Simplify handling of '.' and '-' after unbracketed named references.

* doc/bison.texinfo (Mid-Rule Actions): Mention that periods and
dashes make symbol names less convenient for named references.
* src/scan-code.l:
(handle_action_dollar): New arg textlen.  All callers changed.
(handle_action_at): Likewise.  Also, args are pointers to const.
(ref_tail_fields): Remove; no longer used.
(letter): Now includes '-' and '.', since this is for Bison
identifiers.
(id): Now the simpler traditional defn, since letters now include
'-' and '.'.
(c_letter, c_id): New defns.
(ref): Use c_id for unbracketed IDs.
(<SC_RULE_ACTION>): Simplify, now that the distinction between
Bison and unbracketed IDs are now in the regular expressions.
(VARIANT_BAD_BRACKETING): Remove.
(VARIANT_NOT_VISIBLE_FROM_MIDRULE): Renumber.
(find_prefix_end): Remove, replacing with ....
(identifier_matches): New function.
(variant_add): Use it.  Omit EXPLICIT_BRACKETING arg; no longer
needed.  CP arg is pointer to constant.  All callers changed.
(show_sub_messages): Remove args CP, EXPLICIT_BRACKETING, DOLLAR_OR_AT.
New arg TEXT.  All callers changed.  Do not worry about showing
trailing context.
(parse_ref): Args CP, RULE, TEXT are now pointers to const.  New
arg TEXTLEN.  Remove arg DOLLAR_OR_AT.  All callers changed.
Simplify code now that the regular expressions capture the
restrictions.
* src/scan-gram.l (letter, id): Adjust to match scan-code.l.
* src/symlist.c (symbol_list_null): Arg is now pointer to const.
* src/symlist.h: Likewise.
* tests/named-refs.at (Misleading references): These are now caught
by the C compiler, not by Bison; that's good enough.  Adjust test
to reflect this.
(Many kinds of errors, Unresolved references): Adjust expected
diagnostics to match new behavior.  The same errors are caught,
though the diagnostics are not quite as fancy.
($ or @ followed by . or -): Likewise.  Also, Make the grammar
unambiguous, so that diagnostics are not complicated by ambiguity
warnings.

src/scan-code.l

@@ -47,12 +47,9 @@ YY_DECL;
 
 #define YY_USER_ACTION  location_compute (loc, &loc->end, yytext, yyleng);
 
-static void handle_action_dollar (symbol_list *rule, char *cp,
-				  location dollar_loc);
-static void handle_action_at (symbol_list *rule, char *cp, location at_loc);
-
-/* A string to be pushed to obstack after dollar/at has been handled. */
-static char *ref_tail_fields;
+static void handle_action_dollar (symbol_list *, char *, unsigned, location);
+static void handle_action_at (symbol_list const *, char const *, unsigned,
+			      location);
 
 static location the_location;
 static location *loc = &the_location;
@@ -82,12 +79,16 @@ tag	 [^\0\n>]+
    white space between the backslash and the newline.  */
 splice	 (\\[ \f\t\v]*\n)*
 
-/* C style identifier. Must start with letter. Will be used for
-   named symbol references. Shall be kept synchronized with
-   scan-gram.l "letter" and "id". */
-letter	  [.abcdefghijklmnopqrstuvwxyzABCDEFGHIJKLMNOPQRSTUVWXYZ_]
-id	  -*(-|{letter}({letter}|[-0-9])*)
-ref      -?[0-9]+|{id}|"["{id}"]"|"$"
+/* A Bison identifier.  Keep this synchronized with scan-gram.l "id".  */
+letter	  [-.abcdefghijklmnopqrstuvwxyzABCDEFGHIJKLMNOPQRSTUVWXYZ_]
+id	  {letter}({letter}|[0-9])*
+
+/* An identifier that can appear unbracketed in a reference.
+   This happens to be the same as a C-language identifier.  */
+c_letter  [abcdefghijklmnopqrstuvwxyzABCDEFGHIJKLMNOPQRSTUVWXYZ_]
+c_id	  {c_letter}({c_letter}|[0-9])*
+
+ref	  -?[0-9]+|{c_id}|"["{id}"]"|"$"
 
 %%
 
@@ -182,19 +183,11 @@ ref      -?[0-9]+|{id}|"["{id}"]"|"$"
 <SC_RULE_ACTION>
 {
   "$"("<"{tag}">")?{ref}  {
-    ref_tail_fields = 0;
-    handle_action_dollar (self->rule, yytext, *loc);
-    if (ref_tail_fields) {
-      obstack_sgrow (&obstack_for_string, ref_tail_fields);
-    }
+    handle_action_dollar (self->rule, yytext, yyleng, *loc);
     need_semicolon = true;
   }
   "@"{ref} {
-    ref_tail_fields = 0;
-    handle_action_at (self->rule, yytext, *loc);
-    if (ref_tail_fields) {
-      obstack_sgrow (&obstack_for_string, ref_tail_fields);
-    }
+    handle_action_at (self->rule, yytext, yyleng, *loc);
     need_semicolon = true;
   }
   "$"  {
@@ -338,13 +331,9 @@ typedef struct
    by an explicit symbol reference. */
 #define VARIANT_HIDDEN (1 << 0)
 
-/* Set when the variant refers to a symbol containing
-   dots or dashes. Will require explicit bracketing. */
-#define VARIANT_BAD_BRACKETING (1 << 1)
-
 /* Set when the variant refers to a symbol which is
    not visible from current midrule. */
-#define VARIANT_NOT_VISIBLE_FROM_MIDRULE (1 << 2)
+#define VARIANT_NOT_VISIBLE_FROM_MIDRULE (1 << 1)
 
 static variant *variant_table = 0;
 static unsigned variant_table_size = 0;
@@ -372,31 +361,25 @@ variant_table_free (void)
   variant_table_size = variant_count = 0;
 }
 
-static char *
-find_prefix_end (const char *prefix, char *begin, char *end)
+/* Return TRUE if ID matches the string from CP up to CP_END.
+   The string does not contain null bytes.  */
+static bool
+identifier_matches (char const *id, char const *cp, char const *cp_end)
 {
-  char *ptr = begin;
-
-  for (; *prefix && ptr != end; ++prefix, ++ptr)
-    if (*prefix != *ptr)
-      return 0;
-
-  if (*prefix)
-    return 0;
-
-  return ptr;
+  while (cp != cp_end)
+    if (*id++ != *cp++)
+      return false;
+  return !*id;
 }
 
+/* If scanning ID, return a new variant with that ID, at location
+   ID_LOC with index SYMBOL_INDEX.  Otherwise, return NULL.  The
+   currently scanned identifier starts at CP and ends at CP_END.  */
 static variant *
 variant_add (uniqstr id, location id_loc, unsigned symbol_index,
-	     char *cp, char *cp_end, bool explicit_bracketing)
+	     char const *cp, char const *cp_end)
 {
-  char *prefix_end;
-
-  prefix_end = find_prefix_end (id, cp, cp_end);
-  if (prefix_end &&
-      (prefix_end == cp_end ||
-       (!explicit_bracketing && is_dot_or_dash (*prefix_end))))
+  if (identifier_matches (id, cp, cp_end))
     {
       variant *r = variant_table_grow ();
       r->symbol_index = symbol_index;
@@ -421,11 +404,15 @@ get_at_spec(unsigned symbol_index)
   return at_buf;
 }
 
+/* Show a subsidiary message for a problem with a grammar rule.  TEXT
+   points to the problematic reference.  MIDRULE_RHS_INDEX is the rhs
+   index (1-origin) in the rule.  If IS_WARNING, it is a warning,
+   otherwise a complaint.  Indent the message INDENT spaces.  */
 static void
-show_sub_messages (const char* cp, bool explicit_bracketing,
-                   int midrule_rhs_index, char dollar_or_at,
+show_sub_messages (char const *text, int midrule_rhs_index,
                    bool is_warning, unsigned indent)
 {
+  char dollar_or_at = *text;
   unsigned i;
 
   for (i = 0; i < variant_count; ++i)
@@ -445,8 +432,6 @@ show_sub_messages (const char* cp, bool explicit_bracketing,
       else
 	{
 	  static struct obstack msg_buf;
-	  const char *tail = explicit_bracketing ? "" :
-	    cp + strlen (var->id);
 	  const char *id = var->hidden_by ? var->hidden_by->id :
 	    var->id;
 	  location id_loc = var->hidden_by ? var->hidden_by->loc :
@@ -460,7 +445,6 @@ show_sub_messages (const char* cp, bool explicit_bracketing,
 	    obstack_fgrow1 (&msg_buf, "[%s]", id);
 	  else
 	    obstack_sgrow (&msg_buf, id);
-	  obstack_sgrow (&msg_buf, tail);
 
 	  if (var->err & VARIANT_HIDDEN)
 	    {
@@ -469,7 +453,6 @@ show_sub_messages (const char* cp, bool explicit_bracketing,
 		obstack_fgrow1 (&msg_buf, "[%s]", var->id);
 	      else
 		obstack_sgrow (&msg_buf, var->id);
-	      obstack_sgrow (&msg_buf, tail);
 	    }
 
 	  obstack_fgrow1 (&msg_buf, _(" at %s"), at_spec);
@@ -504,16 +487,22 @@ show_sub_messages (const char* cp, bool explicit_bracketing,
 /* Sub-messages indent. */
 #define SUB_INDENT (4)
 
-/* Parse named or positional reference. In case of positional
-   references, can return negative values for $-n "deep" stack
-   accesses. */
+/* Return the index of a named or positional reference starting at CP
+   for a rule RULE of length RULE_LENGTH.  If MIDRULE_RHS_INDEX is
+   nonzero, this is a generated midrule whose rhs index (1-origin) is
+   MIDRULE_RHS_INDEX in the parent rule.  The entire semantic value
+   containing the reference is TEXT, of length TEXTLEN.  Its location
+   is TEXT_LOC.
+
+   In case of positional references, this can return negative values
+   for $-n "deep" stack accesses.  */
 static long int
-parse_ref (char *cp, symbol_list *rule, int rule_length,
-	   int midrule_rhs_index, char *text, location text_loc,
-	   char dollar_or_at)
+parse_ref (char const *cp, symbol_list const *rule, int rule_length,
+	   int midrule_rhs_index, char const *text, unsigned textlen,
+	   location text_loc)
 {
-  symbol_list *l;
-  char *cp_end;
+  symbol_list const *l;
+  char const *cp_end;
   bool explicit_bracketing;
   unsigned i;
   unsigned valid_variants = 0;
@@ -522,9 +511,9 @@ parse_ref (char *cp, symbol_list *rule, int rule_length,
   if ('$' == *cp)
     return LHS_REF;
 
-  if (c_isdigit (*cp) || (*cp == '-' && c_isdigit (* (cp + 1))))
+  if (c_isdigit (*cp) || *cp == '-')
     {
-      long int num = strtol (cp, &cp, 10);
+      long int num = strtol (cp, NULL, 10);
       if (1 - INT_MAX + rule_length <= num && num <= rule_length)
 	return num;
       else
@@ -535,32 +524,9 @@ parse_ref (char *cp, symbol_list *rule, int rule_length,
 	}
     }
 
-  if ('[' == *cp)
-    {
-      /* Ignore the brackets. */
-      char *p;
-      for (p = ++cp; *p != ']'; ++p)
-	continue;
-      cp_end = p;
-
-      explicit_bracketing = true;
-    }
-  else
-    {
-      /* Take all characters of the name. */
-      char* p;
-      for (p = cp; *p; ++p)
-	if (is_dot_or_dash (*p))
-	  {
-	    ref_tail_fields = p;
-	    break;
-	  }
-      for (p = cp; *p; ++p)
-	continue;
-      cp_end = p;
-
-      explicit_bracketing = false;
-    }
+  explicit_bracketing = (*cp == '[');
+  cp += explicit_bracketing;
+  cp_end = text + textlen - explicit_bracketing;
 
   /* Add all relevant variants. */
   {
@@ -574,13 +540,13 @@ parse_ref (char *cp, symbol_list *rule, int rule_length,
 	  continue;
 
 	var = variant_add (l->content.sym->tag, l->sym_loc,
-                           symbol_index, cp, cp_end, explicit_bracketing);
+                           symbol_index, cp, cp_end);
 	if (var && l->named_ref)
 	  var->hidden_by = l->named_ref;
 
 	if (l->named_ref)
 	  variant_add (l->named_ref->id, l->named_ref->loc,
-                       symbol_index, cp, cp_end, explicit_bracketing);
+                       symbol_index, cp, cp_end);
       }
   }
 
@@ -595,10 +561,6 @@ parse_ref (char *cp, symbol_list *rule, int rule_length,
 	  && (symbol_index == 0 || midrule_rhs_index < symbol_index))
         var->err |= VARIANT_NOT_VISIBLE_FROM_MIDRULE;
 
-      /* Check correct bracketing. */
-      if (!explicit_bracketing && contains_dot_or_dash (var->id))
-        var->err |= VARIANT_BAD_BRACKETING;
-
       /* Check using of hidden symbols. */
       if (var->hidden_by)
         var->err |= VARIANT_HIDDEN;
@@ -614,8 +576,7 @@ parse_ref (char *cp, symbol_list *rule, int rule_length,
     {
     case 0:
       {
-        unsigned len = (explicit_bracketing || !ref_tail_fields) ?
-          cp_end - cp : ref_tail_fields - cp;
+        unsigned len = cp_end - cp;
         unsigned indent = 0;
 
         complain_at_indent (text_loc, &indent, _("invalid reference: %s"),
@@ -629,7 +590,7 @@ parse_ref (char *cp, symbol_list *rule, int rule_length,
             const char *format =
               _("syntax error after `%c', expecting integer, letter,"
                 " `_', `[', or `$'");
-            complain_at_indent (sym_loc, &indent, format, dollar_or_at);
+            complain_at_indent (sym_loc, &indent, format, *text);
           }
         else if (midrule_rhs_index)
           {
@@ -647,8 +608,7 @@ parse_ref (char *cp, symbol_list *rule, int rule_length,
           }
 
         if (variant_count > 0)
-          show_sub_messages (cp, explicit_bracketing, midrule_rhs_index,
-                             dollar_or_at, false, indent);
+          show_sub_messages (text, midrule_rhs_index, false, indent);
         return INVALID_REF;
       }
     case 1:
@@ -658,8 +618,8 @@ parse_ref (char *cp, symbol_list *rule, int rule_length,
           {
             warn_at_indent (text_loc, &indent, _("misleading reference: %s"),
                             quote (text));
-            show_sub_messages (cp, explicit_bracketing, midrule_rhs_index,
-                               dollar_or_at, true, indent + SUB_INDENT);
+            show_sub_messages (text, midrule_rhs_index, true,
+			       indent + SUB_INDENT);
           }
         {
           unsigned symbol_index =
@@ -673,8 +633,8 @@ parse_ref (char *cp, symbol_list *rule, int rule_length,
         unsigned indent = 0;
         complain_at_indent (text_loc, &indent, _("ambiguous reference: %s"),
                             quote (text));
-        show_sub_messages (cp, explicit_bracketing, midrule_rhs_index,
-                           dollar_or_at, false, indent + SUB_INDENT);
+        show_sub_messages (text, midrule_rhs_index, false,
+			   indent + SUB_INDENT);
         return INVALID_REF;
       }
     }
@@ -692,13 +652,14 @@ int max_left_semantic_context = 0;
 /*------------------------------------------------------------------.
 | TEXT is pointing to a wannabee semantic value (i.e., a `$').      |
 |                                                                   |
-| Possible inputs: $[<TYPENAME>]($|integer)                         |
+| Possible inputs: $(<TYPENAME>|)($|integer|c_id|[id])              |
 |                                                                   |
 | Output to OBSTACK_FOR_STRING a reference to this semantic value.  |
 `------------------------------------------------------------------*/
 
 static void
-handle_action_dollar (symbol_list *rule, char *text, location dollar_loc)
+handle_action_dollar (symbol_list *rule, char *text, unsigned textlen,
+		      location dollar_loc)
 {
   char const *type_name = NULL;
   char *cp = text + 1;
@@ -735,7 +696,7 @@ handle_action_dollar (symbol_list *rule, char *text, location dollar_loc)
     }
 
   n = parse_ref (cp, effective_rule, effective_rule_length,
-		 rule->midrule_parent_rhs_index, text, dollar_loc, '$');
+		 rule->midrule_parent_rhs_index, text, textlen, dollar_loc);
 
   if (gt_ptr)
     *gt_ptr = '\0';
@@ -806,10 +767,11 @@ handle_action_dollar (symbol_list *rule, char *text, location dollar_loc)
 `------------------------------------------------------*/
 
 static void
-handle_action_at (symbol_list *rule, char *text, location at_loc)
+handle_action_at (symbol_list const *rule, char const *text, unsigned textlen,
+		  location at_loc)
 {
-  char *cp = text + 1;
-  symbol_list *effective_rule;
+  char const *cp = text + 1;
+  symbol_list const *effective_rule;
   int effective_rule_length;
   int n;
 
@@ -827,7 +789,7 @@ handle_action_at (symbol_list *rule, char *text, location at_loc)
   muscle_percent_define_ensure("locations", at_loc, true);
 
   n = parse_ref (cp, effective_rule, effective_rule_length,
-		       rule->midrule_parent_rhs_index, text, at_loc, '@');
+		 rule->midrule_parent_rhs_index, text, textlen, at_loc);
   switch (n)
     {
     case INVALID_REF:

src/scan-gram.l

@@ -118,8 +118,9 @@ static void unexpected_newline (boundary, char const *);
  /* Bracketed identifiers support. */
 %x SC_BRACKETED_ID SC_RETURN_BRACKETED_ID
 
-letter	  [.abcdefghijklmnopqrstuvwxyzABCDEFGHIJKLMNOPQRSTUVWXYZ_]
-id	  -*(-|{letter}({letter}|[-0-9])*)
+/* A Bison identifier.  Keep this synchronized with scan-code.l "id".  */
+letter	  [-.abcdefghijklmnopqrstuvwxyzABCDEFGHIJKLMNOPQRSTUVWXYZ_]
+id	  {letter}({letter}|[0-9])*
 directive %{id}
 int	  [0-9]+
 

src/symlist.c

@@ -217,7 +217,7 @@ symbol_list_n_type_name_get (symbol_list *l, location loc, int n)
 }
 
 bool
-symbol_list_null (symbol_list *node)
+symbol_list_null (symbol_list const *node)
 {
   return !node ||
     (node->content_type == SYMLIST_SYMBOL && !(node->content.sym));

src/symlist.h

@@ -116,7 +116,7 @@ symbol_list *symbol_list_n_get (symbol_list *l, int n);
 uniqstr symbol_list_n_type_name_get (symbol_list *l, location loc, int n);
 
 /* Check whether the node is a border element of a rule. */
-bool symbol_list_null (symbol_list *node);
+bool symbol_list_null (symbol_list const *node);
 
 /** Set the \c \%destructor for \c node as \c code at \c loc.  */
 void symbol_list_destructor_set (symbol_list *node, char const *code,