1 /* Bison Grammar Parser -*- C -*-
3 Copyright (C) 2002-2015 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/>. */
28 /* On column 0 to please syntax-check. */
38 #include "conflicts.h"
42 #include "named-ref.h"
45 #include "scan-gram.h"
46 #include "scan-code.h"
49 static int current_prec = 0;
50 static location current_lhs_location;
51 static named_ref *current_lhs_named_ref;
52 static symbol *current_lhs_symbol;
53 static symbol_class current_class = unknown_sym;
54 static uniqstr current_type = NULL;
56 /** Set the new current left-hand side symbol, possibly common
57 * to several right-hand side parts of rule.
59 static void current_lhs (symbol *sym, location loc, named_ref *ref);
61 #define YYLLOC_DEFAULT(Current, Rhs, N) \
62 (Current) = lloc_default (Rhs, N)
63 static YYLTYPE lloc_default (YYLTYPE const *, int);
65 #define YY_LOCATION_PRINT(File, Loc) \
66 location_print (Loc, File)
68 /* Strip initial '{' and final '}' (must be first and last characters).
70 static char *strip_braces (char *code);
72 /* Convert CODE by calling code_props_plain_init if PLAIN, otherwise
73 code_props_symbol_action_init. Call
74 gram_scanner_last_string_free to release the latest string from
75 the scanner (should be CODE). */
76 static char const *translate_code (char *code, location loc, bool plain);
78 /* Convert CODE by calling code_props_plain_init after having
79 stripped the first and last characters (expected to be '{', and
80 '}'). Call gram_scanner_last_string_free to release the latest
81 string from the scanner (should be CODE). */
82 static char const *translate_code_braceless (char *code, location loc);
84 static void version_check (location const *loc, char const *version);
86 static void gram_error (location const *, char const *);
88 /* A string that describes a char (e.g., 'a' -> "'a'"). */
89 static char const *char_name (char);
91 #define YYTYPE_INT16 int_fast16_t
92 #define YYTYPE_INT8 int_fast8_t
93 #define YYTYPE_UINT16 uint_fast16_t
94 #define YYTYPE_UINT8 uint_fast8_t
97 %define api.prefix {gram_}
100 %define parse.error verbose
101 %define parse.lac full
109 /* Bison's grammar can initial empty locations, hence a default
110 location is needed. */
111 boundary_set (&@$.start, current_file, 1, 1);
112 boundary_set (&@$.end, current_file, 1, 1);
115 /* Define the tokens together with their human representation. */
116 %token GRAM_EOF 0 "end of file"
117 %token STRING "string"
119 %token PERCENT_TOKEN "%token"
120 %token PERCENT_NTERM "%nterm"
122 %token PERCENT_TYPE "%type"
123 %token PERCENT_DESTRUCTOR "%destructor"
124 %token PERCENT_PRINTER "%printer"
126 %token PERCENT_LEFT "%left"
127 %token PERCENT_RIGHT "%right"
128 %token PERCENT_NONASSOC "%nonassoc"
129 %token PERCENT_PRECEDENCE "%precedence"
131 %token PERCENT_PREC "%prec"
132 %token PERCENT_DPREC "%dprec"
133 %token PERCENT_MERGE "%merge"
135 /*----------------------.
136 | Global Declarations. |
137 `----------------------*/
141 PERCENT_DEFAULT_PREC "%default-prec"
142 PERCENT_DEFINE "%define"
143 PERCENT_DEFINES "%defines"
144 PERCENT_ERROR_VERBOSE "%error-verbose"
145 PERCENT_EXPECT "%expect"
146 PERCENT_EXPECT_RR "%expect-rr"
147 PERCENT_FLAG "%<flag>"
148 PERCENT_FILE_PREFIX "%file-prefix"
149 PERCENT_GLR_PARSER "%glr-parser"
150 PERCENT_INITIAL_ACTION "%initial-action"
151 PERCENT_LANGUAGE "%language"
152 PERCENT_NAME_PREFIX "%name-prefix"
153 PERCENT_NO_DEFAULT_PREC "%no-default-prec"
154 PERCENT_NO_LINES "%no-lines"
155 PERCENT_NONDETERMINISTIC_PARSER
156 "%nondeterministic-parser"
157 PERCENT_OUTPUT "%output"
158 PERCENT_REQUIRE "%require"
159 PERCENT_SKELETON "%skeleton"
160 PERCENT_START "%start"
161 PERCENT_TOKEN_TABLE "%token-table"
162 PERCENT_VERBOSE "%verbose"
166 %token BRACED_CODE "{...}"
167 %token BRACED_PREDICATE "%?{...}"
168 %token BRACKETED_ID "[identifier]"
170 %token EPILOGUE "epilogue"
172 %token ID "identifier"
173 %token ID_COLON "identifier:"
174 %token PERCENT_PERCENT "%%"
176 %token PROLOGUE "%{...%}"
182 %union {unsigned char character;}
183 %type <character> CHAR
184 %printer { fputs (char_name ($$), yyo); } CHAR
186 %union {char *code;};
187 %type <code> "{...}" "%?{...}" "%{...%}" EPILOGUE STRING
188 %printer { fputs (quotearg_style (c_quoting_style, $$), yyo); } STRING
189 %printer { fprintf (yyo, "{\n%s\n}", $$); } <code>
191 %union {uniqstr uniqstr;}
192 %type <uniqstr> BRACKETED_ID ID ID_COLON PERCENT_FLAG TAG tag variable
193 %printer { fputs ($$, yyo); } <uniqstr>
194 %printer { fprintf (yyo, "[%s]", $$); } BRACKETED_ID
195 %printer { fprintf (yyo, "%s:", $$); } ID_COLON
196 %printer { fprintf (yyo, "%%%s", $$); } PERCENT_FLAG
197 %printer { fprintf (yyo, "<%s>", $$); } TAG tag
199 %union {int integer;};
200 %token <integer> INT "integer"
201 %printer { fprintf (yyo, "%d", $$); } <integer>
203 %union {symbol *symbol;}
204 %type <symbol> id id_colon string_as_id symbol symbol.prec
205 %printer { fprintf (yyo, "%s", $$->tag); } <symbol>
206 %printer { fprintf (yyo, "%s:", $$->tag); } id_colon
208 %union {assoc assoc;};
209 %type <assoc> precedence_declarator
211 %union {symbol_list *list;}
212 %type <list> symbols.1 symbols.prec generic_symlist generic_symlist_item
214 %union {named_ref *named_ref;}
215 %type <named_ref> named_ref.opt
226 param_parse = 1 << 1,
227 param_both = param_lex | param_parse
232 /** Add a lex-param and/or a parse-param.
234 * \param type where to push this formal argument.
235 * \param decl the formal argument. Destroyed.
236 * \param loc the location in the source.
238 static void add_param (param_type type, char *decl, location loc);
239 static param_type current_param = param_none;
241 %union {param_type param;}
242 %token <param> PERCENT_PARAM "%param";
247 #define CASE(In, Out) \
248 case param_ ## In: fputs ("%" #Out, yyo); break
249 CASE (lex, lex-param);
250 CASE (parse, parse-param);
253 case param_none: aver (false); break;
264 prologue_declarations "%%" grammar epilogue.opt
268 /*------------------------------------.
269 | Declarations: before the first %%. |
270 `------------------------------------*/
272 prologue_declarations:
274 | prologue_declarations prologue_declaration
277 prologue_declaration:
281 muscle_code_grow (union_seen ? "post_prologue" : "pre_prologue",
282 translate_code ($1, @1, true), @1);
283 code_scanner_last_string_free ();
287 muscle_percent_define_ensure ($1, @1, true);
289 | "%define" variable value
291 muscle_percent_define_insert ($2, @2, $3.kind, $3.chars,
292 MUSCLE_PERCENT_DEFINE_GRAMMAR_FILE);
294 | "%defines" { defines_flag = true; }
298 spec_defines_file = xstrdup ($2);
302 muscle_percent_define_insert ("parse.error", @1, muscle_keyword,
304 MUSCLE_PERCENT_DEFINE_GRAMMAR_FILE);
306 | "%expect" INT { expected_sr_conflicts = $2; }
307 | "%expect-rr" INT { expected_rr_conflicts = $2; }
308 | "%file-prefix" STRING { spec_file_prefix = $2; }
311 nondeterministic_parser = true;
314 | "%initial-action" "{...}"
316 muscle_code_grow ("initial_action", translate_code ($2, @2, false), @2);
317 code_scanner_last_string_free ();
319 | "%language" STRING { language_argmatch ($2, grammar_prio, @1); }
320 | "%name-prefix" STRING { spec_name_prefix = $2; }
321 | "%no-lines" { no_lines_flag = true; }
322 | "%nondeterministic-parser" { nondeterministic_parser = true; }
323 | "%output" STRING { spec_outfile = $2; }
324 | "%param" { current_param = $1; } params { current_param = param_none; }
325 | "%require" STRING { version_check (&@2, $2); }
328 char const *skeleton_user = $2;
329 if (strchr (skeleton_user, '/'))
331 size_t dir_length = strlen (current_file);
332 char *skeleton_build;
333 while (dir_length && current_file[dir_length - 1] != '/')
335 while (dir_length && current_file[dir_length - 1] == '/')
338 xmalloc (dir_length + 1 + strlen (skeleton_user) + 1);
341 memcpy (skeleton_build, current_file, dir_length);
342 skeleton_build[dir_length++] = '/';
344 strcpy (skeleton_build + dir_length, skeleton_user);
345 skeleton_user = uniqstr_new (skeleton_build);
346 free (skeleton_build);
348 skeleton_arg (skeleton_user, grammar_prio, @1);
350 | "%token-table" { token_table_flag = true; }
351 | "%verbose" { report_flag |= report_states; }
352 | "%yacc" { yacc_flag = true; }
353 | /*FIXME: Err? What is this horror doing here? */ ";"
357 params "{...}" { add_param (current_param, $2, @2); }
358 | "{...}" { add_param (current_param, $1, @1); }
362 /*----------------------.
363 | grammar_declaration. |
364 `----------------------*/
367 precedence_declaration
371 grammar_start_symbol_set ($2, @2);
373 | code_props_type "{...}" generic_symlist
376 code_props_symbol_action_init (&code, $2, @2);
377 code_props_translate_code (&code);
380 for (list = $3; list; list = list->next)
381 symbol_list_code_props_set (list, $1, &code);
382 symbol_list_free ($3);
391 default_prec = false;
395 /* Do not invoke muscle_percent_code_grow here since it invokes
396 muscle_user_name_list_grow. */
397 muscle_code_grow ("percent_code()",
398 translate_code_braceless ($2, @2), @2);
399 code_scanner_last_string_free ();
403 muscle_percent_code_grow ($2, @2, translate_code_braceless ($3, @3), @3);
404 code_scanner_last_string_free ();
408 %type <code_type> code_props_type;
409 %union {code_props_type code_type;};
410 %printer { fprintf (yyo, "%s", code_props_type_string ($$)); } <code_type>;
412 "%destructor" { $$ = destructor; }
413 | "%printer" { $$ = printer; }
420 %token PERCENT_UNION "%union";
424 | ID { muscle_percent_define_insert ("api.value.union.name",
425 @1, muscle_keyword, $1,
426 MUSCLE_PERCENT_DEFINE_GRAMMAR_FILE); }
430 "%union" union_name "{...}"
433 muscle_code_grow ("union_members", translate_code_braceless ($3, @3), @3);
434 code_scanner_last_string_free ();
442 "%nterm" { current_class = nterm_sym; } symbol_defs.1
444 current_class = unknown_sym;
447 | "%token" { current_class = token_sym; } symbol_defs.1
449 current_class = unknown_sym;
452 | "%type" TAG symbols.1
456 for (list = $3; list; list = list->next)
457 symbol_type_set (list->content.sym, $2, @2);
458 symbol_list_free ($3);
462 precedence_declaration:
463 precedence_declarator tag.opt symbols.prec
467 for (list = $3; list; list = list->next)
469 symbol_type_set (list->content.sym, current_type, @2);
470 symbol_precedence_set (list->content.sym, current_prec, $1, @1);
472 symbol_list_free ($3);
477 precedence_declarator:
478 "%left" { $$ = left_assoc; }
479 | "%right" { $$ = right_assoc; }
480 | "%nonassoc" { $$ = non_assoc; }
481 | "%precedence" { $$ = precedence_assoc; }
485 %empty { current_type = NULL; }
486 | TAG { current_type = $1; tag_seen = true; }
489 /* Just like symbols.1 but accept INT for the sake of POSIX. */
492 { $$ = symbol_list_sym_new ($1, @1); }
493 | symbols.prec symbol.prec
494 { $$ = symbol_list_append ($1, symbol_list_sym_new ($2, @2)); }
501 symbol_class_set ($1, token_sym, @1, false);
506 symbol_user_token_number_set ($1, $2, @2);
507 symbol_class_set ($1, token_sym, @1, false);
511 /* One or more symbols to be %typed. */
514 { $$ = symbol_list_sym_new ($1, @1); }
516 { $$ = symbol_list_append ($1, symbol_list_sym_new ($2, @2)); }
520 generic_symlist_item { $$ = $1; }
521 | generic_symlist generic_symlist_item { $$ = symbol_list_append ($1, $2); }
524 generic_symlist_item:
525 symbol { $$ = symbol_list_sym_new ($1, @1); }
526 | tag { $$ = symbol_list_type_new ($1, @1); }
531 | "<*>" { $$ = uniqstr_new ("*"); }
532 | "<>" { $$ = uniqstr_new (""); }
535 /* One token definition. */
544 symbol_class_set ($1, current_class, @1, true);
545 symbol_type_set ($1, current_type, @1);
549 symbol_class_set ($1, current_class, @1, true);
550 symbol_type_set ($1, current_type, @1);
551 symbol_user_token_number_set ($1, $2, @2);
555 symbol_class_set ($1, current_class, @1, true);
556 symbol_type_set ($1, current_type, @1);
557 symbol_make_alias ($1, $2, @$);
559 | id INT string_as_id
561 symbol_class_set ($1, current_class, @1, true);
562 symbol_type_set ($1, current_type, @1);
563 symbol_user_token_number_set ($1, $2, @2);
564 symbol_make_alias ($1, $3, @$);
568 /* One or more symbol definitions. */
571 | symbol_defs.1 symbol_def
575 /*------------------------------------------.
576 | The grammar section: between the two %%. |
577 `------------------------------------------*/
580 rules_or_grammar_declaration
581 | grammar rules_or_grammar_declaration
584 /* As a Bison extension, one can use the grammar declarations in the
585 body of the grammar. */
586 rules_or_grammar_declaration:
588 | grammar_declaration ";"
596 id_colon named_ref.opt { current_lhs ($1, @1, $2); } rhses.1
598 /* Free the current lhs. */
599 current_lhs (0, @1, 0);
604 rhs { grammar_current_rule_end (@1); }
605 | rhses.1 "|" rhs { grammar_current_rule_end (@3); }
609 %token PERCENT_EMPTY "%empty";
612 { grammar_current_rule_begin (current_lhs_symbol, current_lhs_location,
613 current_lhs_named_ref); }
614 | rhs symbol named_ref.opt
615 { grammar_current_rule_symbol_append ($2, @2, $3); }
616 | rhs "{...}" named_ref.opt
617 { grammar_current_rule_action_append ($2, @2, $3, false); }
619 { grammar_current_rule_action_append ($2, @2, NULL, true); }
621 { grammar_current_rule_empty_set (@2); }
623 { grammar_current_rule_prec_set ($3, @3); }
625 { grammar_current_rule_dprec_set ($3, @3); }
627 { grammar_current_rule_merge_set ($3, @3); }
632 | BRACKETED_ID { $$ = named_ref_new ($1, @1); }
635 /*---------------------.
636 | variable and value. |
637 `---------------------*/
639 /* The STRING form of variable is deprecated and is not M4-friendly.
640 For example, M4 fails for '%define "[" "value"'. */
643 | STRING { $$ = uniqstr_new ($1); }
646 /* Some content or empty by default. */
647 %code requires {#include "muscle-tab.h"};
661 case muscle_code: fprintf (yyo, "{%s}", $$.chars); break;
662 case muscle_keyword: fprintf (yyo, "%s", $$.chars); break;
663 case muscle_string: fprintf (yyo, "\"%s\"", $$.chars); break;
668 %empty { $$.kind = muscle_keyword; $$.chars = ""; }
669 | ID { $$.kind = muscle_keyword; $$.chars = $1; }
670 | STRING { $$.kind = muscle_string; $$.chars = $1; }
671 | "{...}" { $$.kind = muscle_code; $$.chars = strip_braces ($1); }
679 /* Identifiers are returned as uniqstr values by the scanner.
680 Depending on their use, we may need to make them genuine symbols. */
684 { $$ = symbol_from_uniqstr ($1, @1); }
687 $$ = symbol_get (char_name ($1), @1);
688 symbol_class_set ($$, token_sym, @1, false);
689 symbol_user_token_number_set ($$, $1, @1);
694 ID_COLON { $$ = symbol_from_uniqstr ($1, @1); }
703 /* A string used as an ID: quote it. */
707 $$ = symbol_get (quotearg_style (c_quoting_style, $1), @1);
708 symbol_class_set ($$, token_sym, @1, false);
716 muscle_code_grow ("epilogue", translate_code ($2, @2, true), @2);
717 code_scanner_last_string_free ();
723 /* Return the location of the left-hand side of a rule whose
724 right-hand side is RHS[1] ... RHS[N]. Ignore empty nonterminals in
725 the right-hand side, and return an empty location equal to the end
726 boundary of RHS[0] if the right-hand side is empty. */
729 lloc_default (YYLTYPE const *rhs, int n)
734 /* SGI MIPSpro 7.4.1m miscompiles "loc.start = loc.end = rhs[n].end;".
735 The bug is fixed in 7.4.2m, but play it safe for now. */
736 loc.start = rhs[n].end;
737 loc.end = rhs[n].end;
739 /* Ignore empty nonterminals the start of the right-hand side.
740 Do not bother to ignore them at the end of the right-hand side,
741 since empty nonterminals have the same end as their predecessors. */
742 for (i = 1; i <= n; i++)
743 if (! equal_boundaries (rhs[i].start, rhs[i].end))
745 loc.start = rhs[i].start;
753 char *strip_braces (char *code)
755 code[strlen (code) - 1] = 0;
761 translate_code (char *code, location loc, bool plain)
763 code_props plain_code;
765 code_props_plain_init (&plain_code, code, loc);
767 code_props_symbol_action_init (&plain_code, code, loc);
768 code_props_translate_code (&plain_code);
769 gram_scanner_last_string_free ();
770 return plain_code.code;
775 translate_code_braceless (char *code, location loc)
777 return translate_code (strip_braces (code), loc, true);
781 add_param (param_type type, char *decl, location loc)
783 static char const alphanum[26 + 26 + 1 + 10] =
784 "abcdefghijklmnopqrstuvwxyz"
785 "ABCDEFGHIJKLMNOPQRSTUVWXYZ"
789 char const *name_start = NULL;
792 /* Stop on last actual character. */
793 for (p = decl; p[1]; p++)
795 || ! memchr (alphanum, p[-1], sizeof alphanum))
796 && memchr (alphanum, p[0], sizeof alphanum - 10))
799 /* Strip the surrounding '{' and '}', and any blanks just inside
802 while (c_isspace ((unsigned char) *p))
806 while (c_isspace ((unsigned char) *decl))
811 complain (&loc, complaint, _("missing identifier in parameter declaration"));
814 char *name = xmemdup0 (name_start, strspn (name_start, alphanum));
815 if (type & param_lex)
816 muscle_pair_list_grow ("lex_param", decl, name);
817 if (type & param_parse)
818 muscle_pair_list_grow ("parse_param", decl, name);
822 gram_scanner_last_string_free ();
827 version_check (location const *loc, char const *version)
829 if (strverscmp (version, PACKAGE_VERSION) > 0)
831 complain (loc, complaint, "require bison %s, but have %s",
832 version, PACKAGE_VERSION);
838 gram_error (location const *loc, char const *msg)
840 complain (loc, complaint, "%s", msg);
844 token_name (int type)
846 return yytname[YYTRANSLATE (type)];
857 buf[0] = '\''; buf[1] = c; buf[2] = '\''; buf[3] = '\0';
858 return quotearg_style (escape_quoting_style, buf);
864 current_lhs (symbol *sym, location loc, named_ref *ref)
866 current_lhs_symbol = sym;
867 current_lhs_location = loc;
868 /* In order to simplify memory management, named references for lhs
869 are always assigned by deep copy into the current symbol_list
870 node. This is because a single named-ref in the grammar may
871 result in several uses when the user factors lhs between several
872 rules using "|". Therefore free the parser's original copy. */
873 free (current_lhs_named_ref);
874 current_lhs_named_ref = ref;