char_name ($1), @1);
598 symbol_class_set ($$, token_sym, @1, false);
599 symbol_user_token_number_set ($$, $1, @1);
600 }
601 ;
602
603 id_colon:
604 ID_COLON { $$ = symbol_from_uniqstr ($1, @1); }
605 ;
606
607
608 symbol:
609 id
610 | string_as_id
611 ;
612
613 /* A string used as an ID: quote it. */
614 string_as_id:
615 STRING
616 {
617 $$ = symbol_get (quotearg_style (c_quoting_style, $1), @1);
618 symbol_class_set ($$, token_sym, @1, false);
619 }
620 ;
621
622 epilogue.opt:
623 /* Nothing. */
624 | "%%" EPILOGUE
625 {
626 code_props plain_code;
627 code_props_plain_init (&plain_code, $2, @2);
628 code_props_translate_code (&plain_code);
629 gram_scanner_last_string_free ();
630 muscle_code_grow ("epilogue", plain_code.code, @2);
631 code_scanner_last_string_free ();
632 }
633 ;
634
635 %%
636
637
638 /* Return the location of the left-hand side of a rule whose
639 right-hand side is RHS[1] ... RHS[N]. Ignore empty nonterminals in
640 the right-hand side, and return an empty location equal to the end
641 boundary of RHS[0] if the right-hand side is empty. */
642
643 static YYLTYPE
644 lloc_default (YYLTYPE const *rhs, int n)
645 {
646 int i;
647 YYLTYPE loc;
648
649 /* SGI MIPSpro 7.4.1m miscompiles "loc.start = loc.end = rhs[n].end;".
650 The bug is fixed in 7.4.2m, but play it safe for now. */
651 loc.start = rhs[n].end;
652 loc.end = rhs[n].end;
653
654 /* Ignore empty nonterminals the start of the the right-hand side.
655 Do not bother to ignore them at the end of the right-hand side,
656 since empty nonterminals have the same end as their predecessors. */
657 for (i = 1; i <= n; i++)
658 if (! equal_boundaries (rhs[i].start, rhs[i].end))
659 {
660 loc.start = rhs[i].start;
661 break;
662 }
663
664 return loc;
665 }
666
667
668 /* Add a lex-param or a parse-param (depending on TYPE) with
669 declaration DECL and location LOC. */
670
671 static void
672 add_param (char const *type, char *decl, location loc)
673 {
674 static char const alphanum[26 + 26 + 1 + 10] =
675 "abcdefghijklmnopqrstuvwxyz"
676 "ABCDEFGHIJKLMNOPQRSTUVWXYZ"
677 "_"
678 "0123456789";
679 char const *name_start = NULL;
680 char *p;
681
682 /* Stop on last actual character. */
683 for (p = decl; p[1]; p++)
684 if ((p == decl
685 || ! memchr (alphanum, p[-1], sizeof alphanum))
686 && memchr (alphanum, p[0], sizeof alphanum - 10))
687 name_start = p;
688
689 /* Strip the surrounding '{' and '}', and any blanks just inside
690 the braces. */
691 while (*--p == ' ' || *p == '\t')
692 continue;
693 p[1] = '\0';
694 while (*++decl == ' ' || *decl == '\t')
695 continue;
696
697 if (! name_start)
698 complain_at (loc, _("missing identifier in parameter declaration"));
699 else
700 {
701 char *name;
702 size_t name_len;
703
704 for (name_len = 1;
705 memchr (alphanum, name_start[name_len], sizeof alphanum);
706 name_len++)
707 continue;
708
709 name = xmalloc (name_len + 1);
710 memcpy (name, name_start, name_len);
711 name[name_len] = '\0';
712 muscle_pair_list_grow (type, decl, name);
713 free (name);
714 }
715
716 gram_scanner_last_string_free ();
717 }
718
719
720 static void
721 version_check (location const *loc, char const *version)
722 {
723 if (strverscmp (version, PACKAGE_VERSION) > 0)
724 {
725 complain_at (*loc, "require bison %s, but have %s",
726 version, PACKAGE_VERSION);
727 exit (63);
728 }
729 }
730
731 static void
732 gram_error (location const *loc, char const *msg)
733 {
734 complain_at (*loc, "%s", msg);
735 }
736
737 char const *
738 token_name (int type)
739 {
740 return yytname[YYTRANSLATE (type)];
741 }
742
743 static char const *
744 char_name (char c)
745 {
746 if (c == '\'')
747 return "'\\''";
748 else
749 {
750 char buf[4];
751 buf[0] = '\''; buf[1] = c; buf[2] = '\''; buf[3] = '\0';
752 return quotearg_style (escape_quoting_style, buf);
753 }
754 }