]> git.saurik.com Git - bison.git/blob - src/parse-gram.y
2007-10-17 Wojciech Polak <polak@gnu.org>
[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" { pure_parser = true; }
270 | "%require" STRING { version_check (&@2, $2); }
271 | "%skeleton" STRING
272 {
273 char const *skeleton_user = $2;
274 if (strchr (skeleton_user, '/'))
275 {
276 size_t dir_length = strlen (current_file);
277 char *skeleton_build;
278 while (dir_length && current_file[dir_length - 1] != '/')
279 --dir_length;
280 while (dir_length && current_file[dir_length - 1] == '/')
281 --dir_length;
282 skeleton_build =
283 xmalloc (dir_length + 1 + strlen (skeleton_user) + 1);
284 if (dir_length > 0)
285 {
286 strncpy (skeleton_build, current_file, dir_length);
287 skeleton_build[dir_length++] = '/';
288 }
289 strcpy (skeleton_build + dir_length, skeleton_user);
290 skeleton_user = uniqstr_new (skeleton_build);
291 free (skeleton_build);
292 }
293 skeleton_arg (skeleton_user, 1, &@1);
294 }
295 | "%token-table" { token_table_flag = true; }
296 | "%verbose" { report_flag = report_states; }
297 | "%yacc" { yacc_flag = true; }
298 | /*FIXME: Err? What is this horror doing here? */ ";"
299 ;
300
301 grammar_declaration:
302 precedence_declaration
303 | symbol_declaration
304 | "%start" symbol
305 {
306 grammar_start_symbol_set ($2, @2);
307 }
308 | "%destructor" "{...}" generic_symlist
309 {
310 symbol_list *list;
311 for (list = $3; list; list = list->next)
312 symbol_list_destructor_set (list, $2, @2);
313 symbol_list_free ($3);
314 }
315 | "%printer" "{...}" generic_symlist
316 {
317 symbol_list *list;
318 for (list = $3; list; list = list->next)
319 symbol_list_printer_set (list, $2, @2);
320 symbol_list_free ($3);
321 }
322 | "%default-prec"
323 {
324 default_prec = true;
325 }
326 | "%no-default-prec"
327 {
328 default_prec = false;
329 }
330 | "%code" braceless
331 {
332 /* Do not invoke muscle_percent_code_grow here since it invokes
333 muscle_user_name_list_grow. */
334 muscle_code_grow ("percent_code()", $2, @2);
335 code_scanner_last_string_free ();
336 }
337 | "%code" ID braceless
338 {
339 muscle_percent_code_grow ($2, @2, $3, @3);
340 code_scanner_last_string_free ();
341 }
342 ;
343
344
345 /*----------*
346 | %union. |
347 *----------*/
348
349 %token PERCENT_UNION "%union";
350
351 union_name:
352 /* Nothing. */ {}
353 | ID { muscle_code_grow ("union_name", $1, @1); }
354 ;
355
356 grammar_declaration:
357 "%union" union_name braceless
358 {
359 union_seen = true;
360 muscle_code_grow ("stype", $3, @3);
361 code_scanner_last_string_free ();
362 }
363 ;
364
365
366
367
368 symbol_declaration:
369 "%nterm" { current_class = nterm_sym; } symbol_defs.1
370 {
371 current_class = unknown_sym;
372 current_type = NULL;
373 }
374 | "%token" { current_class = token_sym; } symbol_defs.1
375 {
376 current_class = unknown_sym;
377 current_type = NULL;
378 }
379 | "%type" TYPE symbols.1
380 {
381 symbol_list *list;
382 tag_seen = true;
383 for (list = $3; list; list = list->next)
384 symbol_type_set (list->content.sym, $2, @2);
385 symbol_list_free ($3);
386 }
387 ;
388
389 precedence_declaration:
390 precedence_declarator type.opt symbols.1
391 {
392 symbol_list *list;
393 ++current_prec;
394 for (list = $3; list; list = list->next)
395 {
396 symbol_type_set (list->content.sym, current_type, @2);
397 symbol_precedence_set (list->content.sym, current_prec, $1, @1);
398 }
399 symbol_list_free ($3);
400 current_type = NULL;
401 }
402 ;
403
404 precedence_declarator:
405 "%left" { $$ = left_assoc; }
406 | "%right" { $$ = right_assoc; }
407 | "%nonassoc" { $$ = non_assoc; }
408 ;
409
410 type.opt:
411 /* Nothing. */ { current_type = NULL; }
412 | TYPE { current_type = $1; tag_seen = true; }
413 ;
414
415 /* One or more symbols to be %typed. */
416 symbols.1:
417 symbol
418 { $$ = symbol_list_sym_new ($1, @1); }
419 | symbols.1 symbol
420 { $$ = symbol_list_prepend ($1, symbol_list_sym_new ($2, @2)); }
421 ;
422
423 generic_symlist:
424 generic_symlist_item { $$ = $1; }
425 | generic_symlist generic_symlist_item { $$ = symbol_list_prepend ($1, $2); }
426 ;
427
428 generic_symlist_item:
429 symbol { $$ = symbol_list_sym_new ($1, @1); }
430 | TYPE { $$ = symbol_list_type_new ($1, @1); }
431 | "<*>" { $$ = symbol_list_default_tagged_new (@1); }
432 | "<>" { $$ = symbol_list_default_tagless_new (@1); }
433 ;
434
435 /* One token definition. */
436 symbol_def:
437 TYPE
438 {
439 current_type = $1;
440 tag_seen = true;
441 }
442 | id
443 {
444 symbol_class_set ($1, current_class, @1, true);
445 symbol_type_set ($1, current_type, @1);
446 }
447 | id INT
448 {
449 symbol_class_set ($1, current_class, @1, true);
450 symbol_type_set ($1, current_type, @1);
451 symbol_user_token_number_set ($1, $2, @2);
452 }
453 | id string_as_id
454 {
455 symbol_class_set ($1, current_class, @1, true);
456 symbol_type_set ($1, current_type, @1);
457 symbol_make_alias ($1, $2, @$);
458 }
459 | id INT string_as_id
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 symbol_make_alias ($1, $3, @$);
465 }
466 ;
467
468 /* One or more symbol definitions. */
469 symbol_defs.1:
470 symbol_def
471 | symbol_defs.1 symbol_def
472 ;
473
474
475 /*------------------------------------------.
476 | The grammar section: between the two %%. |
477 `------------------------------------------*/
478
479 grammar:
480 rules_or_grammar_declaration
481 | grammar rules_or_grammar_declaration
482 ;
483
484 /* As a Bison extension, one can use the grammar declarations in the
485 body of the grammar. */
486 rules_or_grammar_declaration:
487 rules
488 | grammar_declaration ";"
489 | error ";"
490 {
491 yyerrok;
492 }
493 ;
494
495 rules:
496 id_colon { current_lhs = $1; current_lhs_location = @1; } rhses.1
497 ;
498
499 rhses.1:
500 rhs { grammar_current_rule_end (@1); }
501 | rhses.1 "|" rhs { grammar_current_rule_end (@3); }
502 | rhses.1 ";"
503 ;
504
505 rhs:
506 /* Nothing. */
507 { grammar_current_rule_begin (current_lhs, current_lhs_location); }
508 | rhs symbol
509 { grammar_current_rule_symbol_append ($2, @2); }
510 | rhs "{...}"
511 { grammar_current_rule_action_append ($2, @2); }
512 | rhs "%prec" symbol
513 { grammar_current_rule_prec_set ($3, @3); }
514 | rhs "%dprec" INT
515 { grammar_current_rule_dprec_set ($3, @3); }
516 | rhs "%merge" TYPE
517 { grammar_current_rule_merge_set ($3, @3); }
518 ;
519
520
521 /*----------------------------*
522 | variable and content.opt. |
523 *---------------------------*/
524
525 variable:
526 ID
527 | STRING { $$ = uniqstr_new ($1); } /* deprecated and not M4-friendly */
528 ;
529
530 /* Some content or empty by default. */
531 content.opt:
532 /* Nothing. */
533 {
534 $$ = "";
535 }
536 | STRING
537 ;
538
539
540 /*-------------*
541 | braceless. |
542 *-------------*/
543
544 braceless:
545 "{...}"
546 {
547 code_props plain_code;
548 $1[strlen ($1) - 1] = '\n';
549 code_props_plain_init (&plain_code, $1+1, @1);
550 code_props_translate_code (&plain_code);
551 gram_scanner_last_string_free ();
552 $$ = plain_code.code;
553 }
554 ;
555
556
557 /*---------------*
558 | Identifiers. |
559 *---------------*/
560
561 /* Identifiers are returned as uniqstr values by the scanner.
562 Depending on their use, we may need to make them genuine symbols. */
563
564 id:
565 ID
566 { $$ = symbol_from_uniqstr ($1, @1); }
567 | CHAR
568 {
569 $$ = symbol_get (char_name ($1), @1);
570 symbol_class_set ($$, token_sym, @1, false);
571 symbol_user_token_number_set ($$, $1, @1);
572 }
573 ;
574
575 id_colon:
576 ID_COLON { $$ = symbol_from_uniqstr ($1, @1); }
577 ;
578
579
580 symbol:
581 id
582 | string_as_id
583 ;
584
585 /* A string used as an ID: quote it. */
586 string_as_id:
587 STRING
588 {
589 $$ = symbol_get (quotearg_style (c_quoting_style, $1), @1);
590 symbol_class_set ($$, token_sym, @1, false);
591 }
592 ;
593
594 epilogue.opt:
595 /* Nothing. */
596 | "%%" EPILOGUE
597 {
598 code_props plain_code;
599 code_props_plain_init (&plain_code, $2, @2);
600 code_props_translate_code (&plain_code);
601 gram_scanner_last_string_free ();
602 muscle_code_grow ("epilogue", plain_code.code, @2);
603 code_scanner_last_string_free ();
604 }
605 ;
606
607 %%
608
609
610 /* Return the location of the left-hand side of a rule whose
611 right-hand side is RHS[1] ... RHS[N]. Ignore empty nonterminals in
612 the right-hand side, and return an empty location equal to the end
613 boundary of RHS[0] if the right-hand side is empty. */
614
615 static YYLTYPE
616 lloc_default (YYLTYPE const *rhs, int n)
617 {
618 int i;
619 YYLTYPE loc;
620
621 /* SGI MIPSpro 7.4.1m miscompiles "loc.start = loc.end = rhs[n].end;".
622 The bug is fixed in 7.4.2m, but play it safe for now. */
623 loc.start = rhs[n].end;
624 loc.end = rhs[n].end;
625
626 /* Ignore empty nonterminals the start of the the right-hand side.
627 Do not bother to ignore them at the end of the right-hand side,
628 since empty nonterminals have the same end as their predecessors. */
629 for (i = 1; i <= n; i++)
630 if (! equal_boundaries (rhs[i].start, rhs[i].end))
631 {
632 loc.start = rhs[i].start;
633 break;
634 }
635
636 return loc;
637 }
638
639
640 /* Add a lex-param or a parse-param (depending on TYPE) with
641 declaration DECL and location LOC. */
642
643 static void
644 add_param (char const *type, char *decl, location loc)
645 {
646 static char const alphanum[26 + 26 + 1 + 10] =
647 "abcdefghijklmnopqrstuvwxyz"
648 "ABCDEFGHIJKLMNOPQRSTUVWXYZ"
649 "_"
650 "0123456789";
651 char const *name_start = NULL;
652 char *p;
653
654 /* Stop on last actual character. */
655 for (p = decl; p[1]; p++)
656 if ((p == decl
657 || ! memchr (alphanum, p[-1], sizeof alphanum))
658 && memchr (alphanum, p[0], sizeof alphanum - 10))
659 name_start = p;
660
661 /* Strip the surrounding '{' and '}', and any blanks just inside
662 the braces. */
663 while (*--p == ' ' || *p == '\t')
664 continue;
665 p[1] = '\0';
666 while (*++decl == ' ' || *decl == '\t')
667 continue;
668
669 if (! name_start)
670 complain_at (loc, _("missing identifier in parameter declaration"));
671 else
672 {
673 char *name;
674 size_t name_len;
675
676 for (name_len = 1;
677 memchr (alphanum, name_start[name_len], sizeof alphanum);
678 name_len++)
679 continue;
680
681 name = xmalloc (name_len + 1);
682 memcpy (name, name_start, name_len);
683 name[name_len] = '\0';
684 muscle_pair_list_grow (type, decl, name);
685 free (name);
686 }
687
688 gram_scanner_last_string_free ();
689 }
690
691
692 static void
693 version_check (location const *loc, char const *version)
694 {
695 if (strverscmp (version, PACKAGE_VERSION) > 0)
696 {
697 complain_at (*loc, "require bison %s, but have %s",
698 version, PACKAGE_VERSION);
699 exit (63);
700 }
701 }
702
703 static void
704 gram_error (location const *loc, char const *msg)
705 {
706 complain_at (*loc, "%s", msg);
707 }
708
709 char const *
710 token_name (int type)
711 {
712 return yytname[YYTRANSLATE (type)];
713 }
714
715 static char const *
716 char_name (char c)
717 {
718 if (c == '\'')
719 return "'\\''";
720 else
721 {
722 char buf[4];
723 buf[0] = '\''; buf[1] = c; buf[2] = '\''; buf[3] = '\0';
724 return quotearg_style (escape_quoting_style, buf);
725 }
726 }