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