X-Git-Url: http://git.megacz.com/?p=sbp.git;a=blobdiff_plain;f=src%2Fedu%2Fberkeley%2Fsbp%2FForest.java;h=1730f607eba82daeb513f88053b53d08137a33e2;hp=98e8e4bb35130aec1f4a4376f6c22f4ad86dea37;hb=0ddfec9ac2783f59115205dc0f3806137265c59c;hpb=035891a122d39983bf9f521acbf4661eb84aacd6

diff --git a/src/edu/berkeley/sbp/Forest.java b/src/edu/berkeley/sbp/Forest.java
index 98e8e4b..1730f60 100644
--- a/src/edu/berkeley/sbp/Forest.java
+++ b/src/edu/berkeley/sbp/Forest.java
@@ -1,3 +1,5 @@
+// Copyright 2006 all rights reserved; see LICENSE file for BSD-style license
+
 package edu.berkeley.sbp;
 import edu.berkeley.sbp.*;
 import edu.berkeley.sbp.Sequence.Position;
@@ -6,177 +8,230 @@ import java.io.*;
 import java.util.*;
 import java.lang.reflect.*;
 
-/** an efficient representation of a collection of trees (Tomita's shared packed parse forest) */
-public abstract class Forest<T> /*extends PrintableTree<Forest.MyBody<T>>*/ implements Visitable<Forest.Body<T>>, IntegerMappable {
-
-    private static int master_idx = 0;
-    private final int idx = master_idx++;
-    public int toInt() { return idx; }
+/**
+ *   <font color=blue>
+ *   An efficient representation of a collection of trees (Tomita's
+ *   shared packed parse forest).
+ *   </font>
+ */
+public abstract class Forest<NodeType> implements GraphViz.ToGraphViz {
 
     /** assume that this forest contains exactly one tree and return it; otherwise throw an exception */
-    public final Tree<T> expand1() throws Ambiguous, ParseFailed {
-        try {
-            Iterator<Tree<T>> it = expand(true).iterator();
-            if (!it.hasNext()) throw new ParseFailed();
-            return it.next();
-        } catch (InnerAmbiguous ia) { throw new Ambiguous(ia.f); }
-    }
+    public abstract Tree<NodeType> expand1() throws Ambiguous;
 
     /** expand this forest into a set of trees */
-    public HashSet<Tree<T>> expand(boolean toss) {
-        //if (toss) scan();
-        final HashSetTreeConsumer<T> ret = new HashSetTreeConsumer<T>();
-        visit(new TreeMaker2<T>(toss, ret), null, null);
-        if (toss && ret.size() > 1) throw new InnerAmbiguous(this);
-        return ret;
-    }
+    public void expand(HashSet<Tree<NodeType>> ht) { expand(ht, new HashSet<Forest<NodeType>>(), null); }
 
-    private static class InnerAmbiguous extends RuntimeException {
-        public final Forest<?> f;
-        public InnerAmbiguous(Forest<?> f) { this.f = f; }
+    static <NodeType> Forest<NodeType> create(Input.Region loc, NodeType head, Forest<NodeType>[] children, boolean lift) {
+        if (loc == null) throw new RuntimeException("invoked Forest.create(null,,,,) -- this should never happen");
+        return new One<NodeType>(loc, head, children, lift);
     }
 
-    public static interface TreeConsumer<T> {
-        public void addTree(Tree<T> t);
-    }
-    public static class HashSetTreeConsumer<T> extends HashSet<Tree<T>> implements TreeConsumer<T> {
-        public void addTree(Tree<T> t) { super.add(t); }
-    }
+    /** create a new forest */
+    public static <NodeType> Forest<NodeType> create(Input.Region loc, NodeType head, Forest<NodeType>[] children) {
+        return Forest.create(loc, head, children, false); }
 
-    static        <T> Forest<T> singleton(Input.Location loc, Position p) {
-        return create(loc, null, new Forest[] { }, false, true, p); }
-    static        <T> Forest<T> singleton(Input.Location loc, Forest<T> body, Position p) {
-        return create(loc, null, new Forest[] { body },  false, true, p); }
-    static        <T> Forest<T> leaf(Input.Location loc, T tag, Position p) { return create(loc, tag, null, false, false, p); }
+    // Package-Private //////////////////////////////////////////////////////////////////////////////
 
-    public static <T> Forest<T> create(Input.Location loc, T tag, Forest<T>[] tokens, boolean unwrap, boolean singleton, Position p) {
-        return new MyBody<T>(loc, tag, tokens, unwrap, singleton, p);
-    }
+    abstract void expand(HashSet<Tree<NodeType>> ht, HashSet<Forest<NodeType>> ignore, Tree<NodeType> bogus);
+    abstract void gather(HashSet<Forest<NodeType>> ignore);
+    abstract void edges(GraphViz.Node n);
+    boolean ambiguous() { return false; }
+    
+    abstract Input.Region getRegion();
 
-    // Body //////////////////////////////////////////////////////////////////////////////
+    // One //////////////////////////////////////////////////////////////////////////////
 
-    protected static interface Body<T> {
-        void expand(int i, TreeMaker<T> h);
-    }
+    /** A "single" forest with a head and child subforests */    
+    private static class One<NodeType> extends Forest<NodeType> {
 
-    protected static class MyBody<T> extends Forest<T> implements Body<T> /* extends PrintableTree<Forest<T>> implements */ {
+        private final Input.Region      location;
+        private final NodeType                head;
+        private final Forest<NodeType>[]       children;
 
-        public <B,C> void visit(Invokable<Forest.Body<T>,B,C> ivbc, B b, C c) {
-            ivbc.invoke(this, b, c);
-        }
+        /** if true, the last child's children are considered children of this node */
+        private final boolean           lift;
 
-        private final Input.Location    location;
-        private final T                 tag;
-        private final Forest<T>[]       tokens;
-        private final boolean           unwrap;
-        private final boolean           singleton;
-        private final Sequence.Position reduction;
+        Input.Region getRegion() { return location; }
 
-        private MyBody(Input.Location loc, T tag, Forest<T>[] tokens, boolean unwrap, boolean singleton, Position reduction) {
+        private One(Input.Region loc, NodeType head, Forest<NodeType>[] children, boolean lift) {
             this.location = loc;
-            this.tag = tag;
-            this.tokens = tokens==null ? emptyForestArray : new Forest[tokens.length];
-            if (tokens != null) System.arraycopy(tokens, 0, this.tokens, 0, tokens.length);
-            if (tokens != null) for(int i=0; i<tokens.length; i++) if (tokens[i]==null) throw new Error(i+"");
-            this.unwrap = unwrap;
-            this.singleton = singleton;
-            this.reduction = reduction;
-        }
-
-        public void expand(final int i, final TreeMaker<T> h) {
-            if (singleton) {
-                tokens[0].visit(h, null, i);
-                return;
-            }
-            if (i==0) h.start(tag, location);
-
-            if (i==tokens.length) {
-                h.finish(tag, location);
+            this.head = head;
+            if (head==null) throw new RuntimeException("invoked Forest.create(,null,,,) -- this should never happen");
+            this.children = children==null ? emptyForestArray : new Forest[children.length];
+            if (children != null) System.arraycopy(children, 0, this.children, 0, children.length);
+            if (children != null) for(int i=0; i<children.length; i++) if (children[i]==null) throw new Error(i+"");
+            this.lift = lift;
+        }
 
-            } else if (unwrap && i==tokens.length-1) {
-                if (tokens[i] != null)
-                    tokens[i].visit(h, null, 0);
+        public Tree<NodeType> expand1() throws Ambiguous {
+            Tree<NodeType>[] ret = new Tree[children.length];
+            for(int i=0; i<children.length; i++) ret[i] = children[i].expand1();
+            return new Tree<NodeType>(location, head, ret, lift);
+        }
 
+        void gather(HashSet<Forest<NodeType>> hf) {
+            hf.add(this);
+            for(Forest<NodeType> f : children) f.gather(hf);
+        }
+        void expand(HashSet<Tree<NodeType>> ht, HashSet<Forest<NodeType>> ignore, Tree<NodeType> bogus) {
+            if (ignore.contains(this)) { ht.add(bogus); return; }
+            expand(0, new Tree[children.length], ht, ignore, bogus);
+        }
+        private void expand(final int i, Tree<NodeType>[] ta, HashSet<Tree<NodeType>> ht, HashSet<Forest<NodeType>> ignore, Tree<NodeType> bogus) {
+            if (i==children.length) {
+                ht.add(new Tree<NodeType>(location, head, ta, lift));
             } else {
-                tokens[i].visit(new TreeMaker<T>(h.toss) {
-                    public void start(T head, Input.Location loc) { }
-                    public void addTree(Tree<T> t) { toks.add(t); }
-                    public void finish(T head, Input.Location loc) {
-                        int old = h.toks.size();
-                        h.addTree(new Tree<T>(loc, head, toks.toArray(tree_hint)));
-                        expand(i+1, h);
-                        while(h.toks.size() > old) h.toks.remove(h.toks.size()-1);
-                    }
-                }, null, null);
+                HashSet<Tree<NodeType>> ht2 = new HashSet<Tree<NodeType>>();
+                children[i].expand(ht2, ignore, bogus);
+                for(Tree<NodeType> tc : ht2) {
+                    ta[i] = tc;
+                    expand(i+1, ta, ht, ignore, bogus);
+                    ta[i] = null;
+                }
             }
         }
 
-        protected String  headToString()         { return null; }
-        protected String  headToJava()           { return null; }
+        // GraphViz, ToInt //////////////////////////////////////////////////////////////////////////////
+
+        public boolean isTransparent() { return false; }
+        public boolean isHidden() { return false; }
+        public GraphViz.Node toGraphViz(GraphViz gv) {
+            if (gv.hasNode(this)) return gv.createNode(this);
+            GraphViz.Node n = gv.createNode(this);
+            n.label = headToString()==null?"":headToString();
+            n.directed = true;
+            edges(n);
+            return n;
+        }
+        boolean edges = false; // FIXME ??
+        public void edges(GraphViz.Node n) {
+            if (edges) return;
+            edges = true;
+            for(int i=0; i<children.length; i++) {
+                if (i==children.length-1 && lift && !children[i].ambiguous()) {
+                    children[i].edges(n);
+                } else {
+                    n.edge(children[i], null);
+                }
+            }
+        }
+
+        protected String  headToString()         { return head==null?null:head.toString(); }
+        protected String  headToJava()           { return "null"; }
         protected String  left()                 { return "{"; }
         protected String  right()                { return "}"; }
         protected boolean ignoreSingleton()      { return false; }
     }
 
 
-    // Ref //////////////////////////////////////////////////////////////////////////////
+    // Many //////////////////////////////////////////////////////////////////////////////
+
+    /** An "ambiguity node"; this is immutable once it has been "looked at" */
+    static class Many<NodeType> extends Forest<NodeType> {
 
-    /**
-     *  This class represents a partially complete collection of
-     *  forests to be viewed as a forest at some later date; once
-     *  viewed, it becomes immutable
-     */
-    static class Ref<T> extends Forest<T> {
-        private FastSet<Forest<T>> hp = new FastSet<Forest<T>>();
-        public Ref() { }
-        public int toInt() {
-            if (hp.size()==1) return hp.iterator().next().toInt();
-            return super.toInt();
+        private FastSet<Forest<NodeType>> hp = new FastSet<Forest<NodeType>>();
+        private boolean touched = false;
+
+        public Many() { }
+
+        Input.Region getRegion() { return hp.iterator().next().getRegion(); } // all should be identical
+
+        public Tree<NodeType> expand1() throws Ambiguous {
+            touched();
+            if (hp.size() > 1) {
+                HashSet<Forest<NodeType>> hf0 = new HashSet<Forest<NodeType>>();
+                Iterator<Forest<NodeType>> ih = hp.iterator();
+                ih.next().gather(hf0);
+                for(Forest<NodeType> f : hp) {
+                    HashSet<Forest<NodeType>> hf1 = new HashSet<Forest<NodeType>>();
+                    f.gather(hf1);
+                    hf0.retainAll(hf1);
+                }
+                HashSet<Tree<NodeType>> ht = new HashSet<Tree<NodeType>>();
+                expand(ht, hf0, new Tree(null, "*"));
+                throw new Ambiguous((Forest<?>)this,
+                                    (HashSet<Tree<?>>)(Object)ht);
+            }
+            return hp.iterator().next().expand1();
         }
-        public void merge(Forest p) { if (p!=this) hp.add(p, true); }
-        public <B,C> void visit(Invokable<Forest.Body<T>,B,C> ivbc, B b, C c) {
-            if (hp==null) return;
-            for(Forest<T> f : hp)
-                f.visit(ivbc, b, c);
+        
+        void gather(HashSet<Forest<NodeType>> ht) {
+            touched();
+
+            // FIXME: do something more sensible here
+            if (ht.contains(this)) {
+                System.err.println("WARNING: grammar produced a circular forest\n" + this);
+                //throw new Error("grammar produced a circular forest:\n" + this);
+                return;
+            }
+
+            ht.add(this);
+            for(Forest<NodeType> f : hp) f.gather(ht);
         }
-        public Forest resolve() { return this; }
-    }
 
-    public abstract <B,C> void visit(Invokable<Forest.Body<T>,B,C> ivbc, B b, C c);
-    private static class TreeMaker2<T> extends TreeMaker<T> {
-        private TreeConsumer<T> tc;
-        public TreeMaker2(boolean toss, TreeConsumer<T> tc) { super(toss); this.tc = tc; }
-        public void finish(T head, Input.Location loc) { tc.addTree(new Tree<T>(loc, head, toks.toArray(tree_hint)));; }
-        public void start(T head, Input.Location loc) { }
-        public void addTree(Tree<T> t) { toks.add(t); }
-    }
-    private static abstract class TreeMaker<T> implements Invokable<Forest.Body<T>,Boolean,Integer>, TreeConsumer<T> {
-        public ArrayList<Tree<T>> toks = new ArrayList<Tree<T>>();
-        private boolean toss;
-        protected T head;
-        public TreeMaker(boolean toss) { this.toss = toss; }
-        public abstract void start(T head, Input.Location loc);
-        public abstract void finish(T head, Input.Location loc);
-        public abstract void addTree(Tree<T> t);
-        public void invoke(Forest.Body<T> bod, Boolean o, Integer i) {
-            if (i==null) {
-                ArrayList<Tree<T>> toks = this.toks;
-                this.toks = new ArrayList<Tree<T>>();
-                bod.expand(0, this);
-                this.toks = toks;
-            } else {
-                bod.expand(i, this);
-            }
+        private void touched() {
+            if (touched) return;
+            touched = true;
+            /*
+            FastSet<Forest<NodeType>> f2 = new FastSet<Forest<NodeType>>();
+            for(Forest f : hp)
+                if (f instanceof Forest.One) f2.add(f);
+                else for(Forest ff : ((Forest.Many<NodeType>)f))
+                    f2.add(ff);
+            hp = f2;
+            */
+        }
+        public boolean contains(Forest f) {
+            touched();
+            return hp.contains(f);
+        }
+        public void merge(Forest p) { 
+            if (touched) throw new RuntimeException("attempt to merge() on a Forest.Many that has already been examined");
+            if (p==this) throw new RuntimeException("attempt to merge() a Forest.Many to itself!");
+            hp.add(p, true);
+        }
+        boolean ambiguous() {
+            touched();
+            if (hp.size()==0) return false;
+            if (hp.size()==1) return hp.iterator().next().ambiguous();
+            return true;
+        }
+
+        void expand(HashSet<Tree<NodeType>> ht, HashSet<Forest<NodeType>> ignore, Tree<NodeType> bogus) {
+            touched();
+            if (ignore.contains(this)) { ht.add(bogus); return; }
+            for (Forest<NodeType> f : hp) f.expand(ht, ignore, bogus);
+        }
+
+
+        // GraphViz, ToInt //////////////////////////////////////////////////////////////////////////////
+
+        public boolean isTransparent() { return hp.size()==1; }
+        public boolean isHidden() { return hp.size()==0; }
+        public void edges(GraphViz.Node n) {
+            if (hp.size()==1) { hp.iterator().next().edges(n); return; }
+            for(Forest f : hp) f.edges(n);
+        }
+        public GraphViz.Node toGraphViz(GraphViz gv) {
+            if (hp.size()==1) return hp.iterator().next().toGraphViz(gv);
+            if (gv.hasNode(this)) return gv.createNode(this);
+            GraphViz.Node n = gv.createNode(this);
+            n.label = "?";
+            n.color = "red";
+            for(Forest f : hp) n.edge(f, null);
+            return n;
         }
     }
 
     // Statics //////////////////////////////////////////////////////////////////////////////
 
     private static Tree[] tree_hint = new Tree[0];
+    private static String[] string_hint = new String[0];
     private static final Forest[] emptyForestArray = new Forest[0];
 
     protected String  headToString()    { return null; }
-    protected String  headToJava()      { return null; }
+    protected String  headToJava()      { return "null"; }
     protected String  left()            { return "<?"; }
     protected String  right()           { return "?>"; }
     protected boolean ignoreSingleton() { return true; }