1 /* Bison Action Scanner -*- C -*-
3 Copyright (C) 2006-2012 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 "flex-scanner.h"
35 #include "scan-code.h"
39 #include <get-errno.h>
42 /* The current calling start condition: SC_RULE_ACTION or
44 # define YY_DECL static char *code_lex (code_props *self, int sc_context)
47 #define YY_USER_ACTION location_compute (loc, &loc->end, yytext, yyleng);
49 static void handle_action_dollar (symbol_list *rule, char *cp,
51 static void handle_action_at (symbol_list *rule, char *cp, location at_loc);
53 /* A string to be pushed to obstack after dollar/at has been handled. */
54 static char *ref_tail_fields;
56 static location the_location;
57 static location *loc = &the_location;
59 /* A string representing the most recent translation. */
60 static char *last_string;
62 /* True if an untyped $$ or $n was seen. */
63 static bool untyped_var_seen;
66 /* C and C++ comments in code. */
67 %x SC_COMMENT SC_LINE_COMMENT
68 /* Strings and characters in code. */
69 %x SC_STRING SC_CHARACTER
70 /* Whether in a rule or symbol action. Specifies the translation
72 %x SC_RULE_ACTION SC_SYMBOL_ACTION
75 /* POSIX says that a tag must be both an id and a C union member, but
76 historically almost any character is allowed in a tag. We disallow
77 NUL and newline, as this simplifies our implementation. */
80 /* Zero or more instances of backslash-newline. Following GCC, allow
81 white space between the backslash and the newline. */
82 splice (\\[ \f\t\v]*\n)*
84 /* C style identifier. Must start with letter. Will be used for
85 named symbol references. Shall be kept synchronized with
86 scan-gram.l "letter" and "id". */
87 letter [.abcdefghijklmnopqrstuvwxyzABCDEFGHIJKLMNOPQRSTUVWXYZ_]
88 id {letter}({letter}|[-0-9])*
89 ref -?[0-9]+|{id}|"["{id}"]"|"$"
94 /* Nesting level of the current code in braces. */
97 /* Whether a semicolon is probably needed.
98 The heuristic is that a semicolon is not needed after '{', '}', ';',
99 or a C preprocessor directive, and that whitespaces and comments
100 do not affect this flag.
101 Note that '{' does not need a semicolon because of '{}'.
102 A semicolon may be needed before a cpp direcive, but don't bother. */
103 bool need_semicolon = false;
105 /* Whether in a C preprocessor directive. Don't use a start condition
106 for this because, at the end of strings and comments, we still need
107 to know whether we're in a directive. */
110 /* This scanner is special: it is invoked only once, henceforth
111 is expected to return only once. This initialization is
112 therefore done once per action to translate. */
113 aver (sc_context == SC_SYMBOL_ACTION
114 || sc_context == SC_RULE_ACTION
115 || sc_context == INITIAL);
119 /*------------------------------------------------------------.
120 | Scanning a C comment. The initial '/ *' is already eaten. |
121 `------------------------------------------------------------*/
125 "*"{splice}"/" STRING_GROW; BEGIN sc_context;
129 /*--------------------------------------------------------------.
130 | Scanning a line comment. The initial '//' is already eaten. |
131 `--------------------------------------------------------------*/
135 "\n" STRING_GROW; BEGIN sc_context;
136 {splice} STRING_GROW;
140 /*--------------------------------------------.
141 | Scanning user-code characters and strings. |
142 `--------------------------------------------*/
144 <SC_CHARACTER,SC_STRING>
146 {splice}|\\{splice}. STRING_GROW;
151 "'" STRING_GROW; BEGIN sc_context;
156 "\"" STRING_GROW; BEGIN sc_context;
160 <SC_RULE_ACTION,SC_SYMBOL_ACTION>{
164 need_semicolon = true;
169 need_semicolon = true;
177 BEGIN SC_LINE_COMMENT;
183 "$"("<"{tag}">")?{ref} {
185 handle_action_dollar (self->rule, yytext, *loc);
186 if (ref_tail_fields) {
187 obstack_sgrow (&obstack_for_string, ref_tail_fields);
189 need_semicolon = true;
193 handle_action_at (self->rule, yytext, *loc);
194 if (ref_tail_fields) {
195 obstack_sgrow (&obstack_for_string, ref_tail_fields);
197 need_semicolon = true;
200 warn_at (*loc, _("stray '$'"));
201 obstack_sgrow (&obstack_for_string, "$][");
202 need_semicolon = true;
205 warn_at (*loc, _("stray '@'"));
206 obstack_sgrow (&obstack_for_string, "@@");
207 need_semicolon = true;
210 obstack_sgrow (&obstack_for_string, "@{");
211 need_semicolon = true;
214 obstack_sgrow (&obstack_for_string, "@}");
215 need_semicolon = true;
218 ";" STRING_GROW; need_semicolon = false;
219 "{" STRING_GROW; ++braces_level; need_semicolon = false;
221 bool outer_brace = --braces_level == 0;
223 /* As an undocumented Bison extension, append ';' before the last
224 brace in braced code, so that the user code can omit trailing
225 ';'. But do not append ';' if emulating Yacc, since Yacc does
227 if (outer_brace && !yacc_flag && language_prio == default_prio
228 && skeleton_prio == default_prio && need_semicolon && ! in_cpp)
230 warn_at (*loc, _("a ';' might be needed at the end of action code"));
231 warn_at (*loc, _("future versions of Bison will not add the ';'"));
232 obstack_1grow (&obstack_for_string, ';');
236 need_semicolon = false;
239 /* Preprocessing directives should only be recognized at the beginning
240 of lines, allowing whitespace including comments, but in C/C++,
241 '#' can only be the start of preprocessor directives or within
242 '#define' directives anyway, so don't bother with begin of line. */
243 "#" STRING_GROW; in_cpp = true;
245 {splice} STRING_GROW;
246 [\n\r] STRING_GROW; if (in_cpp) in_cpp = need_semicolon = false;
249 /* YYFAIL is undocumented and was formally deprecated in Bison
252 STRING_GROW; need_semicolon = true;
253 warn_at (*loc, _("use of YYFAIL, which is deprecated and will be"
257 /* The sole purpose of this is to make sure identifiers that merely
258 contain YYFAIL don't produce the above warning. */
259 [A-Za-z_][0-9A-Za-z_]* STRING_GROW; need_semicolon = true;
261 . STRING_GROW; need_semicolon = true;
267 obstack_sgrow (&obstack_for_string, "]b4_dollar_dollar[");
268 self->is_value_used = true;
271 obstack_sgrow (&obstack_for_string, "]b4_at_dollar[");
272 locations_flag = true;
277 /*-----------------------------------------.
278 | Escape M4 quoting characters in C code. |
279 `-----------------------------------------*/
283 \$ obstack_sgrow (&obstack_for_string, "$][");
284 \@ obstack_sgrow (&obstack_for_string, "@@");
285 \[ obstack_sgrow (&obstack_for_string, "@{");
286 \] obstack_sgrow (&obstack_for_string, "@}");
289 /*-----------------------------------------------------.
290 | By default, grow the string obstack with the input. |
291 `-----------------------------------------------------*/
295 /* End of processing. */
304 is_dot_or_dash (char ch)
306 return ch == '.' || ch == '-';
310 contains_dot_or_dash (const char* p)
313 if (is_dot_or_dash (*p))
318 /* Defines a variant of a symbolic name resolution. */
321 /* Index in symbol list. */
322 unsigned symbol_index;
324 /* Matched symbol id and loc. */
328 /* Hiding named reference. */
329 named_ref* hidden_by;
331 /* Error flags. May contain zero (no errors) or
332 a combination of VARIANT_* values. */
336 /* Set when the variant refers to a symbol hidden
337 by an explicit symbol reference. */
338 #define VARIANT_HIDDEN (1 << 0)
340 /* Set when the variant refers to a symbol containing
341 dots or dashes. Will require explicit bracketing. */
342 #define VARIANT_BAD_BRACKETING (1 << 1)
344 /* Set when the variant refers to a symbol which is
345 not visible from current midrule. */
346 #define VARIANT_NOT_VISIBLE_FROM_MIDRULE (1 << 2)
348 static variant *variant_table = 0;
349 static unsigned variant_table_size = 0;
350 static unsigned variant_count = 0;
353 variant_table_grow (void)
356 if (variant_count > variant_table_size)
358 while (variant_count > variant_table_size)
359 variant_table_size = 2 * variant_table_size + 3;
360 variant_table = xnrealloc (variant_table, variant_table_size,
361 sizeof *variant_table);
363 return &variant_table[variant_count - 1];
367 variant_table_free (void)
369 free (variant_table);
371 variant_table_size = variant_count = 0;
375 find_prefix_end (const char *prefix, char *begin, char *end)
379 for (; *prefix && ptr != end; ++prefix, ++ptr)
390 variant_add (uniqstr id, location id_loc, unsigned symbol_index,
391 char *cp, char *cp_end, bool explicit_bracketing)
395 prefix_end = find_prefix_end (id, cp, cp_end);
397 (prefix_end == cp_end ||
398 (!explicit_bracketing && is_dot_or_dash (*prefix_end))))
400 variant *r = variant_table_grow ();
401 r->symbol_index = symbol_index;
413 get_at_spec(unsigned symbol_index)
415 static char at_buf[20];
416 if (symbol_index == 0)
417 strcpy (at_buf, "$$");
419 snprintf (at_buf, sizeof at_buf, "$%u", symbol_index);
424 show_sub_messages (const char* cp, bool explicit_bracketing,
425 int midrule_rhs_index, char dollar_or_at,
426 bool is_warning, unsigned indent)
430 for (i = 0; i < variant_count; ++i)
432 const variant *var = &variant_table[i];
433 const char *at_spec = get_at_spec (var->symbol_index);
438 warn_at_indent (var->loc, &indent, _("refers to: %c%s at %s"),
439 dollar_or_at, var->id, at_spec);
441 complain_at_indent (var->loc, &indent, _("refers to: %c%s at %s"),
442 dollar_or_at, var->id, at_spec);
446 static struct obstack msg_buf;
447 const char *tail = explicit_bracketing ? "" :
448 cp + strlen (var->id);
449 const char *id = var->hidden_by ? var->hidden_by->id :
451 location id_loc = var->hidden_by ? var->hidden_by->loc :
454 /* Create the explanation message. */
455 obstack_init (&msg_buf);
457 obstack_fgrow1 (&msg_buf, _("possibly meant: %c"), dollar_or_at);
458 if (contains_dot_or_dash (id))
459 obstack_fgrow1 (&msg_buf, "[%s]", id);
461 obstack_sgrow (&msg_buf, id);
462 obstack_sgrow (&msg_buf, tail);
464 if (var->err & VARIANT_HIDDEN)
466 obstack_fgrow1 (&msg_buf, _(", hiding %c"), dollar_or_at);
467 if (contains_dot_or_dash (var->id))
468 obstack_fgrow1 (&msg_buf, "[%s]", var->id);
470 obstack_sgrow (&msg_buf, var->id);
471 obstack_sgrow (&msg_buf, tail);
474 obstack_fgrow1 (&msg_buf, _(" at %s"), at_spec);
476 if (var->err & VARIANT_NOT_VISIBLE_FROM_MIDRULE)
479 _(", cannot be accessed from mid-rule action at $%d");
480 obstack_fgrow1 (&msg_buf, format, midrule_rhs_index);
483 obstack_1grow (&msg_buf, '\0');
485 warn_at_indent (id_loc, &indent, "%s",
486 (char *) obstack_finish (&msg_buf));
488 complain_at_indent (id_loc, &indent, "%s",
489 (char *) obstack_finish (&msg_buf));
490 obstack_free (&msg_buf, 0);
495 /* Returned from "parse_ref" when the reference
497 #define INVALID_REF (INT_MIN)
499 /* Returned from "parse_ref" when the reference
500 points to LHS ($$) of the current rule or midrule. */
501 #define LHS_REF (INT_MIN + 1)
503 /* Sub-messages indent. */
504 #define SUB_INDENT (4)
506 /* Parse named or positional reference. In case of positional
507 references, can return negative values for $-n "deep" stack
510 parse_ref (char *cp, symbol_list *rule, int rule_length,
511 int midrule_rhs_index, char *text, location text_loc,
516 bool explicit_bracketing;
518 unsigned valid_variants = 0;
519 unsigned valid_variant_index = 0;
524 if (c_isdigit (*cp) || (*cp == '-' && c_isdigit (* (cp + 1))))
526 long int num = strtol (cp, &cp, 10);
527 if (1 - INT_MAX + rule_length <= num && num <= rule_length)
531 complain_at (text_loc, _("integer out of range: %s"),
539 /* Ignore the brackets. */
541 for (p = ++cp; *p != ']'; ++p)
545 explicit_bracketing = true;
549 /* Take all characters of the name. */
551 for (p = cp; *p; ++p)
552 if (is_dot_or_dash (*p))
557 for (p = cp; *p; ++p)
561 explicit_bracketing = false;
564 /* Add all relevant variants. */
566 unsigned symbol_index;
568 for (symbol_index = 0, l = rule; !symbol_list_null (l);
569 ++symbol_index, l = l->next)
572 if (l->content_type != SYMLIST_SYMBOL)
575 var = variant_add (l->content.sym->tag, l->sym_loc,
576 symbol_index, cp, cp_end, explicit_bracketing);
577 if (var && l->named_ref)
578 var->hidden_by = l->named_ref;
581 variant_add (l->named_ref->id, l->named_ref->loc,
582 symbol_index, cp, cp_end, explicit_bracketing);
587 for (i = 0; i < variant_count; ++i)
589 variant *var = &variant_table[i];
590 unsigned symbol_index = var->symbol_index;
592 /* Check visibility from mid-rule actions. */
593 if (midrule_rhs_index != 0
594 && (symbol_index == 0 || midrule_rhs_index < symbol_index))
595 var->err |= VARIANT_NOT_VISIBLE_FROM_MIDRULE;
597 /* Check correct bracketing. */
598 if (!explicit_bracketing && contains_dot_or_dash (var->id))
599 var->err |= VARIANT_BAD_BRACKETING;
601 /* Check using of hidden symbols. */
603 var->err |= VARIANT_HIDDEN;
607 valid_variant_index = i;
612 switch (valid_variants)
616 unsigned len = (explicit_bracketing || !ref_tail_fields) ?
617 cp_end - cp : ref_tail_fields - cp;
620 complain_at_indent (text_loc, &indent, _("invalid reference: %s"),
622 indent += SUB_INDENT;
625 location sym_loc = text_loc;
626 sym_loc.start.column += 1;
627 sym_loc.end = sym_loc.start;
629 _("syntax error after '%c', expecting integer, letter,"
630 " '_', '[', or '$'");
631 complain_at_indent (sym_loc, &indent, format, dollar_or_at);
633 else if (midrule_rhs_index)
636 _("symbol not found in production before $%d: %.*s");
637 complain_at_indent (rule->location, &indent, format,
638 midrule_rhs_index, len, cp);
643 _("symbol not found in production: %.*s");
644 complain_at_indent (rule->location, &indent, format,
648 if (variant_count > 0)
649 show_sub_messages (cp, explicit_bracketing, midrule_rhs_index,
650 dollar_or_at, false, indent);
656 if (variant_count > 1)
658 warn_at_indent (text_loc, &indent, _("misleading reference: %s"),
660 show_sub_messages (cp, explicit_bracketing, midrule_rhs_index,
661 dollar_or_at, true, indent + SUB_INDENT);
664 unsigned symbol_index =
665 variant_table[valid_variant_index].symbol_index;
666 return (symbol_index == midrule_rhs_index) ? LHS_REF : symbol_index;
673 complain_at_indent (text_loc, &indent, _("ambiguous reference: %s"),
675 show_sub_messages (cp, explicit_bracketing, midrule_rhs_index,
676 dollar_or_at, false, indent + SUB_INDENT);
685 /* Keeps track of the maximum number of semantic values to the left of
686 a handle (those referenced by $0, $-1, etc.) are required by the
687 semantic actions of this grammar. */
688 int max_left_semantic_context = 0;
691 /*------------------------------------------------------------------.
692 | TEXT is pointing to a wannabee semantic value (i.e., a '$'). |
694 | Possible inputs: $[<TYPENAME>]($|integer) |
696 | Output to OBSTACK_FOR_STRING a reference to this semantic value. |
697 `------------------------------------------------------------------*/
700 handle_action_dollar (symbol_list *rule, char *text, location dollar_loc)
702 char const *type_name = NULL;
705 symbol_list *effective_rule;
706 int effective_rule_length;
709 if (rule->midrule_parent_rule)
711 effective_rule = rule->midrule_parent_rule;
712 effective_rule_length = rule->midrule_parent_rhs_index - 1;
716 effective_rule = rule;
717 effective_rule_length = symbol_list_length (rule->next);
720 /* Get the type name if explicit. */
727 /* The '>' symbol will be later replaced by '\0'. Original
728 'text' is needed for error messages. */
731 if (untyped_var_seen)
732 complain_at (dollar_loc, _("explicit type given in untyped grammar"));
736 n = parse_ref (cp, effective_rule, effective_rule_length,
737 rule->midrule_parent_rhs_index, text, dollar_loc, '$');
749 type_name = symbol_list_n_type_name_get (rule, dollar_loc, 0);
753 if (union_seen | tag_seen)
755 if (rule->midrule_parent_rule)
756 complain_at (dollar_loc,
757 _("$$ for the midrule at $%d of '%s'"
758 " has no declared type"),
759 rule->midrule_parent_rhs_index,
760 effective_rule->content.sym->tag);
762 complain_at (dollar_loc, _("$$ of '%s' has no declared type"),
763 rule->content.sym->tag);
766 untyped_var_seen = true;
770 obstack_fgrow1 (&obstack_for_string,
771 "]b4_lhs_value([%s])[", type_name);
772 rule->action_props.is_value_used = true;
776 if (max_left_semantic_context < 1 - n)
777 max_left_semantic_context = 1 - n;
778 if (!type_name && 0 < n)
780 symbol_list_n_type_name_get (effective_rule, dollar_loc, n);
783 if (union_seen | tag_seen)
784 complain_at (dollar_loc, _("$%s of '%s' has no declared type"),
785 cp, effective_rule->content.sym->tag);
787 untyped_var_seen = true;
791 obstack_fgrow3 (&obstack_for_string,
792 "]b4_rhs_value(%d, %d, [%s])[",
793 effective_rule_length, n, type_name);
795 symbol_list_n_get (effective_rule, n)->action_props.is_value_used =
802 /*------------------------------------------------------.
803 | TEXT is a location token (i.e., a '@...'). Output to |
804 | OBSTACK_FOR_STRING a reference to this location. |
805 `------------------------------------------------------*/
808 handle_action_at (symbol_list *rule, char *text, location at_loc)
811 symbol_list *effective_rule;
812 int effective_rule_length;
815 if (rule->midrule_parent_rule)
817 effective_rule = rule->midrule_parent_rule;
818 effective_rule_length = rule->midrule_parent_rhs_index - 1;
822 effective_rule = rule;
823 effective_rule_length = symbol_list_length (rule->next);
826 locations_flag = true;
828 n = parse_ref (cp, effective_rule, effective_rule_length,
829 rule->midrule_parent_rhs_index, text, at_loc, '@');
836 obstack_sgrow (&obstack_for_string, "]b4_lhs_location[");
840 obstack_fgrow2 (&obstack_for_string, "]b4_rhs_location(%d, %d)[",
841 effective_rule_length, n);
847 /*-------------------------.
848 | Initialize the scanner. |
849 `-------------------------*/
851 /* Translate the dollars and ats in \a self, in the context \a sc_context
852 (SC_RULE_ACTION, SC_SYMBOL_ACTION, INITIAL). */
855 translate_action (code_props *self, int sc_context)
858 static bool initialized = false;
861 obstack_init (&obstack_for_string);
866 loc->start = loc->end = self->location.start;
867 yy_switch_to_buffer (yy_scan_string (self->code));
868 res = code_lex (self, sc_context);
869 yy_delete_buffer (YY_CURRENT_BUFFER);
874 /*------------------------------------------------------------------------.
875 | Implementation of the public interface as documented in "scan-code.h". |
876 `------------------------------------------------------------------------*/
879 code_props_none_init (code_props *self)
881 *self = code_props_none;
884 code_props const code_props_none = CODE_PROPS_NONE_INIT;
887 code_props_plain_init (code_props *self, char const *code,
890 self->kind = CODE_PROPS_PLAIN;
892 self->location = code_loc;
893 self->is_value_used = false;
895 self->named_ref = NULL;
899 code_props_symbol_action_init (code_props *self, char const *code,
902 self->kind = CODE_PROPS_SYMBOL_ACTION;
904 self->location = code_loc;
905 self->is_value_used = false;
907 self->named_ref = NULL;
911 code_props_rule_action_init (code_props *self, char const *code,
912 location code_loc, symbol_list *rule,
915 self->kind = CODE_PROPS_RULE_ACTION;
917 self->location = code_loc;
918 self->is_value_used = false;
920 self->named_ref = name;
924 code_props_translate_code (code_props *self)
928 case CODE_PROPS_NONE:
930 case CODE_PROPS_PLAIN:
931 self->code = translate_action (self, INITIAL);
933 case CODE_PROPS_SYMBOL_ACTION:
934 self->code = translate_action (self, SC_SYMBOL_ACTION);
936 case CODE_PROPS_RULE_ACTION:
937 self->code = translate_action (self, SC_RULE_ACTION);
943 code_scanner_last_string_free (void)
949 code_scanner_free (void)
951 obstack_free (&obstack_for_string, 0);
952 variant_table_free ();
954 /* Reclaim Flex's buffers. */