X-Git-Url: http://git.megacz.com/?p=sbp.git;a=blobdiff_plain;f=src%2Fedu%2Fberkeley%2Fsbp%2FSequence.java;h=1c76aefadda764a09bf478877d764ecb108c0675;hp=e5e984f4e95b307c10b60b61e0c67e5aac07bab1;hb=584cef55d8811e3215858fde22e708d2a3d1cf70;hpb=c48a9498daf7604e41cf4aea7f6f7110388fd659

diff --git a/src/edu/berkeley/sbp/Sequence.java b/src/edu/berkeley/sbp/Sequence.java
index e5e984f..1c76aef 100644
--- a/src/edu/berkeley/sbp/Sequence.java
+++ b/src/edu/berkeley/sbp/Sequence.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.util.*;
 import edu.berkeley.sbp.*;
@@ -7,105 +9,138 @@ import java.util.*;
 import java.lang.reflect.*;
 import java.lang.ref.*;
 
-/** juxtaposition; zero or more adjacent Elements; can specify a rewriting */
-public abstract class Sequence extends Element implements Iterable<Element> {
+/** <font color=green>juxtaposition; zero or more adjacent Elements; can specify a rewriting</font> */
+public abstract class Sequence implements Iterable<Element>, SequenceOrElement {
 
-    // Static Constructors //////////////////////////////////////////////////////////////////////////////
+    protected final Element[] elements;
 
-    abstract Sequence _clone();
-    Sequence dup() {
-        Sequence ret = _clone();
-        for(Sequence s : needs) { ret.needs.add(s); s.needed.add(ret); }
-        for(Sequence s : hates) { ret.hates.add(s); s.hated.add(ret); }
-        ret.follow = follow;
-        return ret;
-    }
+    boolean needed_or_hated = false;
+    boolean in_a_union = false;
+
+    final HashSet<Sequence> needs  = new HashSet<Sequence>();
+    final HashSet<Sequence> hates  = new HashSet<Sequence>();
 
-    /** the empty sequence (matches the empty string) */
-    public static final Sequence empty = new Sequence.Constant.Empty();
+    // FIXME: these are ugly -- migrate into Grammar
+    HashMap<Sequence,Boolean> canNeed = new HashMap<Sequence,Boolean>();
+    HashMap<Sequence,Boolean> canKill = new HashMap<Sequence,Boolean>();
+
+    final Position          firstp;
+
+    Atom follow = null;
+
+    // Static Constructors //////////////////////////////////////////////////////////////////////////////
 
-    /** after matching the sequence, do not add anything to the output tree */
-    public static Sequence drop(Element[] e, boolean lame) { return new Constant.Drop(e, lame); }
+    /** create a sequence of one element */
+    public static Sequence create(Element e) { return create(new Element[] { e }, 0); }
 
-    /** after matching the sequence, insert a constant into the output tree */
-    public static Sequence constant(Element[] e, Object o) { return new Constant(e, o); }
+    /** create a sequence which drops the result of all but one of its element */
+    public static Sequence create(Element[] e, int which) {
+        return new Singleton(e, which); }
 
-    /** after matching the sequence, place the result of the <tt>idx</tt>th match in the output tree */
-    public static Sequence singleton(Element[] e, int idx) { return new Singleton(e, idx); }
-    public static Sequence singleton(Element e) { return singleton(new Element[] { e }, 0); }
+    /** create a sequence which always evaluates to a constant result  */
+    public static Sequence create(Object result, Element[] e) {
+        return new RewritingSequence(result, e, trues(e.length)); }
+
+    private static boolean[] trues(int length) {
+        boolean[] ret = new boolean[length];
+        for(int i=0; i<ret.length; i++) ret[i] = true;
+        return ret;
+    }
 
     /**
-     *  after matching the sequence, create the specified output tree
-     *  @param tag   the tag for the output tree
-     *  @param e     the elements to match
-     *  @param drops only elements of <tt>e</tt> whose corresponding <tt>boolean</tt> in <tt>drops</tt> is <i>false</i> will be included in the output tree
+     *  create a sequence (general form)
+     *  @param head   the head of the output tree
+     *  @param e      the elements to match
+     *  @param drop   only elements of <tt>e</tt> whose corresponding <tt>boolean</tt> in <tt>drops</tt>
+     *                is <i>false</i> will be included in the output tree
+     *  @param lifts  which (if any) child trees to lift
      **/
-    public static Sequence rewritingSequence(Object tag, Element[] e, Object[] labs, boolean[] drops) {
-        return new RewritingSequence(tag, e, labs, drops); }
-
-    ////////////////////////////////////////////////////////////////////////////////
+    public static Sequence create(Object head, Element[] e, boolean[] drop) {
+        return create(head, e, drop, new boolean[e.length]); }
+    public static Sequence create(Object head, Element[] e, boolean[] drop, boolean[] lifts) {
+        if (lifts==null) lifts = new boolean[e.length];
+        return new RewritingSequence(head, e, drop, lifts);
+    }
 
-    public Element follow = null;
-    public final Topology follow() { return follow==null ? null : Atom.toAtom(follow); }
+    /** return a new sequence identical to this one, but with a positive conjunct <tt>s</tt> */
+    public Sequence and(Sequence s) {
+        if (s.in_a_union)
+            throw new RuntimeException("you may not use a sequence as a conjunct if it belongs to a Union");
+        Sequence ret = dup();
+        ret.needs.add(s);
+        s.needed_or_hated=true;
+        return ret;
+    }
 
-    Topology toAtom() {
-        if (elements.length!=1) throw new RuntimeException("cannot invoke toAtom() on a Sequence with " + elements.length + " elements: " + this);
-        return Atom.toAtom(elements[0]);
+    /** return a new sequence identical to this one, but with a negative conjunct <tt>s</tt> */
+    public Sequence andnot(Sequence s) {
+        if (s.in_a_union)
+            throw new RuntimeException("you may not use a sequence as a conjunct if it belongs to a Union");
+        Sequence ret = dup();
+        ret.hates.add(s);
+        s.needed_or_hated=true;
+        return ret;
     }
 
-    public Sequence and(Sequence s) { Sequence ret = dup(); ret.needs.add(s); s.needed.add(ret); return ret; }
-    public Sequence not(Sequence s) { Sequence ret = dup(); ret.hates.add(s); s.hated.add(ret); return ret; }
+    /** return a new sequence identical to this one, but with a follow-set restricted to <tt>a</tt> */
+    public Sequence followedBy(Atom a) { Sequence ret = dup(); ret.follow = a; return ret; }
 
-    protected final Element[] elements;
+    ////////////////////////////////////////////////////////////////////////////////
 
-    final HashSet<Sequence> needed = new HashSet<Sequence>();
-    final HashSet<Sequence> hated = new HashSet<Sequence>();
-    final HashSet<Sequence> needs = new HashSet<Sequence>();
-    final HashSet<Sequence> hates = new HashSet<Sequence>();
-    public boolean           lame  = false;
+    abstract Sequence _clone();
+    private Sequence dup() {
+        Sequence ret = _clone();
+        for(Sequence s : needs) { ret.needs.add(s); }
+        for(Sequence s : hates) { ret.hates.add(s); }
+        ret.follow = follow;
+        return ret;
+    }
+
+    Iterable<Sequence> needs() { return needs; }
+    Iterable<Sequence> hates() { return hates; }
 
-    final Position          firstp;
     Position firstp() { return firstp; }
+    Position lastp() { return firstp().last(); }
 
     public Iterator<Element> iterator()    { return new ArrayIterator<Element>(elements); }
     protected Sequence(Element[] elements) {
         this.elements = elements;
-        this.firstp = new Position(0);
+        for(int i=0; i<elements.length; i++)
+            if (elements[i]==null)
+                throw new RuntimeException("cannot have nulls in a sequence: " + this);
+        this.firstp = new Position(0, null);
     }
 
-    // DO NOT MESS WITH THE FOLLOWING LINE!!!
-    private Forest.Ref epsilonForm = null;
-    Forest epsilonForm() {
-        if (epsilonForm!=null) return epsilonForm;
-        epsilonForm = new Forest.Ref();
-        epsilonForm.merge(firstp().rewrite(null, false));
-        return epsilonForm;
-    }
+    abstract Forest epsilonForm(Input.Region loc, Grammar cache);
 
-    protected abstract <T> Forest<T> postReduce(Input.Location loc, Forest<T>[] args, Position p);
+    protected abstract <T> Forest<T> postReduce(Input.Region loc, Forest<T>[] args, Position p);
 
 
     // Position //////////////////////////////////////////////////////////////////////////////
 
     /** the imaginary position before or after an element of a sequence; corresponds to an "LR item" */
-    public class Position implements IntegerMappable {
+    class Position implements IntegerMappable {
+
+        public int ord = -1;
 
         private Forest zero = null;
-        public Forest zero() {
+        /*
+        public Forest zero(Input.Region reg) {
             if (zero != null) return zero;
-            if (pos > 0) throw new Error();
-            return zero = rewrite(null);
+            if (pos > 0) throw new RuntimeException("Position.zero(): pos>0");
+            return zero = rewrite(reg);
         }
-
-
+        */
                 final int      pos;
         private final Position next;
+        private final Position prev;
                 final Forest[] holder;
         
-        private Position(int pos) {
+        private Position(int pos, Position prev) {
             this.pos      = pos;
-            this.next     = pos==elements.length ? null : new Position(pos+1);
+            this.next     = pos==elements.length ? null : new Position(pos+1, this);
             this.holder   = new Forest[elements.length];
+            this.prev     = prev;
         }
 
         boolean isFirst() { return pos==0; }
@@ -121,20 +156,19 @@ public abstract class Sequence extends Element implements Iterable<Element> {
 
         /** true iff this Position is the last one in the sequence */
         public boolean isLast() { return next()==null; }
+        public Position last() { return isLast() ? this : next().last(); }
+        public Position prev() { return prev; }
 
         // Position /////////////////////////////////////////////////////////////////////////////////
 
-        final <T> Forest<T> rewrite(Input.Location loc) { return rewrite(loc, true); }
-        private final <T> Forest<T> rewrite(Input.Location loc, boolean epsilonCheck) {
-            if (epsilonCheck && this==firstp()) return epsilonForm();
+        final <T> Forest<T> rewrite(Input.Region loc, Grammar cache) {
+            if (this==firstp()) epsilonForm(loc, cache);
             for(int i=0; i<pos; i++) if (holder[i]==null) throw new Error("realbad " + i);
             for(int i=pos; i<elements.length; i++) {
-                if (holder[i]==null) holder[i] = elements[i].epsilonForm();
-                if (holder[i]==null) throw new Error("bad " + i);
+                if (holder[i]==null) holder[i] = ((Union)elements[i]).epsilonForm(loc, cache);
+                if (holder[i]==null) throw new Error("bad");
             }
-            Forest<T> ret = Sequence.this.postReduce(loc, holder, this);
-            for(int k=0; k<pos; k++) holder[k] = null; // to help GC
-            return ret;
+            return Sequence.this.postReduce(loc, holder, this);
         }
 
         public String   toString() {
@@ -160,13 +194,21 @@ public abstract class Sequence extends Element implements Iterable<Element> {
     StringBuffer toString(StringBuffer sb) { return toString(sb, true); }
     StringBuffer toString(StringBuffer sb, boolean spacing) {
         for(int i=0; i<elements.length; i++) {
-            sb.append(elements[i].toString());
+            sb.append(elements[i]+"");
             sb.append(' ');
         }
         if (follow != null) {
             sb.append("-> ");
             sb.append(follow);
         }
+        for(Sequence s : needs) {
+            sb.append("& ");
+            sb.append(s);
+        }
+        for(Sequence s : hates) {
+            sb.append("&~ ");
+            sb.append(s);
+        }
         return sb;
     }
 
@@ -175,71 +217,56 @@ public abstract class Sequence extends Element implements Iterable<Element> {
 
     static class Constant extends Sequence {
         private final Object result;
-        public Constant(Element[] e, Object result) { super(e); this.result = result; }
+        public Constant(Element[] e, Object result) {
+            super(e);
+            if (result==null) throw new Error("constant sequences may not have result==null");
+            this.result = result;
+        }
         Sequence _clone() { return new Constant(elements, result); }
-        public <T> Forest<T> postReduce(Input.Location loc, Forest<T>[] args, Position p) {
-            return (Forest<T>)Forest.leaf(loc, result, p);
+        public <T> Forest<T> postReduce(Input.Region loc, Forest<T>[] args, Position p) {
+            return (Forest<T>)Forest.create(loc, result, null, false);
         }
-        static class Drop extends Constant {
-            Sequence _clone() { return new Drop(elements, lame); }
-            public Drop(Element[] e, boolean lame) {
-                super(e, null);
-                this.lame = lame;
-            }
+        Forest epsilonForm(Input.Region loc, Grammar cache) {
+            return Forest.create(loc, result, null, false);
         }
-        static class Empty extends Sequence.Constant.Drop {
-            Sequence _clone() { return new Empty(); }
-            public Empty() { super(new Element[] { }, false); } }
     }
 
     static class Singleton extends Sequence {
         private final int idx;
         public Singleton(Element e) { this(new Element[] { e }, 0); }
         public Singleton(Element[] e, int idx) { super(e); this.idx = idx; }
-        public <T> Forest<T> postReduce(Input.Location loc, Forest<T>[] args, Position p) { return (Forest<T>)Forest.singleton(loc, args[idx], p); }
+        public <T> Forest<T> postReduce(Input.Region loc, Forest<T>[] args, Position p) { return args[idx]; }
         Sequence _clone() { return new Singleton(elements,idx); }
-    }
-
-    public static Unwrap unwrap(Element[] e, Object tag, boolean[] drops) { return new Unwrap(e, tag, drops); }
-    static class Unwrap extends Sequence {
-        private boolean[] drops;
-        private final Object tag;
-        public Unwrap(Element[] e, Object tag)                  { super(e); this.drops = null; this.tag = tag; }
-        public Unwrap(Element[] e, Object tag, boolean[] drops) { super(e); this.drops = drops; this.tag = tag; }
-        Sequence _clone() { return new Unwrap(elements, drops); }
-        public <T> Forest<T> postReduce(Input.Location loc, Forest<T>[] args, Position p) {
-            for(int i=0; i<args.length; i++) if (args[i]==null) throw new Error();
-            if (drops==null) return Forest.create(loc, (T)tag, args, new Object[args.length], true, false, p);
-            int count = 0;
-            for(int i=0; i<drops.length; i++) if (!drops[i]) count++;
-            Forest<T>[] args2 = new Forest[count];
-            int j = 0;
-            for(int i=0; i<args.length; i++) if (!drops[i]) args2[j++] = args[i];
-            return Forest.create(loc, (T)tag, args2, new Object[args.length], true, false, p);
+        Forest epsilonForm(Input.Region loc, Grammar cache) {
+            return ((Union)elements[idx]).epsilonForm(loc, cache);
         }
     }
 
     static class RewritingSequence extends Sequence {
-        /*private*/public final Object tag;
+        private Object tag;
         private final boolean[] drops;
-        private final Object[] labs;
-        private int count = 0;
-        Sequence _clone() { return new RewritingSequence(tag, elements, labs, drops); }
-        public RewritingSequence(Object tag, Element[] e, Object[] labs) { this(tag, e, labs, null); }
-        public RewritingSequence(Object tag, Element[] e, Object[] labs, boolean[] drops) {
+        private final boolean[] lifts;
+        Sequence _clone() { return new RewritingSequence(tag, elements, drops); }
+        public RewritingSequence(Object tag, Element[] e) { this(tag, e, null); }
+        public RewritingSequence(Object tag, Element[] e, boolean[] drops) { this(tag, e, drops, new boolean[e.length]); }
+        public RewritingSequence(Object tag, Element[] e, boolean[] drops, boolean[] lifts) {
             super(e);
+            if (tag==null) throw new Error();
             this.tag = tag;
             this.drops = drops == null ? new boolean[e.length] : drops;
+            int count = 0;
             for(int i=0; i<this.drops.length; i++) if (!this.drops[i]) count++;
-            this.labs = labs;
+            this.lifts = new boolean[count];
+            int j = 0;
+            for(int i=0; i<this.drops.length; i++)
+                if (!this.drops[i])
+                    this.lifts[j++] = lifts[i];
         }
-        public <T> Forest<T> postReduce(Input.Location loc, Forest<T>[] args, Position p) {
-            Forest<T>[] args2 = new Forest[count];
-            Object[] labs2 = new Object[count];
+        public <T> Forest<T> postReduce(Input.Region loc, Forest<T>[] args, Position p) {
+            Forest<T>[] args2 = new Forest[lifts.length];
             int j = 0;
-            for(int i=0; i<args.length; i++) if (!drops[i]) { labs2[j] = labs==null?null:labs[i]; args2[j++] = args[i]; }
-            //System.out.println("reduce \""+tag+"\"");
-            return Forest.create(loc, (T)tag, args2, labs2, false, false, p);
+            for(int i=0; i<args.length; i++) if (!drops[i]) args2[j++] = args[i];
+            return Forest.create(loc, (T)tag, args2, lifts);
         }
         public StringBuffer toString(StringBuffer sb, boolean spacing) {
             int len = sb.length();
@@ -250,35 +277,8 @@ public abstract class Sequence extends Element implements Iterable<Element> {
             if (spacing) for(int i=0; i<50-len; i++) sb.append(' ');
             return sb;
         }
+        Forest epsilonForm(Input.Region loc, Grammar cache) {
+            return Forest.create(loc, tag, new Forest[0], lifts);
+        }
     }
-
-    // Repeat //////////////////////////////////////////////////////////////////////////////
-
-    /** repeat zero or one times */
-    public  static Repeat maybe(Element e)                             { return new Repeat(e, true, false, null, null); }
-    public  static Repeat maybe(Element e, Object tag)                 { return new Repeat(e, true, false, null, tag); }
-    /** repeat zero or more times */
-    public  static Repeat many0(Element e)                             { return new Repeat(e, true, true, null, null); }
-    public  static Repeat many0(Element e, Object tag)                 { return new Repeat(e, true, true, null, tag); }
-    /** repeat zero or more times, separated by <tt>sep</tt> */
-    public  static Repeat many0(Element e, Element sep)                { return new Repeat(e, true, true, sep, null); }
-    public  static Repeat many0(Element e, Element sep, Object tag)    { return new Repeat(e, true, true, sep, tag); }
-    /** repeat one or more times */
-    public  static Repeat many1(Element e)                             { return new Repeat(e, false, true, null, null); }
-    public  static Repeat many1(Element e, Object tag)                 { return new Repeat(e, false, true, null, tag); }
-    /** repeat one or more times, separated by <tt>sep</tt> */
-    public  static Repeat many1(Element e, Element sep)                { return new Repeat(e, false, true, sep, null); }
-    public  static Repeat many1(Element e, Element sep, Object tag)    { return new Repeat(e, false, true, sep, tag); }
-
-    /** repeat zero or more times, matching a maximal sequence of atoms */
-    public  static Repeat maximal0(Atom e)                             { return new Repeat.Maximal(e, true, true, null); }
-    public  static Repeat maximal0(Atom e, Object tag)                 { return new Repeat.Maximal(e, true, true, tag); }
-    /** repeat one or more times, matching a maximal sequence of atoms */
-    public  static Repeat maximal1(Atom e)                             { return new Repeat.Maximal(e, false, true, null); }
-    public  static Repeat maximal1(Atom e, Object tag)                 { return new Repeat.Maximal(e, false, true, tag); }
-    /** repeat one or more times, separated by an atom <tt>sep</tt>, matching a maximal sequence */
-    public  static Repeat maximal1(Element e, Atom sep)                { return new Repeat.Maximal(e, false, true, sep, null); }
-    public  static Repeat maximal1(Element e, Atom sep, Object tag)    { return new Repeat.Maximal(e, false, true, sep, tag); }
-
-
 }