1 /* Bison Action Scanner -*- C -*-
3 Copyright (C) 2006-2013 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/>. */
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 char *fetch_type_name (char *cp, char const **type_name,
53 static void handle_action_dollar (symbol_list *rule, char *cp,
55 static void handle_action_at (symbol_list *rule, char *cp, location at_loc);
57 /* A string to be pushed to obstack after dollar/at has been handled. */
58 static char *ref_tail_fields;
60 static location the_location;
61 static location *loc = &the_location;
63 /* A string representing the most recent translation. */
64 static char *last_string;
66 /* True if an untyped $$ or $n was seen. */
67 static bool untyped_var_seen;
70 /* C and C++ comments in code. */
71 %x SC_COMMENT SC_LINE_COMMENT
72 /* Strings and characters in code. */
73 %x SC_STRING SC_CHARACTER
74 /* Whether in a rule or symbol action. Specifies the translation
76 %x SC_RULE_ACTION SC_SYMBOL_ACTION
79 /* POSIX says that a tag must be both an id and a C union member, but
80 historically almost any character is allowed in a tag. We disallow
81 NUL and newline, as this simplifies our implementation. */
84 /* Zero or more instances of backslash-newline. Following GCC, allow
85 white space between the backslash and the newline. */
86 splice (\\[ \f\t\v]*\n)*
88 /* C style identifier. Must start with letter. Will be used for
89 named symbol references. Shall be kept synchronized with
90 scan-gram.l "letter" and "id". */
91 letter [.abcdefghijklmnopqrstuvwxyzABCDEFGHIJKLMNOPQRSTUVWXYZ_]
92 id {letter}({letter}|[-0-9])*
93 ref -?[0-9]+|{id}|"["{id}"]"|"$"
98 /* This scanner is special: it is invoked only once, henceforth
99 is expected to return only once. This initialization is
100 therefore done once per action to translate. */
101 aver (sc_context == SC_SYMBOL_ACTION
102 || sc_context == SC_RULE_ACTION
103 || sc_context == INITIAL);
107 /*------------------------------------------------------------.
108 | Scanning a C comment. The initial '/ *' is already eaten. |
109 `------------------------------------------------------------*/
113 "*"{splice}"/" STRING_GROW; BEGIN sc_context;
117 /*--------------------------------------------------------------.
118 | Scanning a line comment. The initial '//' is already eaten. |
119 `--------------------------------------------------------------*/
123 "\n" STRING_GROW; BEGIN sc_context;
124 {splice} STRING_GROW;
128 /*--------------------------------------------.
129 | Scanning user-code characters and strings. |
130 `--------------------------------------------*/
132 <SC_CHARACTER,SC_STRING>
134 {splice}|\\{splice}. STRING_GROW;
139 "'" STRING_GROW; BEGIN sc_context;
144 "\"" STRING_GROW; BEGIN sc_context;
148 <SC_RULE_ACTION,SC_SYMBOL_ACTION>
150 "'" STRING_GROW; BEGIN SC_CHARACTER;
151 "\"" STRING_GROW; BEGIN SC_STRING;
152 "/"{splice}"*" STRING_GROW; BEGIN SC_COMMENT;
153 "/"{splice}"/" STRING_GROW; BEGIN SC_LINE_COMMENT;
156 complain (loc, Wother, _("stray '%s'"), yytext);
157 obstack_escape (&obstack_for_string, yytext);
163 "$"("<"{tag}">")?{ref} {
164 ref_tail_fields = NULL;
165 handle_action_dollar (self->rule, yytext, *loc);
167 obstack_sgrow (&obstack_for_string, ref_tail_fields);
170 ref_tail_fields = NULL;
171 handle_action_at (self->rule, yytext, *loc);
173 obstack_sgrow (&obstack_for_string, ref_tail_fields);
179 "$"("<"{tag}">")?"$" {
180 const char *type_name = NULL;
181 fetch_type_name (yytext + 1, &type_name, *loc)[-1] = 0;
182 obstack_sgrow (&obstack_for_string, "]b4_dollar_dollar(");
183 obstack_quote (&obstack_for_string, type_name);
184 obstack_sgrow (&obstack_for_string, ")[");
185 self->is_value_used = true;
188 obstack_sgrow (&obstack_for_string, "]b4_at_dollar[");
189 muscle_percent_define_ensure("locations", the_location, true);
196 /* Escape M4 quoting characters in C code. */
197 [$@\[\]] obstack_escape (&obstack_for_string, yytext);
199 /* By default, grow the string obstack with the input. */
202 /* End of processing. */
203 <<EOF>> STRING_FINISH; return last_string;
209 is_dot_or_dash (char ch)
211 return ch == '.' || ch == '-';
215 contains_dot_or_dash (const char* p)
218 if (is_dot_or_dash (*p))
223 /* Defines a variant of a symbolic name resolution. */
226 /* Index in symbol list. */
227 unsigned symbol_index;
229 /* Matched symbol id and loc. */
233 /* Hiding named reference. */
234 named_ref* hidden_by;
236 /* Error flags. May contain zero (no errors) or
237 a combination of VARIANT_* values. */
241 /* Set when the variant refers to a symbol hidden
242 by an explicit symbol reference. */
243 #define VARIANT_HIDDEN (1 << 0)
245 /* Set when the variant refers to a symbol containing
246 dots or dashes. Will require explicit bracketing. */
247 #define VARIANT_BAD_BRACKETING (1 << 1)
249 /* Set when the variant refers to a symbol which is
250 not visible from current midrule. */
251 #define VARIANT_NOT_VISIBLE_FROM_MIDRULE (1 << 2)
253 static variant *variant_table = NULL;
254 static unsigned variant_table_size = 0;
255 static unsigned variant_count = 0;
258 variant_table_grow (void)
261 if (variant_count > variant_table_size)
263 while (variant_count > variant_table_size)
264 variant_table_size = 2 * variant_table_size + 3;
265 variant_table = xnrealloc (variant_table, variant_table_size,
266 sizeof *variant_table);
268 return &variant_table[variant_count - 1];
272 variant_table_free (void)
274 free (variant_table);
275 variant_table = NULL;
276 variant_table_size = variant_count = 0;
280 find_prefix_end (const char *prefix, char *begin, char *end)
284 for (; *prefix && ptr != end; ++prefix, ++ptr)
295 variant_add (uniqstr id, location id_loc, unsigned symbol_index,
296 char *cp, char *cp_end, bool explicit_bracketing)
300 prefix_end = find_prefix_end (id, cp, cp_end);
302 (prefix_end == cp_end ||
303 (!explicit_bracketing && is_dot_or_dash (*prefix_end))))
305 variant *r = variant_table_grow ();
306 r->symbol_index = symbol_index;
318 get_at_spec(unsigned symbol_index)
320 static char at_buf[20];
321 if (symbol_index == 0)
322 strcpy (at_buf, "$$");
324 snprintf (at_buf, sizeof at_buf, "$%u", symbol_index);
329 show_sub_message (warnings warning,
330 const char* cp, bool explicit_bracketing,
331 int midrule_rhs_index, char dollar_or_at,
332 unsigned indent, const variant *var)
334 const char *at_spec = get_at_spec (var->symbol_index);
337 complain_indent (&var->loc, warning, &indent,
338 _("refers to: %c%s at %s"), dollar_or_at,
342 static struct obstack msg_buf;
343 const char *tail = explicit_bracketing ? "" : cp + strlen (var->id);
344 const char *id = var->hidden_by ? var->hidden_by->id : var->id;
345 location id_loc = var->hidden_by ? var->hidden_by->loc : var->loc;
347 /* Create the explanation message. */
348 obstack_init (&msg_buf);
350 obstack_printf (&msg_buf, _("possibly meant: %c"), dollar_or_at);
351 if (contains_dot_or_dash (id))
352 obstack_printf (&msg_buf, "[%s]", id);
354 obstack_sgrow (&msg_buf, id);
355 obstack_sgrow (&msg_buf, tail);
357 if (var->err & VARIANT_HIDDEN)
359 obstack_printf (&msg_buf, _(", hiding %c"), dollar_or_at);
360 if (contains_dot_or_dash (var->id))
361 obstack_printf (&msg_buf, "[%s]", var->id);
363 obstack_sgrow (&msg_buf, var->id);
364 obstack_sgrow (&msg_buf, tail);
367 obstack_printf (&msg_buf, _(" at %s"), at_spec);
369 if (var->err & VARIANT_NOT_VISIBLE_FROM_MIDRULE)
370 obstack_printf (&msg_buf,
371 _(", cannot be accessed from mid-rule action at $%d"),
374 complain_indent (&id_loc, warning, &indent, "%s",
375 obstack_finish0 (&msg_buf));
376 obstack_free (&msg_buf, 0);
381 show_sub_messages (warnings warning,
382 const char* cp, bool explicit_bracketing,
383 int midrule_rhs_index, char dollar_or_at,
388 for (i = 0; i < variant_count; ++i)
389 show_sub_message (warning | silent,
390 cp, explicit_bracketing,
391 midrule_rhs_index, dollar_or_at,
392 indent, &variant_table[i]);
395 /* Returned from "parse_ref" when the reference
397 #define INVALID_REF (INT_MIN)
399 /* Returned from "parse_ref" when the reference
400 points to LHS ($$) of the current rule or midrule. */
401 #define LHS_REF (INT_MIN + 1)
403 /* Parse named or positional reference. In case of positional
404 references, can return negative values for $-n "deep" stack
407 parse_ref (char *cp, symbol_list *rule, int rule_length,
408 int midrule_rhs_index, char *text, location text_loc,
413 bool explicit_bracketing;
415 unsigned valid_variants = 0;
416 unsigned valid_variant_index = 0;
421 if (c_isdigit (*cp) || (*cp == '-' && c_isdigit (* (cp + 1))))
423 long int num = strtol (cp, &cp, 10);
424 if (1 - INT_MAX + rule_length <= num && num <= rule_length)
428 complain (&text_loc, complaint, _("integer out of range: %s"),
436 /* Ignore the brackets. */
438 for (p = ++cp; *p != ']'; ++p)
442 explicit_bracketing = true;
446 /* Take all characters of the name. */
448 for (p = cp; *p; ++p)
449 if (is_dot_or_dash (*p))
454 for (p = cp; *p; ++p)
458 explicit_bracketing = false;
461 /* Add all relevant variants. */
463 unsigned symbol_index;
465 for (symbol_index = 0, l = rule; !symbol_list_null (l);
466 ++symbol_index, l = l->next)
469 if (l->content_type != SYMLIST_SYMBOL)
472 var = variant_add (l->content.sym->tag, l->sym_loc,
473 symbol_index, cp, cp_end, explicit_bracketing);
474 if (var && l->named_ref)
475 var->hidden_by = l->named_ref;
478 variant_add (l->named_ref->id, l->named_ref->loc,
479 symbol_index, cp, cp_end, explicit_bracketing);
484 for (i = 0; i < variant_count; ++i)
486 variant *var = &variant_table[i];
487 unsigned symbol_index = var->symbol_index;
489 /* Check visibility from mid-rule actions. */
490 if (midrule_rhs_index != 0
491 && (symbol_index == 0 || midrule_rhs_index < symbol_index))
492 var->err |= VARIANT_NOT_VISIBLE_FROM_MIDRULE;
494 /* Check correct bracketing. */
495 if (!explicit_bracketing && contains_dot_or_dash (var->id))
496 var->err |= VARIANT_BAD_BRACKETING;
498 /* Check using of hidden symbols. */
500 var->err |= VARIANT_HIDDEN;
504 valid_variant_index = i;
509 switch (valid_variants)
513 unsigned len = (explicit_bracketing || !ref_tail_fields) ?
514 cp_end - cp : ref_tail_fields - cp;
517 complain_indent (&text_loc, complaint, &indent,
518 _("invalid reference: %s"), quote (text));
519 indent += SUB_INDENT;
522 location sym_loc = text_loc;
523 sym_loc.start.column += 1;
524 sym_loc.end = sym_loc.start;
525 complain_indent (&sym_loc, complaint, &indent,
526 _("syntax error after '%c', expecting integer, "
527 "letter, '_', '[', or '$'"),
530 else if (midrule_rhs_index)
531 complain_indent (&rule->location, complaint, &indent,
532 _("symbol not found in production before $%d: "
534 midrule_rhs_index, len, cp);
536 complain_indent (&rule->location, complaint, &indent,
537 _("symbol not found in production: %.*s"),
540 if (variant_count > 0)
541 show_sub_messages (complaint,
542 cp, explicit_bracketing, midrule_rhs_index,
543 dollar_or_at, indent);
549 if (variant_count > 1)
551 complain_indent (&text_loc, Wother, &indent,
552 _("misleading reference: %s"), quote (text));
553 show_sub_messages (Wother,
554 cp, explicit_bracketing, midrule_rhs_index,
555 dollar_or_at, indent + SUB_INDENT);
558 unsigned symbol_index =
559 variant_table[valid_variant_index].symbol_index;
560 return (symbol_index == midrule_rhs_index) ? LHS_REF : symbol_index;
567 complain_indent (&text_loc, complaint, &indent,
568 _("ambiguous reference: %s"), quote (text));
569 show_sub_messages (complaint,
570 cp, explicit_bracketing, midrule_rhs_index,
571 dollar_or_at, indent + SUB_INDENT);
580 /* Keeps track of the maximum number of semantic values to the left of
581 a handle (those referenced by $0, $-1, etc.) are required by the
582 semantic actions of this grammar. */
583 int max_left_semantic_context = 0;
586 /* If CP points to a typename (i.e., <.*?>), set TYPE_NAME to its
587 beginning (i.e., after the opening "<", and return the pointer
588 immediately after it. */
592 fetch_type_name (char *cp, char const **type_name,
601 /* The '>' symbol will be later replaced by '\0'. Original
602 'text' is needed for error messages. */
604 if (untyped_var_seen)
605 complain (&dollar_loc, complaint,
606 _("explicit type given in untyped grammar"));
612 /*------------------------------------------------------------------.
613 | TEXT is pointing to a wannabee semantic value (i.e., a '$'). |
615 | Possible inputs: $[<TYPENAME>]($|integer) |
617 | Output to OBSTACK_FOR_STRING a reference to this semantic value. |
618 `------------------------------------------------------------------*/
621 handle_action_dollar (symbol_list *rule, char *text, location dollar_loc)
623 char const *type_name = NULL;
625 symbol_list *effective_rule;
626 int effective_rule_length;
629 if (rule->midrule_parent_rule)
631 effective_rule = rule->midrule_parent_rule;
632 effective_rule_length = rule->midrule_parent_rhs_index - 1;
636 effective_rule = rule;
637 effective_rule_length = symbol_list_length (rule->next);
640 /* Get the type name if explicit. */
641 cp = fetch_type_name (cp, &type_name, dollar_loc);
643 n = parse_ref (cp, effective_rule, effective_rule_length,
644 rule->midrule_parent_rhs_index, text, dollar_loc, '$');
657 type_name = symbol_list_n_type_name_get (rule, dollar_loc, 0);
661 if (union_seen | tag_seen)
663 if (rule->midrule_parent_rule)
664 complain (&dollar_loc, complaint,
665 _("$$ for the midrule at $%d of %s"
666 " has no declared type"),
667 rule->midrule_parent_rhs_index,
668 quote (effective_rule->content.sym->tag));
670 complain (&dollar_loc, complaint,
671 _("$$ of %s has no declared type"),
672 quote (rule->content.sym->tag));
675 untyped_var_seen = true;
678 obstack_sgrow (&obstack_for_string, "]b4_lhs_value(");
679 obstack_quote (&obstack_for_string, type_name);
680 obstack_sgrow (&obstack_for_string, ")[");
681 rule->action_props.is_value_used = true;
685 if (max_left_semantic_context < 1 - n)
686 max_left_semantic_context = 1 - n;
687 if (!type_name && 0 < n)
689 symbol_list_n_type_name_get (effective_rule, dollar_loc, n);
692 if (union_seen | tag_seen)
693 complain (&dollar_loc, complaint,
694 _("$%s of %s has no declared type"), cp,
695 quote (effective_rule->content.sym->tag));
697 untyped_var_seen = true;
700 obstack_printf (&obstack_for_string,
701 "]b4_rhs_value(%d, %d, ", effective_rule_length, n);
702 obstack_quote (&obstack_for_string, type_name);
703 obstack_sgrow (&obstack_for_string, ")[");
705 symbol_list_n_get (effective_rule, n)->action_props.is_value_used =
712 /*------------------------------------------------------.
713 | TEXT is a location token (i.e., a '@...'). Output to |
714 | OBSTACK_FOR_STRING a reference to this location. |
715 `------------------------------------------------------*/
718 handle_action_at (symbol_list *rule, char *text, location at_loc)
721 symbol_list *effective_rule;
722 int effective_rule_length;
725 if (rule->midrule_parent_rule)
727 effective_rule = rule->midrule_parent_rule;
728 effective_rule_length = rule->midrule_parent_rhs_index - 1;
732 effective_rule = rule;
733 effective_rule_length = symbol_list_length (rule->next);
736 muscle_percent_define_ensure("locations", at_loc, true);
738 n = parse_ref (cp, effective_rule, effective_rule_length,
739 rule->midrule_parent_rhs_index, text, at_loc, '@');
746 obstack_sgrow (&obstack_for_string, "]b4_lhs_location[");
750 obstack_printf (&obstack_for_string, "]b4_rhs_location(%d, %d)[",
751 effective_rule_length, n);
757 /*-------------------------.
758 | Initialize the scanner. |
759 `-------------------------*/
761 /* Translate the dollars and ats in \a self, in the context \a sc_context
762 (SC_RULE_ACTION, SC_SYMBOL_ACTION, INITIAL). */
765 translate_action (code_props *self, int sc_context)
768 static bool initialized = false;
771 obstack_init (&obstack_for_string);
776 loc->start = loc->end = self->location.start;
777 yy_switch_to_buffer (yy_scan_string (self->code));
778 res = code_lex (self, sc_context);
779 yy_delete_buffer (YY_CURRENT_BUFFER);
784 /*------------------------------------------------------------------------.
785 | Implementation of the public interface as documented in "scan-code.h". |
786 `------------------------------------------------------------------------*/
789 code_props_none_init (code_props *self)
791 *self = code_props_none;
794 code_props code_props_none = CODE_PROPS_NONE_INIT;
797 code_props_plain_init (code_props *self, char const *code,
800 code_props_none_init (self);
801 self->kind = CODE_PROPS_PLAIN;
803 self->location = code_loc;
807 code_props_symbol_action_init (code_props *self, char const *code,
810 code_props_none_init (self);
811 self->kind = CODE_PROPS_SYMBOL_ACTION;
813 self->location = code_loc;
817 code_props_rule_action_init (code_props *self, char const *code,
818 location code_loc, symbol_list *rule,
819 named_ref *name, bool is_predicate)
821 code_props_none_init (self);
822 self->kind = CODE_PROPS_RULE_ACTION;
824 self->location = code_loc;
826 self->named_ref = name;
827 self->is_predicate = is_predicate;
831 code_props_translate_code (code_props *self)
835 case CODE_PROPS_NONE:
837 case CODE_PROPS_PLAIN:
838 self->code = translate_action (self, INITIAL);
840 case CODE_PROPS_SYMBOL_ACTION:
841 self->code = translate_action (self, SC_SYMBOL_ACTION);
843 case CODE_PROPS_RULE_ACTION:
844 self->code = translate_action (self, SC_RULE_ACTION);
850 code_scanner_last_string_free (void)
856 code_scanner_free (void)
858 obstack_free (&obstack_for_string, 0);
859 variant_table_free ();
861 /* Reclaim Flex's buffers. */