]> git.saurik.com Git - bison.git/blob - src/parse-gram.y
5cefbce158c7fa6801563b5100ba5bfb3dfd77d8
[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}", $$); }
191 braceless content.opt "{...}" "%{...%}" EPILOGUE
192
193 %type <uniqstr> TYPE ID ID_COLON
194 %printer { fprintf (stderr, "<%s>", $$); } 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 char const name_prefix[] = "percent_define_";
240 char *name = xmalloc (sizeof name_prefix + strlen ($2));
241 strcpy (name, name_prefix);
242 strcpy (name + sizeof name_prefix - 1, $2);
243 if (muscle_find_const (name))
244 warn_at (@2, _("%s `%s' redefined"), "%define variable", $2);
245 MUSCLE_INSERT_STRING (uniqstr_new (name), $3);
246 free (name);
247 muscle_grow_user_name_list ("user_percent_define_variables", $2, @2);
248 }
249 | "%defines" { defines_flag = true; }
250 | "%defines" STRING
251 {
252 defines_flag = true;
253 spec_defines_file = xstrdup ($2);
254 }
255 | "%error-verbose" { error_verbose = true; }
256 | "%expect" INT { expected_sr_conflicts = $2; }
257 | "%expect-rr" INT { expected_rr_conflicts = $2; }
258 | "%file-prefix" STRING { spec_file_prefix = $2; }
259 | "%file-prefix" "=" STRING { spec_file_prefix = $3; } /* deprecated */
260 | "%glr-parser"
261 {
262 nondeterministic_parser = true;
263 glr_parser = true;
264 }
265 | "%initial-action" "{...}"
266 {
267 code_props action;
268 code_props_symbol_action_init (&action, $2, @2);
269 code_props_translate_code (&action);
270 gram_scanner_last_string_free ();
271 muscle_code_grow ("initial_action", action.code, @2);
272 code_scanner_last_string_free ();
273 }
274 | "%language" STRING { language_argmatch ($2, 1, &@1); }
275 | "%lex-param" "{...}" { add_param ("lex_param", $2, @2); }
276 | "%locations" { locations_flag = true; }
277 | "%name-prefix" STRING { spec_name_prefix = $2; }
278 | "%name-prefix" "=" STRING { spec_name_prefix = $3; } /* deprecated */
279 | "%no-lines" { no_lines_flag = true; }
280 | "%nondeterministic-parser" { nondeterministic_parser = true; }
281 | "%output" STRING { spec_outfile = $2; }
282 | "%output" "=" STRING { spec_outfile = $3; } /* deprecated */
283 | "%parse-param" "{...}" { add_param ("parse_param", $2, @2); }
284 | "%pure-parser" { pure_parser = true; }
285 | "%push-parser" { push_parser = true; pull_parser = false; }
286 | "%push-pull-parser" { push_parser = true; pull_parser = true; }
287 | "%require" STRING { version_check (&@2, $2); }
288 | "%skeleton" STRING { skeleton_arg ($2, 1, &@1); }
289 | "%token-table" { token_table_flag = true; }
290 | "%verbose" { report_flag = report_states; }
291 | "%yacc" { yacc_flag = true; }
292 | /*FIXME: Err? What is this horror doing here? */ ";"
293 ;
294
295 grammar_declaration:
296 precedence_declaration
297 | symbol_declaration
298 | "%start" symbol
299 {
300 grammar_start_symbol_set ($2, @2);
301 }
302 | "%destructor" "{...}" generic_symlist
303 {
304 symbol_list *list;
305 for (list = $3; list; list = list->next)
306 symbol_list_destructor_set (list, $2, @2);
307 symbol_list_free ($3);
308 }
309 | "%printer" "{...}" generic_symlist
310 {
311 symbol_list *list;
312 for (list = $3; list; list = list->next)
313 symbol_list_printer_set (list, $2, @2);
314 symbol_list_free ($3);
315 }
316 | "%default-prec"
317 {
318 default_prec = true;
319 }
320 | "%no-default-prec"
321 {
322 default_prec = false;
323 }
324 | "%code" braceless
325 {
326 muscle_code_grow ("percent_code", $2, @2);
327 code_scanner_last_string_free ();
328 }
329 | "%code" STRING braceless
330 {
331 /* FIXME: Special characters in $2 may break %code.
332 For example: `['. */
333 char const name_prefix[] = "percent_code_";
334 char *name = xmalloc (sizeof name_prefix + strlen ($2));
335 strcpy (name, name_prefix);
336 strcpy (name + sizeof name_prefix - 1, $2);
337 muscle_code_grow (uniqstr_new (name), $3, @3);
338 free (name);
339 code_scanner_last_string_free ();
340 muscle_grow_user_name_list ("user_percent_code_qualifiers", $2, @2);
341 }
342 ;
343
344
345 /*----------*
346 | %union. |
347 *----------*/
348
349 %token PERCENT_UNION "%union";
350
351 union_name:
352 /* Nothing. */ {}
353 | ID { muscle_code_grow ("union_name", $1, @1); }
354 ;
355
356 grammar_declaration:
357 "%union" union_name "{...}"
358 {
359 char const *body = $3;
360
361 /* Concatenate the %union bodies. If this is the first %union, make sure
362 the synchronization line appears after the opening '{' so as not to
363 confuse Doxygen. Otherwise, turn the previous %union's trailing '}'
364 into '\n', and omit the new %union's leading '{'. */
365 if (!union_seen)
366 {
367 muscle_grow ("stype", "{", "");
368 }
369 else
370 {
371 char *code = muscle_find ("stype");
372 code[strlen (code) - 1] = '\n';
373 }
374 body++;
375
376 union_seen = true;
377 muscle_code_grow ("stype", body, @3);
378 gram_scanner_last_string_free ();
379 }
380 ;
381
382
383
384
385 symbol_declaration:
386 "%nterm" { current_class = nterm_sym; } symbol_defs.1
387 {
388 current_class = unknown_sym;
389 current_type = NULL;
390 }
391 | "%token" { current_class = token_sym; } symbol_defs.1
392 {
393 current_class = unknown_sym;
394 current_type = NULL;
395 }
396 | "%type" TYPE symbols.1
397 {
398 symbol_list *list;
399 tag_seen = true;
400 for (list = $3; list; list = list->next)
401 symbol_type_set (list->content.sym, $2, @2);
402 symbol_list_free ($3);
403 }
404 ;
405
406 precedence_declaration:
407 precedence_declarator type.opt symbols.1
408 {
409 symbol_list *list;
410 ++current_prec;
411 for (list = $3; list; list = list->next)
412 {
413 symbol_type_set (list->content.sym, current_type, @2);
414 symbol_precedence_set (list->content.sym, current_prec, $1, @1);
415 }
416 symbol_list_free ($3);
417 current_type = NULL;
418 }
419 ;
420
421 precedence_declarator:
422 "%left" { $$ = left_assoc; }
423 | "%right" { $$ = right_assoc; }
424 | "%nonassoc" { $$ = non_assoc; }
425 ;
426
427 type.opt:
428 /* Nothing. */ { current_type = NULL; }
429 | TYPE { current_type = $1; tag_seen = true; }
430 ;
431
432 /* One or more symbols to be %typed. */
433 symbols.1:
434 symbol
435 { $$ = symbol_list_sym_new ($1, @1); }
436 | symbols.1 symbol
437 { $$ = symbol_list_prepend ($1, symbol_list_sym_new ($2, @2)); }
438 ;
439
440 generic_symlist:
441 generic_symlist_item { $$ = $1; }
442 | generic_symlist generic_symlist_item { $$ = symbol_list_prepend ($1, $2); }
443 ;
444
445 generic_symlist_item:
446 symbol { $$ = symbol_list_sym_new ($1, @1); }
447 | TYPE { $$ = symbol_list_type_new ($1, @1); }
448 | "<*>" { $$ = symbol_list_default_tagged_new (@1); }
449 | "<>" { $$ = symbol_list_default_tagless_new (@1); }
450 ;
451
452 /* One token definition. */
453 symbol_def:
454 TYPE
455 {
456 current_type = $1;
457 tag_seen = true;
458 }
459 | id
460 {
461 symbol_class_set ($1, current_class, @1, true);
462 symbol_type_set ($1, current_type, @1);
463 }
464 | id INT
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 }
470 | id string_as_id
471 {
472 symbol_class_set ($1, current_class, @1, true);
473 symbol_type_set ($1, current_type, @1);
474 symbol_make_alias ($1, $2, @$);
475 }
476 | id INT string_as_id
477 {
478 symbol_class_set ($1, current_class, @1, true);
479 symbol_type_set ($1, current_type, @1);
480 symbol_user_token_number_set ($1, $2, @2);
481 symbol_make_alias ($1, $3, @$);
482 }
483 ;
484
485 /* One or more symbol definitions. */
486 symbol_defs.1:
487 symbol_def
488 | symbol_defs.1 symbol_def
489 ;
490
491
492 /*------------------------------------------.
493 | The grammar section: between the two %%. |
494 `------------------------------------------*/
495
496 grammar:
497 rules_or_grammar_declaration
498 | grammar rules_or_grammar_declaration
499 ;
500
501 /* As a Bison extension, one can use the grammar declarations in the
502 body of the grammar. */
503 rules_or_grammar_declaration:
504 rules
505 | grammar_declaration ";"
506 | error ";"
507 {
508 yyerrok;
509 }
510 ;
511
512 rules:
513 id_colon { current_lhs = $1; current_lhs_location = @1; } rhses.1
514 ;
515
516 rhses.1:
517 rhs { grammar_current_rule_end (@1); }
518 | rhses.1 "|" rhs { grammar_current_rule_end (@3); }
519 | rhses.1 ";"
520 ;
521
522 rhs:
523 /* Nothing. */
524 { grammar_current_rule_begin (current_lhs, current_lhs_location); }
525 | rhs symbol
526 { grammar_current_rule_symbol_append ($2, @2); }
527 | rhs "{...}"
528 { grammar_current_rule_action_append ($2, @2); }
529 | rhs "%prec" symbol
530 { grammar_current_rule_prec_set ($3, @3); }
531 | rhs "%dprec" INT
532 { grammar_current_rule_dprec_set ($3, @3); }
533 | rhs "%merge" TYPE
534 { grammar_current_rule_merge_set ($3, @3); }
535 ;
536
537
538 /*---------------*
539 | content.opt. |
540 *--------------*/
541
542 /* Some content or "1" by default. */
543 content.opt:
544 /* Nothing. */
545 {
546 static char one[] = "1";
547 $$ = one;
548 }
549 | STRING
550 ;
551
552
553 /*-------------*
554 | braceless. |
555 *-------------*/
556
557 braceless:
558 "{...}"
559 {
560 code_props plain_code;
561 $1[strlen ($1) - 1] = '\n';
562 code_props_plain_init (&plain_code, $1+1, @1);
563 code_props_translate_code (&plain_code);
564 gram_scanner_last_string_free ();
565 $$ = plain_code.code;
566 }
567 ;
568
569
570 /*---------------*
571 | Identifiers. |
572 *---------------*/
573
574 /* Identifiers are returned as uniqstr values by the scanner.
575 Depending on their use, we may need to make them genuine symbols. */
576
577 id:
578 ID
579 { $$ = symbol_from_uniqstr ($1, @1); }
580 | CHAR
581 {
582 $$ = symbol_get (char_name ($1), @1);
583 symbol_class_set ($$, token_sym, @1, false);
584 symbol_user_token_number_set ($$, $1, @1);
585 }
586 ;
587
588 id_colon:
589 ID_COLON { $$ = symbol_from_uniqstr ($1, @1); }
590 ;
591
592
593 symbol:
594 id
595 | string_as_id
596 ;
597
598 /* A string used as an ID: quote it. */
599 string_as_id:
600 STRING
601 {
602 $$ = symbol_get (quotearg_style (c_quoting_style, $1), @1);
603 symbol_class_set ($$, token_sym, @1, false);
604 }
605 ;
606
607 epilogue.opt:
608 /* Nothing. */
609 | "%%" EPILOGUE
610 {
611 code_props plain_code;
612 code_props_plain_init (&plain_code, $2, @2);
613 code_props_translate_code (&plain_code);
614 gram_scanner_last_string_free ();
615 muscle_code_grow ("epilogue", plain_code.code, @2);
616 code_scanner_last_string_free ();
617 }
618 ;
619
620 %%
621
622
623 /* Return the location of the left-hand side of a rule whose
624 right-hand side is RHS[1] ... RHS[N]. Ignore empty nonterminals in
625 the right-hand side, and return an empty location equal to the end
626 boundary of RHS[0] if the right-hand side is empty. */
627
628 static YYLTYPE
629 lloc_default (YYLTYPE const *rhs, int n)
630 {
631 int i;
632 YYLTYPE loc;
633
634 /* SGI MIPSpro 7.4.1m miscompiles "loc.start = loc.end = rhs[n].end;".
635 The bug is fixed in 7.4.2m, but play it safe for now. */
636 loc.start = rhs[n].end;
637 loc.end = rhs[n].end;
638
639 /* Ignore empty nonterminals the start of the the right-hand side.
640 Do not bother to ignore them at the end of the right-hand side,
641 since empty nonterminals have the same end as their predecessors. */
642 for (i = 1; i <= n; i++)
643 if (! equal_boundaries (rhs[i].start, rhs[i].end))
644 {
645 loc.start = rhs[i].start;
646 break;
647 }
648
649 return loc;
650 }
651
652
653 /* Add a lex-param or a parse-param (depending on TYPE) with
654 declaration DECL and location LOC. */
655
656 static void
657 add_param (char const *type, char *decl, location loc)
658 {
659 static char const alphanum[26 + 26 + 1 + 10] =
660 "abcdefghijklmnopqrstuvwxyz"
661 "ABCDEFGHIJKLMNOPQRSTUVWXYZ"
662 "_"
663 "0123456789";
664 char const *name_start = NULL;
665 char *p;
666
667 /* Stop on last actual character. */
668 for (p = decl; p[1]; p++)
669 if ((p == decl
670 || ! memchr (alphanum, p[-1], sizeof alphanum))
671 && memchr (alphanum, p[0], sizeof alphanum - 10))
672 name_start = p;
673
674 /* Strip the surrounding '{' and '}', and any blanks just inside
675 the braces. */
676 while (*--p == ' ' || *p == '\t')
677 continue;
678 p[1] = '\0';
679 while (*++decl == ' ' || *decl == '\t')
680 continue;
681
682 if (! name_start)
683 complain_at (loc, _("missing identifier in parameter declaration"));
684 else
685 {
686 char *name;
687 size_t name_len;
688
689 for (name_len = 1;
690 memchr (alphanum, name_start[name_len], sizeof alphanum);
691 name_len++)
692 continue;
693
694 name = xmalloc (name_len + 1);
695 memcpy (name, name_start, name_len);
696 name[name_len] = '\0';
697 muscle_pair_list_grow (type, decl, name);
698 free (name);
699 }
700
701 gram_scanner_last_string_free ();
702 }
703
704
705 static void
706 version_check (location const *loc, char const *version)
707 {
708 if (strverscmp (version, PACKAGE_VERSION) > 0)
709 {
710 complain_at (*loc, "require bison %s, but have %s",
711 version, PACKAGE_VERSION);
712 exit (63);
713 }
714 }
715
716 static void
717 gram_error (location const *loc, char const *msg)
718 {
719 complain_at (*loc, "%s", msg);
720 }
721
722 char const *
723 token_name (int type)
724 {
725 return yytname[YYTRANSLATE (type)];
726 }
727
728 static char const *
729 char_name (char c)
730 {
731 if (c == '\'')
732 return "'\\''";
733 else
734 {
735 char buf[4];
736 buf[0] = '\''; buf[1] = c; buf[2] = '\''; buf[3] = '\0';
737 return quotearg_style (escape_quoting_style, buf);
738 }
739 }