]> git.saurik.com Git - bison.git/blob - src/parse-gram.y
86c1decea0f16c39efde6fb729773c06748646bb
[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 ($$), stderr); } CHAR
183
184 /* braceless is not to be used for rule or symbol actions, as it
185 calls code_props_plain_init. */
186 %type <chars> STRING "%{...%}" EPILOGUE braceless content.opt
187 %type <code> "{...}"
188 %printer { fputs (quotearg_style (c_quoting_style, $$), stderr); }
189 STRING
190 %printer { fprintf (stderr, "{\n%s\n}", $$); }
191 braceless content.opt "{...}" "%{...%}" EPILOGUE
192
193 %type <uniqstr> TYPE ID ID_COLON variable
194 %printer { fprintf (stderr, "<%s>", $$); } TYPE
195 %printer { fputs ($$, stderr); } ID variable
196 %printer { fprintf (stderr, "%s:", $$); } 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 muscle_code_grow (union_seen ? "post_prologue" : "pre_prologue",
232 plain_code.code, @1);
233 code_scanner_last_string_free ();
234 }
235 | "%debug" { debug_flag = true; }
236 | "%define" variable content.opt
237 {
238 char const name_prefix[] = "percent_define(";
239 size_t length = strlen ($2);
240 char *name = xmalloc (sizeof name_prefix + length + 1);
241 strcpy (name, name_prefix);
242 strcpy (name + sizeof name_prefix - 1, $2);
243 strcpy (name + sizeof name_prefix - 1 + length, ")");
244 if (muscle_find_const (name))
245 warn_at (@2, _("%s `%s' redefined"), "%define variable", $2);
246 MUSCLE_INSERT_STRING (uniqstr_new (name), $3);
247 free (name);
248 muscle_grow_user_name_list ("percent_define_user_variables", $2, @2);
249 }
250 | "%defines" { defines_flag = true; }
251 | "%defines" STRING
252 {
253 defines_flag = true;
254 spec_defines_file = xstrdup ($2);
255 }
256 | "%error-verbose" { error_verbose = true; }
257 | "%expect" INT { expected_sr_conflicts = $2; }
258 | "%expect-rr" INT { expected_rr_conflicts = $2; }
259 | "%file-prefix" STRING { spec_file_prefix = $2; }
260 | "%file-prefix" "=" STRING { spec_file_prefix = $3; } /* deprecated */
261 | "%glr-parser"
262 {
263 nondeterministic_parser = true;
264 glr_parser = true;
265 }
266 | "%initial-action" "{...}"
267 {
268 code_props action;
269 code_props_symbol_action_init (&action, $2, @2);
270 code_props_translate_code (&action);
271 gram_scanner_last_string_free ();
272 muscle_code_grow ("initial_action", action.code, @2);
273 code_scanner_last_string_free ();
274 }
275 | "%language" STRING { language_argmatch ($2, 1, &@1); }
276 | "%lex-param" "{...}" { add_param ("lex_param", $2, @2); }
277 | "%locations" { locations_flag = true; }
278 | "%name-prefix" STRING { spec_name_prefix = $2; }
279 | "%name-prefix" "=" STRING { spec_name_prefix = $3; } /* deprecated */
280 | "%no-lines" { no_lines_flag = true; }
281 | "%nondeterministic-parser" { nondeterministic_parser = true; }
282 | "%output" STRING { spec_outfile = $2; }
283 | "%output" "=" STRING { spec_outfile = $3; } /* deprecated */
284 | "%parse-param" "{...}" { add_param ("parse_param", $2, @2); }
285 | "%pure-parser" { pure_parser = true; }
286 | "%push-parser" { push_parser = true; pull_parser = false; }
287 | "%push-pull-parser" { push_parser = true; pull_parser = true; }
288 | "%require" STRING { version_check (&@2, $2); }
289 | "%skeleton" STRING
290 {
291 char const *skeleton_user = $2;
292 if (strchr (skeleton_user, '/'))
293 {
294 size_t dir_length = strlen (current_file);
295 char *skeleton_build;
296 while (dir_length && current_file[dir_length - 1] != '/')
297 --dir_length;
298 while (dir_length && current_file[dir_length - 1] == '/')
299 --dir_length;
300 skeleton_build =
301 xmalloc (dir_length + 1 + strlen (skeleton_user) + 1);
302 if (dir_length > 0)
303 {
304 strncpy (skeleton_build, current_file, dir_length);
305 skeleton_build[dir_length++] = '/';
306 }
307 strcpy (skeleton_build + dir_length, skeleton_user);
308 skeleton_user = uniqstr_new (skeleton_build);
309 free (skeleton_build);
310 }
311 skeleton_arg (skeleton_user, 1, &@1);
312 }
313 | "%token-table" { token_table_flag = true; }
314 | "%verbose" { report_flag = report_states; }
315 | "%yacc" { yacc_flag = true; }
316 | /*FIXME: Err? What is this horror doing here? */ ";"
317 ;
318
319 grammar_declaration:
320 precedence_declaration
321 | symbol_declaration
322 | "%start" symbol
323 {
324 grammar_start_symbol_set ($2, @2);
325 }
326 | "%destructor" "{...}" generic_symlist
327 {
328 symbol_list *list;
329 for (list = $3; list; list = list->next)
330 symbol_list_destructor_set (list, $2, @2);
331 symbol_list_free ($3);
332 }
333 | "%printer" "{...}" generic_symlist
334 {
335 symbol_list *list;
336 for (list = $3; list; list = list->next)
337 symbol_list_printer_set (list, $2, @2);
338 symbol_list_free ($3);
339 }
340 | "%default-prec"
341 {
342 default_prec = true;
343 }
344 | "%no-default-prec"
345 {
346 default_prec = false;
347 }
348 | "%code" braceless
349 {
350 muscle_code_grow ("percent_code()", $2, @2);
351 code_scanner_last_string_free ();
352 }
353 | "%code" ID braceless
354 {
355 char const name_prefix[] = "percent_code(";
356 size_t length = strlen ($2);
357 char *name = xmalloc (sizeof name_prefix + length + 1);
358 strcpy (name, name_prefix);
359 strcpy (name + sizeof name_prefix - 1, $2);
360 strcpy (name + sizeof name_prefix - 1 + length, ")");
361 muscle_code_grow (uniqstr_new (name), $3, @3);
362 free (name);
363 code_scanner_last_string_free ();
364 muscle_grow_user_name_list ("percent_code_user_qualifiers", $2, @2);
365 }
366 ;
367
368
369 /*----------*
370 | %union. |
371 *----------*/
372
373 %token PERCENT_UNION "%union";
374
375 union_name:
376 /* Nothing. */ {}
377 | ID { muscle_code_grow ("union_name", $1, @1); }
378 ;
379
380 grammar_declaration:
381 "%union" union_name braceless
382 {
383 union_seen = true;
384 muscle_code_grow ("stype", $3, @3);
385 code_scanner_last_string_free ();
386 }
387 ;
388
389
390
391
392 symbol_declaration:
393 "%nterm" { current_class = nterm_sym; } symbol_defs.1
394 {
395 current_class = unknown_sym;
396 current_type = NULL;
397 }
398 | "%token" { current_class = token_sym; } symbol_defs.1
399 {
400 current_class = unknown_sym;
401 current_type = NULL;
402 }
403 | "%type" TYPE symbols.1
404 {
405 symbol_list *list;
406 tag_seen = true;
407 for (list = $3; list; list = list->next)
408 symbol_type_set (list->content.sym, $2, @2);
409 symbol_list_free ($3);
410 }
411 ;
412
413 precedence_declaration:
414 precedence_declarator type.opt symbols.1
415 {
416 symbol_list *list;
417 ++current_prec;
418 for (list = $3; list; list = list->next)
419 {
420 symbol_type_set (list->content.sym, current_type, @2);
421 symbol_precedence_set (list->content.sym, current_prec, $1, @1);
422 }
423 symbol_list_free ($3);
424 current_type = NULL;
425 }
426 ;
427
428 precedence_declarator:
429 "%left" { $$ = left_assoc; }
430 | "%right" { $$ = right_assoc; }
431 | "%nonassoc" { $$ = non_assoc; }
432 ;
433
434 type.opt:
435 /* Nothing. */ { current_type = NULL; }
436 | TYPE { current_type = $1; tag_seen = true; }
437 ;
438
439 /* One or more symbols to be %typed. */
440 symbols.1:
441 symbol
442 { $$ = symbol_list_sym_new ($1, @1); }
443 | symbols.1 symbol
444 { $$ = symbol_list_prepend ($1, symbol_list_sym_new ($2, @2)); }
445 ;
446
447 generic_symlist:
448 generic_symlist_item { $$ = $1; }
449 | generic_symlist generic_symlist_item { $$ = symbol_list_prepend ($1, $2); }
450 ;
451
452 generic_symlist_item:
453 symbol { $$ = symbol_list_sym_new ($1, @1); }
454 | TYPE { $$ = symbol_list_type_new ($1, @1); }
455 | "<*>" { $$ = symbol_list_default_tagged_new (@1); }
456 | "<>" { $$ = symbol_list_default_tagless_new (@1); }
457 ;
458
459 /* One token definition. */
460 symbol_def:
461 TYPE
462 {
463 current_type = $1;
464 tag_seen = true;
465 }
466 | id
467 {
468 symbol_class_set ($1, current_class, @1, true);
469 symbol_type_set ($1, current_type, @1);
470 }
471 | id INT
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 }
477 | id string_as_id
478 {
479 symbol_class_set ($1, current_class, @1, true);
480 symbol_type_set ($1, current_type, @1);
481 symbol_make_alias ($1, $2, @$);
482 }
483 | id INT string_as_id
484 {
485 symbol_class_set ($1, current_class, @1, true);
486 symbol_type_set ($1, current_type, @1);
487 symbol_user_token_number_set ($1, $2, @2);
488 symbol_make_alias ($1, $3, @$);
489 }
490 ;
491
492 /* One or more symbol definitions. */
493 symbol_defs.1:
494 symbol_def
495 | symbol_defs.1 symbol_def
496 ;
497
498
499 /*------------------------------------------.
500 | The grammar section: between the two %%. |
501 `------------------------------------------*/
502
503 grammar:
504 rules_or_grammar_declaration
505 | grammar rules_or_grammar_declaration
506 ;
507
508 /* As a Bison extension, one can use the grammar declarations in the
509 body of the grammar. */
510 rules_or_grammar_declaration:
511 rules
512 | grammar_declaration ";"
513 | error ";"
514 {
515 yyerrok;
516 }
517 ;
518
519 rules:
520 id_colon { current_lhs = $1; current_lhs_location = @1; } rhses.1
521 ;
522
523 rhses.1:
524 rhs { grammar_current_rule_end (@1); }
525 | rhses.1 "|" rhs { grammar_current_rule_end (@3); }
526 | rhses.1 ";"
527 ;
528
529 rhs:
530 /* Nothing. */
531 { grammar_current_rule_begin (current_lhs, current_lhs_location); }
532 | rhs symbol
533 { grammar_current_rule_symbol_append ($2, @2); }
534 | rhs "{...}"
535 { grammar_current_rule_action_append ($2, @2); }
536 | rhs "%prec" symbol
537 { grammar_current_rule_prec_set ($3, @3); }
538 | rhs "%dprec" INT
539 { grammar_current_rule_dprec_set ($3, @3); }
540 | rhs "%merge" TYPE
541 { grammar_current_rule_merge_set ($3, @3); }
542 ;
543
544
545 /*----------------------------*
546 | variable and content.opt. |
547 *---------------------------*/
548
549 variable:
550 ID
551 | STRING { $$ = uniqstr_new ($1); } /* deprecated and not M4-friendly */
552 ;
553
554 /* Some content or empty by default. */
555 content.opt:
556 /* Nothing. */
557 {
558 $$ = "";
559 }
560 | STRING
561 ;
562
563
564 /*-------------*
565 | braceless. |
566 *-------------*/
567
568 braceless:
569 "{...}"
570 {
571 code_props plain_code;
572 $1[strlen ($1) - 1] = '\n';
573 code_props_plain_init (&plain_code, $1+1, @1);
574 code_props_translate_code (&plain_code);
575 gram_scanner_last_string_free ();
576 $$ = plain_code.code;
577 }
578 ;
579
580
581 /*---------------*
582 | Identifiers. |
583 *---------------*/
584
585 /* Identifiers are returned as uniqstr values by the scanner.
586 Depending on their use, we may need to make them genuine symbols. */
587
588 id:
589 ID
590 { $$ = symbol_from_uniqstr ($1, @1); }
591 | CHAR
592 {
593 $$ = symbol_get (char_name ($1), @1);
594 symbol_class_set ($$, token_sym, @1, false);
595 symbol_user_token_number_set ($$, $1, @1);
596 }
597 ;
598
599 id_colon:
600 ID_COLON { $$ = symbol_from_uniqstr ($1, @1); }
601 ;
602
603
604 symbol:
605 id
606 | string_as_id
607 ;
608
609 /* A string used as an ID: quote it. */
610 string_as_id:
611 STRING
612 {
613 $$ = symbol_get (quotearg_style (c_quoting_style, $1), @1);
614 symbol_class_set ($$, token_sym, @1, false);
615 }
616 ;
617
618 epilogue.opt:
619 /* Nothing. */
620 | "%%" EPILOGUE
621 {
622 code_props plain_code;
623 code_props_plain_init (&plain_code, $2, @2);
624 code_props_translate_code (&plain_code);
625 gram_scanner_last_string_free ();
626 muscle_code_grow ("epilogue", plain_code.code, @2);
627 code_scanner_last_string_free ();
628 }
629 ;
630
631 %%
632
633
634 /* Return the location of the left-hand side of a rule whose
635 right-hand side is RHS[1] ... RHS[N]. Ignore empty nonterminals in
636 the right-hand side, and return an empty location equal to the end
637 boundary of RHS[0] if the right-hand side is empty. */
638
639 static YYLTYPE
640 lloc_default (YYLTYPE const *rhs, int n)
641 {
642 int i;
643 YYLTYPE loc;
644
645 /* SGI MIPSpro 7.4.1m miscompiles "loc.start = loc.end = rhs[n].end;".
646 The bug is fixed in 7.4.2m, but play it safe for now. */
647 loc.start = rhs[n].end;
648 loc.end = rhs[n].end;
649
650 /* Ignore empty nonterminals the start of the the right-hand side.
651 Do not bother to ignore them at the end of the right-hand side,
652 since empty nonterminals have the same end as their predecessors. */
653 for (i = 1; i <= n; i++)
654 if (! equal_boundaries (rhs[i].start, rhs[i].end))
655 {
656 loc.start = rhs[i].start;
657 break;
658 }
659
660 return loc;
661 }
662
663
664 /* Add a lex-param or a parse-param (depending on TYPE) with
665 declaration DECL and location LOC. */
666
667 static void
668 add_param (char const *type, char *decl, location loc)
669 {
670 static char const alphanum[26 + 26 + 1 + 10] =
671 "abcdefghijklmnopqrstuvwxyz"
672 "ABCDEFGHIJKLMNOPQRSTUVWXYZ"
673 "_"
674 "0123456789";
675 char const *name_start = NULL;
676 char *p;
677
678 /* Stop on last actual character. */
679 for (p = decl; p[1]; p++)
680 if ((p == decl
681 || ! memchr (alphanum, p[-1], sizeof alphanum))
682 && memchr (alphanum, p[0], sizeof alphanum - 10))
683 name_start = p;
684
685 /* Strip the surrounding '{' and '}', and any blanks just inside
686 the braces. */
687 while (*--p == ' ' || *p == '\t')
688 continue;
689 p[1] = '\0';
690 while (*++decl == ' ' || *decl == '\t')
691 continue;
692
693 if (! name_start)
694 complain_at (loc, _("missing identifier in parameter declaration"));
695 else
696 {
697 char *name;
698 size_t name_len;
699
700 for (name_len = 1;
701 memchr (alphanum, name_start[name_len], sizeof alphanum);
702 name_len++)
703 continue;
704
705 name = xmalloc (name_len + 1);
706 memcpy (name, name_start, name_len);
707 name[name_len] = '\0';
708 muscle_pair_list_grow (type, decl, name);
709 free (name);
710 }
711
712 gram_scanner_last_string_free ();
713 }
714
715
716 static void
717 version_check (location const *loc, char const *version)
718 {
719 if (strverscmp (version, PACKAGE_VERSION) > 0)
720 {
721 complain_at (*loc, "require bison %s, but have %s",
722 version, PACKAGE_VERSION);
723 exit (63);
724 }
725 }
726
727 static void
728 gram_error (location const *loc, char const *msg)
729 {
730 complain_at (*loc, "%s", msg);
731 }
732
733 char const *
734 token_name (int type)
735 {
736 return yytname[YYTRANSLATE (type)];
737 }
738
739 static char const *
740 char_name (char c)
741 {
742 if (c == '\'')
743 return "'\\''";
744 else
745 {
746 char buf[4];
747 buf[0] = '\''; buf[1] = c; buf[2] = '\''; buf[3] = '\0';
748 return quotearg_style (escape_quoting_style, buf);
749 }
750 }