checkpoint

[sbp.git] / src / edu / berkeley / sbp / Parser.java

diff --git a/src/edu/berkeley/sbp/Parser.java b/src/edu/berkeley/sbp/Parser.java
index 300ff27..2344e15 100644 (file)
--- a/src/edu/berkeley/sbp/Parser.java
+++ b/src/edu/berkeley/sbp/Parser.java
@@ -15,9 +15,9 @@ public abstract class Parser<Tok, Result> {
     protected Parser(Table<Tok> pt)               { this.pt = pt; }
 
     /** implement this method to create the output forest corresponding to a lone shifted input token */
-    public abstract Forest<Result> shiftToken(Input.Location oldloc, Tok t, Input.Location newloc);
+    protected abstract Forest<Result> shiftToken(Tok t, Input.Location newloc);
 
-    public boolean helpgc = true;
+    boolean helpgc = true;
 
     public String toString() { return pt.toString(); }
 
@@ -32,7 +32,7 @@ public abstract class Parser<Tok, Result> {
             Input.Location oldloc = loc;
             loc = input.getLocation();
             current.reduce();
-            Forest forest = current.token==null ? null : shiftToken(oldloc, (Tok)current.token, loc);
+            Forest forest = current.token==null ? null : shiftToken((Tok)current.token, loc);
             GSS.Phase next = gss.new Phase<Tok>(current, this, current, input.next(), loc, forest);
             if (!helpgc) {
                 FileOutputStream fos = new FileOutputStream("out-"+idx+".dot");
@@ -82,15 +82,22 @@ public abstract class Parser<Tok, Result> {
             if (hs.contains(e)) return;
             hs.add(e);
             if (e instanceof Atom) return;
-            for(Sequence s : (Union)e) {
-                hs.add(s);
-                for(Position p = s.firstp(); p != null; p = p.next())
-                    walk(p.element(), hs);
-            }
+            for(Sequence s : (Union)e)
+                walk(s, hs);
+        }
+        private void walk(Sequence s, HashSet<Element> hs) {
+            hs.add(s);
+            for(Position p = s.firstp(); p != null; p = p.next())
+                walk(p.element(), hs);
+            for(Sequence ss : s.needs()) walk(ss, hs);
+            for(Sequence ss : s.hates()) walk(ss, hs);
         }
 
         /** the start state */
-        public final State<Tok>   start;
+        public  final State<Tok>   start;
+
+        /** the state from which no reductions can be done */
+        private final State<Tok>   dead_state;
 
         /** used to generate unique values for State.idx */
         private int master_state_idx = 0;
@@ -113,6 +120,8 @@ public abstract class Parser<Tok, Result> {
                 cache.ys.addAll(e, new Walk.YieldSet(e, cache).walk());
             HashSet<Position> hp = new HashSet<Position>();
             reachable(start0, hp);
+
+            this.dead_state = new State<Tok>(new HashSet<Position>(), all_states, all_elements);
             this.start = new State<Tok>(hp, all_states, all_elements);
 
             // for each state, fill in the corresponding "row" of the parse table
@@ -170,15 +179,15 @@ public abstract class Parser<Tok, Result> {
 
             // Interface Methods //////////////////////////////////////////////////////////////////////////////
 
-            boolean             isAccepting()               { return accept; }
-            public Iterator<Position>  iterator()                  { return hs.iterator(); }
+            boolean             isAccepting()           { return accept; }
+            public Iterator<Position>  iterator()       { return hs.iterator(); }
 
-            boolean             canShift(Tok t)           { return oshifts.contains(t); }
+            boolean             canShift(Tok t)         { return oshifts!=null && oshifts.contains(t); }
             <B,C> void          invokeShifts(Tok t, Invokable<State<Tok>,B,C> irbc, B b, C c) {
                 oshifts.invoke(t, irbc, b, c);
             }
 
-            boolean             canReduce(Tok t)          { return t==null ? eofReductions.size()>0 : oreductions.contains(t); }
+            boolean             canReduce(Tok t)        { return oreductions != null && (t==null ? eofReductions.size()>0 : oreductions.contains(t)); }
             <B,C> void          invokeReductions(Tok t, Invokable<Position,B,C> irbc, B b, C c) {
                 if (t==null) for(Position r : eofReductions) irbc.invoke(r, b, c);
                 else         oreductions.invoke(t, irbc, b, c);
@@ -248,6 +257,7 @@ public abstract class Parser<Tok, Result> {
                 //         "yields" [in one or more step] is used instead of "produces" [in exactly one step]
                 //         to avoid having to iteratively construct our set of States as shown in most
                 //         expositions of the algorithm (ie "keep doing XYZ until things stop changing").
+
                 HashMapBag<Element,Position> move = new HashMapBag<Element,Position>();
                 for(Position p : hs) {
                     Element e = p.element();
@@ -261,7 +271,11 @@ public abstract class Parser<Tok, Result> {
                 for(Element y : move) {
                     HashSet<Position> h = move.getAll(y);
                     State<Tok> s = all_states.get(h) == null ? new State<Tok>(h, all_states, all_elements) : all_states.get(h);
-                    gotoSetNonTerminals.put(y, s);
+                    // if a reduction is "lame", it should wind up in the dead_state after reducing
+                    if (y instanceof Sequence && ((Sequence)y).lame)
+                        ((HashMap)gotoSetNonTerminals).put(y, dead_state);
+                    else
+                        gotoSetNonTerminals.put(y, s);
                 }
             }
 
@@ -287,10 +301,15 @@ public abstract class Parser<Tok, Result> {
 
     // Helpers //////////////////////////////////////////////////////////////////////////////
     
+    private static void reachable(Sequence s, HashSet<Position> h) {
+        reachable(s.firstp(), h);
+        for(Sequence ss : s.needs()) reachable(ss, h);
+        for(Sequence ss : s.hates()) reachable(ss, h);
+    }
     private static void reachable(Element e, HashSet<Position> h) {
         if (e instanceof Atom) return;
         for(Sequence s : ((Union)e))
-            reachable(s.firstp(), h);
+            reachable(s, h);
     }
     private static void reachable(Position p, HashSet<Position> h) {
         if (h.contains(p)) return;