]> git.saurik.com Git - bison.git/blob - src/parse-gram.y
53df3a430be2bf4012846cb85d55e2965f19d0c0
[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_DEBUG "%debug"
132 PERCENT_DEFAULT_PREC "%default-prec"
133 PERCENT_DEFINE "%define"
134 PERCENT_DEFINES "%defines"
135 PERCENT_ERROR_VERBOSE "%error-verbose"
136 PERCENT_EXPECT "%expect"
137 PERCENT_EXPECT_RR "%expect-rr"
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}", $$); }
184 braceless content.opt "{...}" "%{...%}" EPILOGUE
185
186 %type <uniqstr> TAG ID ID_COLON variable
187 %printer { fprintf (stderr, "<%s>", $$); } TAG
188 %printer { fputs ($$, stderr); } ID variable
189 %printer { fprintf (stderr, "%s:", $$); } ID_COLON
190
191 %type <integer> INT
192 %printer { fprintf (stderr, "%d", $$); } INT
193
194 %type <symbol> id id_colon symbol symbol.prec string_as_id
195 %printer { fprintf (stderr, "%s", $$->tag); } id symbol string_as_id
196 %printer { fprintf (stderr, "%s:", $$->tag); } id_colon
197
198 %type <assoc> precedence_declarator
199 %type <list> symbols.1 symbols.prec generic_symlist generic_symlist_item
200 %%
201
202 input:
203 prologue_declarations "%%" grammar epilogue.opt
204 ;
205
206
207 /*------------------------------------.
208 | Declarations: before the first %%. |
209 `------------------------------------*/
210
211 prologue_declarations:
212 /* Nothing */
213 | prologue_declarations prologue_declaration
214 ;
215
216 prologue_declaration:
217 grammar_declaration
218 | "%{...%}"
219 {
220 code_props plain_code;
221 code_props_plain_init (&plain_code, $1, @1);
222 code_props_translate_code (&plain_code);
223 gram_scanner_last_string_free ();
224 muscle_code_grow (union_seen ? "post_prologue" : "pre_prologue",
225 plain_code.code, @1);
226 code_scanner_last_string_free ();
227 }
228 | "%debug"
229 {
230 muscle_percent_define_insert ("debug", @$, "");
231 }
232 | "%define" variable content.opt
233 {
234 muscle_percent_define_insert ($2, @2, $3);
235 }
236 | "%defines" { defines_flag = true; }
237 | "%defines" STRING
238 {
239 defines_flag = true;
240 spec_defines_file = xstrdup ($2);
241 }
242 | "%error-verbose"
243 {
244 muscle_percent_define_insert ("error_verbose", @$, "");
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 | "%locations" { locations_flag = true; }
267 | "%name-prefix" STRING { spec_name_prefix = $2; }
268 | "%name-prefix" "=" STRING { spec_name_prefix = $3; } /* deprecated */
269 | "%no-lines" { no_lines_flag = true; }
270 | "%nondeterministic-parser" { nondeterministic_parser = true; }
271 | "%output" STRING { spec_outfile = $2; }
272 | "%output" "=" STRING { spec_outfile = $3; } /* deprecated */
273 | "%parse-param" "{...}" { add_param ("parse_param", $2, @2); }
274 | "%pure-parser"
275 {
276 /* %pure-parser is deprecated in favor of `%define api.pure', so use
277 `%define api.pure' in a backward-compatible manner here. First, don't
278 complain if %pure-parser is specified multiple times. */
279 if (!muscle_find_const ("percent_define(api.pure)"))
280 muscle_percent_define_insert ("api.pure", @1, "");
281 /* In all cases, use api.pure now so that the backend doesn't complain if
282 the skeleton ignores api.pure, but do warn now if there's a previous
283 conflicting definition from an actual %define. */
284 if (!muscle_percent_define_flag_if ("api.pure"))
285 muscle_percent_define_insert ("api.pure", @1, "");
286 }
287 | "%require" STRING { version_check (&@2, $2); }
288 | "%skeleton" STRING
289 {
290 char const *skeleton_user = $2;
291 if (strchr (skeleton_user, '/'))
292 {
293 size_t dir_length = strlen (current_file);
294 char *skeleton_build;
295 while (dir_length && current_file[dir_length - 1] != '/')
296 --dir_length;
297 while (dir_length && current_file[dir_length - 1] == '/')
298 --dir_length;
299 skeleton_build =
300 xmalloc (dir_length + 1 + strlen (skeleton_user) + 1);
301 if (dir_length > 0)
302 {
303 strncpy (skeleton_build, current_file, dir_length);
304 skeleton_build[dir_length++] = '/';
305 }
306 strcpy (skeleton_build + dir_length, skeleton_user);
307 skeleton_user = uniqstr_new (skeleton_build);
308 free (skeleton_build);
309 }
310 skeleton_arg (skeleton_user, grammar_prio, @1);
311 }
312 | "%token-table" { token_table_flag = true; }
313 | "%verbose" { report_flag |= report_states; }
314 | "%yacc" { yacc_flag = true; }
315 | /*FIXME: Err? What is this horror doing here? */ ";"
316 ;
317
318 grammar_declaration:
319 precedence_declaration
320 | symbol_declaration
321 | "%start" symbol
322 {
323 grammar_start_symbol_set ($2, @2);
324 }
325 | "%destructor" "{...}" generic_symlist
326 {
327 symbol_list *list;
328 for (list = $3; list; list = list->next)
329 symbol_list_destructor_set (list, $2, @2);
330 symbol_list_free ($3);
331 }
332 | "%printer" "{...}" generic_symlist
333 {
334 symbol_list *list;
335 for (list = $3; list; list = list->next)
336 symbol_list_printer_set (list, $2, @2);
337 symbol_list_free ($3);
338 }
339 | "%default-prec"
340 {
341 default_prec = true;
342 }
343 | "%no-default-prec"
344 {
345 default_prec = false;
346 }
347 | "%code" braceless
348 {
349 /* Do not invoke muscle_percent_code_grow here since it invokes
350 muscle_user_name_list_grow. */
351 muscle_code_grow ("percent_code()", $2, @2);
352 code_scanner_last_string_free ();
353 }
354 | "%code" ID braceless
355 {
356 muscle_percent_code_grow ($2, @2, $3, @3);
357 code_scanner_last_string_free ();
358 }
359 ;
360
361
362 /*---------.
363 | %union. |
364 `---------*/
365
366 %token PERCENT_UNION "%union";
367
368 union_name:
369 /* Nothing. */ {}
370 | ID { muscle_code_grow ("union_name", $1, @1); }
371 ;
372
373 grammar_declaration:
374 "%union" union_name braceless
375 {
376 union_seen = true;
377 muscle_code_grow ("stype", $3, @3);
378 code_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" TAG 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 tag.opt symbols.prec
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 | "%precedence" { $$ = precedence_assoc; }
426 ;
427
428 tag.opt:
429 /* Nothing. */ { current_type = NULL; }
430 | TAG { current_type = $1; tag_seen = true; }
431 ;
432
433 /* Just like symbols.1 but accept INT for the sake of POSIX. */
434 symbols.prec:
435 symbol.prec
436 { $$ = symbol_list_sym_new ($1, @1); }
437 | symbols.prec symbol.prec
438 { $$ = symbol_list_prepend ($1, symbol_list_sym_new ($2, @2)); }
439 ;
440
441 symbol.prec:
442 symbol { $$ = $1; }
443 | symbol INT { $$ = $1; symbol_user_token_number_set ($1, $2, @2); }
444 ;
445
446 /* One or more symbols to be %typed. */
447 symbols.1:
448 symbol
449 { $$ = symbol_list_sym_new ($1, @1); }
450 | symbols.1 symbol
451 { $$ = symbol_list_prepend ($1, symbol_list_sym_new ($2, @2)); }
452 ;
453
454 generic_symlist:
455 generic_symlist_item { $$ = $1; }
456 | generic_symlist generic_symlist_item { $$ = symbol_list_prepend ($1, $2); }
457 ;
458
459 generic_symlist_item:
460 symbol { $$ = symbol_list_sym_new ($1, @1); }
461 | TAG { $$ = symbol_list_type_new ($1, @1); }
462 | "<*>" { $$ = symbol_list_default_tagged_new (@1); }
463 | "<>" { $$ = symbol_list_default_tagless_new (@1); }
464 ;
465
466 /* One token definition. */
467 symbol_def:
468 TAG
469 {
470 current_type = $1;
471 tag_seen = true;
472 }
473 | id
474 {
475 symbol_class_set ($1, current_class, @1, true);
476 symbol_type_set ($1, current_type, @1);
477 }
478 | id INT
479 {
480 symbol_class_set ($1, current_class, @1, true);
481 symbol_type_set ($1, current_type, @1);
482 symbol_user_token_number_set ($1, $2, @2);
483 }
484 | id string_as_id
485 {
486 symbol_class_set ($1, current_class, @1, true);
487 symbol_type_set ($1, current_type, @1);
488 symbol_make_alias ($1, $2, @$);
489 }
490 | id INT string_as_id
491 {
492 symbol_class_set ($1, current_class, @1, true);
493 symbol_type_set ($1, current_type, @1);
494 symbol_user_token_number_set ($1, $2, @2);
495 symbol_make_alias ($1, $3, @$);
496 }
497 ;
498
499 /* One or more symbol definitions. */
500 symbol_defs.1:
501 symbol_def
502 | symbol_defs.1 symbol_def
503 ;
504
505
506 /*------------------------------------------.
507 | The grammar section: between the two %%. |
508 `------------------------------------------*/
509
510 grammar:
511 rules_or_grammar_declaration
512 | grammar rules_or_grammar_declaration
513 ;
514
515 /* As a Bison extension, one can use the grammar declarations in the
516 body of the grammar. */
517 rules_or_grammar_declaration:
518 rules
519 | grammar_declaration ";"
520 | error ";"
521 {
522 yyerrok;
523 }
524 ;
525
526 rules:
527 id_colon { current_lhs = $1; current_lhs_location = @1; } rhses.1
528 ;
529
530 rhses.1:
531 rhs { grammar_current_rule_end (@1); }
532 | rhses.1 "|" rhs { grammar_current_rule_end (@3); }
533 | rhses.1 ";"
534 ;
535
536 rhs:
537 /* Nothing. */
538 { grammar_current_rule_begin (current_lhs, current_lhs_location); }
539 | rhs symbol
540 { grammar_current_rule_symbol_append ($2, @2); }
541 | rhs "{...}"
542 { grammar_current_rule_action_append ($2, @2); }
543 | rhs "%prec" symbol
544 { grammar_current_rule_prec_set ($3, @3); }
545 | rhs "%dprec" INT
546 { grammar_current_rule_dprec_set ($3, @3); }
547 | rhs "%merge" TAG
548 { grammar_current_rule_merge_set ($3, @3); }
549 ;
550
551
552 /*---------------------------.
553 | variable and content.opt. |
554 `---------------------------*/
555
556 variable:
557 ID
558 | STRING { $$ = uniqstr_new ($1); } /* deprecated and not M4-friendly */
559 ;
560
561 /* Some content or empty by default. */
562 content.opt:
563 /* Nothing. */
564 {
565 $$ = "";
566 }
567 | STRING
568 ;
569
570
571 /*------------.
572 | braceless. |
573 `------------*/
574
575 braceless:
576 "{...}"
577 {
578 code_props plain_code;
579 $1[strlen ($1) - 1] = '\n';
580 code_props_plain_init (&plain_code, $1+1, @1);
581 code_props_translate_code (&plain_code);
582 gram_scanner_last_string_free ();
583 $$ = plain_code.code;
584 }
585 ;
586
587
588 /*--------------.
589 | Identifiers. |
590 `--------------*/
591
592 /* Identifiers are returned as uniqstr values by the scanner.
593 Depending on their use, we may need to make them genuine symbols. */
594
595 id:
596 ID
597 { $$ = symbol_from_uniqstr ($1, @1); }
598 | CHAR
599 {
600 $$ = symbol_get (char_name ($1), @1);
601 symbol_class_set ($$, token_sym, @1, false);
602 symbol_user_token_number_set ($$, $1, @1);
603 }
604 ;
605
606 id_colon:
607 ID_COLON { $$ = symbol_from_uniqstr ($1, @1); }
608 ;
609
610
611 symbol:
612 id
613 | string_as_id
614 ;
615
616 /* A string used as an ID: quote it. */
617 string_as_id:
618 STRING
619 {
620 $$ = symbol_get (quotearg_style (c_quoting_style, $1), @1);
621 symbol_class_set ($$, token_sym, @1, false);
622 }
623 ;
624
625 epilogue.opt:
626 /* Nothing. */
627 | "%%" EPILOGUE
628 {
629 code_props plain_code;
630 code_props_plain_init (&plain_code, $2, @2);
631 code_props_translate_code (&plain_code);
632 gram_scanner_last_string_free ();
633 muscle_code_grow ("epilogue", plain_code.code, @2);
634 code_scanner_last_string_free ();
635 }
636 ;
637
638 %%
639
640
641 /* Return the location of the left-hand side of a rule whose
642 right-hand side is RHS[1] ... RHS[N]. Ignore empty nonterminals in
643 the right-hand side, and return an empty location equal to the end
644 boundary of RHS[0] if the right-hand side is empty. */
645
646 static YYLTYPE
647 lloc_default (YYLTYPE const *rhs, int n)
648 {
649 int i;
650 YYLTYPE loc;
651
652 /* SGI MIPSpro 7.4.1m miscompiles "loc.start = loc.end = rhs[n].end;".
653 The bug is fixed in 7.4.2m, but play it safe for now. */
654 loc.start = rhs[n].end;
655 loc.end = rhs[n].end;
656
657 /* Ignore empty nonterminals the start of the the right-hand side.
658 Do not bother to ignore them at the end of the right-hand side,
659 since empty nonterminals have the same end as their predecessors. */
660 for (i = 1; i <= n; i++)
661 if (! equal_boundaries (rhs[i].start, rhs[i].end))
662 {
663 loc.start = rhs[i].start;
664 break;
665 }
666
667 return loc;
668 }
669
670
671 /* Add a lex-param or a parse-param (depending on TYPE) with
672 declaration DECL and location LOC. */
673
674 static void
675 add_param (char const *type, char *decl, location loc)
676 {
677 static char const alphanum[26 + 26 + 1 + 10] =
678 "abcdefghijklmnopqrstuvwxyz"
679 "ABCDEFGHIJKLMNOPQRSTUVWXYZ"
680 "_"
681 "0123456789";
682 char const *name_start = NULL;
683 char *p;
684
685 /* Stop on last actual character. */
686 for (p = decl; p[1]; p++)
687 if ((p == decl
688 || ! memchr (alphanum, p[-1], sizeof alphanum))
689 && memchr (alphanum, p[0], sizeof alphanum - 10))
690 name_start = p;
691
692 /* Strip the surrounding '{' and '}', and any blanks just inside
693 the braces. */
694 while (*--p == ' ' || *p == '\t')
695 continue;
696 p[1] = '\0';
697 while (*++decl == ' ' || *decl == '\t')
698 continue;
699
700 if (! name_start)
701 complain_at (loc, _("missing identifier in parameter declaration"));
702 else
703 {
704 char *name;
705 size_t name_len;
706
707 for (name_len = 1;
708 memchr (alphanum, name_start[name_len], sizeof alphanum);
709 name_len++)
710 continue;
711
712 name = xmalloc (name_len + 1);
713 memcpy (name, name_start, name_len);
714 name[name_len] = '\0';
715 muscle_pair_list_grow (type, decl, name);
716 free (name);
717 }
718
719 gram_scanner_last_string_free ();
720 }
721
722
723 static void
724 version_check (location const *loc, char const *version)
725 {
726 if (strverscmp (version, PACKAGE_VERSION) > 0)
727 {
728 complain_at (*loc, "require bison %s, but have %s",
729 version, PACKAGE_VERSION);
730 exit (63);
731 }
732 }
733
734 static void
735 gram_error (location const *loc, char const *msg)
736 {
737 complain_at (*loc, "%s", msg);
738 }
739
740 char const *
741 token_name (int type)
742 {
743 return yytname[YYTRANSLATE (type)];
744 }
745
746 static char const *
747 char_name (char c)
748 {
749 if (c == '\'')
750 return "'\\''";
751 else
752 {
753 char buf[4];
754 buf[0] = '\''; buf[1] = c; buf[2] = '\''; buf[3] = '\0';
755 return quotearg_style (escape_quoting_style, buf);
756 }
757 }