gcc/libjava/java/util/Hashtable.java

399 lines
10 KiB
Java
Raw Normal View History

1999-04-07 22:42:40 +08:00
/* Copyright (C) 1998, 1999 Cygnus Solutions
This file is part of libgcj.
This software is copyrighted work licensed under the terms of the
Libgcj License. Please consult the file "LIBGCJ_LICENSE" for
details. */
package java.util;
import java.io.Serializable;
/**
* @author Warren Levy <warrenl@cygnus.com>
* @date September 24, 1998.
*/
/* Written using "Java Class Libraries", 2nd edition, ISBN 0-201-31002-3
* "The Java Language Specification", ISBN 0-201-63451-1
* plus online API docs for JDK 1.2 beta from http://www.javasoft.com.
* Status: Believed complete and correct
*/
class HashtableEntry
{
public Object key;
public Object value;
public HashtableEntry nextEntry = null;
public HashtableEntry(Object key, Object value)
{
this.key = key;
this.value = value;
}
}
class HashtableEnumeration implements Enumeration
{
// TBD: Enumeration is not safe if new elements are put in the table as
// this could cause a rehash and we'd completely lose our place. Even
// without a rehash, it is undetermined if a new element added would
// appear in the enumeration. The spec says nothing about this, but
// the "Java Class Libraries" book infers that modifications to the
// hashtable during enumeration causes indeterminate results. Don't do it!
// A safer way would be to make a copy of the table (e.g. into a vector)
// but this is a fair bit more expensive.
private HashtableEntry[] bucket;
private int bucketIndex;
private HashtableEntry elem;
private int enumCount;
private int size;
private boolean values;
public HashtableEnumeration(HashtableEntry[] bkt, int sz, boolean isValues)
{
bucket = bkt;
bucketIndex = -1;
enumCount = 0;
elem = null;
size = sz;
values = isValues;
}
public boolean hasMoreElements()
{
return enumCount < size;
}
public Object nextElement()
{
if (!hasMoreElements())
throw new NoSuchElementException();
// Find next element
if (elem != null) // In the middle of a bucket
elem = elem.nextEntry;
while (elem == null) // Find the next non-empty bucket
elem = bucket[++bucketIndex];
enumCount++;
return values ? elem.value : elem.key;
}
}
// TBD: The algorithm used here closely reflects what is described in
// the "Java Class Libraries" book. The "Java Language Spec" is much
// less specific about the implementation. Because of this freedom
// provided by the actual spec, hash table algorithms should be
// investigated to see if there is a better alternative to this one.
// TODO12:
// public class Hashtable extends Dictionary
// implements Map, Cloneable, Serializable
public class Hashtable extends Dictionary implements Cloneable, Serializable
{
private HashtableEntry bucket[];
private float loadFactor;
private int hsize = 0;
public Hashtable()
{
// The "Java Class Libraries" book (p. 919) says that initial size in this
// case is 101 (a prime number to increase the odds of even distribution).
this(101, 0.75F);
}
public Hashtable(int initialSize)
{
this(initialSize, 0.75F);
}
public Hashtable(int initialSize, float loadFactor)
{
if (initialSize < 0 || loadFactor <= 0.0 || loadFactor > 1.0)
throw new IllegalArgumentException();
bucket = new HashtableEntry[initialSize];
this.loadFactor = loadFactor;
}
// TODO12:
// public Hashtable(Map t)
// {
// }
public synchronized void clear()
{
// Aid the GC by nulling out the entries in the hash table.
for (int i = 0; i < bucket.length; i++)
{
HashtableEntry elem = bucket[i];
bucket[i] = null; // May already be null.
while (elem != null)
{
HashtableEntry next = elem.nextEntry;
elem.nextEntry = null; // May already be null.
elem = next;
}
}
hsize = 0;
}
public synchronized Object clone()
{
// New hashtable will have same initialCapacity and loadFactor.
Hashtable newTable = new Hashtable(bucket.length, loadFactor);
HashtableEntry newElem, prev = null;
for (int i = 0; i < bucket.length; i++)
for (HashtableEntry elem = bucket[i]; elem != null; elem = elem.nextEntry)
{
// An easy but expensive method is newTable.put(elem.key, elem.value);
// Since the hash tables are the same size, the buckets and collisions
// will be the same in the new one, so we can just clone directly.
// This is much cheaper than using put.
newElem = new HashtableEntry(elem.key, elem.value);
if (newTable.bucket[i] == null)
prev = newTable.bucket[i] = newElem;
else
prev = prev.nextEntry = newElem;
}
newTable.hsize = this.hsize;
return newTable;
}
public synchronized boolean contains(Object value) throws NullPointerException
{
// An exception is thrown here according to the JDK 1.2 doc.
if (value == null)
throw new NullPointerException();
for (int i = 0; i < bucket.length; i++)
for (HashtableEntry elem = bucket[i]; elem != null; elem = elem.nextEntry)
if (elem.value.equals(value))
return true;
return false;
}
public synchronized boolean containsKey(Object key)
{
// The Map interface mandates that we throw this.
if (key == null)
throw new NullPointerException ();
for (HashtableEntry elem = bucket[Math.abs(key.hashCode()
% bucket.length)];
elem != null; elem = elem.nextEntry)
if (elem.key.equals(key))
return true;
return false;
}
public synchronized Enumeration elements()
{
return new HashtableEnumeration(bucket, hsize, true);
}
public synchronized Object get(Object key)
{
// The Dictionary interface mandates that get() throw a
// NullPointerException if key is null.
if (key == null)
throw new NullPointerException ();
for (HashtableEntry elem = bucket[Math.abs (key.hashCode()
% bucket.length)];
elem != null; elem = elem.nextEntry)
if (elem.key.equals(key))
return elem.value;
return null;
}
public boolean isEmpty()
{
return this.hsize <= 0;
}
public synchronized Enumeration keys()
{
return new HashtableEnumeration(bucket, hsize, false);
}
public synchronized Object put(Object key, Object value)
throws NullPointerException
{
if (key == null || value == null)
throw new NullPointerException();
HashtableEntry prevElem = null;
final int index = Math.abs(key.hashCode() % bucket.length);
for (HashtableEntry elem = bucket[index]; elem != null;
prevElem = elem, elem = elem.nextEntry)
if (elem.key.equals(key))
{
// Update with the new value and then return the old one.
Object oldVal = elem.value;
elem.value = value;
return oldVal;
}
// At this point, we know we need to add a new element.
HashtableEntry newElem = new HashtableEntry(key, value);
if (bucket[index] == null)
bucket[index] = newElem;
else
prevElem.nextEntry = newElem;
if (++hsize > loadFactor * bucket.length)
rehash();
return null;
}
protected void rehash()
{
// Create a new table which is twice the size (plus one) of the old.
// One is added to make the new array length odd so it thus has at least
// a (small) possibility of being a prime number.
HashtableEntry oldBucket[] = bucket;
bucket = new HashtableEntry[bucket.length * 2 + 1];
// Copy over each entry into the new table
HashtableEntry elem;
for (int i = 0; i < oldBucket.length; i++)
for (elem = oldBucket[i]; elem != null; elem = elem.nextEntry)
{
// Calling put(elem.key, elem.value); would seem like the easy way
// but it is dangerous since put increases 'hsize' and calls rehash!
// This could become infinite recursion under the right
// circumstances. Instead, we'll add the element directly; this is a
// bit more efficient than put since the data is already verified.
final int index = Math.abs(elem.key.hashCode() % bucket.length);
HashtableEntry newElem = new HashtableEntry(elem.key, elem.value);
if (bucket[index] == null)
bucket[index] = newElem;
else
{
// Since this key can't already be in the table, just add this
// in at the top of the bucket.
newElem.nextEntry = bucket[index];
bucket[index] = newElem;
}
}
}
public synchronized Object remove(Object key)
{
// TBD: Hmm, none of the various docs say to throw an exception here.
if (key == null)
return null;
Object retval;
HashtableEntry prevElem = null;
final int index = Math.abs(key.hashCode() % bucket.length);
for (HashtableEntry elem = bucket[index]; elem != null;
prevElem = elem, elem = elem.nextEntry)
if (elem.key.equals(key))
{
retval = elem.value;
if (prevElem == null)
bucket[index] = elem.nextEntry;
else
prevElem.nextEntry = elem.nextEntry;
--hsize;
return retval;
}
return null;
}
public int size()
{
return this.hsize;
}
public synchronized String toString()
{
// Following the Java Lang Spec 21.5.4 (p. 636).
Enumeration keys = keys();
Enumeration values = elements();
// Prepend first element with open bracket
StringBuffer result = new StringBuffer("{");
// add first element if one exists
// TBD: Seems like it is more efficient to catch the exception than
// to call hasMoreElements each time around.
try
{
result.append(keys.nextElement().toString() + "=" +
values.nextElement().toString());
}
catch (NoSuchElementException ex)
{
}
// Prepend subsequent elements with ", "
try
{
while (true)
result.append(", " + keys.nextElement().toString() + "=" +
values.nextElement().toString());
}
catch (NoSuchElementException ex)
{
}
// Append last element with closing bracket
result.append("}");
return result.toString();
}
// TODO12:
// public Set entrySet()
// {
// }
// TODO12:
// public Set keySet()
// {
// }
// Since JDK 1.2:
// This method is identical to contains but is part of the 1.2 Map interface.
// TBD: Should contains return containsValue instead? Depends on which
// will be called more typically.
public synchronized boolean containsValue(Object value)
{
return this.contains(value);
}
// TODO12:
// public boolean equals(Object o)
// {
// }
// TODO12:
// public boolean hashCode()
// {
// }
// TODO12:
// public void putAll(Map t)
// {
// }
// TODO12:
// public Collection values()
// {
// }
}