]> git.saurik.com Git - bison.git/blob - src/parse-gram.y
Named symbol references.
[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 "named-ref.h"
34 #include "scan-gram.h"
35 #include "scan-code.h"
36
37 #define YYLLOC_DEFAULT(Current, Rhs, N) (Current) = lloc_default (Rhs, N)
38 static YYLTYPE lloc_default (YYLTYPE const *, int);
39
40 #define YY_LOCATION_PRINT(File, Loc) \
41 location_print (File, Loc)
42
43 static void version_check (location const *loc, char const *version);
44
45 /* Request detailed syntax error messages, and pass them to GRAM_ERROR.
46 FIXME: depends on the undocumented availability of YYLLOC. */
47 #undef yyerror
48 #define yyerror(Msg) \
49 gram_error (&yylloc, Msg)
50 static void gram_error (location const *, char const *);
51
52 static char const *char_name (char);
53
54 /** Add a lex-param or a parse-param.
55 *
56 * \param type \a lex_param or \a parse_param
57 * \param decl the formal argument
58 * \param loc the location in the source.
59 */
60 static void add_param (char const *type, char *decl, location loc);
61
62
63 static symbol_class current_class = unknown_sym;
64 static uniqstr current_type = NULL;
65 static symbol *current_lhs;
66 static location current_lhs_location;
67 static named_ref *current_lhs_named_ref;
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 named_ref *named_ref;
104 };
105
106 /* Define the tokens together with their human representation. */
107 %token GRAM_EOF 0 "end of file"
108 %token STRING "string"
109 %token INT "integer"
110
111 %token PERCENT_TOKEN "%token"
112 %token PERCENT_NTERM "%nterm"
113
114 %token PERCENT_TYPE "%type"
115 %token PERCENT_DESTRUCTOR "%destructor"
116 %token PERCENT_PRINTER "%printer"
117
118 %token PERCENT_LEFT "%left"
119 %token PERCENT_RIGHT "%right"
120 %token PERCENT_NONASSOC "%nonassoc"
121 %token PERCENT_PRECEDENCE "%precedence"
122
123 %token PERCENT_PREC "%prec"
124 %token PERCENT_DPREC "%dprec"
125 %token PERCENT_MERGE "%merge"
126
127
128 /*----------------------.
129 | Global Declarations. |
130 `----------------------*/
131
132 %token
133 PERCENT_CODE "%code"
134 PERCENT_DEFAULT_PREC "%default-prec"
135 PERCENT_DEFINE "%define"
136 PERCENT_DEFINES "%defines"
137 PERCENT_EXPECT "%expect"
138 PERCENT_EXPECT_RR "%expect-rr"
139 PERCENT_FLAG "%<flag>"
140 PERCENT_FILE_PREFIX "%file-prefix"
141 PERCENT_GLR_PARSER "%glr-parser"
142 PERCENT_INITIAL_ACTION "%initial-action"
143 PERCENT_LANGUAGE "%language"
144 PERCENT_LEX_PARAM "%lex-param"
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_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 %token BRACKETED_ID "[id]"
174
175 %type <character> CHAR
176 %printer { fputs (char_name ($$), stderr); } CHAR
177
178 /* braceless is not to be used for rule or symbol actions, as it
179 calls code_props_plain_init. */
180 %type <chars> STRING "%{...%}" EPILOGUE braceless content.opt
181 %type <code> "{...}"
182 %printer { fputs (quotearg_style (c_quoting_style, $$), stderr); }
183 STRING
184 %printer { fprintf (stderr, "{\n%s\n}", $$); }
185 braceless content.opt "{...}" "%{...%}" EPILOGUE
186
187 %type <uniqstr> TAG ID ID_COLON BRACKETED_ID PERCENT_FLAG variable
188 %type <named_ref> named_ref.opt
189 %printer { fputs ($$, stderr); } ID variable
190 %printer { fprintf (stderr, "%s:", $$); } ID_COLON
191 %printer { fprintf (stderr, "%%%s", $$); } PERCENT_FLAG
192 %printer { fprintf (stderr, "<%s>", $$); } TAG
193
194 %type <integer> INT
195 %printer { fprintf (stderr, "%d", $$); } INT
196
197 %type <symbol> id id_colon symbol symbol.prec string_as_id
198 %printer { fprintf (stderr, "%s", $$->tag); } id symbol string_as_id
199 %printer { fprintf (stderr, "%s:", $$->tag); } id_colon
200
201 %type <assoc> precedence_declarator
202 %type <list> symbols.1 symbols.prec generic_symlist generic_symlist_item
203 %%
204
205 input:
206 prologue_declarations "%%" grammar epilogue.opt
207 ;
208
209
210 /*------------------------------------.
211 | Declarations: before the first %%. |
212 `------------------------------------*/
213
214 prologue_declarations:
215 /* Nothing */
216 | prologue_declarations prologue_declaration
217 ;
218
219 prologue_declaration:
220 grammar_declaration
221 | "%{...%}"
222 {
223 code_props plain_code;
224 code_props_plain_init (&plain_code, $1, @1);
225 code_props_translate_code (&plain_code);
226 gram_scanner_last_string_free ();
227 muscle_code_grow (union_seen ? "post_prologue" : "pre_prologue",
228 plain_code.code, @1);
229 code_scanner_last_string_free ();
230 }
231 | "%<flag>"
232 {
233 muscle_percent_define_ensure ($1, @1, true);
234 }
235 | "%define" variable content.opt
236 {
237 muscle_percent_define_insert ($2, @2, $3,
238 MUSCLE_PERCENT_DEFINE_GRAMMAR_FILE);
239 }
240 | "%defines" { defines_flag = true; }
241 | "%defines" STRING
242 {
243 defines_flag = true;
244 spec_defines_file = xstrdup ($2);
245 }
246 | "%expect" INT { expected_sr_conflicts = $2; }
247 | "%expect-rr" INT { expected_rr_conflicts = $2; }
248 | "%file-prefix" STRING { spec_file_prefix = $2; }
249 | "%file-prefix" "=" STRING { spec_file_prefix = $3; } /* deprecated */
250 | "%glr-parser"
251 {
252 nondeterministic_parser = true;
253 glr_parser = true;
254 }
255 | "%initial-action" "{...}"
256 {
257 code_props action;
258 code_props_symbol_action_init (&action, $2, @2);
259 code_props_translate_code (&action);
260 gram_scanner_last_string_free ();
261 muscle_code_grow ("initial_action", action.code, @2);
262 code_scanner_last_string_free ();
263 }
264 | "%language" STRING { language_argmatch ($2, grammar_prio, @1); }
265 | "%lex-param" "{...}" { add_param ("lex_param", $2, @2); }
266 | "%name-prefix" STRING { spec_name_prefix = $2; }
267 | "%name-prefix" "=" STRING { spec_name_prefix = $3; } /* deprecated */
268 | "%no-lines" { no_lines_flag = true; }
269 | "%nondeterministic-parser" { nondeterministic_parser = true; }
270 | "%output" STRING { spec_outfile = $2; }
271 | "%output" "=" STRING { spec_outfile = $3; } /* deprecated */
272 | "%parse-param" "{...}" { add_param ("parse_param", $2, @2); }
273 | "%require" STRING { version_check (&@2, $2); }
274 | "%skeleton" STRING
275 {
276 char const *skeleton_user = $2;
277 if (strchr (skeleton_user, '/'))
278 {
279 size_t dir_length = strlen (current_file);
280 char *skeleton_build;
281 while (dir_length && current_file[dir_length - 1] != '/')
282 --dir_length;
283 while (dir_length && current_file[dir_length - 1] == '/')
284 --dir_length;
285 skeleton_build =
286 xmalloc (dir_length + 1 + strlen (skeleton_user) + 1);
287 if (dir_length > 0)
288 {
289 strncpy (skeleton_build, current_file, dir_length);
290 skeleton_build[dir_length++] = '/';
291 }
292 strcpy (skeleton_build + dir_length, skeleton_user);
293 skeleton_user = uniqstr_new (skeleton_build);
294 free (skeleton_build);
295 }
296 skeleton_arg (skeleton_user, grammar_prio, @1);
297 }
298 | "%token-table" { token_table_flag = true; }
299 | "%verbose" { report_flag |= report_states; }
300 | "%yacc" { yacc_flag = true; }
301 | /*FIXME: Err? What is this horror doing here? */ ";"
302 ;
303
304 grammar_declaration:
305 precedence_declaration
306 | symbol_declaration
307 | "%start" symbol
308 {
309 grammar_start_symbol_set ($2, @2);
310 }
311 | "%destructor" "{...}" generic_symlist
312 {
313 symbol_list *list;
314 for (list = $3; list; list = list->next)
315 symbol_list_destructor_set (list, $2, @2);
316 symbol_list_free ($3);
317 }
318 | "%printer" "{...}" generic_symlist
319 {
320 symbol_list *list;
321 for (list = $3; list; list = list->next)
322 symbol_list_printer_set (list, $2, @2);
323 symbol_list_free ($3);
324 }
325 | "%default-prec"
326 {
327 default_prec = true;
328 }
329 | "%no-default-prec"
330 {
331 default_prec = false;
332 }
333 | "%code" braceless
334 {
335 /* Do not invoke muscle_percent_code_grow here since it invokes
336 muscle_user_name_list_grow. */
337 muscle_code_grow ("percent_code()", $2, @2);
338 code_scanner_last_string_free ();
339 }
340 | "%code" ID braceless
341 {
342 muscle_percent_code_grow ($2, @2, $3, @3);
343 code_scanner_last_string_free ();
344 }
345 ;
346
347
348 /*---------.
349 | %union. |
350 `---------*/
351
352 %token PERCENT_UNION "%union";
353
354 union_name:
355 /* Nothing. */ {}
356 | ID { muscle_code_grow ("union_name", $1, @1); }
357 ;
358
359 grammar_declaration:
360 "%union" union_name braceless
361 {
362 union_seen = true;
363 muscle_code_grow ("stype", $3, @3);
364 code_scanner_last_string_free ();
365 }
366 ;
367
368
369
370
371 symbol_declaration:
372 "%nterm" { current_class = nterm_sym; } symbol_defs.1
373 {
374 current_class = unknown_sym;
375 current_type = NULL;
376 }
377 | "%token" { current_class = token_sym; } symbol_defs.1
378 {
379 current_class = unknown_sym;
380 current_type = NULL;
381 }
382 | "%type" TAG symbols.1
383 {
384 symbol_list *list;
385 tag_seen = true;
386 for (list = $3; list; list = list->next)
387 symbol_type_set (list->content.sym, $2, @2);
388 symbol_list_free ($3);
389 }
390 ;
391
392 precedence_declaration:
393 precedence_declarator tag.opt symbols.prec
394 {
395 symbol_list *list;
396 ++current_prec;
397 for (list = $3; list; list = list->next)
398 {
399 symbol_type_set (list->content.sym, current_type, @2);
400 symbol_precedence_set (list->content.sym, current_prec, $1, @1);
401 }
402 symbol_list_free ($3);
403 current_type = NULL;
404 }
405 ;
406
407 precedence_declarator:
408 "%left" { $$ = left_assoc; }
409 | "%right" { $$ = right_assoc; }
410 | "%nonassoc" { $$ = non_assoc; }
411 | "%precedence" { $$ = precedence_assoc; }
412 ;
413
414 tag.opt:
415 /* Nothing. */ { current_type = NULL; }
416 | TAG { current_type = $1; tag_seen = true; }
417 ;
418
419 /* Just like symbols.1 but accept INT for the sake of POSIX. */
420 symbols.prec:
421 symbol.prec
422 { $$ = symbol_list_sym_new ($1, @1); }
423 | symbols.prec symbol.prec
424 { $$ = symbol_list_prepend ($1, symbol_list_sym_new ($2, @2)); }
425 ;
426
427 symbol.prec:
428 symbol { $$ = $1; }
429 | symbol INT { $$ = $1; symbol_user_token_number_set ($1, $2, @2); }
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 { $$ = symbol_list_sym_new ($1, @1); }
447 | TAG { $$ = 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 TAG
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 named_ref.opt { current_lhs = $1; current_lhs_location = @1;
514 current_lhs_named_ref = $2; } rhses.1
515 ;
516
517 rhses.1:
518 rhs { grammar_current_rule_end (@1); }
519 | rhses.1 "|" rhs { grammar_current_rule_end (@3); }
520 | rhses.1 ";"
521 ;
522
523 rhs:
524 /* Nothing. */
525 { grammar_current_rule_begin (current_lhs, current_lhs_location,
526 current_lhs_named_ref); }
527 | rhs symbol named_ref.opt
528 { grammar_current_rule_symbol_append ($2, @2, $3); }
529 | rhs "{...}" named_ref.opt
530 { grammar_current_rule_action_append ($2, @2, $3); }
531 | rhs "%prec" symbol
532 { grammar_current_rule_prec_set ($3, @3); }
533 | rhs "%dprec" INT
534 { grammar_current_rule_dprec_set ($3, @3); }
535 | rhs "%merge" TAG
536 { grammar_current_rule_merge_set ($3, @3); }
537 ;
538
539 named_ref.opt:
540 /* Nothing. */
541 { $$ = 0; }
542 |
543 BRACKETED_ID
544 { $$ = named_ref_new($1, @1); }
545 ;
546
547
548 /*---------------------------.
549 | variable and content.opt. |
550 `---------------------------*/
551
552 /* The STRING form of variable is deprecated and is not M4-friendly.
553 For example, M4 fails for `%define "[" "value"'. */
554 variable:
555 ID
556 | STRING { $$ = uniqstr_new ($1); }
557 ;
558
559 /* Some content or empty by default. */
560 content.opt:
561 /* Nothing. */ { $$ = ""; }
562 | STRING
563 ;
564
565
566 /*------------.
567 | braceless. |
568 `------------*/
569
570 braceless:
571 "{...}"
572 {
573 code_props plain_code;
574 $1[strlen ($1) - 1] = '\n';
575 code_props_plain_init (&plain_code, $1+1, @1);
576 code_props_translate_code (&plain_code);
577 gram_scanner_last_string_free ();
578 $$ = plain_code.code;
579 }
580 ;
581
582
583 /*--------------.
584 | Identifiers. |
585 `--------------*/
586
587 /* Identifiers are returned as uniqstr values by the scanner.
588 Depending on their use, we may need to make them genuine symbols. */
589
590 id:
591 ID
592 { $$ = symbol_from_uniqstr ($1, @1); }
593 | CHAR
594 {
595 $$ = symbol_get (char_name ($1), @1);
596 symbol_class_set ($$, token_sym, @1, false);
597 symbol_user_token_number_set ($$, $1, @1);
598 }
599 ;
600
601 id_colon:
602 ID_COLON { $$ = symbol_from_uniqstr ($1, @1); }
603 ;
604
605
606 symbol:
607 id
608 | string_as_id
609 ;
610
611 /* A string used as an ID: quote it. */
612 string_as_id:
613 STRING
614 {
615 $$ = symbol_get (quotearg_style (c_quoting_style, $1), @1);
616 symbol_class_set ($$, token_sym, @1, false);
617 }
618 ;
619
620 epilogue.opt:
621 /* Nothing. */
622 | "%%" EPILOGUE
623 {
624 code_props plain_code;
625 code_props_plain_init (&plain_code, $2, @2);
626 code_props_translate_code (&plain_code);
627 gram_scanner_last_string_free ();
628 muscle_code_grow ("epilogue", plain_code.code, @2);
629 code_scanner_last_string_free ();
630 }
631 ;
632
633 %%
634
635
636 /* Return the location of the left-hand side of a rule whose
637 right-hand side is RHS[1] ... RHS[N]. Ignore empty nonterminals in
638 the right-hand side, and return an empty location equal to the end
639 boundary of RHS[0] if the right-hand side is empty. */
640
641 static YYLTYPE
642 lloc_default (YYLTYPE const *rhs, int n)
643 {
644 int i;
645 YYLTYPE loc;
646
647 /* SGI MIPSpro 7.4.1m miscompiles "loc.start = loc.end = rhs[n].end;".
648 The bug is fixed in 7.4.2m, but play it safe for now. */
649 loc.start = rhs[n].end;
650 loc.end = rhs[n].end;
651
652 /* Ignore empty nonterminals the start of the the right-hand side.
653 Do not bother to ignore them at the end of the right-hand side,
654 since empty nonterminals have the same end as their predecessors. */
655 for (i = 1; i <= n; i++)
656 if (! equal_boundaries (rhs[i].start, rhs[i].end))
657 {
658 loc.start = rhs[i].start;
659 break;
660 }
661
662 return loc;
663 }
664
665
666 /* Add a lex-param or a parse-param (depending on TYPE) with
667 declaration DECL and location LOC. */
668
669 static void
670 add_param (char const *type, char *decl, location loc)
671 {
672 static char const alphanum[26 + 26 + 1 + 10] =
673 "abcdefghijklmnopqrstuvwxyz"
674 "ABCDEFGHIJKLMNOPQRSTUVWXYZ"
675 "_"
676 "0123456789";
677 char const *name_start = NULL;
678 char *p;
679
680 /* Stop on last actual character. */
681 for (p = decl; p[1]; p++)
682 if ((p == decl
683 || ! memchr (alphanum, p[-1], sizeof alphanum))
684 && memchr (alphanum, p[0], sizeof alphanum - 10))
685 name_start = p;
686
687 /* Strip the surrounding '{' and '}', and any blanks just inside
688 the braces. */
689 while (*--p == ' ' || *p == '\t')
690 continue;
691 p[1] = '\0';
692 while (*++decl == ' ' || *decl == '\t')
693 continue;
694
695 if (! name_start)
696 complain_at (loc, _("missing identifier in parameter declaration"));
697 else
698 {
699 char *name;
700 size_t name_len;
701
702 for (name_len = 1;
703 memchr (alphanum, name_start[name_len], sizeof alphanum);
704 name_len++)
705 continue;
706
707 name = xmalloc (name_len + 1);
708 memcpy (name, name_start, name_len);
709 name[name_len] = '\0';
710 muscle_pair_list_grow (type, decl, name);
711 free (name);
712 }
713
714 gram_scanner_last_string_free ();
715 }
716
717
718 static void
719 version_check (location const *loc, char const *version)
720 {
721 if (strverscmp (version, PACKAGE_VERSION) > 0)
722 {
723 complain_at (*loc, "require bison %s, but have %s",
724 version, PACKAGE_VERSION);
725 exit (63);
726 }
727 }
728
729 static void
730 gram_error (location const *loc, char const *msg)
731 {
732 complain_at (*loc, "%s", msg);
733 }
734
735 char const *
736 token_name (int type)
737 {
738 return yytname[YYTRANSLATE (type)];
739 }
740
741 static char const *
742 char_name (char c)
743 {
744 if (c == '\'')
745 return "'\\''";
746 else
747 {
748 char buf[4];
749 buf[0] = '\''; buf[1] = c; buf[2] = '\''; buf[3] = '\0';
750 return quotearg_style (escape_quoting_style, buf);
751 }
752 }