]> git.saurik.com Git - bison.git/blob - src/parse-gram.y
Work around Java's ``code too large'' problem for parser tables.
[bison.git] / src / parse-gram.y
1 %{/* Bison Grammar Parser -*- C -*-
2
3 Copyright (C) 2002, 2003, 2004, 2005, 2006, 2007, 2008 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 TYPE "type"
171 %token TYPE_TAG_ANY "<*>"
172 %token TYPE_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}", $$); }
184 braceless content.opt "{...}" "%{...%}" EPILOGUE
185
186 %type <uniqstr> TYPE ID ID_COLON variable
187 %printer { fprintf (stderr, "<%s>", $$); } TYPE
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" { debug_flag = true; }
229 | "%define" variable content.opt
230 {
231 muscle_percent_define_insert ($2, @2, $3);
232 }
233 | "%defines" { defines_flag = true; }
234 | "%defines" STRING
235 {
236 defines_flag = true;
237 spec_defines_file = xstrdup ($2);
238 }
239 | "%error-verbose" { error_verbose = true; }
240 | "%expect" INT { expected_sr_conflicts = $2; }
241 | "%expect-rr" INT { expected_rr_conflicts = $2; }
242 | "%file-prefix" STRING { spec_file_prefix = $2; }
243 | "%file-prefix" "=" STRING { spec_file_prefix = $3; } /* deprecated */
244 | "%glr-parser"
245 {
246 nondeterministic_parser = true;
247 glr_parser = true;
248 }
249 | "%initial-action" "{...}"
250 {
251 code_props action;
252 code_props_symbol_action_init (&action, $2, @2);
253 code_props_translate_code (&action);
254 gram_scanner_last_string_free ();
255 muscle_code_grow ("initial_action", action.code, @2);
256 code_scanner_last_string_free ();
257 }
258 | "%language" STRING { language_argmatch ($2, grammar_prio, @1); }
259 | "%lex-param" "{...}" { add_param ("lex_param", $2, @2); }
260 | "%locations" { locations_flag = true; }
261 | "%name-prefix" STRING { spec_name_prefix = $2; }
262 | "%name-prefix" "=" STRING { spec_name_prefix = $3; } /* deprecated */
263 | "%no-lines" { no_lines_flag = true; }
264 | "%nondeterministic-parser" { nondeterministic_parser = true; }
265 | "%output" STRING { spec_outfile = $2; }
266 | "%output" "=" STRING { spec_outfile = $3; } /* deprecated */
267 | "%parse-param" "{...}" { add_param ("parse_param", $2, @2); }
268 | "%pure-parser"
269 {
270 /* %pure-parser is deprecated in favor of `%define api.pure', so use
271 `%define api.pure' in a backward-compatible manner here. First, don't
272 complain if %pure-parser is specified multiple times. */
273 if (!muscle_find_const ("percent_define(api.pure)"))
274 muscle_percent_define_insert ("api.pure", @1, "");
275 /* In all cases, use api.pure now so that the backend doesn't complain if
276 the skeleton ignores api.pure, but do warn now if there's a previous
277 conflicting definition from an actual %define. */
278 if (!muscle_percent_define_flag_if ("api.pure"))
279 muscle_percent_define_insert ("api.pure", @1, "");
280 }
281 | "%require" STRING { version_check (&@2, $2); }
282 | "%skeleton" STRING
283 {
284 char const *skeleton_user = $2;
285 if (strchr (skeleton_user, '/'))
286 {
287 size_t dir_length = strlen (current_file);
288 char *skeleton_build;
289 while (dir_length && current_file[dir_length - 1] != '/')
290 --dir_length;
291 while (dir_length && current_file[dir_length - 1] == '/')
292 --dir_length;
293 skeleton_build =
294 xmalloc (dir_length + 1 + strlen (skeleton_user) + 1);
295 if (dir_length > 0)
296 {
297 strncpy (skeleton_build, current_file, dir_length);
298 skeleton_build[dir_length++] = '/';
299 }
300 strcpy (skeleton_build + dir_length, skeleton_user);
301 skeleton_user = uniqstr_new (skeleton_build);
302 free (skeleton_build);
303 }
304 skeleton_arg (skeleton_user, grammar_prio, @1);
305 }
306 | "%token-table" { token_table_flag = true; }
307 | "%verbose" { report_flag |= report_states; }
308 | "%yacc" { yacc_flag = true; }
309 | /*FIXME: Err? What is this horror doing here? */ ";"
310 ;
311
312 grammar_declaration:
313 precedence_declaration
314 | symbol_declaration
315 | "%start" symbol
316 {
317 grammar_start_symbol_set ($2, @2);
318 }
319 | "%destructor" "{...}" generic_symlist
320 {
321 symbol_list *list;
322 for (list = $3; list; list = list->next)
323 symbol_list_destructor_set (list, $2, @2);
324 symbol_list_free ($3);
325 }
326 | "%printer" "{...}" generic_symlist
327 {
328 symbol_list *list;
329 for (list = $3; list; list = list->next)
330 symbol_list_printer_set (list, $2, @2);
331 symbol_list_free ($3);
332 }
333 | "%default-prec"
334 {
335 default_prec = true;
336 }
337 | "%no-default-prec"
338 {
339 default_prec = false;
340 }
341 | "%code" braceless
342 {
343 /* Do not invoke muscle_percent_code_grow here since it invokes
344 muscle_user_name_list_grow. */
345 muscle_code_grow ("percent_code()", $2, @2);
346 code_scanner_last_string_free ();
347 }
348 | "%code" ID braceless
349 {
350 muscle_percent_code_grow ($2, @2, $3, @3);
351 code_scanner_last_string_free ();
352 }
353 ;
354
355
356 /*---------.
357 | %union. |
358 `---------*/
359
360 %token PERCENT_UNION "%union";
361
362 union_name:
363 /* Nothing. */ {}
364 | ID { muscle_code_grow ("union_name", $1, @1); }
365 ;
366
367 grammar_declaration:
368 "%union" union_name braceless
369 {
370 union_seen = true;
371 muscle_code_grow ("stype", $3, @3);
372 code_scanner_last_string_free ();
373 }
374 ;
375
376
377
378
379 symbol_declaration:
380 "%nterm" { current_class = nterm_sym; } symbol_defs.1
381 {
382 current_class = unknown_sym;
383 current_type = NULL;
384 }
385 | "%token" { current_class = token_sym; } symbol_defs.1
386 {
387 current_class = unknown_sym;
388 current_type = NULL;
389 }
390 | "%type" TYPE symbols.1
391 {
392 symbol_list *list;
393 tag_seen = true;
394 for (list = $3; list; list = list->next)
395 symbol_type_set (list->content.sym, $2, @2);
396 symbol_list_free ($3);
397 }
398 ;
399
400 precedence_declaration:
401 precedence_declarator type.opt symbols.prec
402 {
403 symbol_list *list;
404 ++current_prec;
405 for (list = $3; list; list = list->next)
406 {
407 symbol_type_set (list->content.sym, current_type, @2);
408 symbol_precedence_set (list->content.sym, current_prec, $1, @1);
409 }
410 symbol_list_free ($3);
411 current_type = NULL;
412 }
413 ;
414
415 precedence_declarator:
416 "%left" { $$ = left_assoc; }
417 | "%right" { $$ = right_assoc; }
418 | "%nonassoc" { $$ = non_assoc; }
419 | "%precedence" { $$ = precedence_assoc; }
420 ;
421
422 type.opt:
423 /* Nothing. */ { current_type = NULL; }
424 | TYPE { current_type = $1; tag_seen = true; }
425 ;
426
427 /* Just like symbols.1 but accept INT for the sake of POSIX. */
428 symbols.prec:
429 symbol.prec
430 { $$ = symbol_list_sym_new ($1, @1); }
431 | symbols.prec symbol.prec
432 { $$ = symbol_list_prepend ($1, symbol_list_sym_new ($2, @2)); }
433 ;
434
435 symbol.prec:
436 symbol { $$ = $1; }
437 | symbol INT { $$ = $1; symbol_user_token_number_set ($1, $2, @2); }
438 ;
439
440 /* One or more symbols to be %typed. */
441 symbols.1:
442 symbol
443 { $$ = symbol_list_sym_new ($1, @1); }
444 | symbols.1 symbol
445 { $$ = symbol_list_prepend ($1, symbol_list_sym_new ($2, @2)); }
446 ;
447
448 generic_symlist:
449 generic_symlist_item { $$ = $1; }
450 | generic_symlist generic_symlist_item { $$ = symbol_list_prepend ($1, $2); }
451 ;
452
453 generic_symlist_item:
454 symbol { $$ = symbol_list_sym_new ($1, @1); }
455 | TYPE { $$ = symbol_list_type_new ($1, @1); }
456 | "<*>" { $$ = symbol_list_default_tagged_new (@1); }
457 | "<>" { $$ = symbol_list_default_tagless_new (@1); }
458 ;
459
460 /* One token definition. */
461 symbol_def:
462 TYPE
463 {
464 current_type = $1;
465 tag_seen = true;
466 }
467 | id
468 {
469 symbol_class_set ($1, current_class, @1, true);
470 symbol_type_set ($1, current_type, @1);
471 }
472 | id INT
473 {
474 symbol_class_set ($1, current_class, @1, true);
475 symbol_type_set ($1, current_type, @1);
476 symbol_user_token_number_set ($1, $2, @2);
477 }
478 | id string_as_id
479 {
480 symbol_class_set ($1, current_class, @1, true);
481 symbol_type_set ($1, current_type, @1);
482 symbol_make_alias ($1, $2, @$);
483 }
484 | id INT string_as_id
485 {
486 symbol_class_set ($1, current_class, @1, true);
487 symbol_type_set ($1, current_type, @1);
488 symbol_user_token_number_set ($1, $2, @2);
489 symbol_make_alias ($1, $3, @$);
490 }
491 ;
492
493 /* One or more symbol definitions. */
494 symbol_defs.1:
495 symbol_def
496 | symbol_defs.1 symbol_def
497 ;
498
499
500 /*------------------------------------------.
501 | The grammar section: between the two %%. |
502 `------------------------------------------*/
503
504 grammar:
505 rules_or_grammar_declaration
506 | grammar rules_or_grammar_declaration
507 ;
508
509 /* As a Bison extension, one can use the grammar declarations in the
510 body of the grammar. */
511 rules_or_grammar_declaration:
512 rules
513 | grammar_declaration ";"
514 | error ";"
515 {
516 yyerrok;
517 }
518 ;
519
520 rules:
521 id_colon { current_lhs = $1; current_lhs_location = @1; } rhses.1
522 ;
523
524 rhses.1:
525 rhs { grammar_current_rule_end (@1); }
526 | rhses.1 "|" rhs { grammar_current_rule_end (@3); }
527 | rhses.1 ";"
528 ;
529
530 rhs:
531 /* Nothing. */
532 { grammar_current_rule_begin (current_lhs, current_lhs_location); }
533 | rhs symbol
534 { grammar_current_rule_symbol_append ($2, @2); }
535 | rhs "{...}"
536 { grammar_current_rule_action_append ($2, @2); }
537 | rhs "%prec" symbol
538 { grammar_current_rule_prec_set ($3, @3); }
539 | rhs "%dprec" INT
540 { grammar_current_rule_dprec_set ($3, @3); }
541 | rhs "%merge" TYPE
542 { grammar_current_rule_merge_set ($3, @3); }
543 ;
544
545
546 /*---------------------------.
547 | variable and content.opt. |
548 `---------------------------*/
549
550 variable:
551 ID
552 | STRING { $$ = uniqstr_new ($1); } /* deprecated and not M4-friendly */
553 ;
554
555 /* Some content or empty by default. */
556 content.opt:
557 /* Nothing. */
558 {
559 $$ = "";
560 }
561 | STRING
562 ;
563
564
565 /*------------.
566 | braceless. |
567 `------------*/
568
569 braceless:
570 "{...}"
571 {
572 code_props plain_code;
573 $1[strlen ($1) - 1] = '\n';
574 code_props_plain_init (&plain_code, $1+1, @1);
575 code_props_translate_code (&plain_code);
576 gram_scanner_last_string_free ();
577 $$ = plain_code.code;
578 }
579 ;
580
581
582 /*--------------.
583 | Identifiers. |
584 `--------------*/
585
586 /* Identifiers are returned as uniqstr values by the scanner.
587 Depending on their use, we may need to make them genuine symbols. */
588
589 id:
590 ID
591 { $$ = symbol_from_uniqstr ($1, @1); }
592 | CHAR
593 {
594 $$ = symbol_get (char_name ($1), @1);
595 symbol_class_set ($$, token_sym, @1, false);
596 symbol_user_token_number_set ($$, $1, @1);
597 }
598 ;
599
600 id_colon:
601 ID_COLON { $$ = symbol_from_uniqstr ($1, @1); }
602 ;
603
604
605 symbol:
606 id
607 | string_as_id
608 ;
609
610 /* A string used as an ID: quote it. */
611 string_as_id:
612 STRING
613 {
614 $$ = symbol_get (quotearg_style (c_quoting_style, $1), @1);
615 symbol_class_set ($$, token_sym, @1, false);
616 }
617 ;
618
619 epilogue.opt:
620 /* Nothing. */
621 | "%%" EPILOGUE
622 {
623 code_props plain_code;
624 code_props_plain_init (&plain_code, $2, @2);
625 code_props_translate_code (&plain_code);
626 gram_scanner_last_string_free ();
627 muscle_code_grow ("epilogue", plain_code.code, @2);
628 code_scanner_last_string_free ();
629 }
630 ;
631
632 %%
633
634
635 /* Return the location of the left-hand side of a rule whose
636 right-hand side is RHS[1] ... RHS[N]. Ignore empty nonterminals in
637 the right-hand side, and return an empty location equal to the end
638 boundary of RHS[0] if the right-hand side is empty. */
639
640 static YYLTYPE
641 lloc_default (YYLTYPE const *rhs, int n)
642 {
643 int i;
644 YYLTYPE loc;
645
646 /* SGI MIPSpro 7.4.1m miscompiles "loc.start = loc.end = rhs[n].end;".
647 The bug is fixed in 7.4.2m, but play it safe for now. */
648 loc.start = rhs[n].end;
649 loc.end = rhs[n].end;
650
651 /* Ignore empty nonterminals the start of the the right-hand side.
652 Do not bother to ignore them at the end of the right-hand side,
653 since empty nonterminals have the same end as their predecessors. */
654 for (i = 1; i <= n; i++)
655 if (! equal_boundaries (rhs[i].start, rhs[i].end))
656 {
657 loc.start = rhs[i].start;
658 break;
659 }
660
661 return loc;
662 }
663
664
665 /* Add a lex-param or a parse-param (depending on TYPE) with
666 declaration DECL and location LOC. */
667
668 static void
669 add_param (char const *type, char *decl, location loc)
670 {
671 static char const alphanum[26 + 26 + 1 + 10] =
672 "abcdefghijklmnopqrstuvwxyz"
673 "ABCDEFGHIJKLMNOPQRSTUVWXYZ"
674 "_"
675 "0123456789";
676 char const *name_start = NULL;
677 char *p;
678
679 /* Stop on last actual character. */
680 for (p = decl; p[1]; p++)
681 if ((p == decl
682 || ! memchr (alphanum, p[-1], sizeof alphanum))
683 && memchr (alphanum, p[0], sizeof alphanum - 10))
684 name_start = p;
685
686 /* Strip the surrounding '{' and '}', and any blanks just inside
687 the braces. */
688 while (*--p == ' ' || *p == '\t')
689 continue;
690 p[1] = '\0';
691 while (*++decl == ' ' || *decl == '\t')
692 continue;
693
694 if (! name_start)
695 complain_at (loc, _("missing identifier in parameter declaration"));
696 else
697 {
698 char *name;
699 size_t name_len;
700
701 for (name_len = 1;
702 memchr (alphanum, name_start[name_len], sizeof alphanum);
703 name_len++)
704 continue;
705
706 name = xmalloc (name_len + 1);
707 memcpy (name, name_start, name_len);
708 name[name_len] = '\0';
709 muscle_pair_list_grow (type, decl, name);
710 free (name);
711 }
712
713 gram_scanner_last_string_free ();
714 }
715
716
717 static void
718 version_check (location const *loc, char const *version)
719 {
720 if (strverscmp (version, PACKAGE_VERSION) > 0)
721 {
722 complain_at (*loc, "require bison %s, but have %s",
723 version, PACKAGE_VERSION);
724 exit (63);
725 }
726 }
727
728 static void
729 gram_error (location const *loc, char const *msg)
730 {
731 complain_at (*loc, "%s", msg);
732 }
733
734 char const *
735 token_name (int type)
736 {
737 return yytname[YYTRANSLATE (type)];
738 }
739
740 static char const *
741 char_name (char c)
742 {
743 if (c == '\'')
744 return "'\\''";
745 else
746 {
747 char buf[4];
748 buf[0] = '\''; buf[1] = c; buf[2] = '\''; buf[3] = '\0';
749 return quotearg_style (escape_quoting_style, buf);
750 }
751 }