]> git.saurik.com Git - bison.git/blob - src/parse-gram.y
Use aver not assert.
[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 "named-ref.h"
31 #include "quotearg.h"
32 #include "reader.h"
33 #include "symlist.h"
34 #include "scan-gram.h"
35 #include "scan-code.h"
36
37 #define YYLLOC_DEFAULT(Current, Rhs, N) (Current) = lloc_default (Rhs, N)
38 static YYLTYPE lloc_default (YYLTYPE const *, int);
39
40 #define YY_LOCATION_PRINT(File, Loc) \
41 location_print (File, Loc)
42
43 static void version_check (location const *loc, char const *version);
44
45 /* Request detailed syntax error messages, and pass them to GRAM_ERROR.
46 FIXME: depends on the undocumented availability of YYLLOC. */
47 #undef yyerror
48 #define yyerror(Msg) \
49 gram_error (&yylloc, Msg)
50 static void gram_error (location const *, char const *);
51
52 static char const *char_name (char);
53
54 /** Add a lex-param or a parse-param.
55 *
56 * \param type \a lex_param or \a parse_param
57 * \param decl the formal argument
58 * \param loc the location in the source.
59 */
60 static void add_param (char const *type, char *decl, location loc);
61
62
63 static symbol_class current_class = unknown_sym;
64 static uniqstr current_type = NULL;
65 static symbol *current_lhs;
66 static location current_lhs_location;
67 static named_ref *current_lhs_named_ref;
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 %define parse.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 named_ref *named_ref;
104 };
105
106 /* Define the tokens together with their human representation. */
107 %token GRAM_EOF 0 "end of file"
108 %token STRING "string"
109 %token INT "integer"
110
111 %token PERCENT_TOKEN "%token"
112 %token PERCENT_NTERM "%nterm"
113
114 %token PERCENT_TYPE "%type"
115 %token PERCENT_DESTRUCTOR "%destructor"
116 %token PERCENT_PRINTER "%printer"
117
118 %token PERCENT_LEFT "%left"
119 %token PERCENT_RIGHT "%right"
120 %token PERCENT_NONASSOC "%nonassoc"
121 %token PERCENT_PRECEDENCE "%precedence"
122
123 %token PERCENT_PREC "%prec"
124 %token PERCENT_DPREC "%dprec"
125 %token PERCENT_MERGE "%merge"
126
127
128 /*----------------------.
129 | Global Declarations. |
130 `----------------------*/
131
132 %token
133 PERCENT_CODE "%code"
134 PERCENT_DEFAULT_PREC "%default-prec"
135 PERCENT_DEFINE "%define"
136 PERCENT_DEFINES "%defines"
137 PERCENT_ERROR_VERBOSE "%error-verbose"
138 PERCENT_EXPECT "%expect"
139 PERCENT_EXPECT_RR "%expect-rr"
140 PERCENT_FLAG "%<flag>"
141 PERCENT_FILE_PREFIX "%file-prefix"
142 PERCENT_GLR_PARSER "%glr-parser"
143 PERCENT_INITIAL_ACTION "%initial-action"
144 PERCENT_LANGUAGE "%language"
145 PERCENT_LEX_PARAM "%lex-param"
146 PERCENT_NAME_PREFIX "%name-prefix"
147 PERCENT_NO_DEFAULT_PREC "%no-default-prec"
148 PERCENT_NO_LINES "%no-lines"
149 PERCENT_NONDETERMINISTIC_PARSER
150 "%nondeterministic-parser"
151 PERCENT_OUTPUT "%output"
152 PERCENT_PARSE_PARAM "%parse-param"
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 TAG "<tag>"
172 %token TAG_ANY "<*>"
173 %token TAG_NONE "<>"
174 %token BRACKETED_ID "[identifier]"
175
176 %type <character> CHAR
177 %printer { fputs (char_name ($$), stderr); } CHAR
178
179 /* braceless is not to be used for rule or symbol actions, as it
180 calls code_props_plain_init. */
181 %type <chars> STRING "%{...%}" EPILOGUE braceless content.opt
182 %type <code> "{...}"
183 %printer { fputs (quotearg_style (c_quoting_style, $$), stderr); }
184 STRING
185 %printer { fprintf (stderr, "{\n%s\n}", $$); }
186 braceless content.opt "{...}" "%{...%}" EPILOGUE
187
188 %type <uniqstr> TAG ID ID_COLON BRACKETED_ID PERCENT_FLAG variable
189 %type <named_ref> named_ref.opt
190 %printer { fputs ($$, stderr); } ID variable
191 %printer { fprintf (stderr, "%s:", $$); } ID_COLON
192 %printer { fprintf (stderr, "%%%s", $$); } PERCENT_FLAG
193 %printer { fprintf (stderr, "<%s>", $$); } TAG
194
195 %type <integer> INT
196 %printer { fprintf (stderr, "%d", $$); } INT
197
198 %type <symbol> id id_colon symbol symbol.prec string_as_id
199 %printer { fprintf (stderr, "%s", $$->tag); } id symbol string_as_id
200 %printer { fprintf (stderr, "%s:", $$->tag); } id_colon
201
202 %type <assoc> precedence_declarator
203 %type <list> symbols.1 symbols.prec generic_symlist generic_symlist_item
204 %%
205
206 input:
207 prologue_declarations "%%" grammar epilogue.opt
208 ;
209
210
211 /*------------------------------------.
212 | Declarations: before the first %%. |
213 `------------------------------------*/
214
215 prologue_declarations:
216 /* Nothing */
217 | prologue_declarations prologue_declaration
218 ;
219
220 prologue_declaration:
221 grammar_declaration
222 | "%{...%}"
223 {
224 code_props plain_code;
225 code_props_plain_init (&plain_code, $1, @1);
226 code_props_translate_code (&plain_code);
227 gram_scanner_last_string_free ();
228 muscle_code_grow (union_seen ? "post_prologue" : "pre_prologue",
229 plain_code.code, @1);
230 code_scanner_last_string_free ();
231 }
232 | "%<flag>"
233 {
234 muscle_percent_define_ensure ($1, @1, true);
235 }
236 | "%define" variable content.opt
237 {
238 muscle_percent_define_insert ($2, @2, $3,
239 MUSCLE_PERCENT_DEFINE_GRAMMAR_FILE);
240 }
241 | "%defines" { defines_flag = true; }
242 | "%defines" STRING
243 {
244 defines_flag = true;
245 spec_defines_file = xstrdup ($2);
246 }
247 | "%error-verbose"
248 {
249 muscle_percent_define_insert ("parse.error", @1, "verbose",
250 MUSCLE_PERCENT_DEFINE_GRAMMAR_FILE);
251 }
252 | "%expect" INT { expected_sr_conflicts = $2; }
253 | "%expect-rr" INT { expected_rr_conflicts = $2; }
254 | "%file-prefix" STRING { spec_file_prefix = $2; }
255 | "%file-prefix" "=" STRING { spec_file_prefix = $3; } /* deprecated */
256 | "%glr-parser"
257 {
258 nondeterministic_parser = true;
259 glr_parser = true;
260 }
261 | "%initial-action" "{...}"
262 {
263 code_props action;
264 code_props_symbol_action_init (&action, $2, @2);
265 code_props_translate_code (&action);
266 gram_scanner_last_string_free ();
267 muscle_code_grow ("initial_action", action.code, @2);
268 code_scanner_last_string_free ();
269 }
270 | "%language" STRING { language_argmatch ($2, grammar_prio, @1); }
271 | "%lex-param" "{...}" { add_param ("lex_param", $2, @2); }
272 | "%name-prefix" STRING { spec_name_prefix = $2; }
273 | "%name-prefix" "=" STRING { spec_name_prefix = $3; } /* deprecated */
274 | "%no-lines" { no_lines_flag = true; }
275 | "%nondeterministic-parser" { nondeterministic_parser = true; }
276 | "%output" STRING { spec_outfile = $2; }
277 | "%output" "=" STRING { spec_outfile = $3; } /* deprecated */
278 | "%parse-param" "{...}" { add_param ("parse_param", $2, @2); }
279 | "%require" STRING { version_check (&@2, $2); }
280 | "%skeleton" STRING
281 {
282 char const *skeleton_user = $2;
283 if (strchr (skeleton_user, '/'))
284 {
285 size_t dir_length = strlen (current_file);
286 char *skeleton_build;
287 while (dir_length && current_file[dir_length - 1] != '/')
288 --dir_length;
289 while (dir_length && current_file[dir_length - 1] == '/')
290 --dir_length;
291 skeleton_build =
292 xmalloc (dir_length + 1 + strlen (skeleton_user) + 1);
293 if (dir_length > 0)
294 {
295 strncpy (skeleton_build, current_file, dir_length);
296 skeleton_build[dir_length++] = '/';
297 }
298 strcpy (skeleton_build + dir_length, skeleton_user);
299 skeleton_user = uniqstr_new (skeleton_build);
300 free (skeleton_build);
301 }
302 skeleton_arg (skeleton_user, grammar_prio, @1);
303 }
304 | "%token-table" { token_table_flag = true; }
305 | "%verbose" { report_flag |= report_states; }
306 | "%yacc" { yacc_flag = true; }
307 | /*FIXME: Err? What is this horror doing here? */ ";"
308 ;
309
310 grammar_declaration:
311 precedence_declaration
312 | symbol_declaration
313 | "%start" symbol
314 {
315 grammar_start_symbol_set ($2, @2);
316 }
317 | "%destructor" "{...}" generic_symlist
318 {
319 symbol_list *list;
320 for (list = $3; list; list = list->next)
321 symbol_list_destructor_set (list, $2, @2);
322 symbol_list_free ($3);
323 }
324 | "%printer" "{...}" generic_symlist
325 {
326 symbol_list *list;
327 for (list = $3; list; list = list->next)
328 symbol_list_printer_set (list, $2, @2);
329 symbol_list_free ($3);
330 }
331 | "%default-prec"
332 {
333 default_prec = true;
334 }
335 | "%no-default-prec"
336 {
337 default_prec = false;
338 }
339 | "%code" braceless
340 {
341 /* Do not invoke muscle_percent_code_grow here since it invokes
342 muscle_user_name_list_grow. */
343 muscle_code_grow ("percent_code()", $2, @2);
344 code_scanner_last_string_free ();
345 }
346 | "%code" ID braceless
347 {
348 muscle_percent_code_grow ($2, @2, $3, @3);
349 code_scanner_last_string_free ();
350 }
351 ;
352
353
354 /*---------.
355 | %union. |
356 `---------*/
357
358 %token PERCENT_UNION "%union";
359
360 union_name:
361 /* Nothing. */ {}
362 | ID { muscle_code_grow ("union_name", $1, @1); }
363 ;
364
365 grammar_declaration:
366 "%union" union_name braceless
367 {
368 union_seen = true;
369 muscle_code_grow ("stype", $3, @3);
370 code_scanner_last_string_free ();
371 }
372 ;
373
374
375
376
377 symbol_declaration:
378 "%nterm" { current_class = nterm_sym; } symbol_defs.1
379 {
380 current_class = unknown_sym;
381 current_type = NULL;
382 }
383 | "%token" { current_class = token_sym; } symbol_defs.1
384 {
385 current_class = unknown_sym;
386 current_type = NULL;
387 }
388 | "%type" TAG symbols.1
389 {
390 symbol_list *list;
391 tag_seen = true;
392 for (list = $3; list; list = list->next)
393 symbol_type_set (list->content.sym, $2, @2);
394 symbol_list_free ($3);
395 }
396 ;
397
398 precedence_declaration:
399 precedence_declarator tag.opt symbols.prec
400 {
401 symbol_list *list;
402 ++current_prec;
403 for (list = $3; list; list = list->next)
404 {
405 symbol_type_set (list->content.sym, current_type, @2);
406 symbol_precedence_set (list->content.sym, current_prec, $1, @1);
407 }
408 symbol_list_free ($3);
409 current_type = NULL;
410 }
411 ;
412
413 precedence_declarator:
414 "%left" { $$ = left_assoc; }
415 | "%right" { $$ = right_assoc; }
416 | "%nonassoc" { $$ = non_assoc; }
417 | "%precedence" { $$ = precedence_assoc; }
418 ;
419
420 tag.opt:
421 /* Nothing. */ { current_type = NULL; }
422 | TAG { current_type = $1; tag_seen = true; }
423 ;
424
425 /* Just like symbols.1 but accept INT for the sake of POSIX. */
426 symbols.prec:
427 symbol.prec
428 { $$ = symbol_list_sym_new ($1, @1); }
429 | symbols.prec symbol.prec
430 { $$ = symbol_list_prepend ($1, symbol_list_sym_new ($2, @2)); }
431 ;
432
433 symbol.prec:
434 symbol { $$ = $1; }
435 | symbol INT { $$ = $1; symbol_user_token_number_set ($1, $2, @2); }
436 ;
437
438 /* One or more symbols to be %typed. */
439 symbols.1:
440 symbol
441 { $$ = symbol_list_sym_new ($1, @1); }
442 | symbols.1 symbol
443 { $$ = symbol_list_prepend ($1, symbol_list_sym_new ($2, @2)); }
444 ;
445
446 generic_symlist:
447 generic_symlist_item { $$ = $1; }
448 | generic_symlist generic_symlist_item { $$ = symbol_list_prepend ($1, $2); }
449 ;
450
451 generic_symlist_item:
452 symbol { $$ = symbol_list_sym_new ($1, @1); }
453 | TAG { $$ = symbol_list_type_new ($1, @1); }
454 | "<*>" { $$ = symbol_list_default_tagged_new (@1); }
455 | "<>" { $$ = symbol_list_default_tagless_new (@1); }
456 ;
457
458 /* One token definition. */
459 symbol_def:
460 TAG
461 {
462 current_type = $1;
463 tag_seen = true;
464 }
465 | id
466 {
467 symbol_class_set ($1, current_class, @1, true);
468 symbol_type_set ($1, current_type, @1);
469 }
470 | id INT
471 {
472 symbol_class_set ($1, current_class, @1, true);
473 symbol_type_set ($1, current_type, @1);
474 symbol_user_token_number_set ($1, $2, @2);
475 }
476 | id string_as_id
477 {
478 symbol_class_set ($1, current_class, @1, true);
479 symbol_type_set ($1, current_type, @1);
480 symbol_make_alias ($1, $2, @$);
481 }
482 | id INT string_as_id
483 {
484 symbol_class_set ($1, current_class, @1, true);
485 symbol_type_set ($1, current_type, @1);
486 symbol_user_token_number_set ($1, $2, @2);
487 symbol_make_alias ($1, $3, @$);
488 }
489 ;
490
491 /* One or more symbol definitions. */
492 symbol_defs.1:
493 symbol_def
494 | symbol_defs.1 symbol_def
495 ;
496
497
498 /*------------------------------------------.
499 | The grammar section: between the two %%. |
500 `------------------------------------------*/
501
502 grammar:
503 rules_or_grammar_declaration
504 | grammar rules_or_grammar_declaration
505 ;
506
507 /* As a Bison extension, one can use the grammar declarations in the
508 body of the grammar. */
509 rules_or_grammar_declaration:
510 rules
511 | grammar_declaration ";"
512 | error ";"
513 {
514 yyerrok;
515 }
516 ;
517
518 rules:
519 id_colon named_ref.opt { current_lhs = $1; current_lhs_location = @1;
520 current_lhs_named_ref = $2; } rhses.1
521 ;
522
523 rhses.1:
524 rhs { grammar_current_rule_end (@1); }
525 | rhses.1 "|" rhs { grammar_current_rule_end (@3); }
526 | rhses.1 ";"
527 ;
528
529 rhs:
530 /* Nothing. */
531 { grammar_current_rule_begin (current_lhs, current_lhs_location,
532 current_lhs_named_ref); }
533 | rhs symbol named_ref.opt
534 { grammar_current_rule_symbol_append ($2, @2, $3); }
535 | rhs "{...}" named_ref.opt
536 { grammar_current_rule_action_append ($2, @2, $3); }
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" TAG
542 { grammar_current_rule_merge_set ($3, @3); }
543 ;
544
545 named_ref.opt:
546 /* Nothing. */ { $$ = 0; }
547 |
548 BRACKETED_ID { $$ = named_ref_new($1, @1); }
549 ;
550
551
552 /*---------------------------.
553 | variable and content.opt. |
554 `---------------------------*/
555
556 /* The STRING form of variable is deprecated and is not M4-friendly.
557 For example, M4 fails for `%define "[" "value"'. */
558 variable:
559 ID
560 | STRING { $$ = uniqstr_new ($1); }
561 ;
562
563 /* Some content or empty by default. */
564 content.opt:
565 /* Nothing. */ { $$ = ""; }
566 | ID { $$ = $1; }
567 | STRING
568 ;
569
570
571 /*------------.
572 | braceless. |
573 `------------*/
574
575 braceless:
576 "{...}"
577 {
578 code_props plain_code;
579 $1[strlen ($1) - 1] = '\n';
580 code_props_plain_init (&plain_code, $1+1, @1);
581 code_props_translate_code (&plain_code);
582 gram_scanner_last_string_free ();
583 $$ = 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 }