1 %{/* Bison Grammar Parser -*- C -*-
3 Copyright (C) 2002-2013 Free Software Foundation, Inc.
5 This file is part of Bison, the GNU Compiler Compiler.
7 This program is free software: you can redistribute it and/or modify
8 it under the terms of the GNU General Public License as published by
9 the Free Software Foundation, either version 3 of the License, or
10 (at your option) any later version.
12 This program is distributed in the hope that it will be useful,
13 but WITHOUT ANY WARRANTY; without even the implied warranty of
14 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
15 GNU General Public License for more details.
17 You should have received a copy of the GNU General Public License
18 along with this program. If not, see <http://www.gnu.org/licenses/>. */
25 #include "conflicts.h"
29 #include "named-ref.h"
34 #include "scan-gram.h"
35 #include "scan-code.h"
38 #define YYLLOC_DEFAULT(Current, Rhs, N) (Current) = lloc_default (Rhs, N)
39 static YYLTYPE lloc_default (YYLTYPE const *, int);
41 #define YY_LOCATION_PRINT(File, Loc) \
42 location_print (Loc, File)
44 static void version_check (location const *loc, char const *version);
46 static void gram_error (location const *, char const *);
48 /* A string that describes a char (e.g., 'a' -> "'a'"). */
49 static char const *char_name (char);
54 static int current_prec = 0;
55 static location current_lhs_location;
56 static named_ref *current_lhs_named_ref;
57 static symbol *current_lhs_symbol;
58 static symbol_class current_class = unknown_sym;
59 static uniqstr current_type = NULL;
61 /** Set the new current left-hand side symbol, possibly common
62 * to several right-hand side parts of rule.
66 current_lhs (symbol *sym, location loc, named_ref *ref)
68 current_lhs_symbol = sym;
69 current_lhs_location = loc;
70 /* In order to simplify memory management, named references for lhs
71 are always assigned by deep copy into the current symbol_list
72 node. This is because a single named-ref in the grammar may
73 result in several uses when the user factors lhs between several
74 rules using "|". Therefore free the parser's original copy. */
75 free (current_lhs_named_ref);
76 current_lhs_named_ref = ref;
79 #define YYTYPE_INT16 int_fast16_t
80 #define YYTYPE_INT8 int_fast8_t
81 #define YYTYPE_UINT16 uint_fast16_t
82 #define YYTYPE_UINT8 uint_fast8_t
85 %define api.prefix "gram_"
88 %define parse.error verbose
89 %define parse.lac full
97 /* Bison's grammar can initial empty locations, hence a default
98 location is needed. */
99 boundary_set (&@$.start, current_file, 1, 1);
100 boundary_set (&@$.end, current_file, 1, 1);
103 /* Define the tokens together with their human representation. */
104 %token GRAM_EOF 0 "end of file"
105 %token STRING "string"
107 %token PERCENT_TOKEN "%token"
108 %token PERCENT_NTERM "%nterm"
110 %token PERCENT_TYPE "%type"
111 %token PERCENT_DESTRUCTOR "%destructor"
112 %token PERCENT_PRINTER "%printer"
114 %token PERCENT_LEFT "%left"
115 %token PERCENT_RIGHT "%right"
116 %token PERCENT_NONASSOC "%nonassoc"
117 %token PERCENT_PRECEDENCE "%precedence"
119 %token PERCENT_PREC "%prec"
120 %token PERCENT_DPREC "%dprec"
121 %token PERCENT_MERGE "%merge"
123 /*----------------------.
124 | Global Declarations. |
125 `----------------------*/
129 PERCENT_DEFAULT_PREC "%default-prec"
130 PERCENT_DEFINE "%define"
131 PERCENT_DEFINES "%defines"
132 PERCENT_ERROR_VERBOSE "%error-verbose"
133 PERCENT_EXPECT "%expect"
134 PERCENT_EXPECT_RR "%expect-rr"
135 PERCENT_FLAG "%<flag>"
136 PERCENT_FILE_PREFIX "%file-prefix"
137 PERCENT_GLR_PARSER "%glr-parser"
138 PERCENT_INITIAL_ACTION "%initial-action"
139 PERCENT_LANGUAGE "%language"
140 PERCENT_NAME_PREFIX "%name-prefix"
141 PERCENT_NO_DEFAULT_PREC "%no-default-prec"
142 PERCENT_NO_LINES "%no-lines"
143 PERCENT_NONDETERMINISTIC_PARSER
144 "%nondeterministic-parser"
145 PERCENT_OUTPUT "%output"
146 PERCENT_REQUIRE "%require"
147 PERCENT_SKELETON "%skeleton"
148 PERCENT_START "%start"
149 PERCENT_TOKEN_TABLE "%token-table"
150 PERCENT_VERBOSE "%verbose"
154 %token BRACED_CODE "{...}"
155 %token BRACED_PREDICATE "%?{...}"
156 %token BRACKETED_ID "[identifier]"
158 %token EPILOGUE "epilogue"
160 %token ID "identifier"
161 %token ID_COLON "identifier:"
162 %token PERCENT_PERCENT "%%"
164 %token PROLOGUE "%{...%}"
170 %union {unsigned char character;}
171 %type <character> CHAR
172 %printer { fputs (char_name ($$), yyo); } CHAR
174 /* braceless is not to be used for rule or symbol actions, as it
175 calls code_props_plain_init. */
181 %type <chars> STRING "%{...%}" EPILOGUE braceless
182 %type <code> "{...}" "%?{...}"
183 %printer { fputs (quotearg_style (c_quoting_style, $$), yyo); }
185 %printer { fprintf (yyo, "{\n%s\n}", $$); }
186 braceless "{...}" "%{...%}" EPILOGUE
188 %union {uniqstr uniqstr;}
189 %type <uniqstr> BRACKETED_ID ID ID_COLON PERCENT_FLAG TAG tag variable
190 %printer { fputs ($$, yyo); } <uniqstr>
191 %printer { fprintf (yyo, "[%s]", $$); } BRACKETED_ID
192 %printer { fprintf (yyo, "%s:", $$); } ID_COLON
193 %printer { fprintf (yyo, "%%%s", $$); } PERCENT_FLAG
194 %printer { fprintf (yyo, "<%s>", $$); } TAG tag
196 %union {int integer;};
197 %token <integer> INT "integer"
198 %printer { fprintf (yyo, "%d", $$); } <integer>
200 %union {symbol *symbol;}
201 %type <symbol> id id_colon string_as_id symbol symbol.prec
202 %printer { fprintf (yyo, "%s", $$->tag); } <symbol>
203 %printer { fprintf (yyo, "%s:", $$->tag); } id_colon
205 %union {assoc assoc;};
206 %type <assoc> precedence_declarator
208 %union {symbol_list *list;}
209 %type <list> symbols.1 symbols.prec generic_symlist generic_symlist_item
211 %union {named_ref *named_ref;}
212 %type <named_ref> named_ref.opt
223 param_parse = 1 << 1,
224 param_both = param_lex | param_parse
229 /** Add a lex-param and/or a parse-param.
231 * \param type where to push this formal argument.
232 * \param decl the formal argument. Destroyed.
233 * \param loc the location in the source.
235 static void add_param (param_type type, char *decl, location loc);
236 static param_type current_param = param_none;
238 %union {param_type param;}
239 %token <param> PERCENT_PARAM "%param";
244 #define CASE(In, Out) \
245 case param_ ## In: fputs ("%" #Out, yyo); break
246 CASE (lex, lex-param);
247 CASE (parse, parse-param);
250 case param_none: aver (false); break;
261 prologue_declarations "%%" grammar epilogue.opt
265 /*------------------------------------.
266 | Declarations: before the first %%. |
267 `------------------------------------*/
269 prologue_declarations:
271 | prologue_declarations prologue_declaration
274 prologue_declaration:
278 code_props plain_code;
279 code_props_plain_init (&plain_code, $1, @1);
280 code_props_translate_code (&plain_code);
281 gram_scanner_last_string_free ();
282 muscle_code_grow (union_seen ? "post_prologue" : "pre_prologue",
283 plain_code.code, @1);
284 code_scanner_last_string_free ();
288 muscle_percent_define_ensure ($1, @1, true);
290 | "%define" variable value
292 muscle_percent_define_insert ($2, @2, $3.kind, $3.chars,
293 MUSCLE_PERCENT_DEFINE_GRAMMAR_FILE);
295 | "%defines" { defines_flag = true; }
299 spec_defines_file = xstrdup ($2);
303 muscle_percent_define_insert ("parse.error", @1, muscle_keyword,
305 MUSCLE_PERCENT_DEFINE_GRAMMAR_FILE);
307 | "%expect" INT { expected_sr_conflicts = $2; }
308 | "%expect-rr" INT { expected_rr_conflicts = $2; }
309 | "%file-prefix" STRING { spec_file_prefix = $2; }
312 nondeterministic_parser = true;
315 | "%initial-action" "{...}"
318 code_props_symbol_action_init (&action, $2, @2);
319 code_props_translate_code (&action);
320 gram_scanner_last_string_free ();
321 muscle_code_grow ("initial_action", action.code, @2);
322 code_scanner_last_string_free ();
324 | "%language" STRING { language_argmatch ($2, grammar_prio, @1); }
325 | "%name-prefix" STRING { spec_name_prefix = $2; }
326 | "%no-lines" { no_lines_flag = true; }
327 | "%nondeterministic-parser" { nondeterministic_parser = true; }
328 | "%output" STRING { spec_outfile = $2; }
329 | "%param" { current_param = $1; } params { current_param = param_none; }
330 | "%require" STRING { version_check (&@2, $2); }
333 char const *skeleton_user = $2;
334 if (strchr (skeleton_user, '/'))
336 size_t dir_length = strlen (current_file);
337 char *skeleton_build;
338 while (dir_length && current_file[dir_length - 1] != '/')
340 while (dir_length && current_file[dir_length - 1] == '/')
343 xmalloc (dir_length + 1 + strlen (skeleton_user) + 1);
346 memcpy (skeleton_build, current_file, dir_length);
347 skeleton_build[dir_length++] = '/';
349 strcpy (skeleton_build + dir_length, skeleton_user);
350 skeleton_user = uniqstr_new (skeleton_build);
351 free (skeleton_build);
353 skeleton_arg (skeleton_user, grammar_prio, @1);
355 | "%token-table" { token_table_flag = true; }
356 | "%verbose" { report_flag |= report_states; }
357 | "%yacc" { yacc_flag = true; }
358 | /*FIXME: Err? What is this horror doing here? */ ";"
362 params "{...}" { add_param (current_param, $2, @2); }
363 | "{...}" { add_param (current_param, $1, @1); }
367 /*----------------------.
368 | grammar_declaration. |
369 `----------------------*/
372 precedence_declaration
376 grammar_start_symbol_set ($2, @2);
378 | code_props_type "{...}" generic_symlist
381 code_props_symbol_action_init (&code, $2, @2);
382 code_props_translate_code (&code);
385 for (list = $3; list; list = list->next)
386 symbol_list_code_props_set (list, $1, &code);
387 symbol_list_free ($3);
396 default_prec = false;
400 /* Do not invoke muscle_percent_code_grow here since it invokes
401 muscle_user_name_list_grow. */
402 muscle_code_grow ("percent_code()", $2, @2);
403 code_scanner_last_string_free ();
405 | "%code" ID braceless
407 muscle_percent_code_grow ($2, @2, $3, @3);
408 code_scanner_last_string_free ();
412 %type <code_type> code_props_type;
413 %union {code_props_type code_type;};
414 %printer { fprintf (yyo, "%s", code_props_type_string ($$)); } <code_type>;
416 "%destructor" { $$ = destructor; }
417 | "%printer" { $$ = printer; }
424 %token PERCENT_UNION "%union";
428 | ID { muscle_code_grow ("union_name", $1, @1); }
432 "%union" union_name braceless
435 muscle_code_grow ("union_members", $3, @3);
436 code_scanner_last_string_free ();
444 "%nterm" { current_class = nterm_sym; } symbol_defs.1
446 current_class = unknown_sym;
449 | "%token" { current_class = token_sym; } symbol_defs.1
451 current_class = unknown_sym;
454 | "%type" TAG symbols.1
458 for (list = $3; list; list = list->next)
459 symbol_type_set (list->content.sym, $2, @2);
460 symbol_list_free ($3);
464 precedence_declaration:
465 precedence_declarator tag.opt symbols.prec
469 for (list = $3; list; list = list->next)
471 symbol_type_set (list->content.sym, current_type, @2);
472 symbol_precedence_set (list->content.sym, current_prec, $1, @1);
474 symbol_list_free ($3);
479 precedence_declarator:
480 "%left" { $$ = left_assoc; }
481 | "%right" { $$ = right_assoc; }
482 | "%nonassoc" { $$ = non_assoc; }
483 | "%precedence" { $$ = precedence_assoc; }
487 %empty { current_type = NULL; }
488 | TAG { current_type = $1; tag_seen = true; }
491 /* Just like symbols.1 but accept INT for the sake of POSIX. */
494 { $$ = symbol_list_sym_new ($1, @1); }
495 | symbols.prec symbol.prec
496 { $$ = symbol_list_append ($1, symbol_list_sym_new ($2, @2)); }
503 symbol_class_set ($1, token_sym, @1, false);
508 symbol_user_token_number_set ($1, $2, @2);
509 symbol_class_set ($1, token_sym, @1, false);
513 /* One or more symbols to be %typed. */
516 { $$ = symbol_list_sym_new ($1, @1); }
518 { $$ = symbol_list_append ($1, symbol_list_sym_new ($2, @2)); }
522 generic_symlist_item { $$ = $1; }
523 | generic_symlist generic_symlist_item { $$ = symbol_list_append ($1, $2); }
526 generic_symlist_item:
527 symbol { $$ = symbol_list_sym_new ($1, @1); }
528 | tag { $$ = symbol_list_type_new ($1, @1); }
533 | "<*>" { $$ = uniqstr_new ("*"); }
534 | "<>" { $$ = uniqstr_new (""); }
537 /* One token definition. */
546 symbol_class_set ($1, current_class, @1, true);
547 symbol_type_set ($1, current_type, @1);
551 symbol_class_set ($1, current_class, @1, true);
552 symbol_type_set ($1, current_type, @1);
553 symbol_user_token_number_set ($1, $2, @2);
557 symbol_class_set ($1, current_class, @1, true);
558 symbol_type_set ($1, current_type, @1);
559 symbol_make_alias ($1, $2, @$);
561 | id INT string_as_id
563 symbol_class_set ($1, current_class, @1, true);
564 symbol_type_set ($1, current_type, @1);
565 symbol_user_token_number_set ($1, $2, @2);
566 symbol_make_alias ($1, $3, @$);
570 /* One or more symbol definitions. */
573 | symbol_defs.1 symbol_def
577 /*------------------------------------------.
578 | The grammar section: between the two %%. |
579 `------------------------------------------*/
582 rules_or_grammar_declaration
583 | grammar rules_or_grammar_declaration
586 /* As a Bison extension, one can use the grammar declarations in the
587 body of the grammar. */
588 rules_or_grammar_declaration:
590 | grammar_declaration ";"
598 id_colon named_ref.opt { current_lhs ($1, @1, $2); } rhses.1
600 /* Free the current lhs. */
601 current_lhs (0, @1, 0);
606 rhs { grammar_current_rule_end (@1); }
607 | rhses.1 "|" rhs { grammar_current_rule_end (@3); }
611 %token PERCENT_EMPTY "%empty";
614 { grammar_current_rule_begin (current_lhs_symbol, current_lhs_location,
615 current_lhs_named_ref); }
616 | rhs symbol named_ref.opt
617 { grammar_current_rule_symbol_append ($2, @2, $3); }
618 | rhs "{...}" named_ref.opt
619 { grammar_current_rule_action_append ($2, @2, $3, false); }
621 { grammar_current_rule_action_append ($2, @2, NULL, true); }
623 { grammar_current_rule_empty_set (@2); }
625 { grammar_current_rule_prec_set ($3, @3); }
627 { grammar_current_rule_dprec_set ($3, @3); }
629 { grammar_current_rule_merge_set ($3, @3); }
634 | BRACKETED_ID { $$ = named_ref_new($1, @1); }
637 /*---------------------.
638 | variable and value. |
639 `---------------------*/
641 /* The STRING form of variable is deprecated and is not M4-friendly.
642 For example, M4 fails for '%define "[" "value"'. */
645 | STRING { $$ = uniqstr_new ($1); }
648 /* Some content or empty by default. */
649 %code requires {#include "muscle-tab.h"};
663 case muscle_code: fprintf (yyo, "{%s}", $$.chars); break;
664 case muscle_keyword: fprintf (yyo, "%s", $$.chars); break;
665 case muscle_string: fprintf (yyo, "\"%s\"", $$.chars); break;
670 %empty { $$.kind = muscle_keyword; $$.chars = ""; }
671 | ID { $$.kind = muscle_keyword; $$.chars = $1; }
672 | STRING { $$.kind = muscle_string; $$.chars = $1; }
673 | braceless { $$.kind = muscle_code; $$.chars = $1; }
684 code_props plain_code;
685 $1[strlen ($1) - 1] = '\0';
686 code_props_plain_init (&plain_code, $1+1, @1);
687 code_props_translate_code (&plain_code);
688 gram_scanner_last_string_free ();
689 $$ = plain_code.code;
698 /* Identifiers are returned as uniqstr values by the scanner.
699 Depending on their use, we may need to make them genuine symbols. */
703 { $$ = symbol_from_uniqstr ($1, @1); }
706 $$ = symbol_get (char_name ($1), @1);
707 symbol_class_set ($$, token_sym, @1, false);
708 symbol_user_token_number_set ($$, $1, @1);
713 ID_COLON { $$ = symbol_from_uniqstr ($1, @1); }
722 /* A string used as an ID: quote it. */
726 $$ = symbol_get (quotearg_style (c_quoting_style, $1), @1);
727 symbol_class_set ($$, token_sym, @1, false);
735 code_props plain_code;
736 code_props_plain_init (&plain_code, $2, @2);
737 code_props_translate_code (&plain_code);
738 gram_scanner_last_string_free ();
739 muscle_code_grow ("epilogue", plain_code.code, @2);
740 code_scanner_last_string_free ();
746 /* Return the location of the left-hand side of a rule whose
747 right-hand side is RHS[1] ... RHS[N]. Ignore empty nonterminals in
748 the right-hand side, and return an empty location equal to the end
749 boundary of RHS[0] if the right-hand side is empty. */
752 lloc_default (YYLTYPE const *rhs, int n)
757 /* SGI MIPSpro 7.4.1m miscompiles "loc.start = loc.end = rhs[n].end;".
758 The bug is fixed in 7.4.2m, but play it safe for now. */
759 loc.start = rhs[n].end;
760 loc.end = rhs[n].end;
762 /* Ignore empty nonterminals the start of the right-hand side.
763 Do not bother to ignore them at the end of the right-hand side,
764 since empty nonterminals have the same end as their predecessors. */
765 for (i = 1; i <= n; i++)
766 if (! equal_boundaries (rhs[i].start, rhs[i].end))
768 loc.start = rhs[i].start;
777 add_param (param_type type, char *decl, location loc)
779 static char const alphanum[26 + 26 + 1 + 10] =
780 "abcdefghijklmnopqrstuvwxyz"
781 "ABCDEFGHIJKLMNOPQRSTUVWXYZ"
785 char const *name_start = NULL;
788 /* Stop on last actual character. */
789 for (p = decl; p[1]; p++)
791 || ! memchr (alphanum, p[-1], sizeof alphanum))
792 && memchr (alphanum, p[0], sizeof alphanum - 10))
795 /* Strip the surrounding '{' and '}', and any blanks just inside
798 while (c_isspace ((unsigned char) *p))
802 while (c_isspace ((unsigned char) *decl))
807 complain (&loc, complaint, _("missing identifier in parameter declaration"));
810 char *name = xmemdup0 (name_start, strspn (name_start, alphanum));
811 if (type & param_lex)
812 muscle_pair_list_grow ("lex_param", decl, name);
813 if (type & param_parse)
814 muscle_pair_list_grow ("parse_param", decl, name);
818 gram_scanner_last_string_free ();
823 version_check (location const *loc, char const *version)
825 if (strverscmp (version, PACKAGE_VERSION) > 0)
827 complain (loc, complaint, "require bison %s, but have %s",
828 version, PACKAGE_VERSION);
834 gram_error (location const *loc, char const *msg)
836 complain (loc, complaint, "%s", msg);
840 token_name (int type)
842 return yytname[YYTRANSLATE (type)];
853 buf[0] = '\''; buf[1] = c; buf[2] = '\''; buf[3] = '\0';
854 return quotearg_style (escape_quoting_style, buf);