* src/lalr.c (F): Now a bitset*.

[bison.git] / src / lalr.c

diff --git a/src/lalr.c b/src/lalr.c
index eb226f4483321439f1ac816df6bd4b6914f85432..87e91cd5f553e9857ab345cba7c43bb07eae3123 100644 (file)
--- a/src/lalr.c
+++ b/src/lalr.c
@@ -24,8 +24,11 @@
    tokens they accept.  */
 
 #include "system.h"
    tokens they accept.  */
 
 #include "system.h"

+#include "bitset.h"
+#include "reader.h"

 #include "types.h"
 #include "LR0.h"
 #include "types.h"
 #include "LR0.h"

+#include "symtab.h"

 #include "gram.h"
 #include "complain.h"
 #include "lalr.h"
 #include "gram.h"
 #include "complain.h"
 #include "lalr.h"


@@ -33,44 +36,45 @@
 #include "derives.h"
 #include "getargs.h"
 
 #include "derives.h"
 #include "getargs.h"
 

-/* All the decorated states, indexed by the state number.  Warning:
-   there is a state_TABLE in LR0.c, but it is different and static.
-   */
-state_t *state_table = NULL;
+/* All the decorated states, indexed by the state number.  */
+state_t **states = NULL;

 
 int tokensetsize;
 
 int tokensetsize;

-short *LAruleno;
-unsigned *LA;
+short *LAruleno = NULL;
+bitset *LA = NULL;
+size_t nLA;

 
 

-short *goto_map;
-short *from_state;
-short *to_state;
-
-extern void berror PARAMS ((const char *));
-
-static int infinity;

 static int ngotos;
 static int ngotos;

+short *goto_map = NULL;
+short *from_state = NULL;
+short *to_state = NULL;

 
 /* And for the famous F variable, which name is so descriptive that a
    comment is hardly needed.  <grin>.  */
 
 /* And for the famous F variable, which name is so descriptive that a
    comment is hardly needed.  <grin>.  */

-static unsigned *F = NULL;
-#define F(Rule)  (F + (Rule) * tokensetsize)
+static bitset *F = NULL;

 
 static short **includes;
 static shorts **lookback;
 
 static short **includes;
 static shorts **lookback;

+
+
+/*---------------------------------------------------------------.
+| digraph & traverse.                                            |
+|                                                                |
+| The following variables are used as common storage between the |
+| two.                                                           |
+`---------------------------------------------------------------*/
+

 static short **R;
 static short *INDEX;
 static short *VERTICES;
 static int top;
 static short **R;
 static short *INDEX;
 static short *VERTICES;
 static int top;

-
+static int infinity;

 
 static void
 traverse (int i)
 {
   int j;
 
 static void
 traverse (int i)
 {
   int j;

-  size_t k;

   int height;
   int height;

-  size_t size = F (i + 1) - F(i);

 
   VERTICES[++top] = i;
   INDEX[i] = height = top;
 
   VERTICES[++top] = i;
   INDEX[i] = height = top;


@@ -84,8 +88,7 @@ traverse (int i)
        if (INDEX[i] > INDEX[R[i][j]])
          INDEX[i] = INDEX[R[i][j]];
 
        if (INDEX[i] > INDEX[R[i][j]])
          INDEX[i] = INDEX[R[i][j]];
 

-       for (k = 0; k < size; ++k)
-         F (i)[k] |= F (R[i][j])[k];
+       bitset_or (F[i], F[i], F[R[i][j]]);

       }
 
   if (INDEX[i] == height)
       }
 
   if (INDEX[i] == height)


@@ -97,8 +100,7 @@ traverse (int i)
        if (i == j)
          break;
 
        if (i == j)
          break;
 

-       for (k = 0; k < size; ++k)
-         F (i)[k] = F (j)[k];
+       bitset_copy (F[j], F[i]);

       }
 }
 
       }
 }
 


@@ -127,161 +129,84 @@ digraph (short **relation)
 }
 
 
 }
 
 

-/*--------------------.
-| Build STATE_TABLE.  |
-`--------------------*/
-
-static void
-set_state_table (void)
-{
-  /* NSTATES + 1 because lookahead for the pseudo state number NSTATES
-     might be used (see conflicts.c).  It is too opaque for me to
-     provide a probably less hacky implementation. --akim */
-  state_table = XCALLOC (state_t, nstates + 1);
-
-  {
-    core *sp;
-    for (sp = first_state; sp; sp = sp->next)
-      {
-       state_table[sp->number].state = sp;
-       state_table[sp->number].accessing_symbol = sp->accessing_symbol;
-      }
-  }
-
-  {
-    shifts *sp;
-    for (sp = first_shift; sp; sp = sp->next)
-      state_table[sp->number].shift_table = sp;
-  }
-
-  {
-    reductions *rp;
-    for (rp = first_reduction; rp; rp = rp->next)
-      state_table[rp->number].reduction_table = rp;
-  }
-
-  /* Initializing the lookaheads members.  Please note that it must be
-     performed after having set some of the other members which are
-     used below.  Change with extreme caution.  */
-  {
-    int i;
-    int count = 0;
-    for (i = 0; i < nstates; i++)
-      {
-       int k;
-       reductions *rp = state_table[i].reduction_table;
-       shifts *sp = state_table[i].shift_table;
-
-       state_table[i].lookaheads = count;
-
-       if (rp
-           && (rp->nreds > 1
-               || (sp && !ISVAR (state_table[sp->shifts[0]].accessing_symbol))))
-         count += rp->nreds;
-       else
-         state_table[i].consistent = 1;
-
-       if (sp)
-         for (k = 0; k < sp->nshifts; k++)
-           if (state_table[sp->shifts[k]].accessing_symbol
-               == error_token_number)
-             {
-               state_table[i].consistent = 0;
-               break;
-             }
-      }
-     state_table[nstates].lookaheads = count;
-  }
-}
-
-

 static void
 initialize_LA (void)
 {
 static void
 initialize_LA (void)
 {

-  int i;
+  size_t i;

   int j;
   short *np;
   int j;
   short *np;

-  reductions *rp;

 
 

-  size_t nLA = state_table[nstates].lookaheads;
+  /* Avoid having to special case 0.  */

   if (!nLA)
     nLA = 1;
 
   if (!nLA)
     nLA = 1;
 

-  LA = XCALLOC (unsigned, nLA * tokensetsize);
+  LA = XCALLOC (bitset, nLA);
+  for (i = 0; i < nLA; ++i)
+    {
+      LA[i] = bitset_create (ntokens, BITSET_FIXED);
+      bitset_zero (LA[i]);
+    }

   LAruleno = XCALLOC (short, nLA);
   lookback = XCALLOC (shorts *, nLA);
 
   np = LAruleno;
   for (i = 0; i < nstates; i++)
   LAruleno = XCALLOC (short, nLA);
   lookback = XCALLOC (shorts *, nLA);
 
   np = LAruleno;
   for (i = 0; i < nstates; i++)

-    if (!state_table[i].consistent)
-      if ((rp = state_table[i].reduction_table))
-       for (j = 0; j < rp->nreds; j++)
-         *np++ = rp->rules[j];
+    if (!states[i]->consistent)
+      for (j = 0; j < states[i]->reductions->nreds; j++)
+       *np++ = states[i]->reductions->rules[j];

 }
 
 
 static void
 set_goto_map (void)
 {
 }
 
 
 static void
 set_goto_map (void)
 {

-  shifts *sp;
+  size_t state;

   int i;
   int i;

-  int symbol;
-  int k;

   short *temp_map;
   short *temp_map;

-  int state2;
-  int state1;

 
   goto_map = XCALLOC (short, nvars + 1) - ntokens;
   temp_map = XCALLOC (short, nvars + 1) - ntokens;
 
   ngotos = 0;
 
   goto_map = XCALLOC (short, nvars + 1) - ntokens;
   temp_map = XCALLOC (short, nvars + 1) - ntokens;
 
   ngotos = 0;

-  for (sp = first_shift; sp; sp = sp->next)
+  for (state = 0; state < nstates; ++state)

     {
     {

-      for (i = sp->nshifts - 1; i >= 0; i--)
+      shifts *sp = states[state]->shifts;
+      for (i = sp->nshifts - 1; i >= 0 && SHIFT_IS_GOTO (sp, i); --i)

        {
        {

-         symbol = state_table[sp->shifts[i]].accessing_symbol;
-
-         if (ISTOKEN (symbol))
-           break;
-

          if (ngotos == MAXSHORT)
            fatal (_("too many gotos (max %d)"), MAXSHORT);
 
          ngotos++;
          if (ngotos == MAXSHORT)
            fatal (_("too many gotos (max %d)"), MAXSHORT);
 
          ngotos++;

-         goto_map[symbol]++;
+         goto_map[SHIFT_SYMBOL (sp, i)]++;

        }
     }
 
        }
     }
 

-  k = 0;
-  for (i = ntokens; i < nsyms; i++)
-    {
-      temp_map[i] = k;
-      k += goto_map[i];
-    }
+  {
+    int k = 0;
+    for (i = ntokens; i < nsyms; i++)
+      {
+       temp_map[i] = k;
+       k += goto_map[i];
+      }

 
 

-  for (i = ntokens; i < nsyms; i++)
-    goto_map[i] = temp_map[i];
+    for (i = ntokens; i < nsyms; i++)
+      goto_map[i] = temp_map[i];

 
 

-  goto_map[nsyms] = ngotos;
-  temp_map[nsyms] = ngotos;
+    goto_map[nsyms] = ngotos;
+    temp_map[nsyms] = ngotos;
+  }

 
   from_state = XCALLOC (short, ngotos);
   to_state = XCALLOC (short, ngotos);
 
 
   from_state = XCALLOC (short, ngotos);
   to_state = XCALLOC (short, ngotos);
 

-  for (sp = first_shift; sp; sp = sp->next)
+  for (state = 0; state < nstates; ++state)

     {
     {

-      state1 = sp->number;
-      for (i = sp->nshifts - 1; i >= 0; i--)
+      shifts *sp = states[state]->shifts;
+      for (i = sp->nshifts - 1; i >= 0 && SHIFT_IS_GOTO (sp, i); --i)

        {
        {

-         state2 = sp->shifts[i];
-         symbol = state_table[state2].accessing_symbol;
-
-         if (ISTOKEN (symbol))
-           break;
-
-         k = temp_map[symbol]++;
-         from_state[k] = state1;
-         to_state[k] = state2;
+         int k = temp_map[SHIFT_SYMBOL (sp, i)]++;
+         from_state[k] = state;
+         to_state[k] = sp->shifts[i];

        }
     }
 
        }
     }
 


@@ -332,38 +257,35 @@ initialize_F (void)
 
   int i;
 
 
   int i;
 

-  F = XCALLOC (unsigned, ngotos * tokensetsize);
+  F = XCALLOC (bitset, ngotos);
+  for (i = 0; i < ngotos; ++i)
+    {
+      F[i] = bitset_create (ntokens, BITSET_FIXED);
+      bitset_zero (F[i]);
+    }

 
   for (i = 0; i < ngotos; i++)
     {
       int stateno = to_state[i];
 
   for (i = 0; i < ngotos; i++)
     {
       int stateno = to_state[i];

-      shifts *sp = state_table[stateno].shift_table;
+      shifts *sp = states[stateno]->shifts;

 
 

-      if (sp)
-       {
-         int j;
-         for (j = 0; j < sp->nshifts; j++)
-           {
-             int symbol = state_table[sp->shifts[j]].accessing_symbol;
-             if (ISVAR (symbol))
-               break;
-             SETBIT (F + i * tokensetsize, symbol);
-           }
+      int j;
+      for (j = 0; j < sp->nshifts && SHIFT_IS_SHIFT (sp, j); j++)
+       bitset_set (F[i], SHIFT_SYMBOL (sp, j));

 
 

-         for (; j < sp->nshifts; j++)
-           {
-             int symbol = state_table[sp->shifts[j]].accessing_symbol;
-             if (nullable[symbol])
-               edge[nedges++] = map_goto (stateno, symbol);
-           }
+      for (; j < sp->nshifts; j++)
+       {
+         int symbol = SHIFT_SYMBOL (sp, j);
+         if (nullable[symbol])
+           edge[nedges++] = map_goto (stateno, symbol);
+       }

 
 

-         if (nedges)
-           {
-             reads[i] = XCALLOC (short, nedges + 1);
-             shortcpy (reads[i], edge, nedges);
-             reads[i][nedges] = -1;
-             nedges = 0;
-           }
+      if (nedges)
+       {
+         reads[i] = XCALLOC (short, nedges + 1);
+         shortcpy (reads[i], edge, nedges);
+         reads[i][nedges] = -1;
+         nedges = 0;

        }
     }
 
        }
     }
 


@@ -378,30 +300,21 @@ initialize_F (void)
 
 
 static void
 
 
 static void

-add_lookback_edge (int stateno, int ruleno, int gotono)
+add_lookback_edge (state_t *state, int ruleno, int gotono)

 {
   int i;
 {
   int i;

-  int k;
-  int found;

   shorts *sp;
 
   shorts *sp;
 

-  i = state_table[stateno].lookaheads;
-  k = state_table[stateno + 1].lookaheads;
-  found = 0;
-  while (!found && i < k)
-    {
-      if (LAruleno[i] == ruleno)
-       found = 1;
-      else
-       i++;
-    }
+  for (i = 0; i < state->nlookaheads; ++i)
+    if (LAruleno[state->lookaheadsp + i] == ruleno)
+      break;

 
 

-  assert (found);
+  assert (LAruleno[state->lookaheadsp + i] == ruleno);

 
   sp = XCALLOC (shorts, 1);
 
   sp = XCALLOC (shorts, 1);

-  sp->next = lookback[i];
+  sp->next = lookback[state->lookaheadsp + i];

   sp->value = gotono;
   sp->value = gotono;

-  lookback[i] = sp;
+  lookback[state->lookaheadsp + i] = sp;

 }
 
 
 }
 
 


@@ -495,7 +408,7 @@ static void
 build_relations (void)
 {
   short *edge = XCALLOC (short, ngotos + 1);
 build_relations (void)
 {
   short *edge = XCALLOC (short, ngotos + 1);

-  short *states = XCALLOC (short, ritem_longest_rhs () + 1);
+  short *states1 = XCALLOC (short, ritem_longest_rhs () + 1);

   int i;
 
   includes = XCALLOC (short *, ngotos);
   int i;
 
   includes = XCALLOC (short *, ngotos);


@@ -503,34 +416,33 @@ build_relations (void)
   for (i = 0; i < ngotos; i++)
     {
       int nedges = 0;
   for (i = 0; i < ngotos; i++)
     {
       int nedges = 0;

-      int state1 = from_state[i];
-      int symbol1 = state_table[to_state[i]].accessing_symbol;
+      int symbol1 = states[to_state[i]]->accessing_symbol;

       short *rulep;
 
       for (rulep = derives[symbol1]; *rulep > 0; rulep++)
        {
          int done;
          int length = 1;
       short *rulep;
 
       for (rulep = derives[symbol1]; *rulep > 0; rulep++)
        {
          int done;
          int length = 1;

-         int stateno = state1;

          short *rp;
          short *rp;

-         states[0] = state1;
+         state_t *state = states[from_state[i]];
+         states1[0] = state->number;

 
 

-         for (rp = ritem + rule_table[*rulep].rhs; *rp > 0; rp++)
+         for (rp = &ritem[rules[*rulep].rhs]; *rp >= 0; rp++)

            {
            {

-             shifts *sp = state_table[stateno].shift_table;
+             shifts *sp = state->shifts;

              int j;
              for (j = 0; j < sp->nshifts; j++)
                {
              int j;
              for (j = 0; j < sp->nshifts; j++)
                {

-                 stateno = sp->shifts[j];
-                 if (state_table[stateno].accessing_symbol == *rp)
+                 state = states[sp->shifts[j]];
+                 if (state->accessing_symbol == *rp)

                    break;
                }
 
                    break;
                }
 

-             states[length++] = stateno;
+             states1[length++] = state->number;

            }
 
            }
 

-         if (!state_table[stateno].consistent)
-           add_lookback_edge (stateno, *rulep, i);
+         if (!state->consistent)
+           add_lookback_edge (state, *rulep, i);

 
          length--;
          done = 0;
 
          length--;
          done = 0;


@@ -541,8 +453,7 @@ build_relations (void)
              /* JF added rp>=ritem &&   I hope to god its right! */
              if (rp >= ritem && ISVAR (*rp))
                {
              /* JF added rp>=ritem &&   I hope to god its right! */
              if (rp >= ritem && ISVAR (*rp))
                {

-                 stateno = states[--length];
-                 edge[nedges++] = map_goto (stateno, *rp);
+                 edge[nedges++] = map_goto (states1[--length], *rp);

                  if (nullable[*rp])
                    done = 0;
                }
                  if (nullable[*rp])
                    done = 0;
                }


@@ -560,7 +471,7 @@ build_relations (void)
     }
 
   XFREE (edge);
     }
 
   XFREE (edge);

-  XFREE (states);
+  XFREE (states1);

 
   includes = transpose (includes, ngotos);
 }
 
   includes = transpose (includes, ngotos);
 }


@@ -584,37 +495,102 @@ compute_FOLLOWS (void)
 static void
 compute_lookaheads (void)
 {
 static void
 compute_lookaheads (void)
 {

-  int i;
+  size_t i;

   shorts *sp;
 
   shorts *sp;
 

-  for (i = 0; i < state_table[nstates].lookaheads; i++)
+  for (i = 0; i < nLA; i++)

     for (sp = lookback[i]; sp; sp = sp->next)
     for (sp = lookback[i]; sp; sp = sp->next)

-      {
-       int size = LA (i + 1) - LA (i);
-       int j;
-       for (j = 0; j < size; ++j)
-         LA (i)[j] |= F (sp->value)[j];
-      }
+      bitset_or (LA[i], LA[i], F[sp->value]);

 
   /* Free LOOKBACK. */
 
   /* Free LOOKBACK. */

-  for (i = 0; i < state_table[nstates].lookaheads; i++)
+  for (i = 0; i < nLA; i++)

     LIST_FREE (shorts, lookback[i]);
 
   XFREE (lookback);
     LIST_FREE (shorts, lookback[i]);
 
   XFREE (lookback);

+  for (i = 0; i < (unsigned) ngotos; ++i)
+    bitset_free (F[i]);

   XFREE (F);
 }
 
 
   XFREE (F);
 }
 
 

+/*--------------------------------------.
+| Initializing the lookaheads members.  |
+`--------------------------------------*/
+
+static void
+initialize_lookaheads (void)
+{
+  size_t i;
+  nLA = 0;
+  for (i = 0; i < nstates; i++)
+    {
+      int k;
+      int nlookaheads = 0;
+      reductions *rp = states[i]->reductions;
+      shifts *sp = states[i]->shifts;
+
+      /* We need a lookahead either to distinguish different
+        reductions (i.e., there are two or more), or to distinguish a
+        reduction from a shift.  Otherwise, it is straightforward,
+        and the state is `consistent'.  */
+      if (rp->nreds > 1
+         || (rp->nreds == 1 && sp->nshifts && SHIFT_IS_SHIFT (sp, 0)))
+       nlookaheads += rp->nreds;
+      else
+       states[i]->consistent = 1;
+
+      for (k = 0; k < sp->nshifts; k++)
+       if (SHIFT_IS_ERROR (sp, k))
+         {
+           states[i]->consistent = 0;
+           break;
+         }
+
+      states[i]->nlookaheads = nlookaheads;
+      states[i]->lookaheadsp = nLA;
+      nLA += nlookaheads;
+    }
+}
+
+
+/*---------------------------------------.
+| Output the lookaheads for each state.  |
+`---------------------------------------*/
+
+static void
+lookaheads_print (FILE *out)
+{
+  size_t i;
+  int j, k;
+  fprintf (out, "Lookaheads: BEGIN\n");
+  for (i = 0; i < nstates; ++i)
+    {
+      fprintf (out, "State %d: %d lookaheads\n",
+              i, states[i]->nlookaheads);
+
+      for (j = 0; j < states[i]->nlookaheads; ++j)
+       for (k = 0; k < ntokens; ++k)
+         if (bitset_test (LA[states[i]->lookaheadsp + j], j))
+           fprintf (out, "   on %d (%s) -> rule %d\n",
+                    k, symbols[k]->tag,
+                    -LAruleno[states[i]->lookaheadsp + j] - 1);
+    }
+  fprintf (out, "Lookaheads: END\n");
+}
+

 void
 lalr (void)
 {
   tokensetsize = WORDSIZE (ntokens);
 
 void
 lalr (void)
 {
   tokensetsize = WORDSIZE (ntokens);
 

-  set_state_table ();
+  initialize_lookaheads ();

   initialize_LA ();
   set_goto_map ();
   initialize_F ();
   build_relations ();
   compute_FOLLOWS ();
   compute_lookaheads ();
   initialize_LA ();
   set_goto_map ();
   initialize_F ();
   build_relations ();
   compute_FOLLOWS ();
   compute_lookaheads ();

+
+  if (trace_flag)
+    lookaheads_print (stderr);

 }
 }