]> git.saurik.com Git - bison.git/blob - src/parse-gram.y
Regen.
[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_DEFAULT_PREC "%default-prec"
132 PERCENT_DEFINE "%define"
133 PERCENT_DEFINES "%defines"
134 PERCENT_EXPECT "%expect"
135 PERCENT_EXPECT_RR "%expect-rr"
136 PERCENT_FLAG "%<flag>"
137 PERCENT_FILE_PREFIX "%file-prefix"
138 PERCENT_GLR_PARSER "%glr-parser"
139 PERCENT_INITIAL_ACTION "%initial-action"
140 PERCENT_LANGUAGE "%language"
141 PERCENT_LEX_PARAM "%lex-param"
142 PERCENT_LOCATIONS "%locations"
143 PERCENT_NAME_PREFIX "%name-prefix"
144 PERCENT_NO_DEFAULT_PREC "%no-default-prec"
145 PERCENT_NO_LINES "%no-lines"
146 PERCENT_NONDETERMINISTIC_PARSER
147 "%nondeterministic-parser"
148 PERCENT_OUTPUT "%output"
149 PERCENT_PARSE_PARAM "%parse-param"
150 PERCENT_REQUIRE "%require"
151 PERCENT_SKELETON "%skeleton"
152 PERCENT_START "%start"
153 PERCENT_TOKEN_TABLE "%token-table"
154 PERCENT_VERBOSE "%verbose"
155 PERCENT_YACC "%yacc"
156 ;
157
158 %token BRACED_CODE "{...}"
159 %token CHAR "char"
160 %token EPILOGUE "epilogue"
161 %token EQUAL "="
162 %token ID "identifier"
163 %token ID_COLON "identifier:"
164 %token PERCENT_PERCENT "%%"
165 %token PIPE "|"
166 %token PROLOGUE "%{...%}"
167 %token SEMICOLON ";"
168 %token TAG "<tag>"
169 %token TAG_ANY "<*>"
170 %token TAG_NONE "<>"
171
172 %type <character> CHAR
173 %printer { fputs (char_name ($$), stderr); } CHAR
174
175 /* braceless is not to be used for rule or symbol actions, as it
176 calls code_props_plain_init. */
177 %type <chars> STRING "%{...%}" EPILOGUE braceless content.opt
178 %type <code> "{...}"
179 %printer { fputs (quotearg_style (c_quoting_style, $$), stderr); }
180 STRING
181 %printer { fprintf (stderr, "{\n%s\n}", $$); }
182 braceless content.opt "{...}" "%{...%}" EPILOGUE
183
184 %type <uniqstr> TAG ID ID_COLON PERCENT_FLAG variable
185 %printer { fputs ($$, stderr); } ID variable
186 %printer { fprintf (stderr, "%s:", $$); } ID_COLON
187 %printer { fprintf (stderr, "%%%s", $$); } PERCENT_FLAG
188 %printer { fprintf (stderr, "<%s>", $$); } TAG
189
190 %type <integer> INT
191 %printer { fprintf (stderr, "%d", $$); } INT
192
193 %type <symbol> id id_colon symbol symbol.prec string_as_id
194 %printer { fprintf (stderr, "%s", $$->tag); } id symbol string_as_id
195 %printer { fprintf (stderr, "%s:", $$->tag); } id_colon
196
197 %type <assoc> precedence_declarator
198 %type <list> symbols.1 symbols.prec generic_symlist generic_symlist_item
199 %%
200
201 input:
202 prologue_declarations "%%" grammar epilogue.opt
203 ;
204
205
206 /*------------------------------------.
207 | Declarations: before the first %%. |
208 `------------------------------------*/
209
210 prologue_declarations:
211 /* Nothing */
212 | prologue_declarations prologue_declaration
213 ;
214
215 prologue_declaration:
216 grammar_declaration
217 | "%{...%}"
218 {
219 code_props plain_code;
220 code_props_plain_init (&plain_code, $1, @1);
221 code_props_translate_code (&plain_code);
222 gram_scanner_last_string_free ();
223 muscle_code_grow (union_seen ? "post_prologue" : "pre_prologue",
224 plain_code.code, @1);
225 code_scanner_last_string_free ();
226 }
227 | "%<flag>"
228 {
229 muscle_percent_define_ensure ($1, @1, true);
230 }
231 | "%define" variable content.opt
232 {
233 muscle_percent_define_insert ($2, @2, $3);
234 }
235 | "%defines" { defines_flag = true; }
236 | "%defines" STRING
237 {
238 defines_flag = true;
239 spec_defines_file = xstrdup ($2);
240 }
241 | "%expect" INT { expected_sr_conflicts = $2; }
242 | "%expect-rr" INT { expected_rr_conflicts = $2; }
243 | "%file-prefix" STRING { spec_file_prefix = $2; }
244 | "%file-prefix" "=" STRING { spec_file_prefix = $3; } /* deprecated */
245 | "%glr-parser"
246 {
247 nondeterministic_parser = true;
248 glr_parser = true;
249 }
250 | "%initial-action" "{...}"
251 {
252 code_props action;
253 code_props_symbol_action_init (&action, $2, @2);
254 code_props_translate_code (&action);
255 gram_scanner_last_string_free ();
256 muscle_code_grow ("initial_action", action.code, @2);
257 code_scanner_last_string_free ();
258 }
259 | "%language" STRING { language_argmatch ($2, grammar_prio, @1); }
260 | "%lex-param" "{...}" { add_param ("lex_param", $2, @2); }
261 | "%locations" { locations_flag = true; }
262 | "%name-prefix" STRING { spec_name_prefix = $2; }
263 | "%name-prefix" "=" STRING { spec_name_prefix = $3; } /* deprecated */
264 | "%no-lines" { no_lines_flag = true; }
265 | "%nondeterministic-parser" { nondeterministic_parser = true; }
266 | "%output" STRING { spec_outfile = $2; }
267 | "%output" "=" STRING { spec_outfile = $3; } /* deprecated */
268 | "%parse-param" "{...}" { add_param ("parse_param", $2, @2); }
269 | "%require" STRING { version_check (&@2, $2); }
270 | "%skeleton" STRING
271 {
272 char const *skeleton_user = $2;
273 if (strchr (skeleton_user, '/'))
274 {
275 size_t dir_length = strlen (current_file);
276 char *skeleton_build;
277 while (dir_length && current_file[dir_length - 1] != '/')
278 --dir_length;
279 while (dir_length && current_file[dir_length - 1] == '/')
280 --dir_length;
281 skeleton_build =
282 xmalloc (dir_length + 1 + strlen (skeleton_user) + 1);
283 if (dir_length > 0)
284 {
285 strncpy (skeleton_build, current_file, dir_length);
286 skeleton_build[dir_length++] = '/';
287 }
288 strcpy (skeleton_build + dir_length, skeleton_user);
289 skeleton_user = uniqstr_new (skeleton_build);
290 free (skeleton_build);
291 }
292 skeleton_arg (skeleton_user, grammar_prio, @1);
293 }
294 | "%token-table" { token_table_flag = true; }
295 | "%verbose" { report_flag |= report_states; }
296 | "%yacc" { yacc_flag = true; }
297 | /*FIXME: Err? What is this horror doing here? */ ";"
298 ;
299
300 grammar_declaration:
301 precedence_declaration
302 | symbol_declaration
303 | "%start" symbol
304 {
305 grammar_start_symbol_set ($2, @2);
306 }
307 | "%destructor" "{...}" generic_symlist
308 {
309 symbol_list *list;
310 for (list = $3; list; list = list->next)
311 symbol_list_destructor_set (list, $2, @2);
312 symbol_list_free ($3);
313 }
314 | "%printer" "{...}" generic_symlist
315 {
316 symbol_list *list;
317 for (list = $3; list; list = list->next)
318 symbol_list_printer_set (list, $2, @2);
319 symbol_list_free ($3);
320 }
321 | "%default-prec"
322 {
323 default_prec = true;
324 }
325 | "%no-default-prec"
326 {
327 default_prec = false;
328 }
329 | "%code" braceless
330 {
331 /* Do not invoke muscle_percent_code_grow here since it invokes
332 muscle_user_name_list_grow. */
333 muscle_code_grow ("percent_code()", $2, @2);
334 code_scanner_last_string_free ();
335 }
336 | "%code" ID braceless
337 {
338 muscle_percent_code_grow ($2, @2, $3, @3);
339 code_scanner_last_string_free ();
340 }
341 ;
342
343
344 /*---------.
345 | %union. |
346 `---------*/
347
348 %token PERCENT_UNION "%union";
349
350 union_name:
351 /* Nothing. */ {}
352 | ID { muscle_code_grow ("union_name", $1, @1); }
353 ;
354
355 grammar_declaration:
356 "%union" union_name braceless
357 {
358 union_seen = true;
359 muscle_code_grow ("stype", $3, @3);
360 code_scanner_last_string_free ();
361 }
362 ;
363
364
365
366
367 symbol_declaration:
368 "%nterm" { current_class = nterm_sym; } symbol_defs.1
369 {
370 current_class = unknown_sym;
371 current_type = NULL;
372 }
373 | "%token" { current_class = token_sym; } symbol_defs.1
374 {
375 current_class = unknown_sym;
376 current_type = NULL;
377 }
378 | "%type" TAG symbols.1
379 {
380 symbol_list *list;
381 tag_seen = true;
382 for (list = $3; list; list = list->next)
383 symbol_type_set (list->content.sym, $2, @2);
384 symbol_list_free ($3);
385 }
386 ;
387
388 precedence_declaration:
389 precedence_declarator tag.opt symbols.prec
390 {
391 symbol_list *list;
392 ++current_prec;
393 for (list = $3; list; list = list->next)
394 {
395 symbol_type_set (list->content.sym, current_type, @2);
396 symbol_precedence_set (list->content.sym, current_prec, $1, @1);
397 }
398 symbol_list_free ($3);
399 current_type = NULL;
400 }
401 ;
402
403 precedence_declarator:
404 "%left" { $$ = left_assoc; }
405 | "%right" { $$ = right_assoc; }
406 | "%nonassoc" { $$ = non_assoc; }
407 | "%precedence" { $$ = precedence_assoc; }
408 ;
409
410 tag.opt:
411 /* Nothing. */ { current_type = NULL; }
412 | TAG { current_type = $1; tag_seen = true; }
413 ;
414
415 /* Just like symbols.1 but accept INT for the sake of POSIX. */
416 symbols.prec:
417 symbol.prec
418 { $$ = symbol_list_sym_new ($1, @1); }
419 | symbols.prec symbol.prec
420 { $$ = symbol_list_prepend ($1, symbol_list_sym_new ($2, @2)); }
421 ;
422
423 symbol.prec:
424 symbol { $$ = $1; }
425 | symbol INT { $$ = $1; symbol_user_token_number_set ($1, $2, @2); }
426 ;
427
428 /* One or more symbols to be %typed. */
429 symbols.1:
430 symbol
431 { $$ = symbol_list_sym_new ($1, @1); }
432 | symbols.1 symbol
433 { $$ = symbol_list_prepend ($1, symbol_list_sym_new ($2, @2)); }
434 ;
435
436 generic_symlist:
437 generic_symlist_item { $$ = $1; }
438 | generic_symlist generic_symlist_item { $$ = symbol_list_prepend ($1, $2); }
439 ;
440
441 generic_symlist_item:
442 symbol { $$ = symbol_list_sym_new ($1, @1); }
443 | TAG { $$ = symbol_list_type_new ($1, @1); }
444 | "<*>" { $$ = symbol_list_default_tagged_new (@1); }
445 | "<>" { $$ = symbol_list_default_tagless_new (@1); }
446 ;
447
448 /* One token definition. */
449 symbol_def:
450 TAG
451 {
452 current_type = $1;
453 tag_seen = true;
454 }
455 | id
456 {
457 symbol_class_set ($1, current_class, @1, true);
458 symbol_type_set ($1, current_type, @1);
459 }
460 | id INT
461 {
462 symbol_class_set ($1, current_class, @1, true);
463 symbol_type_set ($1, current_type, @1);
464 symbol_user_token_number_set ($1, $2, @2);
465 }
466 | id string_as_id
467 {
468 symbol_class_set ($1, current_class, @1, true);
469 symbol_type_set ($1, current_type, @1);
470 symbol_make_alias ($1, $2, @$);
471 }
472 | id INT string_as_id
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 symbol_make_alias ($1, $3, @$);
478 }
479 ;
480
481 /* One or more symbol definitions. */
482 symbol_defs.1:
483 symbol_def
484 | symbol_defs.1 symbol_def
485 ;
486
487
488 /*------------------------------------------.
489 | The grammar section: between the two %%. |
490 `------------------------------------------*/
491
492 grammar:
493 rules_or_grammar_declaration
494 | grammar rules_or_grammar_declaration
495 ;
496
497 /* As a Bison extension, one can use the grammar declarations in the
498 body of the grammar. */
499 rules_or_grammar_declaration:
500 rules
501 | grammar_declaration ";"
502 | error ";"
503 {
504 yyerrok;
505 }
506 ;
507
508 rules:
509 id_colon { current_lhs = $1; current_lhs_location = @1; } rhses.1
510 ;
511
512 rhses.1:
513 rhs { grammar_current_rule_end (@1); }
514 | rhses.1 "|" rhs { grammar_current_rule_end (@3); }
515 | rhses.1 ";"
516 ;
517
518 rhs:
519 /* Nothing. */
520 { grammar_current_rule_begin (current_lhs, current_lhs_location); }
521 | rhs symbol
522 { grammar_current_rule_symbol_append ($2, @2); }
523 | rhs "{...}"
524 { grammar_current_rule_action_append ($2, @2); }
525 | rhs "%prec" symbol
526 { grammar_current_rule_prec_set ($3, @3); }
527 | rhs "%dprec" INT
528 { grammar_current_rule_dprec_set ($3, @3); }
529 | rhs "%merge" TAG
530 { grammar_current_rule_merge_set ($3, @3); }
531 ;
532
533
534 /*---------------------------.
535 | variable and content.opt. |
536 `---------------------------*/
537
538 variable:
539 ID
540 | STRING { $$ = uniqstr_new ($1); } /* deprecated and not M4-friendly */
541 ;
542
543 /* Some content or empty by default. */
544 content.opt:
545 /* Nothing. */
546 {
547 $$ = "";
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 }