New file: skeleton for GLR (Generalized LR) parser.

[bison.git] / src / closure.c

diff --git a/src/closure.c b/src/closure.c
index ab086342dc67c4140b586098740f970efb398617..ea64250ea5aa9cf1741b26bc6a81c058f5974fac 100644 (file)
--- a/src/closure.c
+++ b/src/closure.c
@@ -1,5 +1,5 @@
 /* Subroutines for bison
 /* Subroutines for bison

-   Copyright 1984, 1989, 2000, 2001 Free Software Foundation, Inc.
+   Copyright (C) 1984, 1989, 2000, 2001, 2002 Free Software Foundation, Inc.

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


@@ -19,31 +19,30 @@
    02111-1307, USA.  */
 
 #include "system.h"
    02111-1307, USA.  */
 
 #include "system.h"

+#include "quotearg.h"
+#include "bitset.h"
+#include "bitsetv.h"
+#include "bitsetv-print.h"

 #include "getargs.h"
 #include "getargs.h"

+#include "symtab.h"

 #include "gram.h"
 #include "reader.h"
 #include "closure.h"
 #include "derives.h"
 #include "gram.h"
 #include "reader.h"
 #include "closure.h"
 #include "derives.h"

-#include "warshall.h"

 
 

-/* ITEMSETSIZE is the size of the array ITEMSET.  */
-short *itemset;
-size_t itemsetsize;
+/* NITEMSET is the size of the array ITEMSET.  */
+item_number_t *itemset;
+int nritemset;

 
 

-static unsigned *ruleset;
+static bitset ruleset;

 
 /* internal data.  See comments before set_fderives and set_firsts.  */
 
 /* internal data.  See comments before set_fderives and set_firsts.  */

-static unsigned *fderives;
-static unsigned *firsts;
+static bitsetv fderives = NULL;
+static bitsetv firsts = NULL;

 
 

-#define FDERIVES(Symbol)   (fderives + (Symbol) * rulesetsize)
-#define   FIRSTS(Symbol)   (firsts   + (Symbol) * varsetsize)
-
-/* number of words required to hold a bit for each rule */
-static int rulesetsize;
-
-/* number of words required to hold a bit for each variable */
-static int varsetsize;
+/* Retrieve the FDERIVES/FIRSTS sets of the nonterminals numbered Var.  */
+#define FDERIVES(Var)   fderives[(Var) - ntokens]
+#define   FIRSTS(Var)   firsts[(Var) - ntokens]

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


@@ -51,34 +50,35 @@ static int varsetsize;
 `-----------------*/
 
 static void
 `-----------------*/
 
 static void

-print_closure (int n)
+print_closure (const char *title, item_number_t *array, size_t size)

 {
 {

-  int i;
-  fprintf (stderr, "n = %d\n", n);
-  for (i = 0; i < itemsetsize; ++i)
-    fprintf (stderr, "   %d\n", itemset[i]);
-  fprintf (stderr, "\n\n");
+  size_t i;
+  fprintf (stderr, "Closure: %s\n", title);
+  for (i = 0; i < size; ++i)
+    {
+      item_number_t *rp;
+      fprintf (stderr, "  %2d: .", array[i]);
+      for (rp = &ritem[array[i]]; *rp >= 0; ++rp)
+       fprintf (stderr, " %s", symbol_tag_get (symbols[*rp]));
+      fprintf (stderr, "  (rule %d)\n", -*rp - 1);
+    }
+  fputs ("\n\n", stderr);

 }
 
 
 static void
 print_firsts (void)
 {
 }
 
 
 static void
 print_firsts (void)
 {

-  int i;
-  int j;
-  unsigned *rowp;
+  int i, j;

 
   fprintf (stderr, "FIRSTS\n");
 
   fprintf (stderr, "FIRSTS\n");

-

   for (i = ntokens; i < nsyms; i++)
     {
   for (i = ntokens; i < nsyms; i++)
     {

-      fprintf (stderr, "\t%s firsts\n", tags[i]);
-
-      rowp = FIRSTS (i - ntokens);
-
+      fprintf (stderr, "\t%s firsts\n", symbol_tag_get (symbols[i]));

       for (j = 0; j < nvars; j++)
       for (j = 0; j < nvars; j++)

-       if (BITISSET (rowp, j))
-         fprintf (stderr, "\t\t%d (%s)\n", j + ntokens, tags[j + ntokens]);
+       if (bitset_test (FIRSTS (i), j))
+         fprintf (stderr, "\t\t%s\n",
+                  symbol_tag_get (symbols[j + ntokens]));

     }
   fprintf (stderr, "\n\n");
 }
     }
   fprintf (stderr, "\n\n");
 }


@@ -87,72 +87,56 @@ print_firsts (void)
 static void
 print_fderives (void)
 {
 static void
 print_fderives (void)
 {

-  int i;
-  int j;
-  unsigned *rp;
+  int i, j;

 
   fprintf (stderr, "FDERIVES\n");
 
   fprintf (stderr, "FDERIVES\n");

-

   for (i = ntokens; i < nsyms; i++)
     {
   for (i = ntokens; i < nsyms; i++)
     {

-      fprintf (stderr, "\t%s derives\n", tags[i]);
-      rp = FDERIVES (i);
-
-      for (j = 0; j <= nrules; j++)
-       if (BITISSET (rp, j))
+      fprintf (stderr, "\t%s derives\n", symbol_tag_get (symbols[i]));
+      for (j = 0; j < nrules + 1; j++)
+       if (bitset_test (FDERIVES (i), j))

          {
          {

-           short *rhsp;
-           fprintf (stderr, "\t\t%d:", j);
-           for (rhsp = ritem + rule_table[j].rhs; *rhsp > 0; ++rhsp)
-             fprintf (stderr, " %s", tags[*rhsp]);
+           item_number_t *rhsp;
+           fprintf (stderr, "\t\t%d:", j - 1);
+           for (rhsp = rules[j].rhs; *rhsp >= 0; ++rhsp)
+             fprintf (stderr, " %s", symbol_tag_get (symbols[*rhsp]));

            fputc ('\n', stderr);
          }
     }
   fprintf (stderr, "\n\n");
 }
 \f
            fputc ('\n', stderr);
          }
     }
   fprintf (stderr, "\n\n");
 }
 \f

-/*-------------------------------------------------------------------.
-| Set FIRSTS to be an NVARS by NVARS bit matrix indicating which     |
-| items can represent the beginning of the input corresponding to    |
-| which other items.                                                 |
-|                                                                    |
-| For example, if some rule expands symbol 5 into the sequence of    |
-| symbols 8 3 20, the symbol 8 can be the beginning of the data for  |
-| symbol 5, so the bit [8 - ntokens, 5 - ntokens] in firsts is set.  |
-`-------------------------------------------------------------------*/
+/*------------------------------------------------------------------.
+| Set FIRSTS to be an NVARS array of NVARS bitsets indicating which |
+| items can represent the beginning of the input corresponding to   |
+| which other items.                                                |
+|                                                                   |
+| For example, if some rule expands symbol 5 into the sequence of   |
+| symbols 8 3 20, the symbol 8 can be the beginning of the data for |
+| symbol 5, so the bit [8 - ntokens] in first[5 - ntokens] (= FIRST |
+| (5)) is set.                                                      |
+`------------------------------------------------------------------*/

 
 static void
 set_firsts (void)
 {
 
 static void
 set_firsts (void)
 {

-  unsigned *row;
-  int symbol;
-  short *sp;
-  int rowsize;
-
-  int i;
-
-  varsetsize = rowsize = WORDSIZE (nvars);
+  int i, j;

 
 

-  firsts = XCALLOC (unsigned, nvars * rowsize);
+  firsts = bitsetv_create (nvars, nvars, BITSET_FIXED);

 
 

-  row = firsts;

   for (i = ntokens; i < nsyms; i++)
   for (i = ntokens; i < nsyms; i++)

-    {
-      sp = derives[i];
-      while (*sp >= 0)
-       {
-         symbol = ritem[rule_table[*sp++].rhs];
-         if (ISVAR (symbol))
-           {
-             symbol -= ntokens;
-             SETBIT (row, symbol);
-           }
-       }
-
-      row += rowsize;
-    }
+    for (j = 0; derives[i][j] >= 0; ++j)
+      {
+       int symbol = rules[derives[i][j]].rhs[0];
+       if (ISVAR (symbol))
+         bitset_set (FIRSTS (i), symbol - ntokens);
+      }

 
 

-  RTC (firsts, nvars);
+  if (trace_flag)
+    bitsetv_matrix_dump (stderr, "RTC: Firsts Input", firsts);
+  bitsetv_reflexive_transitive_closure (firsts);
+  if (trace_flag)
+    bitsetv_matrix_dump (stderr, "RTC: Firsts Output", firsts);

 
   if (trace_flag)
     print_firsts ();
 
   if (trace_flag)
     print_firsts ();


@@ -171,61 +155,32 @@ set_firsts (void)
 static void
 set_fderives (void)
 {
 static void
 set_fderives (void)
 {

-  unsigned *rrow;
-  unsigned *vrow;
-  int j;
-  unsigned cword;
-  short *rp;
-  int b;
+  int i, j, k;

 
 

-  int ruleno;
-  int i;
-
-  fderives = XCALLOC (unsigned, nvars * rulesetsize) - ntokens * rulesetsize;
+  fderives = bitsetv_create (nvars, nrules + 1, BITSET_FIXED);

 
   set_firsts ();
 
 
   set_firsts ();
 

-  rrow = FDERIVES (ntokens);
-
-  for (i = ntokens; i < nsyms; i++)
-    {
-      vrow = FIRSTS (i - ntokens);
-      cword = *vrow++;
-      b = 0;
-      for (j = ntokens; j < nsyms; j++)
-       {
-         if (cword & (1 << b))
-           {
-             rp = derives[j];
-             while ((ruleno = *rp++) > 0)
-               SETBIT (rrow, ruleno);
-           }
-
-         b++;
-         if (b >= BITS_PER_WORD && j + 1 < nsyms)
-           {
-             cword = *vrow++;
-             b = 0;
-           }
-       }
-
-      rrow += rulesetsize;
-    }
+  for (i = ntokens; i < nsyms; ++i)
+    for (j = ntokens; j < nsyms; ++j)
+      if (bitset_test (FIRSTS (i), j - ntokens))
+       for (k = 0; derives[j][k] > 0; ++k)
+         bitset_set (FDERIVES (i), derives[j][k]);

 
   if (trace_flag)
     print_fderives ();
 
 
   if (trace_flag)
     print_fderives ();
 

-  XFREE (firsts);
+  bitsetv_free (firsts);

 }
 }

+

 \f
 
 void
 new_closure (int n)
 {
 \f
 
 void
 new_closure (int n)
 {

-  itemset = XCALLOC (short, n);
+  itemset = XCALLOC (item_number_t, n);

 
 

-  rulesetsize = WORDSIZE (nrules + 1);
-  ruleset = XCALLOC (unsigned, rulesetsize);
+  ruleset = bitset_create (nrules + 1, BITSET_FIXED);

 
   set_fderives ();
 }
 
   set_fderives ();
 }


@@ -233,78 +188,48 @@ new_closure (int n)
 
 
 void
 
 
 void

-closure (short *core, int n)
+closure (item_number_t *core, int n)

 {
 {

-  int ruleno;
-  short *csp;
+  /* Index over CORE. */
+  int c;

 
 

-  int itemno;
-  int i;
+  /* A bit index over RULESET. */
+  int ruleno;

 
   if (trace_flag)
 
   if (trace_flag)

-    {
-      fprintf (stderr, "Entering closure (items = {");
-      for (i = 0; i < n; ++i)
-       fprintf (stderr, " %d ", core[i]);
-      fprintf (stderr, "}, nitems = %d)\n", n);
-    }
+    print_closure ("input", core, n);

 
 

-  if (n == 0)
-    {
-      for (i = 0; i < rulesetsize; ++i)
-       ruleset[i] = FDERIVES (start_symbol)[i];
-    }
-  else
-    {
-      for (i = 0; i < rulesetsize; ++i)
-       ruleset[i] = 0;
-
-      for (i = 0; i < n; ++i)
-       {
-         int symbol = ritem[core[i]];
-         if (ISVAR (symbol))
-           {
-             int j;
-             for (j = 0; j < rulesetsize; ++j)
-               ruleset[j] |= FDERIVES (symbol)[j];
-           }
-       }
-    }
+  bitset_zero (ruleset);
+
+  for (c = 0; c < n; ++c)
+    if (ISVAR (ritem[core[c]]))
+      bitset_or (ruleset, ruleset, FDERIVES (ritem[core[c]]));

 
 

-  ruleno = 0;
-  itemsetsize = 0;
-  csp = core;
-  for (i = 0; i < rulesetsize; ++i)
+  nritemset = 0;
+  c = 0;
+  for (ruleno = 0; ruleno < nrules + 1; ++ruleno)
+    if (bitset_test (ruleset, ruleno))
+      {
+       item_number_t itemno = rules[ruleno].rhs - ritem;
+       while (c < n && core[c] < itemno)
+         {
+           itemset[nritemset] = core[c];
+           nritemset++;
+           c++;
+         }
+       itemset[nritemset] = itemno;
+       nritemset++;
+      }
+
+  while (c < n)

     {
     {

-      int word = ruleset[i];
-      if (word == 0)
-       {
-         ruleno += BITS_PER_WORD;
-       }
-      else
-       {
-         int b;
-
-         for (b = 0; b < BITS_PER_WORD; b++)
-           {
-             if (word & (1 << b))
-               {
-                 itemno = rule_table[ruleno].rhs;
-                 while (csp < (core + n) && *csp < itemno)
-                   itemset[itemsetsize++] = *csp++;
-                 itemset[itemsetsize++] = itemno;
-               }
-
-             ruleno++;
-           }
-       }
+      itemset[nritemset] = core[c];
+      nritemset++;
+      c++;

     }
 
     }
 

-  while (csp < (core + n))
-    itemset[itemsetsize++] = *csp++;
-

   if (trace_flag)
   if (trace_flag)

-    print_closure (n);
+    print_closure ("output", itemset, nritemset);

 }
 
 
 }
 
 


@@ -312,6 +237,6 @@ void
 free_closure (void)
 {
   XFREE (itemset);
 free_closure (void)
 {
   XFREE (itemset);

-  XFREE (ruleset);
-  XFREE (fderives + ntokens * rulesetsize);
+  bitset_free (ruleset);
+  bitsetv_free (fderives);

 }
 }