src/edu/berkeley/sbp/Forest.java

   1 // Copyright 2006-2007 all rights reserved; see LICENSE file for BSD-style license
   2
   3 package edu.berkeley.sbp;
   4 import edu.berkeley.sbp.util.*;
   5 import java.io.*;
   6 import java.util.*;
   7
   8 /**
   9  *   <font color=blue>
  10  *   An efficient representation of a collection of trees (Tomita's
  11  *   shared packed parse forest).
  12  *   </font>
  13  */
  14 public abstract class Forest<NodeType> implements GraphViz.ToGraphViz {
  15
  16     /** assume that this forest contains exactly one tree and return it; otherwise throw an exception */
  17     public abstract Tree<NodeType> expand1() throws Ambiguous;
  18
  19     /** expand this forest into a set of trees */
  20     public Iterable<Tree<NodeType>> expand() {
  21         HashSet<Tree<NodeType>> ht = new HashSet<Tree<NodeType>>();
  22         expand(ht, new HashSet<Forest<NodeType>>(), null);
  23         return ht;
  24     }
  25
  26     /** returns the input Region which this Forest was parsed from */
  27     public abstract Input.Region getRegion();
  28
  29     // Package-Private //////////////////////////////////////////////////////////////////////////////
  30
  31     public static <NodeType> Forest<NodeType> create(Input.Region region, NodeType head, Forest<NodeType>[] children) {
  32         return create(region, head, children, new boolean[children==null ? 0 : children.length]); }
  33     public static <NodeType> Forest<NodeType> create(Input.Region region, NodeType head, Forest<NodeType>[] children, boolean[] lifts) {
  34         if (region == null) throw new RuntimeException("invoked Forest.create(region=null) -- this should never happen");
  35         return new One<NodeType>(region, head, children, lifts);
  36     }
  37
  38     abstract void expand(HashSet<Tree<NodeType>> ht, HashSet<Forest<NodeType>> ignore, Tree<NodeType> bogus);
  39     abstract void gather(HashSet<Forest<NodeType>> ignore);
  40     abstract void edges(GraphViz.StateNode n);
  41     boolean ambiguous() { return false; }
  42
  43     // One //////////////////////////////////////////////////////////////////////////////
  44
  45     /** A "single" forest with a head and child subforests */
  46     private static class One<NodeType> extends Forest<NodeType> {
  47
  48         private final Input.Region      location;
  49         private final NodeType                head;
  50         private final Forest<NodeType>[]       children;
  51
  52         /** if true, the last child's children are considered children of this node */
  53         private final boolean[]         lifts;
  54
  55         public Input.Region getRegion() { return location; }
  56
  57         private One(Input.Region loc, NodeType head, Forest<NodeType>[] children, boolean[] lifts) {
  58             this.location = loc;
  59             this.head = head;
  60             if (head==null) throw new RuntimeException("invoked Forest.create(,null,,,) -- this should never happen");
  61             this.children = children==null ? emptyForestArray : new Forest[children.length];
  62             if (children != null) System.arraycopy(children, 0, this.children, 0, children.length);
  63             if (children != null) for(int i=0; i<children.length; i++) if (children[i]==null) throw new Error(i+"");
  64             this.lifts = lifts;
  65         }
  66
  67         public Tree<NodeType> expand1() throws Ambiguous {
  68             if (children.length == 0)
  69                 return new Tree<NodeType>(location, head);
  70             else
  71                 return expand1ForOneAndMany((Forest<NodeType>)this);
  72         }
  73
  74         void gather(HashSet<Forest<NodeType>> hf) {
  75             hf.add(this);
  76             for(Forest<NodeType> f : children) f.gather(hf);
  77         }
  78         void expand(HashSet<Tree<NodeType>> ht, HashSet<Forest<NodeType>> ignore, Tree<NodeType> bogus) {
  79             if (ignore.contains(this)) { ht.add(bogus); return; }
  80             expand(0, new Tree[children.length], ht, ignore, bogus);
  81         }
  82         private void expand(final int i, Tree<NodeType>[] ta, HashSet<Tree<NodeType>> ht, HashSet<Forest<NodeType>> ignore,
  83                             Tree<NodeType> bogus) {
  84             if (i==children.length) {
  85                 ht.add(new Tree<NodeType>(location, head, ta, lifts));
  86             } else {
  87                 HashSet<Tree<NodeType>> ht2 = new HashSet<Tree<NodeType>>();
  88                 children[i].expand(ht2, ignore, bogus);
  89                 for(Tree<NodeType> tc : ht2) {
  90                     ta[i] = tc;
  91                     expand(i+1, ta, ht, ignore, bogus);
  92                     ta[i] = null;
  93                 }
  94             }
  95         }
  96
  97         // GraphViz, ToInt //////////////////////////////////////////////////////////////////////////////
  98
  99         public boolean isTransparent() { return false; }
 100         public boolean isHidden() { return false; }
 101         public GraphViz.StateNode toGraphViz(GraphViz gv) {
 102             if (gv.hasNode(this)) return gv.createNode(this);
 103             GraphViz.StateNode n = gv.createNode(this);
 104             n.label = headToString()==null?"":headToString();
 105             n.directed = true;
 106             edges(n);
 107             return n;
 108         }
 109         boolean edges = false; // FIXME ??
 110         public void edges(GraphViz.StateNode n) {
 111             if (edges) return;
 112             edges = true;
 113             for(int i=0; i<children.length; i++) {
 114                 if (lifts[i] && !children[i].ambiguous()) {
 115                     children[i].edges(n);
 116                 } else {
 117                     n.edge(children[i], null);
 118                 }
 119             }
 120         }
 121
 122         protected String  headToString()         { return head==null?null:head.toString(); }
 123         protected String  headToJava()           { return "null"; }
 124         protected String  left()                 { return "{"; }
 125         protected String  right()                { return "}"; }
 126         protected boolean ignoreSingleton()      { return false; }
 127     }
 128
 129
 130     // Many //////////////////////////////////////////////////////////////////////////////
 131
 132     /** An "ambiguity node"; this is immutable once it has been "looked at" */
 133     static class Many<NodeType> extends Forest<NodeType> {
 134
 135         private FastSet<Forest<NodeType>> hp = new FastSet<Forest<NodeType>>();
 136         private boolean touched = false;
 137
 138         public Many() { }
 139
 140         public Input.Region getRegion() { return hp.iterator().next().getRegion(); } // all should be identical
 141
 142         Forest<NodeType> simplify() throws Ambiguous {
 143             touched();
 144             if (hp.size() > 1) {
 145                 HashSet<Forest<NodeType>> hf0 = new HashSet<Forest<NodeType>>();
 146                 Iterator<Forest<NodeType>> ih = hp.iterator();
 147                 ih.next().gather(hf0);
 148                 for(Forest<NodeType> f : hp) {
 149                     HashSet<Forest<NodeType>> hf1 = new HashSet<Forest<NodeType>>();
 150                     f.gather(hf1);
 151                     hf0.retainAll(hf1);
 152                 }
 153                 HashSet<Tree<NodeType>> ht = new HashSet<Tree<NodeType>>();
 154                 expand(ht, hf0, new Tree(null, "*"));
 155                 throw new Ambiguous((Forest<?>)this,
 156                                     (HashSet<Tree<?>>)(Object)ht);
 157             }
 158
 159             return hp.iterator().next();
 160         }
 161
 162         public Tree<NodeType> expand1() throws Ambiguous {
 163             return simplify().expand1();
 164         }
 165
 166         void gather(HashSet<Forest<NodeType>> ht) {
 167             touched();
 168
 169             // FIXME: do something more sensible here
 170             if (ht.contains(this)) {
 171                 System.err.println("WARNING: grammar produced a circular forest\n" + this);
 172                 //throw new Error("grammar produced a circular forest:\n" + this);
 173                 return;
 174             }
 175
 176             ht.add(this);
 177             for(Forest<NodeType> f : hp) f.gather(ht);
 178         }
 179
 180         private void touched() {
 181             if (touched) return;
 182             touched = true;
 183             /*
 184             FastSet<Forest<NodeType>> f2 = new FastSet<Forest<NodeType>>();
 185             for(Forest f : hp)
 186                 if (f instanceof Forest.One) f2.add(f);
 187                 else for(Forest ff : ((Forest.Many<NodeType>)f))
 188                     f2.add(ff);
 189             hp = f2;
 190             */
 191         }
 192         public boolean contains(Forest f) {
 193             touched();
 194             return hp.contains(f);
 195         }
 196         public void merge(Forest p) {
 197             if (touched) throw new RuntimeException("attempt to merge() on a Forest.Many that has already been examined");
 198             if (p==this) throw new RuntimeException("attempt to merge() a Forest.Many to itself!");
 199             hp.add(p, true);
 200         }
 201         boolean ambiguous() {
 202             touched();
 203             if (hp.size()==0) return false;
 204             if (hp.size()==1) return hp.iterator().next().ambiguous();
 205             return true;
 206         }
 207
 208         void expand(HashSet<Tree<NodeType>> ht, HashSet<Forest<NodeType>> ignore, Tree<NodeType> bogus) {
 209             touched();
 210             if (ignore.contains(this)) { ht.add(bogus); return; }
 211             for (Forest<NodeType> f : hp) f.expand(ht, ignore, bogus);
 212         }
 213
 214
 215         // GraphViz, ToInt //////////////////////////////////////////////////////////////////////////////
 216
 217         public boolean isTransparent() { return hp.size()==1; }
 218         public boolean isHidden() { return hp.size()==0; }
 219         public void edges(GraphViz.StateNode n) {
 220             if (hp.size()==1) { hp.iterator().next().edges(n); return; }
 221             for(Forest f : hp) f.edges(n);
 222         }
 223         public GraphViz.StateNode toGraphViz(GraphViz gv) {
 224             if (hp.size()==1) return hp.iterator().next().toGraphViz(gv);
 225             if (gv.hasNode(this)) return gv.createNode(this);
 226             GraphViz.StateNode n = gv.createNode(this);
 227             n.label = "?";
 228             n.color = "red";
 229             for(Forest f : hp) n.edge(f, null);
 230             return n;
 231         }
 232     }
 233
 234
 235     // Statics //////////////////////////////////////////////////////////////////////////////
 236
 237     /** Depth-first expansion, handling .One and .Many without recursion. */
 238     private static <NodeType> Tree<NodeType> expand1ForOneAndMany(
 239             Forest<NodeType> forest) throws Ambiguous {
 240         List<Tree<NodeType>> nodes = new ArrayList<Tree<NodeType>>();
 241         List<Tree<NodeType>> build = new ArrayList<Tree<NodeType>>();
 242
 243         // path stack
 244         List<Forest.One<NodeType>> path = new ArrayList<Forest.One<NodeType>>();
 245         List<Integer> sizes = new ArrayList<Integer>(),
 246                       poss  = new ArrayList<Integer>();
 247
 248         // handle the left-most leaf
 249         expand1Descend(path, nodes, sizes, poss, forest);
 250
 251         for (Forest.One<NodeType> f; path.size() > 0;) {
 252             // up one
 253             f = pop(path);
 254             int size = pop(sizes) + 1;
 255             int pos = pop(poss) + 1;
 256             Forest<NodeType>[] children = f.children;
 257             if (pos < children.length) {
 258                 // down the next branch
 259                 path.add(f);
 260                 sizes.add(new Integer(size));
 261                 poss.add(new Integer(pos));
 262
 263                 // handle the left-most leaf
 264                 expand1Descend(path, nodes, sizes, poss, children[pos]);
 265             } else {
 266                 if (path.size() > 0 && peek(path).lifts[peek(poss)]) {
 267                     // skip assembling this node, lift children
 268                     sizes.add(pop(sizes) + size - 1);
 269                     continue;
 270                 }
 271                 // assemble this node
 272                 for (int i=size; i > 0; i--)
 273                     build.add(nodes.remove(nodes.size() - i));
 274                 nodes.add(new Tree<NodeType>(f.location, f.head, build));
 275                 build.clear();
 276             }
 277         }
 278
 279         return pop(nodes);
 280     }
 281
 282     /** Descend to the left-most leaf, building the path. */
 283     private static <NodeType> void expand1Descend(
 284             List<Forest.One<NodeType>> path, List<Tree<NodeType>> nodes,
 285             List<Integer> sizes, List<Integer> poss, Forest<NodeType> f)
 286             throws Ambiguous {
 287         while (true) {
 288             if (f instanceof Forest.Many)
 289                 f = ((Forest.Many<NodeType>)f).simplify();
 290             else if (f instanceof Forest.One) {
 291                 Forest.One<NodeType> one = (Forest.One<NodeType>)f;
 292                 if (one.children.length == 0)
 293                     break;
 294                 path.add(one);
 295                 sizes.add(0);
 296                 poss.add(0);
 297                 f = one.children[0];
 298             } else {
 299                 nodes.add(f.expand1());
 300                 return;
 301             }
 302         }
 303
 304         if (path.size() > 0 && peek(path).lifts[peek(poss)])
 305             sizes.add(pop(sizes) - 1); // ignore lifted leafs
 306         else
 307             nodes.add(f.expand1());
 308     }
 309
 310     private static <T> T pop(List<T> list) {
 311         return list.remove(list.size() - 1); }
 312     private static <T> T peek(List<T> list) {
 313         return list.get(list.size() - 1); }
 314
 315     private static Tree[] tree_hint = new Tree[0];
 316     private static String[] string_hint = new String[0];
 317     private static final Forest[] emptyForestArray = new Forest[0];
 318
 319     protected String  headToString()    { return null; }
 320     protected String  headToJava()      { return "null"; }
 321     protected String  left()            { return "<?"; }
 322     protected String  right()           { return "?>"; }
 323     protected boolean ignoreSingleton() { return true; }
 324 }