1 /* Bison Action Scanner -*- C -*-
3 Copyright (C) 2006-2010 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/>. */
20 %option debug nodefault noinput nounput noyywrap never-interactive
21 %option prefix="code_" outfile="lex.yy.c"
24 /* Work around a bug in flex 2.5.31. See Debian bug 333231
25 <http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=333231>. */
29 #define FLEX_PREFIX(Id) code_ ## Id
30 #include <src/flex-scanner.h>
32 #include <src/complain.h>
33 #include <src/reader.h>
34 #include <src/getargs.h>
35 #include <src/muscle-tab.h>
36 #include <src/scan-code.h>
37 #include <src/symlist.h>
40 #include <get-errno.h>
43 /* The current calling start condition: SC_RULE_ACTION or
45 # define YY_DECL static char *code_lex (code_props *self, int sc_context)
48 #define YY_USER_ACTION location_compute (loc, &loc->end, yytext, yyleng);
50 static void handle_action_dollar (symbol_list *rule, char *cp,
52 static void handle_action_at (symbol_list *rule, char *cp, location at_loc);
54 /* A string to be pushed to obstack after dollar/at has been handled. */
55 static char *ref_tail_fields;
57 static location the_location;
58 static location *loc = &the_location;
60 /* A string representing the most recent translation. */
61 static char *last_string;
63 /* True if an untyped $$ or $n was seen. */
64 static bool untyped_var_seen;
67 /* C and C++ comments in code. */
68 %x SC_COMMENT SC_LINE_COMMENT
69 /* Strings and characters in code. */
70 %x SC_STRING SC_CHARACTER
71 /* Whether in a rule or symbol action. Specifies the translation
73 %x SC_RULE_ACTION SC_SYMBOL_ACTION
76 /* POSIX says that a tag must be both an id and a C union member, but
77 historically almost any character is allowed in a tag. We disallow
78 NUL and newline, as this simplifies our implementation. */
81 /* Zero or more instances of backslash-newline. Following GCC, allow
82 white space between the backslash and the newline. */
83 splice (\\[ \f\t\v]*\n)*
85 /* C style identifier. Must start with letter. Will be used for
86 named symbol references. Shall be kept synchronized with
87 scan-gram.l "letter" and "id". */
88 letter [-.abcdefghijklmnopqrstuvwxyzABCDEFGHIJKLMNOPQRSTUVWXYZ_]
89 id {letter}({letter}|[0-9])*
90 ref -?[0-9]+|{id}|"["{id}"]"|"$"
95 /* Nesting level of the current code in braces. */
98 /* Whether a semicolon is probably needed.
99 The heuristic is that a semicolon is not needed after `{', `}', `;',
100 or a C preprocessor directive, and that whitespaces and comments
101 do not affect this flag.
102 Note that `{' does not need a semicolon because of `{}'.
103 A semicolon may be needed before a cpp direcive, but don't bother. */
104 bool need_semicolon = false;
106 /* Whether in a C preprocessor directive. Don't use a start condition
107 for this because, at the end of strings and comments, we still need
108 to know whether we're in a directive. */
111 /* This scanner is special: it is invoked only once, henceforth
112 is expected to return only once. This initialization is
113 therefore done once per action to translate. */
114 aver (sc_context == SC_SYMBOL_ACTION
115 || sc_context == SC_RULE_ACTION
116 || sc_context == INITIAL);
120 /*------------------------------------------------------------.
121 | Scanning a C comment. The initial `/ *' is already eaten. |
122 `------------------------------------------------------------*/
126 "*"{splice}"/" STRING_GROW; BEGIN sc_context;
130 /*--------------------------------------------------------------.
131 | Scanning a line comment. The initial `//' is already eaten. |
132 `--------------------------------------------------------------*/
136 "\n" STRING_GROW; BEGIN sc_context;
137 {splice} STRING_GROW;
141 /*--------------------------------------------.
142 | Scanning user-code characters and strings. |
143 `--------------------------------------------*/
145 <SC_CHARACTER,SC_STRING>
147 {splice}|\\{splice}. STRING_GROW;
152 "'" STRING_GROW; BEGIN sc_context;
157 "\"" STRING_GROW; BEGIN sc_context;
161 <SC_RULE_ACTION,SC_SYMBOL_ACTION>{
165 need_semicolon = true;
170 need_semicolon = true;
178 BEGIN SC_LINE_COMMENT;
184 "$"("<"{tag}">")?{ref} {
186 handle_action_dollar (self->rule, yytext, *loc);
187 if (ref_tail_fields) {
188 obstack_sgrow (&obstack_for_string, ref_tail_fields);
190 need_semicolon = true;
194 handle_action_at (self->rule, yytext, *loc);
195 if (ref_tail_fields) {
196 obstack_sgrow (&obstack_for_string, ref_tail_fields);
198 need_semicolon = true;
201 warn_at (*loc, _("stray `$'"));
202 obstack_sgrow (&obstack_for_string, "$][");
203 need_semicolon = true;
206 warn_at (*loc, _("stray `@'"));
207 obstack_sgrow (&obstack_for_string, "@@");
208 need_semicolon = true;
211 obstack_sgrow (&obstack_for_string, "@{");
212 need_semicolon = true;
215 obstack_sgrow (&obstack_for_string, "@}");
216 need_semicolon = true;
219 ";" STRING_GROW; need_semicolon = false;
220 "{" STRING_GROW; ++braces_level; need_semicolon = false;
222 bool outer_brace = --braces_level == 0;
224 /* As an undocumented Bison extension, append `;' before the last
225 brace in braced code, so that the user code can omit trailing
226 `;'. But do not append `;' if emulating Yacc, since Yacc does
228 if (outer_brace && !yacc_flag && language_prio == default_prio
229 && skeleton_prio == default_prio && need_semicolon && ! in_cpp)
231 warn_at (*loc, _("a `;' might be needed at the end of action code"));
232 warn_at (*loc, _("future versions of Bison will not add the `;'"));
233 obstack_1grow (&obstack_for_string, ';');
237 need_semicolon = false;
240 /* Preprocessing directives should only be recognized at the beginning
241 of lines, allowing whitespace including comments, but in C/C++,
242 `#' can only be the start of preprocessor directives or within
243 `#define' directives anyway, so don't bother with begin of line. */
244 "#" STRING_GROW; in_cpp = true;
246 {splice} STRING_GROW;
247 [\n\r] STRING_GROW; if (in_cpp) in_cpp = need_semicolon = false;
250 /* YYFAIL is undocumented and was formally deprecated in Bison
253 STRING_GROW; need_semicolon = true;
254 warn_at (*loc, _("use of YYFAIL, which is deprecated and will be"
258 /* The sole purpose of this is to make sure identifiers that merely
259 contain YYFAIL don't produce the above warning. */
260 [A-Za-z_][0-9A-Za-z_]* STRING_GROW; need_semicolon = true;
262 . STRING_GROW; need_semicolon = true;
268 obstack_sgrow (&obstack_for_string, "]b4_dollar_dollar[");
269 self->is_value_used = true;
272 obstack_sgrow (&obstack_for_string, "]b4_at_dollar[");
273 muscle_percent_define_ensure("locations", the_location, true);
278 /*-----------------------------------------.
279 | Escape M4 quoting characters in C code. |
280 `-----------------------------------------*/
284 \$ obstack_sgrow (&obstack_for_string, "$][");
285 \@ obstack_sgrow (&obstack_for_string, "@@");
286 \[ obstack_sgrow (&obstack_for_string, "@{");
287 \] obstack_sgrow (&obstack_for_string, "@}");
290 /*-----------------------------------------------------.
291 | By default, grow the string obstack with the input. |
292 `-----------------------------------------------------*/
296 /* End of processing. */
305 is_dot_or_dash (char ch)
307 return ch == '.' || ch == '-';
311 contains_dot_or_dash (const char* p)
314 if (is_dot_or_dash (*p))
319 /* Defines a variant of a symbolic name resolution. */
322 /* Index in symbol list. */
323 unsigned symbol_index;
325 /* Matched symbol id and loc. */
329 /* Hiding named reference. */
330 named_ref* hidden_by;
332 /* Error flags. May contain zero (no errors) or
333 a combination of VARIANT_* values. */
337 /* Set when the variant refers to a symbol hidden
338 by an explicit symbol reference. */
339 #define VARIANT_HIDDEN (1 << 0)
341 /* Set when the variant refers to a symbol containing
342 dots or dashes. Will require explicit bracketing. */
343 #define VARIANT_BAD_BRACKETING (1 << 1)
345 /* Set when the variant refers to a symbol which is
346 not visible from current midrule. */
347 #define VARIANT_NOT_VISIBLE_FROM_MIDRULE (1 << 2)
349 static variant *variant_table = 0;
350 static unsigned variant_table_size = 0;
351 static unsigned variant_count = 0;
354 variant_table_grow (void)
357 if (variant_count > variant_table_size)
359 while (variant_count > variant_table_size)
360 variant_table_size = 2 * variant_table_size + 3;
361 variant_table = xnrealloc (variant_table, variant_table_size,
362 sizeof *variant_table);
364 return &variant_table[variant_count - 1];
368 variant_table_free (void)
370 free (variant_table);
372 variant_table_size = variant_count = 0;
376 find_prefix_end (const char *prefix, char *begin, char *end)
380 for (; *prefix && ptr != end; ++prefix, ++ptr)
391 variant_add (uniqstr id, location id_loc, unsigned symbol_index,
392 char *cp, char *cp_end, bool explicit_bracketing)
396 prefix_end = find_prefix_end (id, cp, cp_end);
398 (prefix_end == cp_end ||
399 (!explicit_bracketing && is_dot_or_dash (*prefix_end))))
401 variant *r = variant_table_grow ();
402 r->symbol_index = symbol_index;
414 get_at_spec(unsigned symbol_index)
416 static char at_buf[20];
417 if (symbol_index == 0)
418 strcpy (at_buf, "$$");
420 snprintf (at_buf, sizeof at_buf, "$%u", symbol_index);
425 show_sub_messages (const char* cp, bool explicit_bracketing,
426 int midrule_rhs_index, char dollar_or_at,
427 bool is_warning, unsigned indent)
431 for (i = 0; i < variant_count; ++i)
433 const variant *var = &variant_table[i];
434 const char *at_spec = get_at_spec (var->symbol_index);
439 warn_at_indent (var->loc, &indent, _("refers to: %c%s at %s"),
440 dollar_or_at, var->id, at_spec);
442 complain_at_indent (var->loc, &indent, _("refers to: %c%s at %s"),
443 dollar_or_at, var->id, at_spec);
447 static struct obstack msg_buf;
448 const char *tail = explicit_bracketing ? "" :
449 cp + strlen (var->id);
450 const char *id = var->hidden_by ? var->hidden_by->id :
452 location id_loc = var->hidden_by ? var->hidden_by->loc :
455 /* Create the explanation message. */
456 obstack_init (&msg_buf);
458 obstack_fgrow1 (&msg_buf, _("possibly meant: %c"), dollar_or_at);
459 if (contains_dot_or_dash (id))
460 obstack_fgrow1 (&msg_buf, "[%s]", id);
462 obstack_sgrow (&msg_buf, id);
463 obstack_sgrow (&msg_buf, tail);
465 if (var->err & VARIANT_HIDDEN)
467 obstack_fgrow1 (&msg_buf, _(", hiding %c"), dollar_or_at);
468 if (contains_dot_or_dash (var->id))
469 obstack_fgrow1 (&msg_buf, "[%s]", var->id);
471 obstack_sgrow (&msg_buf, var->id);
472 obstack_sgrow (&msg_buf, tail);
475 obstack_fgrow1 (&msg_buf, _(" at %s"), at_spec);
477 if (var->err & VARIANT_NOT_VISIBLE_FROM_MIDRULE)
480 _(", cannot be accessed from mid-rule action at $%d");
481 obstack_fgrow1 (&msg_buf, format, midrule_rhs_index);
484 obstack_1grow (&msg_buf, '\0');
486 warn_at_indent (id_loc, &indent, "%s",
487 (char *) obstack_finish (&msg_buf));
489 complain_at_indent (id_loc, &indent, "%s",
490 (char *) obstack_finish (&msg_buf));
491 obstack_free (&msg_buf, 0);
496 /* Returned from "parse_ref" when the reference
498 #define INVALID_REF (INT_MIN)
500 /* Returned from "parse_ref" when the reference
501 points to LHS ($$) of the current rule or midrule. */
502 #define LHS_REF (INT_MIN + 1)
504 /* Sub-messages indent. */
505 #define SUB_INDENT (4)
507 /* Parse named or positional reference. In case of positional
508 references, can return negative values for $-n "deep" stack
511 parse_ref (char *cp, symbol_list *rule, int rule_length,
512 int midrule_rhs_index, char *text, location text_loc,
517 bool explicit_bracketing;
519 unsigned valid_variants = 0;
520 unsigned valid_variant_index = 0;
525 if (c_isdigit (*cp) || (*cp == '-' && c_isdigit (* (cp + 1))))
527 long int num = strtol (cp, &cp, 10);
528 if (1 - INT_MAX + rule_length <= num && num <= rule_length)
532 complain_at (text_loc, _("integer out of range: %s"),
540 /* Ignore the brackets. */
542 for (p = ++cp; *p != ']'; ++p)
546 explicit_bracketing = true;
550 /* Take all characters of the name. */
552 for (p = cp; *p; ++p)
553 if (is_dot_or_dash (*p))
558 for (p = cp; *p; ++p)
562 explicit_bracketing = false;
565 /* Add all relevant variants. */
567 unsigned symbol_index;
569 for (symbol_index = 0, l = rule; !symbol_list_null (l);
570 ++symbol_index, l = l->next)
573 if (l->content_type != SYMLIST_SYMBOL)
576 var = variant_add (l->content.sym->tag, l->sym_loc,
577 symbol_index, cp, cp_end, explicit_bracketing);
578 if (var && l->named_ref)
579 var->hidden_by = l->named_ref;
582 variant_add (l->named_ref->id, l->named_ref->loc,
583 symbol_index, cp, cp_end, explicit_bracketing);
588 for (i = 0; i < variant_count; ++i)
590 variant *var = &variant_table[i];
591 unsigned symbol_index = var->symbol_index;
593 /* Check visibility from mid-rule actions. */
594 if (midrule_rhs_index != 0
595 && (symbol_index == 0 || midrule_rhs_index < symbol_index))
596 var->err |= VARIANT_NOT_VISIBLE_FROM_MIDRULE;
598 /* Check correct bracketing. */
599 if (!explicit_bracketing && contains_dot_or_dash (var->id))
600 var->err |= VARIANT_BAD_BRACKETING;
602 /* Check using of hidden symbols. */
604 var->err |= VARIANT_HIDDEN;
608 valid_variant_index = i;
613 switch (valid_variants)
617 unsigned len = (explicit_bracketing || !ref_tail_fields) ?
618 cp_end - cp : ref_tail_fields - cp;
621 complain_at_indent (text_loc, &indent, _("invalid reference: %s"),
623 indent += SUB_INDENT;
624 if (midrule_rhs_index)
627 _("symbol not found in production before $%d: %.*s");
628 complain_at_indent (rule->location, &indent, format,
629 midrule_rhs_index, len, cp);
634 _("symbol not found in production: %.*s");
635 complain_at_indent (rule->location, &indent, format,
639 if (variant_count > 0)
640 show_sub_messages (cp, explicit_bracketing, midrule_rhs_index,
641 dollar_or_at, false, indent);
647 if (variant_count > 1)
649 warn_at_indent (text_loc, &indent, _("misleading reference: %s"),
651 show_sub_messages (cp, explicit_bracketing, midrule_rhs_index,
652 dollar_or_at, true, indent + SUB_INDENT);
655 unsigned symbol_index =
656 variant_table[valid_variant_index].symbol_index;
657 return (symbol_index == midrule_rhs_index) ? LHS_REF : symbol_index;
664 complain_at_indent (text_loc, &indent, _("ambiguous reference: %s"),
666 show_sub_messages (cp, explicit_bracketing, midrule_rhs_index,
667 dollar_or_at, false, indent + SUB_INDENT);
676 /* Keeps track of the maximum number of semantic values to the left of
677 a handle (those referenced by $0, $-1, etc.) are required by the
678 semantic actions of this grammar. */
679 int max_left_semantic_context = 0;
682 /*------------------------------------------------------------------.
683 | TEXT is pointing to a wannabee semantic value (i.e., a `$'). |
685 | Possible inputs: $[<TYPENAME>]($|integer) |
687 | Output to OBSTACK_FOR_STRING a reference to this semantic value. |
688 `------------------------------------------------------------------*/
691 handle_action_dollar (symbol_list *rule, char *text, location dollar_loc)
693 char const *type_name = NULL;
696 symbol_list *effective_rule;
697 int effective_rule_length;
700 if (rule->midrule_parent_rule)
702 effective_rule = rule->midrule_parent_rule;
703 effective_rule_length = rule->midrule_parent_rhs_index - 1;
707 effective_rule = rule;
708 effective_rule_length = symbol_list_length (rule->next);
711 /* Get the type name if explicit. */
718 /* The '>' symbol will be later replaced by '\0'. Original
719 'text' is needed for error messages. */
722 if (untyped_var_seen)
723 complain_at (dollar_loc, _("explicit type given in untyped grammar"));
727 n = parse_ref (cp, effective_rule, effective_rule_length,
728 rule->midrule_parent_rhs_index, text, dollar_loc, '$');
740 type_name = symbol_list_n_type_name_get (rule, dollar_loc, 0);
744 if (union_seen | tag_seen)
746 if (rule->midrule_parent_rule)
747 complain_at (dollar_loc,
748 _("$$ for the midrule at $%d of `%s'"
749 " has no declared type"),
750 rule->midrule_parent_rhs_index,
751 effective_rule->content.sym->tag);
753 complain_at (dollar_loc, _("$$ of `%s' has no declared type"),
754 rule->content.sym->tag);
757 untyped_var_seen = true;
761 obstack_fgrow1 (&obstack_for_string,
762 "]b4_lhs_value([%s])[", type_name);
763 rule->action_props.is_value_used = true;
767 if (max_left_semantic_context < 1 - n)
768 max_left_semantic_context = 1 - n;
769 if (!type_name && 0 < n)
771 symbol_list_n_type_name_get (effective_rule, dollar_loc, n);
774 if (union_seen | tag_seen)
775 complain_at (dollar_loc, _("$%s of `%s' has no declared type"),
776 cp, effective_rule->content.sym->tag);
778 untyped_var_seen = true;
782 obstack_fgrow3 (&obstack_for_string,
783 "]b4_rhs_value(%d, %d, [%s])[",
784 effective_rule_length, n, type_name);
786 symbol_list_n_get (effective_rule, n)->action_props.is_value_used =
793 /*------------------------------------------------------.
794 | TEXT is a location token (i.e., a `@...'). Output to |
795 | OBSTACK_FOR_STRING a reference to this location. |
796 `------------------------------------------------------*/
799 handle_action_at (symbol_list *rule, char *text, location at_loc)
802 symbol_list *effective_rule;
803 int effective_rule_length;
806 if (rule->midrule_parent_rule)
808 effective_rule = rule->midrule_parent_rule;
809 effective_rule_length = rule->midrule_parent_rhs_index - 1;
813 effective_rule = rule;
814 effective_rule_length = symbol_list_length (rule->next);
817 muscle_percent_define_ensure("locations", at_loc, true);
819 n = parse_ref (cp, effective_rule, effective_rule_length,
820 rule->midrule_parent_rhs_index, text, at_loc, '@');
827 obstack_sgrow (&obstack_for_string, "]b4_lhs_location[");
831 obstack_fgrow2 (&obstack_for_string, "]b4_rhs_location(%d, %d)[",
832 effective_rule_length, n);
838 /*-------------------------.
839 | Initialize the scanner. |
840 `-------------------------*/
842 /* Translate the dollars and ats in \a self, in the context \a sc_context
843 (SC_RULE_ACTION, SC_SYMBOL_ACTION, INITIAL). */
846 translate_action (code_props *self, int sc_context)
849 static bool initialized = false;
852 obstack_init (&obstack_for_string);
857 loc->start = loc->end = self->location.start;
858 yy_switch_to_buffer (yy_scan_string (self->code));
859 res = code_lex (self, sc_context);
860 yy_delete_buffer (YY_CURRENT_BUFFER);
865 /*------------------------------------------------------------------------.
866 | Implementation of the public interface as documented in "scan-code.h". |
867 `------------------------------------------------------------------------*/
870 code_props_none_init (code_props *self)
872 *self = code_props_none;
875 code_props const code_props_none = CODE_PROPS_NONE_INIT;
878 code_props_plain_init (code_props *self, char const *code,
881 self->kind = CODE_PROPS_PLAIN;
883 self->location = code_loc;
884 self->is_value_used = false;
886 self->named_ref = NULL;
890 code_props_symbol_action_init (code_props *self, char const *code,
893 self->kind = CODE_PROPS_SYMBOL_ACTION;
895 self->location = code_loc;
896 self->is_value_used = false;
898 self->named_ref = NULL;
902 code_props_rule_action_init (code_props *self, char const *code,
903 location code_loc, symbol_list *rule,
906 self->kind = CODE_PROPS_RULE_ACTION;
908 self->location = code_loc;
909 self->is_value_used = false;
911 self->named_ref = name;
915 code_props_translate_code (code_props *self)
919 case CODE_PROPS_NONE:
921 case CODE_PROPS_PLAIN:
922 self->code = translate_action (self, INITIAL);
924 case CODE_PROPS_SYMBOL_ACTION:
925 self->code = translate_action (self, SC_SYMBOL_ACTION);
927 case CODE_PROPS_RULE_ACTION:
928 self->code = translate_action (self, SC_RULE_ACTION);
934 code_scanner_last_string_free (void)
940 code_scanner_free (void)
942 obstack_free (&obstack_for_string, 0);
943 variant_table_free ();
945 /* Reclaim Flex's buffers. */