Inform m4 whether a tag is a valid id.

[bison.git] / src / output.c

diff --git a/src/output.c b/src/output.c
index 5391903dd62963d3c307406ae0bf8e3c3d02ef91..afcec94956b05cce53ea896b3533438c82c5d076 100644 (file)
--- a/src/output.c
+++ b/src/output.c
@@ -283,6 +283,58 @@ 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;
+  symbol **syms = symbols_by_type_name ();
+  fputs ("m4_define([b4_type_names],\n[", out);
+  for (i = 0; i < nsyms; /* nothing */)
+    {
+      // The index of the first symbol of the current type-name.
+      int i0 = i;
+      fputs (i ? ",\n[" : "[", out);
+      for (; i < nsyms && syms[i]->type_name == syms[i0]->type_name; ++i)
+        fprintf (out, "%s%d", i != i0 ? ", " : "", syms[i]->number);
+      fputs ("]", out);
+    }
+  fputs ("])\n\n", out);
+  free (syms);
+}
+
 
 /*-------------------------------------.
 | The list of all the symbol numbers.  |
@@ -344,6 +396,26 @@ merger_output (FILE *out)
 }
 
 
+/*----------------------------------.
+| Whether S is a valid identifier.  |
+`----------------------------------*/
+
+static bool
+is_identifier (uniqstr s)
+{
+  static char const alphanum[26 + 26 + 1 + 10] =
+    "abcdefghijklmnopqrstuvwxyz"
+    "ABCDEFGHIJKLMNOPQRSTUVWXYZ"
+    "_"
+    "0123456789";
+  if (!s || ! memchr (alphanum, *s, sizeof alphanum - 10))
+    return false;
+  for (++s; *s; ++s)
+    if (! memchr (alphanum, *s, sizeof alphanum))
+      return false;
+  return true;
+}
+
 /*---------------------------------------.
 | Output the symbol definitions to OUT.  |
 `---------------------------------------*/
@@ -362,15 +434,27 @@ symbol_definitions_output (FILE *out)
       obstack_1grow (&format_obstack, 0);                               \
       key = obstack_finish (&format_obstack);
 
+      // Whether the tag is a valid identifier.
+      SET_KEY("tag_is_id");
+      MUSCLE_INSERT_INT (key, is_identifier(sym->tag));
+
+      // The inner tag.
       SET_KEY("tag");
       MUSCLE_INSERT_STRING (key, sym->tag);
 
       SET_KEY("user_number");
       MUSCLE_INSERT_INT (key, sym->user_token_number);
 
+      SET_KEY("is_token");
+      MUSCLE_INSERT_INT (key,
+                         i < ntokens && sym != errtoken && sym != undeftoken);
+
       SET_KEY("number");
       MUSCLE_INSERT_INT (key, sym->number);
 
+      SET_KEY("has_type_name");
+      MUSCLE_INSERT_INT (key, !!sym->type_name);
+
       SET_KEY("type_name");
       MUSCLE_INSERT_STRING (key, sym->type_name ? sym->type_name : "");
 
@@ -523,15 +607,15 @@ 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);
 }
 \f
@@ -630,7 +714,6 @@ output_skeleton (void)
   free (full_m4bison);
   free (full_skeleton);
 
-
   if (trace_flag & trace_muscles)
     muscles_output (stderr);
   {