General purpose merging iterator implementation

git-svn-id: file:///humgen/gsa-scr1/gsa-engineering/svn_contents/trunk@1593 348d0f76-0448-11de-a6fe-93d51630548a
This commit is contained in:
depristo 2009-09-12 19:06:15 +00:00
parent 275707f5f6
commit 17ab1d8b25
2 changed files with 228 additions and 0 deletions

View File

@ -0,0 +1,109 @@
package org.broadinstitute.sting.utils;
import org.broadinstitute.sting.gatk.iterators.PeekingIterator;
import org.broadinstitute.sting.gatk.refdata.ReferenceOrderedDatum;
import java.util.*;
public class MergingIterator<E extends Comparable<E>> implements Iterator<E>, PeekingIterator<E>, Iterable<E> {
PriorityQueue<Element> queue = new PriorityQueue<Element>();
private class Element implements Comparable<Element> {
public Iterator<E> it = null;
public E value = null;
public Element(Iterator<E> it) {
this.it = it;
update();
}
public Element update() {
if ( ! it.hasNext() )
throw new RuntimeException("it is empty");
E prev = value;
value = it.next();
//System.out.printf("Updating %s to prev=%s, next=%s%n", this, prev, value);
return this;
}
public int compareTo(Element other) {
return value.compareTo(other.value);
}
}
public Iterator<E> iterator() {
return this;
}
public MergingIterator() {
;
}
public MergingIterator(Iterator<E> it) {
add(it);
}
public MergingIterator(Collection<Iterator<E>> its) {
for ( Iterator<E> it : its ) {
add(it);
}
}
public void add(Iterator<E> it) {
if ( it.hasNext() )
queue.add(new Element(it));
}
public boolean hasNext() {
return ! queue.isEmpty();
}
public E next() {
Element e = queue.poll();
E value = e.value;
if ( e.it != null && e.it.hasNext() )
queue.add(new Element(e.it));
//System.out.printf("Element is %s%n", e.value);
return value;
}
public E peek() {
return queue.peek().value;
}
public Collection<E> allElementsLTE(E elt) {
return allElementsLTE(elt, true);
}
public Collection<E> allElementsLTE(E elt, boolean includeElt) {
LinkedList<E> all = new LinkedList<E>();
if ( includeElt ) all.add(elt);
while ( hasNext() ) {
E x = peek();
//System.out.printf("elt.compareTo(x) == %d%n", elt.compareTo(x));
//System.out.printf("In allElementLTE%n");
int cmp = elt.compareTo(x);
//System.out.printf("x=%s%n elt=%s%n => elt.compareTo(x) == %d%n", x, elt, cmp);
if ( cmp >= 0 ) {
//System.out.printf(" Adding element x=%s, size = %d%n", x, all.size());
all.add(next());
//System.out.printf(" Added size = %d%n", all.size());
}
else {
//System.out.printf("breaking...%n");
break;
}
}
return all;
}
public void remove() {
throw new UnsupportedOperationException();
}
}

View File

@ -0,0 +1,119 @@
// our package
package org.broadinstitute.sting.utils;
// the imports for unit testing.
import org.junit.Assert;
import org.junit.Test;
import org.junit.Before;
import org.broadinstitute.sting.BaseTest;
import org.broadinstitute.sting.gatk.walkers.recalibration.RecalData;
import org.broadinstitute.sting.utils.QualityUtils;
import java.util.*;
/**
* Basic unit test for RecalData
*/
public class MergingIteratorTest extends BaseTest {
List<Integer> one_to_ten = Arrays.asList(1, 2, 3, 4, 5, 6, 7, 8, 9, 10);
List<Integer> evens = Arrays.asList(2, 4, 6, 8, 10);
List<Integer> three_four_five = Arrays.asList(3, 4, 5);
List<Integer> bigs = Arrays.asList(100, 200, 500);
MergingIterator<Integer> x = null;
@Before
public void before() {
x = null;
}
@Test
public void testOneToOne() {
logger.warn("Executing testOneToOne");
x = new MergingIterator<Integer>(one_to_ten.iterator());
for ( int i = 1; i < 11; i++ ) {
int elt = x.next();
logger.warn(String.format("Testing %d == %d", i, elt));
Assert.assertEquals(elt, i);
}
Assert.assertFalse(x.hasNext());
}
@Test
public void testMerging10Evens() {
logger.warn("Executing testMerging10Evens");
x = new MergingIterator<Integer>(one_to_ten.iterator());
x.add(evens.iterator());
List<Integer> expected = Arrays.asList(1, 2, 2, 3, 4, 4, 5, 6, 6, 7, 8, 8, 9, 10, 10);
testExpected( x, expected );
Assert.assertFalse(x.hasNext());
}
@Test
public void testMerging3() {
logger.warn("Executing testMerging3");
x = new MergingIterator<Integer>(Arrays.asList(bigs.iterator(), evens.iterator(), three_four_five.iterator()));
List<Integer> expected = Arrays.asList(2, 3, 4, 4, 5, 6, 8, 10, 100, 200, 500);
testExpected( x, expected );
Assert.assertFalse(x.hasNext());
}
@Test
public void testCollectLTE() {
logger.warn("Executing testCollectLTE");
x = new MergingIterator<Integer>(Arrays.asList(bigs.iterator(), evens.iterator(), three_four_five.iterator()));
//for ( int i : x ) {
// System.out.printf("i=%d%n", i);
//}
logger.warn("Testing next element...");
Assert.assertEquals( 2, (int)x.next());
logger.warn("Testing allElementsLTE(2) = 2 ...");
assertListsEqual( x.allElementsLTE(2), Arrays.asList(2) );
logger.warn("Testing allElementsLTE(3) = 3 ...");
Assert.assertEquals(3, (int)x.next());
assertListsEqual( x.allElementsLTE(3), Arrays.asList(3) );
logger.warn("Testing allElementsLTE(4) = 4, 4 ...");
Assert.assertEquals(4, (int)x.next());
assertListsEqual( x.allElementsLTE(4), Arrays.asList(4, 4) );
logger.warn("Testing allElementsLTE(8) = 5, 6, 8 ...");
assertListsEqual( x.allElementsLTE(8, false), Arrays.asList(5, 6, 8) );
logger.warn("Testing remaining...");
testExpected( x, Arrays.asList(10, 100, 200, 500) );
Assert.assertFalse(x.hasNext());
}
private void testExpected( MergingIterator<Integer> mi, List<Integer> expected ) {
for ( int expect : expected ) {
int elt = mi.next();
logger.warn(String.format("Testing elt=%d == expected=%d", elt, expect));
Assert.assertEquals(expect, elt);
}
}
private void assertListsEqual( Collection<Integer> l, List<Integer> expected ) {
logger.warn("Size is " + l.size());
Iterator<Integer> lit = l.iterator();
for ( int expect : expected ) {
Assert.assertTrue(lit.hasNext());
int elt = lit.next();
logger.warn(String.format("Testing elt=%d == expected=%d", elt, expect));
Assert.assertEquals(expect, elt);
}
}
}