1 /* Bison Action Scanner -*- C -*-
3 Copyright (C) 2006, 2007, 2008-2009 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, "$][");
202 need_semicolon = true;
205 warn_at (*loc, _("stray `@'"));
206 obstack_sgrow (&obstack_for_string, "@@");
207 need_semicolon = true;
210 obstack_sgrow (&obstack_for_string, "@{");
211 need_semicolon = true;
214 obstack_sgrow (&obstack_for_string, "@}");
215 need_semicolon = true;
218 ";" STRING_GROW; need_semicolon = false;
219 "{" STRING_GROW; ++braces_level; need_semicolon = false;
221 bool outer_brace = --braces_level == 0;
223 /* As an undocumented Bison extension, append `;' before the last
224 brace in braced code, so that the user code can omit trailing
225 `;'. But do not append `;' if emulating Yacc, since Yacc does
227 if (outer_brace && !yacc_flag && language_prio == default_prio
228 && skeleton_prio == default_prio && need_semicolon && ! in_cpp)
230 warn_at (*loc, _("a `;' might be needed at the end of action code"));
231 warn_at (*loc, _("future versions of Bison will not add the `;'"));
232 obstack_1grow (&obstack_for_string, ';');
236 need_semicolon = false;
239 /* Preprocessing directives should only be recognized at the beginning
240 of lines, allowing whitespace including comments, but in C/C++,
241 `#' can only be the start of preprocessor directives or within
242 `#define' directives anyway, so don't bother with begin of line. */
243 "#" STRING_GROW; in_cpp = true;
245 {splice} STRING_GROW;
246 [\n\r] STRING_GROW; if (in_cpp) in_cpp = need_semicolon = false;
248 . STRING_GROW; need_semicolon = true;
254 obstack_sgrow (&obstack_for_string, "]b4_dollar_dollar[");
255 self->is_value_used = true;
258 obstack_sgrow (&obstack_for_string, "]b4_at_dollar[");
259 locations_flag = true;
264 /*-----------------------------------------.
265 | Escape M4 quoting characters in C code. |
266 `-----------------------------------------*/
270 \$ obstack_sgrow (&obstack_for_string, "$][");
271 \@ obstack_sgrow (&obstack_for_string, "@@");
272 \[ obstack_sgrow (&obstack_for_string, "@{");
273 \] obstack_sgrow (&obstack_for_string, "@}");
276 /*-----------------------------------------------------.
277 | By default, grow the string obstack with the input. |
278 `-----------------------------------------------------*/
282 /* End of processing. */
291 is_dot_or_dash (char ch)
293 return ch == '.' || ch == '-';
297 contains_dot_or_dash (const char* p)
300 if (is_dot_or_dash (*p))
305 /* Defines a variant of a symbolic name resolution. */
308 /* Index in symbol list. */
309 unsigned symbol_index;
311 /* Matched symbol id and loc. */
315 /* Hiding named reference. */
316 named_ref* hidden_by;
318 /* Error flags. May contain zero (no errors) or
319 a combination of VARIANT_* values. */
323 /* Set when the variant refers to a symbol hidden
324 by an explicit symbol reference. */
325 #define VARIANT_HIDDEN (1 << 0)
327 /* Set when the variant refers to a symbol containing
328 dots or dashes. Will require explicit bracketing. */
329 #define VARIANT_BAD_BRACKETING (1 << 1)
331 /* Set when the variant refers to a symbol which is
332 not visible from current midrule. */
333 #define VARIANT_NOT_VISIBLE_FROM_MIDRULE (1 << 2)
335 static variant *variant_table = 0;
336 static unsigned variant_table_size = 0;
337 static unsigned variant_count = 0;
340 variant_table_grow (void)
343 if (variant_count > variant_table_size)
345 while (variant_count > variant_table_size)
346 variant_table_size = 2 * variant_table_size + 3;
347 variant_table = xnrealloc (variant_table, variant_table_size,
348 sizeof *variant_table);
350 return &variant_table[variant_count - 1];
354 variant_table_free (void)
356 free (variant_table);
358 variant_table_size = variant_count = 0;
362 find_prefix_end (const char *prefix, char *begin, char *end)
366 for (; *prefix && ptr != end; ++prefix, ++ptr)
377 variant_add (uniqstr id, location id_loc, unsigned symbol_index,
378 char *cp, char *cp_end, bool explicit_bracketing)
382 prefix_end = find_prefix_end (id, cp, cp_end);
384 (prefix_end == cp_end ||
385 (!explicit_bracketing && is_dot_or_dash (*prefix_end))))
387 variant *r = variant_table_grow ();
388 r->symbol_index = symbol_index;
400 get_at_spec(unsigned symbol_index)
402 static char at_buf[20];
403 if (symbol_index == 0)
404 strcpy (at_buf, "$$");
406 snprintf (at_buf, sizeof at_buf, "$%u", symbol_index);
411 show_sub_messages (const char* cp, bool explicit_bracketing,
412 int midrule_rhs_index, char dollar_or_at,
413 bool is_warning, unsigned indent)
417 for (i = 0; i < variant_count; ++i)
419 const variant *var = &variant_table[i];
420 const char *at_spec = get_at_spec (var->symbol_index);
425 warn_at_indent (var->loc, &indent, _("refers to: %c%s at %s"),
426 dollar_or_at, var->id, at_spec);
428 complain_at_indent (var->loc, &indent, _("refers to: %c%s at %s"),
429 dollar_or_at, var->id, at_spec);
433 static struct obstack msg_buf;
434 const char *tail = explicit_bracketing ? "" :
435 cp + strlen (var->id);
436 const char *id = var->hidden_by ? var->hidden_by->id :
438 location id_loc = var->hidden_by ? var->hidden_by->loc :
441 /* Create the explanation message. */
442 obstack_init (&msg_buf);
444 obstack_fgrow1 (&msg_buf, _("possibly meant: %c"), dollar_or_at);
445 if (contains_dot_or_dash (id))
446 obstack_fgrow1 (&msg_buf, "[%s]", id);
448 obstack_sgrow (&msg_buf, id);
449 obstack_sgrow (&msg_buf, tail);
451 if (var->err & VARIANT_HIDDEN)
453 obstack_fgrow1 (&msg_buf, _(", hiding %c"), dollar_or_at);
454 if (contains_dot_or_dash (var->id))
455 obstack_fgrow1 (&msg_buf, "[%s]", var->id);
457 obstack_sgrow (&msg_buf, var->id);
458 obstack_sgrow (&msg_buf, tail);
461 obstack_fgrow1 (&msg_buf, _(" at %s"), at_spec);
463 if (var->err & VARIANT_NOT_VISIBLE_FROM_MIDRULE)
466 _(", cannot be accessed from mid-rule action at $%d");
467 obstack_fgrow1 (&msg_buf, format, midrule_rhs_index);
470 obstack_1grow (&msg_buf, '\0');
472 warn_at_indent (id_loc, &indent, "%s",
473 (char *) obstack_finish (&msg_buf));
475 complain_at_indent (id_loc, &indent, "%s",
476 (char *) obstack_finish (&msg_buf));
477 obstack_free (&msg_buf, 0);
482 /* Returned from "parse_ref" when the reference
484 #define INVALID_REF (INT_MIN)
486 /* Returned from "parse_ref" when the reference
487 points to LHS ($$) of the current rule or midrule. */
488 #define LHS_REF (INT_MIN + 1)
490 /* Sub-messages indent. */
491 #define SUB_INDENT (4)
493 /* Parse named or positional reference. In case of positional
494 references, can return negative values for $-n "deep" stack
497 parse_ref (char *cp, symbol_list *rule, int rule_length,
498 int midrule_rhs_index, char *text, location text_loc,
503 bool explicit_bracketing;
505 unsigned valid_variants = 0;
506 unsigned valid_variant_index = 0;
511 if (c_isdigit (*cp) || (*cp == '-' && c_isdigit (* (cp + 1))))
513 long int num = strtol (cp, &cp, 10);
514 if (1 - INT_MAX + rule_length <= num && num <= rule_length)
518 complain_at (text_loc, _("integer out of range: %s"),
526 /* Ignore the brackets. */
528 for (p = ++cp; *p != ']'; ++p)
532 explicit_bracketing = true;
536 /* Take all characters of the name. */
538 for (p = cp; *p; ++p)
539 if (is_dot_or_dash (*p))
544 for (p = cp; *p; ++p)
548 explicit_bracketing = false;
551 /* Add all relevant variants. */
553 unsigned symbol_index;
555 for (symbol_index = 0, l = rule; !symbol_list_null (l);
556 ++symbol_index, l = l->next)
559 if (l->content_type != SYMLIST_SYMBOL)
562 var = variant_add (l->content.sym->tag, l->sym_loc,
563 symbol_index, cp, cp_end, explicit_bracketing);
564 if (var && l->named_ref)
565 var->hidden_by = l->named_ref;
568 variant_add (l->named_ref->id, l->named_ref->loc,
569 symbol_index, cp, cp_end, explicit_bracketing);
574 for (i = 0; i < variant_count; ++i)
576 variant *var = &variant_table[i];
577 unsigned symbol_index = var->symbol_index;
579 /* Check visibility from mid-rule actions. */
580 if (midrule_rhs_index != 0
581 && (symbol_index == 0 || midrule_rhs_index < symbol_index))
582 var->err |= VARIANT_NOT_VISIBLE_FROM_MIDRULE;
584 /* Check correct bracketing. */
585 if (!explicit_bracketing && contains_dot_or_dash (var->id))
586 var->err |= VARIANT_BAD_BRACKETING;
588 /* Check using of hidden symbols. */
590 var->err |= VARIANT_HIDDEN;
594 valid_variant_index = i;
599 switch (valid_variants)
603 unsigned len = (explicit_bracketing || !ref_tail_fields) ?
604 cp_end - cp : ref_tail_fields - cp;
607 complain_at_indent (text_loc, &indent, _("invalid reference: %s"),
609 indent += SUB_INDENT;
610 if (midrule_rhs_index)
613 _("symbol not found in production before $%d: %.*s");
614 complain_at_indent (rule->location, &indent, format,
615 midrule_rhs_index, len, cp);
620 _("symbol not found in production: %.*s");
621 complain_at_indent (rule->location, &indent, format,
625 if (variant_count > 0)
626 show_sub_messages (cp, explicit_bracketing, midrule_rhs_index,
627 dollar_or_at, false, indent);
633 if (variant_count > 1)
635 warn_at_indent (text_loc, &indent, _("misleading reference: %s"),
637 show_sub_messages (cp, explicit_bracketing, midrule_rhs_index,
638 dollar_or_at, true, indent + SUB_INDENT);
641 unsigned symbol_index =
642 variant_table[valid_variant_index].symbol_index;
643 return (symbol_index == midrule_rhs_index) ? LHS_REF : symbol_index;
650 complain_at_indent (text_loc, &indent, _("ambiguous reference: %s"),
652 show_sub_messages (cp, explicit_bracketing, midrule_rhs_index,
653 dollar_or_at, false, indent + SUB_INDENT);
662 /* Keeps track of the maximum number of semantic values to the left of
663 a handle (those referenced by $0, $-1, etc.) are required by the
664 semantic actions of this grammar. */
665 int max_left_semantic_context = 0;
668 /*------------------------------------------------------------------.
669 | TEXT is pointing to a wannabee semantic value (i.e., a `$'). |
671 | Possible inputs: $[<TYPENAME>]($|integer) |
673 | Output to OBSTACK_FOR_STRING a reference to this semantic value. |
674 `------------------------------------------------------------------*/
677 handle_action_dollar (symbol_list *rule, char *text, location dollar_loc)
679 char const *type_name = NULL;
682 symbol_list *effective_rule;
683 int effective_rule_length;
686 if (rule->midrule_parent_rule)
688 effective_rule = rule->midrule_parent_rule;
689 effective_rule_length = rule->midrule_parent_rhs_index - 1;
693 effective_rule = rule;
694 effective_rule_length = symbol_list_length (rule->next);
697 /* Get the type name if explicit. */
704 /* The '>' symbol will be later replaced by '\0'. Original
705 'text' is needed for error messages. */
708 if (untyped_var_seen)
709 complain_at (dollar_loc, _("explicit type given in untyped grammar"));
713 n = parse_ref (cp, effective_rule, effective_rule_length,
714 rule->midrule_parent_rhs_index, text, dollar_loc, '$');
726 type_name = symbol_list_n_type_name_get (rule, dollar_loc, 0);
730 if (union_seen | tag_seen)
732 if (rule->midrule_parent_rule)
733 complain_at (dollar_loc,
734 _("$$ for the midrule at $%d of `%s'"
735 " has no declared type"),
736 rule->midrule_parent_rhs_index,
737 effective_rule->content.sym->tag);
739 complain_at (dollar_loc, _("$$ of `%s' has no declared type"),
740 rule->content.sym->tag);
743 untyped_var_seen = true;
747 obstack_fgrow1 (&obstack_for_string,
748 "]b4_lhs_value([%s])[", type_name);
749 rule->action_props.is_value_used = true;
753 if (max_left_semantic_context < 1 - n)
754 max_left_semantic_context = 1 - n;
755 if (!type_name && 0 < n)
757 symbol_list_n_type_name_get (effective_rule, dollar_loc, n);
760 if (union_seen | tag_seen)
761 complain_at (dollar_loc, _("$%s of `%s' has no declared type"),
762 cp, effective_rule->content.sym->tag);
764 untyped_var_seen = true;
768 obstack_fgrow3 (&obstack_for_string,
769 "]b4_rhs_value(%d, %d, [%s])[",
770 effective_rule_length, n, type_name);
772 symbol_list_n_get (effective_rule, n)->action_props.is_value_used =
779 /*------------------------------------------------------.
780 | TEXT is a location token (i.e., a `@...'). Output to |
781 | OBSTACK_FOR_STRING a reference to this location. |
782 `------------------------------------------------------*/
785 handle_action_at (symbol_list *rule, char *text, location at_loc)
788 symbol_list *effective_rule;
789 int effective_rule_length;
792 if (rule->midrule_parent_rule)
794 effective_rule = rule->midrule_parent_rule;
795 effective_rule_length = rule->midrule_parent_rhs_index - 1;
799 effective_rule = rule;
800 effective_rule_length = symbol_list_length (rule->next);
803 locations_flag = true;
805 n = parse_ref (cp, effective_rule, effective_rule_length,
806 rule->midrule_parent_rhs_index, text, at_loc, '@');
813 obstack_sgrow (&obstack_for_string, "]b4_lhs_location[");
817 obstack_fgrow2 (&obstack_for_string, "]b4_rhs_location(%d, %d)[",
818 effective_rule_length, n);
824 /*-------------------------.
825 | Initialize the scanner. |
826 `-------------------------*/
828 /* Translate the dollars and ats in \a self, in the context \a sc_context
829 (SC_RULE_ACTION, SC_SYMBOL_ACTION, INITIAL). */
832 translate_action (code_props *self, int sc_context)
835 static bool initialized = false;
838 obstack_init (&obstack_for_string);
843 loc->start = loc->end = self->location.start;
844 yy_switch_to_buffer (yy_scan_string (self->code));
845 res = code_lex (self, sc_context);
846 yy_delete_buffer (YY_CURRENT_BUFFER);
851 /*------------------------------------------------------------------------.
852 | Implementation of the public interface as documented in "scan-code.h". |
853 `------------------------------------------------------------------------*/
856 code_props_none_init (code_props *self)
858 *self = code_props_none;
861 code_props const code_props_none = CODE_PROPS_NONE_INIT;
864 code_props_plain_init (code_props *self, char const *code,
867 self->kind = CODE_PROPS_PLAIN;
869 self->location = code_loc;
870 self->is_value_used = false;
872 self->named_ref = NULL;
876 code_props_symbol_action_init (code_props *self, char const *code,
879 self->kind = CODE_PROPS_SYMBOL_ACTION;
881 self->location = code_loc;
882 self->is_value_used = false;
884 self->named_ref = NULL;
888 code_props_rule_action_init (code_props *self, char const *code,
889 location code_loc, symbol_list *rule,
892 self->kind = CODE_PROPS_RULE_ACTION;
894 self->location = code_loc;
895 self->is_value_used = false;
897 self->named_ref = name;
901 code_props_translate_code (code_props *self)
905 case CODE_PROPS_NONE:
907 case CODE_PROPS_PLAIN:
908 self->code = translate_action (self, INITIAL);
910 case CODE_PROPS_SYMBOL_ACTION:
911 self->code = translate_action (self, SC_SYMBOL_ACTION);
913 case CODE_PROPS_RULE_ACTION:
914 self->code = translate_action (self, SC_RULE_ACTION);
920 code_scanner_last_string_free (void)
926 code_scanner_free (void)
928 obstack_free (&obstack_for_string, 0);
929 variant_table_free ();
931 /* Reclaim Flex's buffers. */