]> git.saurik.com Git - bison.git/blob - src/parse-gram.y
Deprecate %pure-parser and add `%define api.pure'. Discussed starting
[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 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 <strverscmp.h>
25
26 #include "complain.h"
27 #include "conflicts.h"
28 #include "files.h"
29 #include "getargs.h"
30 #include "gram.h"
31 #include "muscle_tab.h"
32 #include "quotearg.h"
33 #include "reader.h"
34 #include "symlist.h"
35 #include "scan-gram.h"
36 #include "scan-code.h"
37
38 #define YYLLOC_DEFAULT(Current, Rhs, N) (Current) = lloc_default (Rhs, N)
39 static YYLTYPE lloc_default (YYLTYPE const *, int);
40
41 #define YY_LOCATION_PRINT(File, Loc) \
42 location_print (File, Loc)
43
44 static void version_check (location const *loc, char const *version);
45
46 /* Request detailed syntax error messages, and pass them to GRAM_ERROR.
47 FIXME: depends on the undocumented availability of YYLLOC. */
48 #undef yyerror
49 #define yyerror(Msg) \
50 gram_error (&yylloc, Msg)
51 static void gram_error (location const *, char const *);
52
53 static char const *char_name (char);
54
55 /** Add a lex-param or a parse-param.
56 *
57 * \param type \a lex_param or \a parse_param
58 * \param decl the formal argument
59 * \param loc the location in the source.
60 */
61 static void add_param (char const *type, char *decl, location loc);
62
63
64 static symbol_class current_class = unknown_sym;
65 static uniqstr current_type = NULL;
66 static symbol *current_lhs;
67 static location current_lhs_location;
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 };
104
105 /* Define the tokens together with their human representation. */
106 %token GRAM_EOF 0 "end of file"
107 %token STRING "string"
108 %token INT "integer"
109
110 %token PERCENT_TOKEN "%token"
111 %token PERCENT_NTERM "%nterm"
112
113 %token PERCENT_TYPE "%type"
114 %token PERCENT_DESTRUCTOR "%destructor"
115 %token PERCENT_PRINTER "%printer"
116
117 %token PERCENT_LEFT "%left"
118 %token PERCENT_RIGHT "%right"
119 %token PERCENT_NONASSOC "%nonassoc"
120
121 %token PERCENT_PREC "%prec"
122 %token PERCENT_DPREC "%dprec"
123 %token PERCENT_MERGE "%merge"
124
125
126 /*----------------------.
127 | Global Declarations. |
128 `----------------------*/
129
130 %token
131 PERCENT_CODE "%code"
132 PERCENT_DEBUG "%debug"
133 PERCENT_DEFAULT_PREC "%default-prec"
134 PERCENT_DEFINE "%define"
135 PERCENT_DEFINES "%defines"
136 PERCENT_ERROR_VERBOSE "%error-verbose"
137 PERCENT_EXPECT "%expect"
138 PERCENT_EXPECT_RR "%expect-rr"
139 PERCENT_FILE_PREFIX "%file-prefix"
140 PERCENT_GLR_PARSER "%glr-parser"
141 PERCENT_INITIAL_ACTION "%initial-action"
142 PERCENT_LANGUAGE "%language"
143 PERCENT_LEX_PARAM "%lex-param"
144 PERCENT_LOCATIONS "%locations"
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_PURE_PARSER "%pure-parser"
153 PERCENT_REQUIRE "%require"
154 PERCENT_SKELETON "%skeleton"
155 PERCENT_START "%start"
156 PERCENT_TOKEN_TABLE "%token-table"
157 PERCENT_VERBOSE "%verbose"
158 PERCENT_YACC "%yacc"
159 ;
160
161 %token BRACED_CODE "{...}"
162 %token CHAR "char"
163 %token EPILOGUE "epilogue"
164 %token EQUAL "="
165 %token ID "identifier"
166 %token ID_COLON "identifier:"
167 %token PERCENT_PERCENT "%%"
168 %token PIPE "|"
169 %token PROLOGUE "%{...%}"
170 %token SEMICOLON ";"
171 %token TYPE "type"
172 %token TYPE_TAG_ANY "<*>"
173 %token TYPE_TAG_NONE "<>"
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}", $$); }
185 braceless content.opt "{...}" "%{...%}" EPILOGUE
186
187 %type <uniqstr> TYPE ID ID_COLON variable
188 %printer { fprintf (stderr, "<%s>", $$); } TYPE
189 %printer { fputs ($$, stderr); } ID variable
190 %printer { fprintf (stderr, "%s:", $$); } ID_COLON
191
192 %type <integer> INT
193 %printer { fprintf (stderr, "%d", $$); } INT
194
195 %type <symbol> id id_colon symbol string_as_id
196 %printer { fprintf (stderr, "%s", $$->tag); } id symbol string_as_id
197 %printer { fprintf (stderr, "%s:", $$->tag); } id_colon
198
199 %type <assoc> precedence_declarator
200 %type <list> symbols.1 generic_symlist generic_symlist_item
201 %%
202
203 input:
204 prologue_declarations "%%" grammar epilogue.opt
205 ;
206
207
208 /*------------------------------------.
209 | Declarations: before the first %%. |
210 `------------------------------------*/
211
212 prologue_declarations:
213 /* Nothing */
214 | prologue_declarations prologue_declaration
215 ;
216
217 prologue_declaration:
218 grammar_declaration
219 | "%{...%}"
220 {
221 code_props plain_code;
222 code_props_plain_init (&plain_code, $1, @1);
223 code_props_translate_code (&plain_code);
224 gram_scanner_last_string_free ();
225 muscle_code_grow (union_seen ? "post_prologue" : "pre_prologue",
226 plain_code.code, @1);
227 code_scanner_last_string_free ();
228 }
229 | "%debug" { debug_flag = true; }
230 | "%define" variable content.opt
231 {
232 muscle_percent_define_insert ($2, @2, $3);
233 }
234 | "%defines" { defines_flag = true; }
235 | "%defines" STRING
236 {
237 defines_flag = true;
238 spec_defines_file = xstrdup ($2);
239 }
240 | "%error-verbose" { error_verbose = true; }
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, 1, &@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 | "%pure-parser"
270 {
271 /* %pure-parser is deprecated in favor of `%define api.pure', so use
272 `%define api.pure' in a backward-compatible manner here. First, don't
273 complain if %pure-parser is specified multiple times. */
274 if (!muscle_find_const ("percent_define(api.pure)"))
275 muscle_percent_define_insert ("api.pure", @1, "");
276 /* In all cases, use api.pure now so that the backend doesn't complain if
277 the skeleton ignores api.pure, but do warn now if there's a previous
278 conflicting definition from an actual %define. */
279 if (!muscle_percent_define_flag_if ("api.pure"))
280 muscle_percent_define_insert ("api.pure", @1, "");
281 }
282 | "%require" STRING { version_check (&@2, $2); }
283 | "%skeleton" STRING
284 {
285 char const *skeleton_user = $2;
286 if (strchr (skeleton_user, '/'))
287 {
288 size_t dir_length = strlen (current_file);
289 char *skeleton_build;
290 while (dir_length && current_file[dir_length - 1] != '/')
291 --dir_length;
292 while (dir_length && current_file[dir_length - 1] == '/')
293 --dir_length;
294 skeleton_build =
295 xmalloc (dir_length + 1 + strlen (skeleton_user) + 1);
296 if (dir_length > 0)
297 {
298 strncpy (skeleton_build, current_file, dir_length);
299 skeleton_build[dir_length++] = '/';
300 }
301 strcpy (skeleton_build + dir_length, skeleton_user);
302 skeleton_user = uniqstr_new (skeleton_build);
303 free (skeleton_build);
304 }
305 skeleton_arg (skeleton_user, 1, &@1);
306 }
307 | "%token-table" { token_table_flag = true; }
308 | "%verbose" { report_flag = report_states; }
309 | "%yacc" { yacc_flag = true; }
310 | /*FIXME: Err? What is this horror doing here? */ ";"
311 ;
312
313 grammar_declaration:
314 precedence_declaration
315 | symbol_declaration
316 | "%start" symbol
317 {
318 grammar_start_symbol_set ($2, @2);
319 }
320 | "%destructor" "{...}" generic_symlist
321 {
322 symbol_list *list;
323 for (list = $3; list; list = list->next)
324 symbol_list_destructor_set (list, $2, @2);
325 symbol_list_free ($3);
326 }
327 | "%printer" "{...}" generic_symlist
328 {
329 symbol_list *list;
330 for (list = $3; list; list = list->next)
331 symbol_list_printer_set (list, $2, @2);
332 symbol_list_free ($3);
333 }
334 | "%default-prec"
335 {
336 default_prec = true;
337 }
338 | "%no-default-prec"
339 {
340 default_prec = false;
341 }
342 | "%code" braceless
343 {
344 /* Do not invoke muscle_percent_code_grow here since it invokes
345 muscle_user_name_list_grow. */
346 muscle_code_grow ("percent_code()", $2, @2);
347 code_scanner_last_string_free ();
348 }
349 | "%code" ID braceless
350 {
351 muscle_percent_code_grow ($2, @2, $3, @3);
352 code_scanner_last_string_free ();
353 }
354 ;
355
356
357 /*----------*
358 | %union. |
359 *----------*/
360
361 %token PERCENT_UNION "%union";
362
363 union_name:
364 /* Nothing. */ {}
365 | ID { muscle_code_grow ("union_name", $1, @1); }
366 ;
367
368 grammar_declaration:
369 "%union" union_name braceless
370 {
371 union_seen = true;
372 muscle_code_grow ("stype", $3, @3);
373 code_scanner_last_string_free ();
374 }
375 ;
376
377
378
379
380 symbol_declaration:
381 "%nterm" { current_class = nterm_sym; } symbol_defs.1
382 {
383 current_class = unknown_sym;
384 current_type = NULL;
385 }
386 | "%token" { current_class = token_sym; } symbol_defs.1
387 {
388 current_class = unknown_sym;
389 current_type = NULL;
390 }
391 | "%type" TYPE symbols.1
392 {
393 symbol_list *list;
394 tag_seen = true;
395 for (list = $3; list; list = list->next)
396 symbol_type_set (list->content.sym, $2, @2);
397 symbol_list_free ($3);
398 }
399 ;
400
401 precedence_declaration:
402 precedence_declarator type.opt symbols.1
403 {
404 symbol_list *list;
405 ++current_prec;
406 for (list = $3; list; list = list->next)
407 {
408 symbol_type_set (list->content.sym, current_type, @2);
409 symbol_precedence_set (list->content.sym, current_prec, $1, @1);
410 }
411 symbol_list_free ($3);
412 current_type = NULL;
413 }
414 ;
415
416 precedence_declarator:
417 "%left" { $$ = left_assoc; }
418 | "%right" { $$ = right_assoc; }
419 | "%nonassoc" { $$ = non_assoc; }
420 ;
421
422 type.opt:
423 /* Nothing. */ { current_type = NULL; }
424 | TYPE { current_type = $1; tag_seen = true; }
425 ;
426
427 /* One or more symbols to be %typed. */
428 symbols.1:
429 symbol
430 { $$ = symbol_list_sym_new ($1, @1); }
431 | symbols.1 symbol
432 { $$ = symbol_list_prepend ($1, symbol_list_sym_new ($2, @2)); }
433 ;
434
435 generic_symlist:
436 generic_symlist_item { $$ = $1; }
437 | generic_symlist generic_symlist_item { $$ = symbol_list_prepend ($1, $2); }
438 ;
439
440 generic_symlist_item:
441 symbol { $$ = symbol_list_sym_new ($1, @1); }
442 | TYPE { $$ = symbol_list_type_new ($1, @1); }
443 | "<*>" { $$ = symbol_list_default_tagged_new (@1); }
444 | "<>" { $$ = symbol_list_default_tagless_new (@1); }
445 ;
446
447 /* One token definition. */
448 symbol_def:
449 TYPE
450 {
451 current_type = $1;
452 tag_seen = true;
453 }
454 | id
455 {
456 symbol_class_set ($1, current_class, @1, true);
457 symbol_type_set ($1, current_type, @1);
458 }
459 | id INT
460 {
461 symbol_class_set ($1, current_class, @1, true);
462 symbol_type_set ($1, current_type, @1);
463 symbol_user_token_number_set ($1, $2, @2);
464 }
465 | id string_as_id
466 {
467 symbol_class_set ($1, current_class, @1, true);
468 symbol_type_set ($1, current_type, @1);
469 symbol_make_alias ($1, $2, @$);
470 }
471 | id INT string_as_id
472 {
473 symbol_class_set ($1, current_class, @1, true);
474 symbol_type_set ($1, current_type, @1);
475 symbol_user_token_number_set ($1, $2, @2);
476 symbol_make_alias ($1, $3, @$);
477 }
478 ;
479
480 /* One or more symbol definitions. */
481 symbol_defs.1:
482 symbol_def
483 | symbol_defs.1 symbol_def
484 ;
485
486
487 /*------------------------------------------.
488 | The grammar section: between the two %%. |
489 `------------------------------------------*/
490
491 grammar:
492 rules_or_grammar_declaration
493 | grammar rules_or_grammar_declaration
494 ;
495
496 /* As a Bison extension, one can use the grammar declarations in the
497 body of the grammar. */
498 rules_or_grammar_declaration:
499 rules
500 | grammar_declaration ";"
501 | error ";"
502 {
503 yyerrok;
504 }
505 ;
506
507 rules:
508 id_colon { current_lhs = $1; current_lhs_location = @1; } rhses.1
509 ;
510
511 rhses.1:
512 rhs { grammar_current_rule_end (@1); }
513 | rhses.1 "|" rhs { grammar_current_rule_end (@3); }
514 | rhses.1 ";"
515 ;
516
517 rhs:
518 /* Nothing. */
519 { grammar_current_rule_begin (current_lhs, current_lhs_location); }
520 | rhs symbol
521 { grammar_current_rule_symbol_append ($2, @2); }
522 | rhs "{...}"
523 { grammar_current_rule_action_append ($2, @2); }
524 | rhs "%prec" symbol
525 { grammar_current_rule_prec_set ($3, @3); }
526 | rhs "%dprec" INT
527 { grammar_current_rule_dprec_set ($3, @3); }
528 | rhs "%merge" TYPE
529 { grammar_current_rule_merge_set ($3, @3); }
530 ;
531
532
533 /*----------------------------*
534 | variable and content.opt. |
535 *---------------------------*/
536
537 variable:
538 ID
539 | STRING { $$ = uniqstr_new ($1); } /* deprecated and not M4-friendly */
540 ;
541
542 /* Some content or empty by default. */
543 content.opt:
544 /* Nothing. */
545 {
546 $$ = "";
547 }
548 | STRING
549 ;
550
551
552 /*-------------*
553 | braceless. |
554 *-------------*/
555
556 braceless:
557 "{...}"
558 {
559 code_props plain_code;
560 $1[strlen ($1) - 1] = '\n';
561 code_props_plain_init (&plain_code, $1+1, @1);
562 code_props_translate_code (&plain_code);
563 gram_scanner_last_string_free ();
564 $$ = plain_code.code;
565 }
566 ;
567
568
569 /*---------------*
570 | Identifiers. |
571 *---------------*/
572
573 /* Identifiers are returned as uniqstr values by the scanner.
574 Depending on their use, we may need to make them genuine symbols. */
575
576 id:
577 ID
578 { $$ = symbol_from_uniqstr ($1, @1); }
579 | CHAR
580 {
581 $$ = symbol_get (char_name ($1), @1);
582 symbol_class_set ($$, token_sym, @1, false);
583 symbol_user_token_number_set ($$, $1, @1);
584 }
585 ;
586
587 id_colon:
588 ID_COLON { $$ = symbol_from_uniqstr ($1, @1); }
589 ;
590
591
592 symbol:
593 id
594 | string_as_id
595 ;
596
597 /* A string used as an ID: quote it. */
598 string_as_id:
599 STRING
600 {
601 $$ = symbol_get (quotearg_style (c_quoting_style, $1), @1);
602 symbol_class_set ($$, token_sym, @1, false);
603 }
604 ;
605
606 epilogue.opt:
607 /* Nothing. */
608 | "%%" EPILOGUE
609 {
610 code_props plain_code;
611 code_props_plain_init (&plain_code, $2, @2);
612 code_props_translate_code (&plain_code);
613 gram_scanner_last_string_free ();
614 muscle_code_grow ("epilogue", plain_code.code, @2);
615 code_scanner_last_string_free ();
616 }
617 ;
618
619 %%
620
621
622 /* Return the location of the left-hand side of a rule whose
623 right-hand side is RHS[1] ... RHS[N]. Ignore empty nonterminals in
624 the right-hand side, and return an empty location equal to the end
625 boundary of RHS[0] if the right-hand side is empty. */
626
627 static YYLTYPE
628 lloc_default (YYLTYPE const *rhs, int n)
629 {
630 int i;
631 YYLTYPE loc;
632
633 /* SGI MIPSpro 7.4.1m miscompiles "loc.start = loc.end = rhs[n].end;".
634 The bug is fixed in 7.4.2m, but play it safe for now. */
635 loc.start = rhs[n].end;
636 loc.end = rhs[n].end;
637
638 /* Ignore empty nonterminals the start of the the right-hand side.
639 Do not bother to ignore them at the end of the right-hand side,
640 since empty nonterminals have the same end as their predecessors. */
641 for (i = 1; i <= n; i++)
642 if (! equal_boundaries (rhs[i].start, rhs[i].end))
643 {
644 loc.start = rhs[i].start;
645 break;
646 }
647
648 return loc;
649 }
650
651
652 /* Add a lex-param or a parse-param (depending on TYPE) with
653 declaration DECL and location LOC. */
654
655 static void
656 add_param (char const *type, char *decl, location loc)
657 {
658 static char const alphanum[26 + 26 + 1 + 10] =
659 "abcdefghijklmnopqrstuvwxyz"
660 "ABCDEFGHIJKLMNOPQRSTUVWXYZ"
661 "_"
662 "0123456789";
663 char const *name_start = NULL;
664 char *p;
665
666 /* Stop on last actual character. */
667 for (p = decl; p[1]; p++)
668 if ((p == decl
669 || ! memchr (alphanum, p[-1], sizeof alphanum))
670 && memchr (alphanum, p[0], sizeof alphanum - 10))
671 name_start = p;
672
673 /* Strip the surrounding '{' and '}', and any blanks just inside
674 the braces. */
675 while (*--p == ' ' || *p == '\t')
676 continue;
677 p[1] = '\0';
678 while (*++decl == ' ' || *decl == '\t')
679 continue;
680
681 if (! name_start)
682 complain_at (loc, _("missing identifier in parameter declaration"));
683 else
684 {
685 char *name;
686 size_t name_len;
687
688 for (name_len = 1;
689 memchr (alphanum, name_start[name_len], sizeof alphanum);
690 name_len++)
691 continue;
692
693 name = xmalloc (name_len + 1);
694 memcpy (name, name_start, name_len);
695 name[name_len] = '\0';
696 muscle_pair_list_grow (type, decl, name);
697 free (name);
698 }
699
700 gram_scanner_last_string_free ();
701 }
702
703
704 static void
705 version_check (location const *loc, char const *version)
706 {
707 if (strverscmp (version, PACKAGE_VERSION) > 0)
708 {
709 complain_at (*loc, "require bison %s, but have %s",
710 version, PACKAGE_VERSION);
711 exit (63);
712 }
713 }
714
715 static void
716 gram_error (location const *loc, char const *msg)
717 {
718 complain_at (*loc, "%s", msg);
719 }
720
721 char const *
722 token_name (int type)
723 {
724 return yytname[YYTRANSLATE (type)];
725 }
726
727 static char const *
728 char_name (char c)
729 {
730 if (c == '\'')
731 return "'\\''";
732 else
733 {
734 char buf[4];
735 buf[0] = '\''; buf[1] = c; buf[2] = '\''; buf[3] = '\0';
736 return quotearg_style (escape_quoting_style, buf);
737 }
738 }