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 <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;
 
 249   .         STRING_GROW; need_semicolon = true;
 
 255     obstack_sgrow (&obstack_for_string, "]b4_dollar_dollar[");
 
 256     self->is_value_used = true;
 
 259     obstack_sgrow (&obstack_for_string, "]b4_at_dollar[");
 
 260     muscle_percent_define_ensure("locations", the_location, true);
 
 265   /*-----------------------------------------.
 
 266   | Escape M4 quoting characters in C code.  |
 
 267   `-----------------------------------------*/
 
 271   \$    obstack_sgrow (&obstack_for_string, "$][");
 
 272   \@    obstack_sgrow (&obstack_for_string, "@@");
 
 273   \[    obstack_sgrow (&obstack_for_string, "@{");
 
 274   \]    obstack_sgrow (&obstack_for_string, "@}");
 
 277   /*-----------------------------------------------------.
 
 278   | By default, grow the string obstack with the input.  |
 
 279   `-----------------------------------------------------*/
 
 283  /* End of processing. */
 
 292 is_dot_or_dash (char ch)
 
 294   return ch == '.' || ch == '-';
 
 298 contains_dot_or_dash (const char* p)
 
 301     if (is_dot_or_dash (*p))
 
 306 /* Defines a variant of a symbolic name resolution. */
 
 309   /* Index in symbol list. */
 
 310   unsigned symbol_index;
 
 312   /* Matched symbol id and loc. */
 
 316   /* Hiding named reference. */
 
 317   named_ref* hidden_by;
 
 319   /* Error flags. May contain zero (no errors) or
 
 320      a combination of VARIANT_* values. */
 
 324 /* Set when the variant refers to a symbol hidden
 
 325    by an explicit symbol reference. */
 
 326 #define VARIANT_HIDDEN (1 << 0)
 
 328 /* Set when the variant refers to a symbol containing
 
 329    dots or dashes. Will require explicit bracketing. */
 
 330 #define VARIANT_BAD_BRACKETING (1 << 1)
 
 332 /* Set when the variant refers to a symbol which is
 
 333    not visible from current midrule. */
 
 334 #define VARIANT_NOT_VISIBLE_FROM_MIDRULE (1 << 2)
 
 336 static variant *variant_table = 0;
 
 337 static unsigned variant_table_size = 0;
 
 338 static unsigned variant_count = 0;
 
 341 variant_table_grow (void)
 
 344   if (variant_count > variant_table_size)
 
 346       while (variant_count > variant_table_size)
 
 347         variant_table_size = 2 * variant_table_size + 3;
 
 348       variant_table = xnrealloc (variant_table, variant_table_size,
 
 349                                  sizeof *variant_table);
 
 351   return &variant_table[variant_count - 1];
 
 355 variant_table_free (void)
 
 357   free (variant_table);
 
 359   variant_table_size = variant_count = 0;
 
 363 find_prefix_end (const char *prefix, char *begin, char *end)
 
 367   for (; *prefix && ptr != end; ++prefix, ++ptr)
 
 378 variant_add (uniqstr id, location id_loc, unsigned symbol_index,
 
 379              char *cp, char *cp_end, bool exact_mode)
 
 383   prefix_end = find_prefix_end (id, cp, cp_end);
 
 385       (prefix_end == cp_end ||
 
 386        (!exact_mode && is_dot_or_dash (*prefix_end))))
 
 388       variant *r = variant_table_grow ();
 
 389       r->symbol_index = symbol_index;
 
 401 get_at_spec(unsigned symbol_index)
 
 403   static char at_buf[20];
 
 404   if (symbol_index == 0)
 
 405     strcpy (at_buf, "$$");
 
 407     snprintf (at_buf, sizeof at_buf, "$%u", symbol_index);
 
 412 show_sub_messages (const char* cp, bool exact_mode, int midrule_rhs_index,
 
 413                    char dollar_or_at, bool is_warning)
 
 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 (var->loc, _("  refers to: %c%s at %s"),
 
 426                      dollar_or_at, var->id, at_spec);
 
 428             complain_at (var->loc, _("  refers to: %c%s at %s"),
 
 429                          dollar_or_at, var->id, at_spec);
 
 433           static struct obstack msg_buf;
 
 434           const char *tail = exact_mode ? "" :
 
 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)
 
 464             obstack_fgrow1 (&msg_buf, ", cannot be accessed from "
 
 465                             "mid-rule action at $%d", midrule_rhs_index);
 
 467           obstack_1grow (&msg_buf, '\0');
 
 469             warn_at (id_loc, _("%s"), (char *) obstack_finish (&msg_buf));
 
 471             complain_at (id_loc, _("%s"), (char *) obstack_finish (&msg_buf));
 
 472           obstack_free (&msg_buf, 0);
 
 477 /* Returned from "parse_ref" when the reference
 
 479 #define INVALID_REF (INT_MIN)
 
 481 /* Returned from "parse_ref" when the reference
 
 482    points to LHS ($$) of the current rule or midrule. */
 
 483 #define LHS_REF (INT_MIN + 1)
 
 485 /* Parse named or positional reference. In case of positional
 
 486    references, can return negative values for $-n "deep" stack
 
 489 parse_ref (char *cp, symbol_list *rule, int rule_length,
 
 490            int midrule_rhs_index, char *text, location text_loc,
 
 497   unsigned valid_variants = 0;
 
 498   unsigned valid_variant_index = 0;
 
 503   if (c_isdigit (*cp) || (*cp == '-' && c_isdigit (* (cp + 1))))
 
 505       long int num = strtol (cp, &cp, 10);
 
 506       if (1 - INT_MAX + rule_length <= num && num <= rule_length)
 
 510           complain_at (text_loc, _("integer out of range: %s"),
 
 518       /* Ignore the brackets. */
 
 520       for (p = ++cp; *p != ']'; ++p)
 
 528       /* Take all characters of the name. */
 
 530       for (p = cp; *p; ++p)
 
 531         if (is_dot_or_dash (*p))
 
 536       for (p = cp; *p; ++p)
 
 543   /* Add all relevant variants. */
 
 545     unsigned symbol_index;
 
 547     for (symbol_index = 0, l = rule; !symbol_list_null (l);
 
 548          ++symbol_index, l = l->next)
 
 551         if (l->content_type != SYMLIST_SYMBOL)
 
 554         var = variant_add (l->content.sym->tag, l->sym_loc,
 
 555                            symbol_index, cp, cp_end, exact_mode);
 
 556         if (var && l->named_ref)
 
 557           var->hidden_by = l->named_ref;
 
 560           variant_add (l->named_ref->id, l->named_ref->loc,
 
 561                        symbol_index, cp, cp_end, exact_mode);
 
 566   for (i = 0; i < variant_count; ++i)
 
 568       variant *var = &variant_table[i];
 
 569       unsigned symbol_index = var->symbol_index;
 
 571       /* Check visibility from mid-rule actions. */
 
 572       if (midrule_rhs_index != 0
 
 573           && (symbol_index == 0 || midrule_rhs_index < symbol_index))
 
 574         var->err |= VARIANT_NOT_VISIBLE_FROM_MIDRULE;
 
 576       /* Check correct bracketing. */
 
 577       if (!exact_mode && contains_dot_or_dash (var->id))
 
 578         var->err |= VARIANT_BAD_BRACKETING;
 
 580       /* Check using of hidden symbols. */
 
 582         var->err |= VARIANT_HIDDEN;
 
 586           valid_variant_index = i;
 
 591   switch (valid_variants)
 
 594       if (variant_count == 0)
 
 595         complain_at (text_loc, _("invalid reference: %s, symbol not found"),
 
 599           complain_at (text_loc, _("invalid reference: %s"),
 
 601           show_sub_messages (cp, exact_mode, midrule_rhs_index,
 
 602                              dollar_or_at, false);
 
 607         if (variant_count > 1)
 
 609             warn_at (text_loc, _("misleading reference: %s"),
 
 611             show_sub_messages (cp, exact_mode, midrule_rhs_index,
 
 615           unsigned symbol_index =
 
 616             variant_table[valid_variant_index].symbol_index;
 
 617           return (symbol_index == midrule_rhs_index) ? LHS_REF : symbol_index;
 
 622       complain_at (text_loc, _("ambiguous reference: %s"),
 
 624       show_sub_messages (cp, exact_mode, midrule_rhs_index,
 
 625                          dollar_or_at, false);
 
 633 /* Keeps track of the maximum number of semantic values to the left of
 
 634    a handle (those referenced by $0, $-1, etc.) are required by the
 
 635    semantic actions of this grammar. */
 
 636 int max_left_semantic_context = 0;
 
 639 /*------------------------------------------------------------------.
 
 640 | TEXT is pointing to a wannabee semantic value (i.e., a `$').      |
 
 642 | Possible inputs: $[<TYPENAME>]($|integer)                         |
 
 644 | Output to OBSTACK_FOR_STRING a reference to this semantic value.  |
 
 645 `------------------------------------------------------------------*/
 
 648 handle_action_dollar (symbol_list *rule, char *text, location dollar_loc)
 
 650   char const *type_name = NULL;
 
 653   symbol_list *effective_rule;
 
 654   int effective_rule_length;
 
 657   if (rule->midrule_parent_rule)
 
 659       effective_rule = rule->midrule_parent_rule;
 
 660       effective_rule_length = rule->midrule_parent_rhs_index - 1;
 
 664       effective_rule = rule;
 
 665       effective_rule_length = symbol_list_length (rule->next);
 
 668   /* Get the type name if explicit. */
 
 675       /* The '>' symbol will be later replaced by '\0'. Original
 
 676          'text' is needed for error messages. */
 
 679       if (untyped_var_seen)
 
 680         complain_at (dollar_loc, _("explicit type given in untyped grammar"));
 
 684   n = parse_ref (cp, effective_rule, effective_rule_length,
 
 685                  rule->midrule_parent_rhs_index, text, dollar_loc, '$');
 
 697         type_name = symbol_list_n_type_name_get (rule, dollar_loc, 0);
 
 701           if (union_seen | tag_seen)
 
 703               if (rule->midrule_parent_rule)
 
 704                 complain_at (dollar_loc,
 
 705                              _("$$ for the midrule at $%d of `%s'"
 
 706                                " has no declared type"),
 
 707                              rule->midrule_parent_rhs_index,
 
 708                              effective_rule->content.sym->tag);
 
 710                 complain_at (dollar_loc, _("$$ of `%s' has no declared type"),
 
 711                              rule->content.sym->tag);
 
 714             untyped_var_seen = true;
 
 718       obstack_fgrow1 (&obstack_for_string,
 
 719                       "]b4_lhs_value([%s])[", type_name);
 
 720       rule->action_props.is_value_used = true;
 
 724       if (max_left_semantic_context < 1 - n)
 
 725         max_left_semantic_context = 1 - n;
 
 726       if (!type_name && 0 < n)
 
 728           symbol_list_n_type_name_get (effective_rule, dollar_loc, n);
 
 731           if (union_seen | tag_seen)
 
 732             complain_at (dollar_loc, _("$%s of `%s' has no declared type"),
 
 733                          cp, effective_rule->content.sym->tag);
 
 735             untyped_var_seen = true;
 
 739       obstack_fgrow3 (&obstack_for_string,
 
 740                       "]b4_rhs_value(%d, %d, [%s])[",
 
 741                       effective_rule_length, n, type_name);
 
 743         symbol_list_n_get (effective_rule, n)->action_props.is_value_used =
 
 750 /*------------------------------------------------------.
 
 751 | TEXT is a location token (i.e., a `@...').  Output to |
 
 752 | OBSTACK_FOR_STRING a reference to this location.      |
 
 753 `------------------------------------------------------*/
 
 756 handle_action_at (symbol_list *rule, char *text, location at_loc)
 
 759   symbol_list *effective_rule;
 
 760   int effective_rule_length;
 
 763   if (rule->midrule_parent_rule)
 
 765       effective_rule = rule->midrule_parent_rule;
 
 766       effective_rule_length = rule->midrule_parent_rhs_index - 1;
 
 770       effective_rule = rule;
 
 771       effective_rule_length = symbol_list_length (rule->next);
 
 774   muscle_percent_define_ensure("locations", at_loc, true);
 
 776   n = parse_ref (cp, effective_rule, effective_rule_length,
 
 777                        rule->midrule_parent_rhs_index, text, at_loc, '@');
 
 784       obstack_sgrow (&obstack_for_string, "]b4_lhs_location[");
 
 788       obstack_fgrow2 (&obstack_for_string, "]b4_rhs_location(%d, %d)[",
 
 789                       effective_rule_length, n);
 
 795 /*-------------------------.
 
 796 | Initialize the scanner.  |
 
 797 `-------------------------*/
 
 799 /* Translate the dollars and ats in \a self, in the context \a sc_context
 
 800    (SC_RULE_ACTION, SC_SYMBOL_ACTION, INITIAL).  */
 
 803 translate_action (code_props *self, int sc_context)
 
 806   static bool initialized = false;
 
 809       obstack_init (&obstack_for_string);
 
 814   loc->start = loc->end = self->location.start;
 
 815   yy_switch_to_buffer (yy_scan_string (self->code));
 
 816   res = code_lex (self, sc_context);
 
 817   yy_delete_buffer (YY_CURRENT_BUFFER);
 
 822 /*------------------------------------------------------------------------.
 
 823 | Implementation of the public interface as documented in "scan-code.h".  |
 
 824 `------------------------------------------------------------------------*/
 
 827 code_props_none_init (code_props *self)
 
 829   *self = code_props_none;
 
 832 code_props const code_props_none = CODE_PROPS_NONE_INIT;
 
 835 code_props_plain_init (code_props *self, char const *code,
 
 838   self->kind = CODE_PROPS_PLAIN;
 
 840   self->location = code_loc;
 
 841   self->is_value_used = false;
 
 843   self->named_ref = NULL;
 
 847 code_props_symbol_action_init (code_props *self, char const *code,
 
 850   self->kind = CODE_PROPS_SYMBOL_ACTION;
 
 852   self->location = code_loc;
 
 853   self->is_value_used = false;
 
 855   self->named_ref = NULL;
 
 859 code_props_rule_action_init (code_props *self, char const *code,
 
 860                              location code_loc, symbol_list *rule,
 
 863   self->kind = CODE_PROPS_RULE_ACTION;
 
 865   self->location = code_loc;
 
 866   self->is_value_used = false;
 
 868   self->named_ref = name;
 
 872 code_props_translate_code (code_props *self)
 
 876       case CODE_PROPS_NONE:
 
 878       case CODE_PROPS_PLAIN:
 
 879         self->code = translate_action (self, INITIAL);
 
 881       case CODE_PROPS_SYMBOL_ACTION:
 
 882         self->code = translate_action (self, SC_SYMBOL_ACTION);
 
 884       case CODE_PROPS_RULE_ACTION:
 
 885         self->code = translate_action (self, SC_RULE_ACTION);
 
 891 code_scanner_last_string_free (void)
 
 897 code_scanner_free (void)
 
 899   obstack_free (&obstack_for_string, 0);
 
 900   variant_table_free ();
 
 902   /* Reclaim Flex's buffers.  */