Fix a longstanding bug uncovered by bro-0.9a9/src/parse.y, which I

[bison.git] / src / reader.c

diff --git a/src/reader.c b/src/reader.c
index e90cd2d73b315372556f7e09f51f491bc069aafa..75a23692dd8c91ffe8e0125805445accc776491f 100644 (file)
--- a/src/reader.c
+++ b/src/reader.c
@@ -1,6 +1,7 @@
 /* Input parser for Bison
 /* Input parser for Bison

-   Copyright (C) 1984, 1986, 1989, 1992, 1998, 2000, 2001, 2002
-   Free Software Foundation, Inc.
+
+   Copyright (C) 1984, 1986, 1989, 1992, 1998, 2000, 2001, 2002, 2003,
+   2005, 2006 Free Software Foundation, Inc.

 
    This file is part of Bison, the GNU Compiler Compiler.
 
 
    This file is part of Bison, the GNU Compiler Compiler.
 


@@ -16,9 +17,10 @@
 
    You should have received a copy of the GNU General Public License
    along with Bison; see the file COPYING.  If not, write to
 
    You should have received a copy of the GNU General Public License
    along with Bison; see the file COPYING.  If not, write to

-   the Free Software Foundation, Inc., 59 Temple Place - Suite 330,
-   Boston, MA 02111-1307, USA.  */
+   the Free Software Foundation, Inc., 51 Franklin Street, Fifth Floor,
+   Boston, MA 02110-1301, USA.  */

 
 

+#include <config.h>

 #include "system.h"
 
 #include <quotearg.h>
 #include "system.h"
 
 #include <quotearg.h>


@@ -34,26 +36,31 @@
 #include "symlist.h"
 #include "symtab.h"
 
 #include "symlist.h"
 #include "symtab.h"
 

+static void check_and_convert_grammar (void);
+

 static symbol_list *grammar = NULL;
 static symbol_list *grammar = NULL;

-static int start_flag = 0;
+static bool start_flag = false;

 merger_list *merge_functions;
 
 merger_list *merge_functions;
 

-/* Nonzero if %union has been seen.  */
-int typed = 0;
+/* Was %union seen?  */
+bool typed = false;
+
+/* Should rules have a default precedence?  */
+bool default_prec = true;

 \f
 /*-----------------------.
 | Set the start symbol.  |
 `-----------------------*/
 
 void
 \f
 /*-----------------------.
 | Set the start symbol.  |
 `-----------------------*/
 
 void

-grammar_start_symbol_set (symbol *s, location loc)
+grammar_start_symbol_set (symbol *sym, location loc)

 {
   if (start_flag)
     complain_at (loc, _("multiple %s declarations"), "%start");
   else
     {
 {
   if (start_flag)
     complain_at (loc, _("multiple %s declarations"), "%start");
   else
     {

-      start_flag = 1;
-      startsymbol = s;
+      start_flag = true;
+      startsymbol = sym;

       startsymbol_location = loc;
     }
 }
       startsymbol_location = loc;
     }
 }


@@ -70,36 +77,13 @@ prologue_augment (const char *prologue, location loc)
   struct obstack *oout =
     !typed ? &pre_prologue_obstack : &post_prologue_obstack;
 
   struct obstack *oout =
     !typed ? &pre_prologue_obstack : &post_prologue_obstack;
 

-  obstack_fgrow1 (oout, "]b4_syncline([[%d]], [[", loc.start.line);
+  obstack_fgrow1 (oout, "]b4_syncline(%d, [[", loc.start.line);

   MUSCLE_OBSTACK_SGROW (oout,
                        quotearg_style (c_quoting_style, loc.start.file));
   obstack_sgrow (oout, "]])[\n");
   obstack_sgrow (oout, prologue);
 }
 
   MUSCLE_OBSTACK_SGROW (oout,
                        quotearg_style (c_quoting_style, loc.start.file));
   obstack_sgrow (oout, "]])[\n");
   obstack_sgrow (oout, prologue);
 }
 

-
-
-
-/*----------------------.
-| Handle the epilogue.  |
-`----------------------*/
-
-void
-epilogue_augment (const char *epilogue, location loc)
-{
-  char *extension = NULL;
-  obstack_fgrow1 (&muscle_obstack, "]b4_syncline([[%d]], [[", loc.start.line);
-  MUSCLE_OBSTACK_SGROW (&muscle_obstack,
-                       quotearg_style (c_quoting_style, loc.start.file));
-  obstack_sgrow (&muscle_obstack, "]])[\n");
-  obstack_sgrow (&muscle_obstack, epilogue);
-  obstack_1grow (&muscle_obstack, 0);
-  extension = obstack_finish (&muscle_obstack);
-  muscle_grow ("epilogue", extension, "");
-  obstack_free (&muscle_obstack, extension);
-}
-
-

 \f
 
 /*-------------------------------------------------------------------.
 \f
 
 /*-------------------------------------------------------------------.


@@ -122,12 +106,12 @@ get_merge_function (uniqstr name, uniqstr type, location loc)
     type = uniqstr_new ("");
 
   head.next = merge_functions;
     type = uniqstr_new ("");
 
   head.next = merge_functions;

-  for (syms = &head, n = 1; syms->next != NULL; syms = syms->next, n += 1)
+  for (syms = &head, n = 1; syms->next; syms = syms->next, n += 1)

     if (UNIQSTR_EQ (name, syms->next->name))
       break;
   if (syms->next == NULL)
     {
     if (UNIQSTR_EQ (name, syms->next->name))
       break;
   if (syms->next == NULL)
     {

-      syms->next = XMALLOC (merger_list, 1);
+      syms->next = xmalloc (sizeof syms->next[0]);

       syms->next->name = uniqstr_new (name);
       syms->next->type = uniqstr_new (type);
       syms->next->next = NULL;
       syms->next->name = uniqstr_new (name);
       syms->next->type = uniqstr_new (type);
       syms->next->next = NULL;


@@ -146,11 +130,8 @@ get_merge_function (uniqstr name, uniqstr type, location loc)
 void
 free_merger_functions (void)
 {
 void
 free_merger_functions (void)
 {

-  merger_list *L0;
-  if (! glr_parser)
-    return;
-  L0 = merge_functions;
-  while (L0 != NULL)
+  merger_list *L0 = merge_functions;
+  while (L0)

     {
       merger_list *L1 = L0->next;
       free (L0);
     {
       merger_list *L1 = L0->next;
       free (L0);


@@ -168,20 +149,13 @@ free_merger_functions (void)
 |                                                                    |
 | All actions are copied out, labelled by the rule number they apply |
 | to.                                                                |
 |                                                                    |
 | All actions are copied out, labelled by the rule number they apply |
 | to.                                                                |

-|                                                                    |
-| Bison used to allow some %directives in the rules sections, but    |
-| this is no longer consider appropriate: (i) the documented grammar |
-| doesn't claim it, (ii), it would promote bad style, (iii), error   |
-| recovery for %directives consists in skipping the junk until a `%' |
-| is seen and helrp synchronizing.  This scheme is definitely wrong  |
-| in the rules section.                                              |

 `-------------------------------------------------------------------*/
 
 /* The (currently) last symbol of GRAMMAR. */
 `-------------------------------------------------------------------*/
 
 /* The (currently) last symbol of GRAMMAR. */

-symbol_list *grammar_end = NULL;
+static symbol_list *grammar_end = NULL;

 
 

-/* Append S to the GRAMMAR. */
-void
+/* Append SYM to the grammar.  */
+static void

 grammar_symbol_append (symbol *sym, location loc)
 {
   symbol_list *p = symbol_list_new (sym, loc);
 grammar_symbol_append (symbol *sym, location loc)
 {
   symbol_list *p = symbol_list_new (sym, loc);


@@ -192,13 +166,18 @@ grammar_symbol_append (symbol *sym, location loc)
     grammar = p;
 
   grammar_end = p;
     grammar = p;
 
   grammar_end = p;

+
+  /* A null SYM stands for an end of rule; it is not an actual
+     part of it.  */
+  if (sym)
+    ++nritems;

 }
 
 /* The rule currently being defined, and the previous rule.
    CURRENT_RULE points to the first LHS of the current rule, while
    PREVIOUS_RULE_END points to the *end* of the previous rule (NULL).  */
 symbol_list *current_rule = NULL;
 }
 
 /* The rule currently being defined, and the previous rule.
    CURRENT_RULE points to the first LHS of the current rule, while
    PREVIOUS_RULE_END points to the *end* of the previous rule (NULL).  */
 symbol_list *current_rule = NULL;

-symbol_list *previous_rule_end = NULL;
+static symbol_list *previous_rule_end = NULL;

 
 
 /*----------------------------------------------.
 
 
 /*----------------------------------------------.


@@ -206,25 +185,22 @@ symbol_list *previous_rule_end = NULL;
 `----------------------------------------------*/
 
 void
 `----------------------------------------------*/
 
 void

-grammar_rule_begin (symbol *lhs, location loc)
+grammar_current_rule_begin (symbol *lhs, location loc)

 {
   if (!start_flag)
     {
       startsymbol = lhs;
       startsymbol_location = loc;
 {
   if (!start_flag)
     {
       startsymbol = lhs;
       startsymbol_location = loc;

-      start_flag = 1;
+      start_flag = true;

     }
 
   /* Start a new rule and record its lhs.  */
   ++nrules;
     }
 
   /* Start a new rule and record its lhs.  */
   ++nrules;

-  ++nritems;
-

   previous_rule_end = grammar_end;
   grammar_symbol_append (lhs, loc);
   current_rule = grammar_end;
 
   /* Mark the rule's lhs as a nonterminal if not already so.  */
   previous_rule_end = grammar_end;
   grammar_symbol_append (lhs, loc);
   current_rule = grammar_end;
 
   /* Mark the rule's lhs as a nonterminal if not already so.  */

-

   if (lhs->class == unknown_sym)
     {
       lhs->class = nterm_sym;
   if (lhs->class == unknown_sym)
     {
       lhs->class = nterm_sym;


@@ -235,39 +211,71 @@ grammar_rule_begin (symbol *lhs, location loc)
     complain_at (loc, _("rule given for %s, which is a token"), lhs->tag);
 }
 
     complain_at (loc, _("rule given for %s, which is a token"), lhs->tag);
 }
 

-/* Check that the last rule (CURRENT_RULE) is properly defined.  For
-   instance, there should be no type clash on the default action.  */
+
+/*----------------------------------------------------------------------.
+| A symbol should be used if it has a destructor, or if it is a         |
+| mid-rule symbol (i.e., the generated LHS replacing a mid-rule         |
+| action) that was assigned to, as in "exp: { $$ = 1; } { $$ = $1; }".  |
+`----------------------------------------------------------------------*/
+
+static bool
+symbol_should_be_used (symbol_list const *s)
+{
+  return (s->sym->destructor
+         || (s->midrule && s->midrule->used));
+}
+
+/*----------------------------------------------------------------.
+| Check that the rule R is properly defined.  For instance, there |
+| should be no type clash on the default action.                  |
+`----------------------------------------------------------------*/

 
 static void
 
 static void

-grammar_current_rule_check (void)
+grammar_rule_check (const symbol_list *r)

 {
 {

-  symbol *lhs = current_rule->sym;
-  char const *lhs_type = lhs->type_name;
-  symbol *first_rhs = current_rule->next->sym;
+  /* Type check.

 
 

-  /* If there is an action, then there is nothing we can do: the user
-     is allowed to shoot herself in the foot.  */
-  if (current_rule->action)
-    return;
+     If there is an action, then there is nothing we can do: the user
+     is allowed to shoot herself in the foot.

 
 

-  /* Don't worry about the default action if $$ is untyped, since $$'s
+     Don't worry about the default action if $$ is untyped, since $$'s

      value can't be used.  */
      value can't be used.  */

-  if (! lhs_type)
-    return;
-
-  /* If $$ is being set in default way, report if any type mismatch.  */
-  if (first_rhs)
+  if (!r->action && r->sym->type_name)

     {
     {

-      const char *rhs_type = first_rhs->type_name ? first_rhs->type_name : "";
-      if (!UNIQSTR_EQ (lhs_type, rhs_type))
-       warn_at (current_rule->location,
-                _("type clash on default action: <%s> != <%s>"),
-                lhs_type, rhs_type);
+      symbol *first_rhs = r->next->sym;
+      /* If $$ is being set in default way, report if any type mismatch.  */
+      if (first_rhs)
+       {
+         char const *lhs_type = r->sym->type_name;
+         const char *rhs_type =
+           first_rhs->type_name ? first_rhs->type_name : "";
+         if (!UNIQSTR_EQ (lhs_type, rhs_type))
+           warn_at (r->location,
+                    _("type clash on default action: <%s> != <%s>"),
+                    lhs_type, rhs_type);
+       }
+      /* Warn if there is no default for $$ but we need one.  */
+      else
+       warn_at (r->location,
+                _("empty rule for typed nonterminal, and no action"));

     }
     }

-  /* Warn if there is no default for $$ but we need one.  */
-  else
-    warn_at (current_rule->location,
-            _("empty rule for typed nonterminal, and no action"));
+
+  /* Check that symbol values that should be used are in fact used.  */
+  {
+    symbol_list const *l = r;
+    int n = 0;
+    for (; l && l->sym; l = l->next, ++n)
+      if (! (l->used
+            || !symbol_should_be_used (l)
+            /* The default action, $$ = $1, `uses' both.  */
+            || (!r->action && (n == 0 || n == 1))))
+       {
+         if (n)
+           warn_at (r->location, _("unused value: $%d"), n);
+         else
+           warn_at (r->location, _("unset value: $$"));
+       }
+  }

 }
 
 
 }
 
 


@@ -276,12 +284,12 @@ grammar_current_rule_check (void)
 `-------------------------------------*/
 
 void
 `-------------------------------------*/
 
 void

-grammar_rule_end (location loc)
+grammar_current_rule_end (location loc)

 {
   /* Put an empty link in the list to mark the end of this rule  */
   grammar_symbol_append (NULL, grammar_end->location);
   current_rule->location = loc;
 {
   /* Put an empty link in the list to mark the end of this rule  */
   grammar_symbol_append (NULL, grammar_end->location);
   current_rule->location = loc;

-  grammar_current_rule_check ();
+  grammar_rule_check (current_rule);

 }
 
 
 }
 
 


@@ -314,6 +322,10 @@ grammar_midrule_action (void)
   midrule->action = current_rule->action;
   midrule->action_location = dummy_location;
   current_rule->action = NULL;
   midrule->action = current_rule->action;
   midrule->action_location = dummy_location;
   current_rule->action = NULL;

+  /* If $$ was used in the action, the LHS of the enclosing rule was
+     incorrectly flagged as used.  */
+  midrule->used = current_rule->used;
+  current_rule->used = false;

 
   if (previous_rule_end)
     previous_rule_end->next = midrule;
 
   if (previous_rule_end)
     previous_rule_end->next = midrule;


@@ -321,14 +333,16 @@ grammar_midrule_action (void)
     grammar = midrule;
 
   /* End the dummy's rule.  */
     grammar = midrule;
 
   /* End the dummy's rule.  */

-  previous_rule_end = symbol_list_new (NULL, dummy_location);
-  previous_rule_end->next = current_rule;
+  midrule->next = symbol_list_new (NULL, dummy_location);
+  grammar_rule_check (midrule);
+  midrule->next->next = current_rule;

 
 

-  midrule->next = previous_rule_end;
+  previous_rule_end = midrule->next;

 
   /* Insert the dummy nonterminal replacing the midrule action into
 
   /* Insert the dummy nonterminal replacing the midrule action into

-     the current rule.  */
+     the current rule.  Bind it to its dedicated rule.  */

   grammar_current_rule_symbol_append (dummy, dummy_location);
   grammar_current_rule_symbol_append (dummy, dummy_location);

+  grammar_end->midrule = midrule;

 }
 
 /* Set the precedence symbol of the current rule to PRECSYM. */
 }
 
 /* Set the precedence symbol of the current rule to PRECSYM. */


@@ -377,18 +391,16 @@ grammar_current_rule_symbol_append (symbol *sym, location loc)
 {
   if (current_rule->action)
     grammar_midrule_action ();
 {
   if (current_rule->action)
     grammar_midrule_action ();

-  ++nritems;

   grammar_symbol_append (sym, loc);
 }
 
   grammar_symbol_append (sym, loc);
 }
 

-/* Attach an ACTION to the current rule.  If needed, move the previous
-   action as a mid-rule action.  */
+/* Attach an ACTION to the current rule.  */

 
 void
 grammar_current_rule_action_append (const char *action, location loc)
 {
 
 void
 grammar_current_rule_action_append (const char *action, location loc)
 {

-  if (current_rule->action)
-    grammar_midrule_action ();
+  /* There's no need to invoke grammar_midrule_action here, since the
+     scanner already did it if necessary.  */

   current_rule->action = action;
   current_rule->action_location = loc;
 }
   current_rule->action = action;
   current_rule->action_location = loc;
 }


@@ -406,8 +418,12 @@ packgram (void)
   rule_number ruleno = 0;
   symbol_list *p = grammar;
 
   rule_number ruleno = 0;
   symbol_list *p = grammar;
 

-  ritem = XCALLOC (item_number, nritems);
-  rules = XCALLOC (rule, nrules);
+  ritem = xnmalloc (nritems + 1, sizeof *ritem);
+
+  /* This sentinel is used by build_relations in gram.c.  */
+  *ritem++ = 0;
+
+  rules = xnmalloc (nrules, sizeof *rules);

 
   while (p)
     {
 
   while (p)
     {


@@ -416,12 +432,14 @@ packgram (void)
       rules[ruleno].number = ruleno;
       rules[ruleno].lhs = p->sym;
       rules[ruleno].rhs = ritem + itemno;
       rules[ruleno].number = ruleno;
       rules[ruleno].lhs = p->sym;
       rules[ruleno].rhs = ritem + itemno;

+      rules[ruleno].prec = NULL;
+      rules[ruleno].dprec = p->dprec;
+      rules[ruleno].merger = p->merger;
+      rules[ruleno].precsym = NULL;

       rules[ruleno].location = p->location;
       rules[ruleno].useful = true;
       rules[ruleno].action = p->action;
       rules[ruleno].action_location = p->action_location;
       rules[ruleno].location = p->location;
       rules[ruleno].useful = true;
       rules[ruleno].action = p->action;
       rules[ruleno].action_location = p->action_location;

-      rules[ruleno].dprec = p->dprec;
-      rules[ruleno].merger = p->merger;

 
       p = p->next;
       while (p && p->sym)
 
       p = p->next;
       while (p && p->sym)


@@ -431,7 +449,7 @@ packgram (void)
          ritem[itemno++] = symbol_number_as_item_number (p->sym->number);
          /* A rule gets by default the precedence and associativity
             of the last token in it.  */
          ritem[itemno++] = symbol_number_as_item_number (p->sym->number);
          /* A rule gets by default the precedence and associativity
             of the last token in it.  */

-         if (p->sym->class == token_sym)
+         if (p->sym->class == token_sym && default_prec)

            rules[ruleno].prec = p->sym;
          if (p)
            p = p->next;
            rules[ruleno].prec = p->sym;
          if (p)
            p = p->next;


@@ -451,8 +469,7 @@ packgram (void)
        p = p->next;
     }
 
        p = p->next;
     }
 

-  if (itemno != nritems)
-    abort ();
+  assert (itemno == nritems);

 
   if (trace_flag & trace_sets)
     ritem_print (stderr);
 
   if (trace_flag & trace_sets)
     ritem_print (stderr);


@@ -491,20 +508,29 @@ reader (void)
   obstack_init (&pre_prologue_obstack);
   obstack_init (&post_prologue_obstack);
 
   obstack_init (&pre_prologue_obstack);
   obstack_init (&post_prologue_obstack);
 

-  finput = xfopen (grammar_file, "r");
-  gram_in = finput;
+  gram_in = xfopen (grammar_file, "r");

 
   gram__flex_debug = trace_flag & trace_scan;
   gram_debug = trace_flag & trace_parse;
   scanner_initialize ();
   gram_parse ();
 
 
   gram__flex_debug = trace_flag & trace_scan;
   gram_debug = trace_flag & trace_parse;
   scanner_initialize ();
   gram_parse ();
 

-  /* If something went wrong during the parsing, don't try to
-     continue.  */
-  if (complaint_issued)
-    return;
+  if (! complaint_issued)
+    check_and_convert_grammar ();

 
 

-  /* Grammar has been read.  Do some checking */
+  xfclose (gram_in);
+}
+
+
+/*-------------------------------------------------------------.
+| Check the grammar that has just been read, and convert it to |
+| internal form.                                              |
+`-------------------------------------------------------------*/
+
+static void
+check_and_convert_grammar (void)
+{
+  /* Grammar has been read.  Do some checking.  */

   if (nrules == 0)
     fatal (_("no rules in the input grammar"));
 
   if (nrules == 0)
     fatal (_("no rules in the input grammar"));
 


@@ -521,7 +547,7 @@ reader (void)
       endtoken->user_token_number = 0;
     }
 
       endtoken->user_token_number = 0;
     }
 

-  /* Insert the initial rule, which line is that of the first rule
+  /* Insert the initial rule, whose line is that of the first rule

      (not that of the start symbol):
 
      accept: %start EOF.  */
      (not that of the start symbol):
 
      accept: %start EOF.  */


@@ -537,10 +563,7 @@ reader (void)
     grammar = p;
   }
 
     grammar = p;
   }
 

-  if (! (nsyms <= SYMBOL_NUMBER_MAXIMUM && nsyms == ntokens + nvars))
-    abort ();
-
-  xfclose (finput);
+  assert (nsyms <= SYMBOL_NUMBER_MAXIMUM && nsyms == ntokens + nvars);

 
   /* Assign the symbols their symbol numbers.  Write #defines for the
      token symbols into FDEFINES if requested.  */
 
   /* Assign the symbols their symbol numbers.  Write #defines for the
      token symbols into FDEFINES if requested.  */