[PATCH] Classify symbols by type-name.

[Akim Demaille]

        * src/uniqstr.h (UNIQSTR_CMP): New.
        * src/output.c (symbol_type_name_cmp, symbols_by_type_name)
        (type_names_output): New.
        (muscles_output): Use it.
        * data/lalr1.cc (b4_symbol_action_): Remove.
        (b4_symbol_case_, b4_type_action_): New.
        Adjust uses of b4_symbol_action_ to use b4_type_action_.
---
 ChangeLog     |   11 +++++++++
 data/lalr1.cc |   24 +++++++++++++------
 src/output.c  |   68 ++++++++++++++++++++++++++++++++++++++++++++++++++++++--
 src/uniqstr.h |    6 ++++-
 4 files changed, 97 insertions(+), 12 deletions(-)

diff --git a/ChangeLog b/ChangeLog
index b785bf9..a2e08e0 100644
--- a/ChangeLog
+++ b/ChangeLog
@@ -1,5 +1,16 @@
 2008-11-10  Akim Demaille  <address@hidden>
 
+       Classify symbols by type-name.
+       * src/uniqstr.h (UNIQSTR_CMP): New.
+       * src/output.c (symbol_type_name_cmp, symbols_by_type_name)
+       (type_names_output): New.
+       (muscles_output): Use it.
+       * data/lalr1.cc (b4_symbol_action_): Remove.
+       (b4_symbol_case_, b4_type_action_): New.
+       Adjust uses of b4_symbol_action_ to use b4_type_action_.
+
+2008-11-10  Akim Demaille  <address@hidden>
+
        Change the handling of the symbols in the skeletons.
        Before we were using tables which lines were the symbols and which
        columns were things like number, tag, type-name etc.  It is was
diff --git a/data/lalr1.cc b/data/lalr1.cc
index 7a04815..52aae20 100644
--- a/data/lalr1.cc
+++ b/data/lalr1.cc
@@ -132,17 +132,25 @@ m4_popdef([b4_dollar_dollar])dnl
 ])
 
 
-# b4_symbol_action_(NUM)
-# ----------------------
-# Invoke b4_dollar_dollar(SYMBOL_TYPENAME) for each symbol.
-m4_define([b4_symbol_action_],
-[m4_ifval(b4_symbol([$1], [type_name]),
+# b4_symbol_case_(SYMBOL-NUM)
+# ---------------------------
+# Issue a "case NUM" for SYMBOL-NUM.
+m4_define([b4_symbol_case_],
 [      case b4_symbol([$1], [number]): // b4_symbol([$1], [tag])
+])
+
+
+# b4_type_action_(NUMS)
+# ---------------------
+# Run actions for the symbol NUMS that all have the same type-name.
+m4_define([b4_type_action_],
+[m4_map([b4_symbol_case_], address@hidden)[]dnl
         b4_dollar_dollar([b4_symbol([$1], [number])],
                          [b4_symbol([$1], [tag])],
                          [b4_symbol([$1], [type_name])]);
        break;
-])])
+
+])
 
 
 # b4_symbol_variant(YYTYPE, YYVAL, ACTION, [ARGS])
@@ -154,7 +162,7 @@ m4_define([b4_symbol_variant],
             [$2.$3<$][3>(m4_shift3($@))])dnl
   switch ($1)
     {
-m4_map([b4_symbol_action_], m4_defn([b4_symbol_numbers]))
+m4_map([b4_type_action_], m4_defn([b4_type_names]))[]dnl
      default:
        break;
     }
@@ -860,7 +868,7 @@ b4_percent_code_get[]dnl
       }]b4_variant_if([
 
     // Type destructor.
-    b4_symbol_variant([[yytype]], [[yysym.value]], [[template destroy]])])[
+  b4_symbol_variant([[yytype]], [[yysym.value]], [[template destroy]])])[
   }
 
 #if YYDEBUG
diff --git a/src/output.c b/src/output.c
index 5391903..5950b4f 100644
--- a/src/output.c
+++ b/src/output.c
@@ -283,6 +283,69 @@ prepare_states (void)
 }
 
 
+/*-------------------------------------------------------.
+| Compare two symbols by type-name, and then by number.  |
+`-------------------------------------------------------*/
+
+int
+symbol_type_name_cmp (const symbol **lhs, const symbol **rhs)
+{
+  int res = UNIQSTR_CMP((*lhs)->type_name, (*rhs)->type_name);
+  if (res)
+    return res;
+  return (*lhs)->number - (*rhs)->number;
+}
+
+
+/*----------------------------------------------------------------.
+| Return a (malloc'ed) table of the symbols sorted by type-name.  |
+`----------------------------------------------------------------*/
+
+static symbol **
+symbols_by_type_name ()
+{
+  typedef int (*qcmp_type) (const void *, const void *);
+  symbol **res = xmemdup (symbols, nsyms * sizeof *res);
+  qsort (res, nsyms, sizeof *res, (qcmp_type) &symbol_type_name_cmp);
+  return res;
+}
+
+
+/*------------------------------------------------------------------.
+| Define b4_type_names, which is a list of (lists of the numbers of |
+| symbols with same type-name).                                     |
+`------------------------------------------------------------------*/
+
+static void
+type_names_output (FILE *out)
+{
+  int i;
+  const char *isep = "";
+  symbol **syms = symbols_by_type_name ();
+  fputs ("m4_define([b4_type_names],\n[", out);
+  for (i = 0; i < nsyms; )
+    if (syms[i]->type_name)
+      {
+        int j;
+        const char *jsep = "";
+        fprintf (out, "%s[", isep);
+        isep = ",\n";
+        for (j = i; j < nsyms; ++j)
+          {
+            if (syms[i]->type_name != syms[j]->type_name)
+              break;
+            fprintf (out, "%s%d", jsep, syms[j]->number);
+            jsep = ", ";
+          }
+        fputs ("]", out);
+        i = j;
+      }
+    else
+      ++i;
+  fputs ("])\n\n", out);
+  free (syms);
+}
+
 
 /*-------------------------------------.
 | The list of all the symbol numbers.  |
@@ -523,15 +586,15 @@ static void
 muscles_output (FILE *out)
 {
   fputs ("m4_init()\n", out);
-
   merger_output (out);
   symbol_code_props_output (out, "destructors", &symbol_destructor_get);
   symbol_code_props_output (out, "printers", &symbol_printer_get);
   symbol_definitions_output (out);
   symbol_numbers_output (out);
   token_definitions_output (out);
+  type_names_output (out);
   user_actions_output (out);
-
+  // Must be last.
   muscles_m4_output (out);
 }
 
@@ -630,7 +693,6 @@ output_skeleton (void)
   free (full_m4bison);
   free (full_skeleton);
 
-
   if (trace_flag & trace_muscles)
     muscles_output (stderr);
   {
diff --git a/src/uniqstr.h b/src/uniqstr.h
index 3eb152d..c986df0 100644
--- a/src/uniqstr.h
+++ b/src/uniqstr.h
@@ -1,6 +1,6 @@
 /* Keeping a unique copy of strings.
 
-   Copyright (C) 2002, 2003 Free Software Foundation, Inc.
+   Copyright (C) 2002, 2003, 2008 Free Software Foundation, Inc.
 
    This file is part of Bison, the GNU Compiler Compiler.
 
@@ -32,6 +32,10 @@ uniqstr uniqstr_new (char const *str);
 /* Two uniqstr values have the same value iff they are the same.  */
 #define UNIQSTR_EQ(USTR1, USTR2) ((USTR1) == (USTR2))
 
+/* Compare two uniqstr a la strlen: negative for <, nul for =, and
+   positive for >.  */
+#define UNIQSTR_CMP(USTR1, USTR2) ((USTR1) - (USTR2))
+
 /*--------------------------------------.
 | Initializing, destroying, debugging.  |
 `--------------------------------------*/
-- 
1.6.0.2.588.g3102