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.  We allow
 
  82    "->" as a means to dereference a pointer.  */
 
  85 /* Zero or more instances of backslash-newline.  Following GCC, allow
 
  86    white space between the backslash and the newline.  */
 
  87 splice   (\\[ \f\t\v]*\n)*
 
  89 /* C style identifier. Must start with letter. Will be used for
 
  90    named symbol references. Shall be kept synchronized with
 
  91    scan-gram.l "letter" and "id". */
 
  92 letter    [.abcdefghijklmnopqrstuvwxyzABCDEFGHIJKLMNOPQRSTUVWXYZ_]
 
  93 id        {letter}({letter}|[-0-9])*
 
  94 ref      -?[0-9]+|{id}|"["{id}"]"|"$"
 
  99   /* This scanner is special: it is invoked only once, henceforth
 
 100      is expected to return only once.  This initialization is
 
 101      therefore done once per action to translate. */
 
 102   aver (sc_context == SC_SYMBOL_ACTION
 
 103         || sc_context == SC_RULE_ACTION
 
 104         || sc_context == INITIAL);
 
 108   /*------------------------------------------------------------.
 
 109   | Scanning a C comment.  The initial '/ *' is already eaten.  |
 
 110   `------------------------------------------------------------*/
 
 114   "*"{splice}"/"  STRING_GROW; BEGIN sc_context;
 
 118   /*--------------------------------------------------------------.
 
 119   | Scanning a line comment.  The initial '//' is already eaten.  |
 
 120   `--------------------------------------------------------------*/
 
 124   "\n"           STRING_GROW; BEGIN sc_context;
 
 125   {splice}       STRING_GROW;
 
 129   /*--------------------------------------------.
 
 130   | Scanning user-code characters and strings.  |
 
 131   `--------------------------------------------*/
 
 133 <SC_CHARACTER,SC_STRING>
 
 135   {splice}|\\{splice}.  STRING_GROW;
 
 140   "'"           STRING_GROW; BEGIN sc_context;
 
 145   "\""          STRING_GROW; BEGIN sc_context;
 
 149 <SC_RULE_ACTION,SC_SYMBOL_ACTION>
 
 151   "'"              STRING_GROW; BEGIN SC_CHARACTER;
 
 152   "\""             STRING_GROW; BEGIN SC_STRING;
 
 153   "/"{splice}"*"   STRING_GROW; BEGIN SC_COMMENT;
 
 154   "/"{splice}"/"   STRING_GROW; BEGIN SC_LINE_COMMENT;
 
 157     complain (loc, Wother, _("stray '%s'"), yytext);
 
 158     obstack_escape (&obstack_for_string, yytext);
 
 164   "$"("<"{tag}">")?{ref}  {
 
 165     ref_tail_fields = NULL;
 
 166     handle_action_dollar (self->rule, yytext, *loc);
 
 168       obstack_sgrow (&obstack_for_string, ref_tail_fields);
 
 171     ref_tail_fields = NULL;
 
 172     handle_action_at (self->rule, yytext, *loc);
 
 174       obstack_sgrow (&obstack_for_string, ref_tail_fields);
 
 180   "$"("<"{tag}">")?"$" {
 
 181     const char *type_name = NULL;
 
 182     fetch_type_name (yytext + 1, &type_name, *loc)[-1] = 0;
 
 183     obstack_sgrow (&obstack_for_string, "]b4_dollar_dollar(");
 
 184     obstack_quote (&obstack_for_string, type_name);
 
 185     obstack_sgrow (&obstack_for_string, ")[");
 
 186     self->is_value_used = true;
 
 189     obstack_sgrow (&obstack_for_string, "]b4_at_dollar[");
 
 190     muscle_percent_define_ensure("locations", the_location, true);
 
 197   /* Escape M4 quoting characters in C code.  */
 
 198   [$@\[\]]    obstack_escape (&obstack_for_string, yytext);
 
 200   /* By default, grow the string obstack with the input.  */
 
 203  /* End of processing. */
 
 204   <<EOF>>     STRING_FINISH; return last_string;
 
 210 is_dot_or_dash (char ch)
 
 212   return ch == '.' || ch == '-';
 
 216 contains_dot_or_dash (const char* p)
 
 219     if (is_dot_or_dash (*p))
 
 224 /* Defines a variant of a symbolic name resolution. */
 
 227   /* Index in symbol list. */
 
 228   unsigned symbol_index;
 
 230   /* Matched symbol id and loc. */
 
 234   /* Hiding named reference. */
 
 235   named_ref* hidden_by;
 
 237   /* Error flags. May contain zero (no errors) or
 
 238      a combination of VARIANT_* values. */
 
 242 /* Set when the variant refers to a symbol hidden
 
 243    by an explicit symbol reference. */
 
 244 #define VARIANT_HIDDEN (1 << 0)
 
 246 /* Set when the variant refers to a symbol containing
 
 247    dots or dashes. Will require explicit bracketing. */
 
 248 #define VARIANT_BAD_BRACKETING (1 << 1)
 
 250 /* Set when the variant refers to a symbol which is
 
 251    not visible from current midrule. */
 
 252 #define VARIANT_NOT_VISIBLE_FROM_MIDRULE (1 << 2)
 
 254 static variant *variant_table = NULL;
 
 255 static unsigned variant_table_size = 0;
 
 256 static unsigned variant_count = 0;
 
 259 variant_table_grow (void)
 
 262   if (variant_count > variant_table_size)
 
 264       while (variant_count > variant_table_size)
 
 265         variant_table_size = 2 * variant_table_size + 3;
 
 266       variant_table = xnrealloc (variant_table, variant_table_size,
 
 267                                  sizeof *variant_table);
 
 269   return &variant_table[variant_count - 1];
 
 273 variant_table_free (void)
 
 275   free (variant_table);
 
 276   variant_table = NULL;
 
 277   variant_table_size = variant_count = 0;
 
 281 find_prefix_end (const char *prefix, char *begin, char *end)
 
 285   for (; *prefix && ptr != end; ++prefix, ++ptr)
 
 296 variant_add (uniqstr id, location id_loc, unsigned symbol_index,
 
 297              char *cp, char *cp_end, bool explicit_bracketing)
 
 301   prefix_end = find_prefix_end (id, cp, cp_end);
 
 303       (prefix_end == cp_end ||
 
 304        (!explicit_bracketing && is_dot_or_dash (*prefix_end))))
 
 306       variant *r = variant_table_grow ();
 
 307       r->symbol_index = symbol_index;
 
 319 get_at_spec(unsigned symbol_index)
 
 321   static char at_buf[20];
 
 322   if (symbol_index == 0)
 
 323     strcpy (at_buf, "$$");
 
 325     snprintf (at_buf, sizeof at_buf, "$%u", symbol_index);
 
 330 show_sub_message (warnings warning,
 
 331                   const char* cp, bool explicit_bracketing,
 
 332                   int midrule_rhs_index, char dollar_or_at,
 
 333                   unsigned indent, const variant *var)
 
 335   const char *at_spec = get_at_spec (var->symbol_index);
 
 338     complain_indent (&var->loc, warning, &indent,
 
 339                      _("refers to: %c%s at %s"), dollar_or_at,
 
 343       static struct obstack msg_buf;
 
 344       const char *tail = explicit_bracketing ? "" : cp + strlen (var->id);
 
 345       const char *id = var->hidden_by ? var->hidden_by->id : var->id;
 
 346       location id_loc = var->hidden_by ? var->hidden_by->loc : var->loc;
 
 348       /* Create the explanation message. */
 
 349       obstack_init (&msg_buf);
 
 351       obstack_printf (&msg_buf, _("possibly meant: %c"), dollar_or_at);
 
 352       if (contains_dot_or_dash (id))
 
 353         obstack_printf (&msg_buf, "[%s]", id);
 
 355         obstack_sgrow (&msg_buf, id);
 
 356       obstack_sgrow (&msg_buf, tail);
 
 358       if (var->err & VARIANT_HIDDEN)
 
 360           obstack_printf (&msg_buf, _(", hiding %c"), dollar_or_at);
 
 361           if (contains_dot_or_dash (var->id))
 
 362             obstack_printf (&msg_buf, "[%s]", var->id);
 
 364             obstack_sgrow (&msg_buf, var->id);
 
 365           obstack_sgrow (&msg_buf, tail);
 
 368       obstack_printf (&msg_buf, _(" at %s"), at_spec);
 
 370       if (var->err & VARIANT_NOT_VISIBLE_FROM_MIDRULE)
 
 371         obstack_printf (&msg_buf,
 
 372                         _(", cannot be accessed from mid-rule action at $%d"),
 
 375       complain_indent (&id_loc, warning, &indent, "%s",
 
 376                         obstack_finish0 (&msg_buf));
 
 377       obstack_free (&msg_buf, 0);
 
 382 show_sub_messages (warnings warning,
 
 383                    const char* cp, bool explicit_bracketing,
 
 384                    int midrule_rhs_index, char dollar_or_at,
 
 389   for (i = 0; i < variant_count; ++i)
 
 390     show_sub_message (warning | silent,
 
 391                       cp, explicit_bracketing,
 
 392                       midrule_rhs_index, dollar_or_at,
 
 393                       indent, &variant_table[i]);
 
 396 /* Returned from "parse_ref" when the reference
 
 398 #define INVALID_REF (INT_MIN)
 
 400 /* Returned from "parse_ref" when the reference
 
 401    points to LHS ($$) of the current rule or midrule. */
 
 402 #define LHS_REF (INT_MIN + 1)
 
 404 /* Parse named or positional reference. In case of positional
 
 405    references, can return negative values for $-n "deep" stack
 
 408 parse_ref (char *cp, symbol_list *rule, int rule_length,
 
 409            int midrule_rhs_index, char *text, location text_loc,
 
 414   bool explicit_bracketing;
 
 416   unsigned valid_variants = 0;
 
 417   unsigned valid_variant_index = 0;
 
 422   if (c_isdigit (*cp) || (*cp == '-' && c_isdigit (* (cp + 1))))
 
 424       long int num = strtol (cp, &cp, 10);
 
 425       if (1 - INT_MAX + rule_length <= num && num <= rule_length)
 
 429           complain (&text_loc, complaint, _("integer out of range: %s"),
 
 437       /* Ignore the brackets. */
 
 439       for (p = ++cp; *p != ']'; ++p)
 
 443       explicit_bracketing = true;
 
 447       /* Take all characters of the name. */
 
 449       for (p = cp; *p; ++p)
 
 450         if (is_dot_or_dash (*p))
 
 455       for (p = cp; *p; ++p)
 
 459       explicit_bracketing = false;
 
 462   /* Add all relevant variants. */
 
 464     unsigned symbol_index;
 
 466     for (symbol_index = 0, l = rule; !symbol_list_null (l);
 
 467          ++symbol_index, l = l->next)
 
 470         if (l->content_type != SYMLIST_SYMBOL)
 
 473         var = variant_add (l->content.sym->tag, l->sym_loc,
 
 474                            symbol_index, cp, cp_end, explicit_bracketing);
 
 475         if (var && l->named_ref)
 
 476           var->hidden_by = l->named_ref;
 
 479           variant_add (l->named_ref->id, l->named_ref->loc,
 
 480                        symbol_index, cp, cp_end, explicit_bracketing);
 
 485   for (i = 0; i < variant_count; ++i)
 
 487       variant *var = &variant_table[i];
 
 488       unsigned symbol_index = var->symbol_index;
 
 490       /* Check visibility from mid-rule actions. */
 
 491       if (midrule_rhs_index != 0
 
 492           && (symbol_index == 0 || midrule_rhs_index < symbol_index))
 
 493         var->err |= VARIANT_NOT_VISIBLE_FROM_MIDRULE;
 
 495       /* Check correct bracketing. */
 
 496       if (!explicit_bracketing && contains_dot_or_dash (var->id))
 
 497         var->err |= VARIANT_BAD_BRACKETING;
 
 499       /* Check using of hidden symbols. */
 
 501         var->err |= VARIANT_HIDDEN;
 
 505           valid_variant_index = i;
 
 510   switch (valid_variants)
 
 514         unsigned len = (explicit_bracketing || !ref_tail_fields) ?
 
 515           cp_end - cp : ref_tail_fields - cp;
 
 518         complain_indent (&text_loc, complaint, &indent,
 
 519                          _("invalid reference: %s"), quote (text));
 
 520         indent += SUB_INDENT;
 
 523             location sym_loc = text_loc;
 
 524             sym_loc.start.column += 1;
 
 525             sym_loc.end = sym_loc.start;
 
 526             complain_indent (&sym_loc, complaint, &indent,
 
 527                              _("syntax error after '%c', expecting integer, "
 
 528                                "letter, '_', '[', or '$'"),
 
 531         else if (midrule_rhs_index)
 
 532           complain_indent (&rule->location, complaint, &indent,
 
 533                            _("symbol not found in production before $%d: "
 
 535                            midrule_rhs_index, len, cp);
 
 537           complain_indent (&rule->location, complaint, &indent,
 
 538                            _("symbol not found in production: %.*s"),
 
 541         if (variant_count > 0)
 
 542           show_sub_messages (complaint,
 
 543                              cp, explicit_bracketing, midrule_rhs_index,
 
 544                              dollar_or_at, indent);
 
 550         if (variant_count > 1)
 
 552             complain_indent (&text_loc, Wother, &indent,
 
 553                              _("misleading reference: %s"), quote (text));
 
 554             show_sub_messages (Wother,
 
 555                                cp, explicit_bracketing, midrule_rhs_index,
 
 556                                dollar_or_at, indent + SUB_INDENT);
 
 559           unsigned symbol_index =
 
 560             variant_table[valid_variant_index].symbol_index;
 
 561           return (symbol_index == midrule_rhs_index) ? LHS_REF : symbol_index;
 
 568         complain_indent (&text_loc, complaint, &indent,
 
 569                          _("ambiguous reference: %s"), quote (text));
 
 570         show_sub_messages (complaint,
 
 571                            cp, explicit_bracketing, midrule_rhs_index,
 
 572                            dollar_or_at, indent + SUB_INDENT);
 
 581 /* Keeps track of the maximum number of semantic values to the left of
 
 582    a handle (those referenced by $0, $-1, etc.) are required by the
 
 583    semantic actions of this grammar. */
 
 584 int max_left_semantic_context = 0;
 
 587 /* If CP points to a typename (i.e., <.*?>), set TYPE_NAME to its
 
 588    beginning (i.e., after the opening "<", and return the pointer
 
 589    immediately after it.  */
 
 593 fetch_type_name (char *cp, char const **type_name,
 
 599       /* Series of non-'>' or "->".  */
 
 600       while (*cp != '>' || cp[-1] == '-')
 
 603       /* The '>' symbol will be later replaced by '\0'. Original
 
 604          'text' is needed for error messages. */
 
 606       if (untyped_var_seen)
 
 607         complain (&dollar_loc, complaint,
 
 608                   _("explicit type given in untyped grammar"));
 
 614 /*------------------------------------------------------------------.
 
 615 | TEXT is pointing to a wannabee semantic value (i.e., a '$').      |
 
 617 | Possible inputs: $[<TYPENAME>]($|integer)                         |
 
 619 | Output to OBSTACK_FOR_STRING a reference to this semantic value.  |
 
 620 `------------------------------------------------------------------*/
 
 623 handle_action_dollar (symbol_list *rule, char *text, location dollar_loc)
 
 625   char const *type_name = NULL;
 
 627   symbol_list *effective_rule;
 
 628   int effective_rule_length;
 
 631   if (rule->midrule_parent_rule)
 
 633       effective_rule = rule->midrule_parent_rule;
 
 634       effective_rule_length = rule->midrule_parent_rhs_index - 1;
 
 638       effective_rule = rule;
 
 639       effective_rule_length = symbol_list_length (rule->next);
 
 642   /* Get the type name if explicit. */
 
 643   cp = fetch_type_name (cp, &type_name, dollar_loc);
 
 645   n = parse_ref (cp, effective_rule, effective_rule_length,
 
 646                  rule->midrule_parent_rhs_index, text, dollar_loc, '$');
 
 659         type_name = symbol_list_n_type_name_get (rule, dollar_loc, 0);
 
 663           if (union_seen | tag_seen)
 
 665               if (rule->midrule_parent_rule)
 
 666                 complain (&dollar_loc, complaint,
 
 667                              _("$$ for the midrule at $%d of %s"
 
 668                                " has no declared type"),
 
 669                              rule->midrule_parent_rhs_index,
 
 670                              quote (effective_rule->content.sym->tag));
 
 672                 complain (&dollar_loc, complaint,
 
 673                           _("$$ of %s has no declared type"),
 
 674                           quote (rule->content.sym->tag));
 
 677             untyped_var_seen = true;
 
 680       obstack_sgrow (&obstack_for_string, "]b4_lhs_value(");
 
 681       obstack_quote (&obstack_for_string, type_name);
 
 682       obstack_sgrow (&obstack_for_string, ")[");
 
 683       rule->action_props.is_value_used = true;
 
 687       if (max_left_semantic_context < 1 - n)
 
 688         max_left_semantic_context = 1 - n;
 
 689       if (!type_name && 0 < n)
 
 691           symbol_list_n_type_name_get (effective_rule, dollar_loc, n);
 
 694           if (union_seen | tag_seen)
 
 695             complain (&dollar_loc, complaint,
 
 696                       _("$%s of %s has no declared type"), cp,
 
 697                       quote (effective_rule->content.sym->tag));
 
 699             untyped_var_seen = true;
 
 702       obstack_printf (&obstack_for_string,
 
 703                       "]b4_rhs_value(%d, %d, ", effective_rule_length, n);
 
 704       obstack_quote (&obstack_for_string, type_name);
 
 705       obstack_sgrow (&obstack_for_string, ")[");
 
 707         symbol_list_n_get (effective_rule, n)->action_props.is_value_used =
 
 714 /*------------------------------------------------------.
 
 715 | TEXT is a location token (i.e., a '@...').  Output to |
 
 716 | OBSTACK_FOR_STRING a reference to this location.      |
 
 717 `------------------------------------------------------*/
 
 720 handle_action_at (symbol_list *rule, char *text, location at_loc)
 
 723   symbol_list *effective_rule;
 
 724   int effective_rule_length;
 
 727   if (rule->midrule_parent_rule)
 
 729       effective_rule = rule->midrule_parent_rule;
 
 730       effective_rule_length = rule->midrule_parent_rhs_index - 1;
 
 734       effective_rule = rule;
 
 735       effective_rule_length = symbol_list_length (rule->next);
 
 738   muscle_percent_define_ensure("locations", at_loc, true);
 
 740   n = parse_ref (cp, effective_rule, effective_rule_length,
 
 741                        rule->midrule_parent_rhs_index, text, at_loc, '@');
 
 748       obstack_sgrow (&obstack_for_string, "]b4_lhs_location[");
 
 752       obstack_printf (&obstack_for_string, "]b4_rhs_location(%d, %d)[",
 
 753                       effective_rule_length, n);
 
 759 /*-------------------------.
 
 760 | Initialize the scanner.  |
 
 761 `-------------------------*/
 
 763 /* Translate the dollars and ats in \a self, in the context \a sc_context
 
 764    (SC_RULE_ACTION, SC_SYMBOL_ACTION, INITIAL).  */
 
 767 translate_action (code_props *self, int sc_context)
 
 770   static bool initialized = false;
 
 773       obstack_init (&obstack_for_string);
 
 778   loc->start = loc->end = self->location.start;
 
 779   yy_switch_to_buffer (yy_scan_string (self->code));
 
 780   res = code_lex (self, sc_context);
 
 781   yy_delete_buffer (YY_CURRENT_BUFFER);
 
 786 /*------------------------------------------------------------------------.
 
 787 | Implementation of the public interface as documented in "scan-code.h".  |
 
 788 `------------------------------------------------------------------------*/
 
 791 code_props_none_init (code_props *self)
 
 793   *self = code_props_none;
 
 796 code_props code_props_none = CODE_PROPS_NONE_INIT;
 
 799 code_props_plain_init (code_props *self, char const *code,
 
 802   code_props_none_init (self);
 
 803   self->kind = CODE_PROPS_PLAIN;
 
 805   self->location = code_loc;
 
 809 code_props_symbol_action_init (code_props *self, char const *code,
 
 812   code_props_none_init (self);
 
 813   self->kind = CODE_PROPS_SYMBOL_ACTION;
 
 815   self->location = code_loc;
 
 819 code_props_rule_action_init (code_props *self, char const *code,
 
 820                              location code_loc, symbol_list *rule,
 
 821                              named_ref *name, bool is_predicate)
 
 823   code_props_none_init (self);
 
 824   self->kind = CODE_PROPS_RULE_ACTION;
 
 826   self->location = code_loc;
 
 828   self->named_ref = name;
 
 829   self->is_predicate = is_predicate;
 
 833 code_props_translate_code (code_props *self)
 
 837       case CODE_PROPS_NONE:
 
 839       case CODE_PROPS_PLAIN:
 
 840         self->code = translate_action (self, INITIAL);
 
 842       case CODE_PROPS_SYMBOL_ACTION:
 
 843         self->code = translate_action (self, SC_SYMBOL_ACTION);
 
 845       case CODE_PROPS_RULE_ACTION:
 
 846         self->code = translate_action (self, SC_RULE_ACTION);
 
 852 code_scanner_last_string_free (void)
 
 858 code_scanner_free (void)
 
 860   obstack_free (&obstack_for_string, 0);
 
 861   variant_table_free ();
 
 863   /* Reclaim Flex's buffers.  */