/*
 * Copyright (c) 2016 Chris K Wensel <chris@wensel.net>. All Rights Reserved.
 * Copyright (c) 2007-2017 Xplenty, Inc. All Rights Reserved.
 *
 * Project and contact information: http://www.cascading.org/
 *
 * This file is part of the Cascading project.
 *
 * Licensed under the Apache License, Version 2.0 (the "License");
 * you may not use this file except in compliance with the License.
 * You may obtain a copy of the License at
 *
 *     http://www.apache.org/licenses/LICENSE-2.0
 *
 * Unless required by applicable law or agreed to in writing, software
 * distributed under the License is distributed on an "AS IS" BASIS,
 * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
 * See the License for the specific language governing permissions and
 * limitations under the License.
 */

package cascading;

import java.io.Serializable;
import java.util.ArrayList;
import java.util.Collections;
import java.util.Comparator;
import java.util.HashMap;
import java.util.HashSet;
import java.util.List;
import java.util.Map;
import java.util.Set;

import cascading.flow.Flow;
import cascading.flow.FlowDef;
import cascading.flow.FlowStep;
import cascading.flow.planner.graph.ElementGraph;
import cascading.operation.Aggregator;
import cascading.operation.Function;
import cascading.operation.Identity;
import cascading.operation.aggregator.Count;
import cascading.operation.aggregator.First;
import cascading.operation.expression.ExpressionFunction;
import cascading.operation.regex.RegexFilter;
import cascading.operation.regex.RegexSplitter;
import cascading.pipe.Checkpoint;
import cascading.pipe.CoGroup;
import cascading.pipe.Each;
import cascading.pipe.Every;
import cascading.pipe.GroupBy;
import cascading.pipe.HashJoin;
import cascading.pipe.Merge;
import cascading.pipe.Pipe;
import cascading.pipe.assembly.Rename;
import cascading.pipe.joiner.InnerJoin;
import cascading.pipe.joiner.Joiner;
import cascading.pipe.joiner.LeftJoin;
import cascading.pipe.joiner.MixedJoin;
import cascading.pipe.joiner.OuterJoin;
import cascading.pipe.joiner.RightJoin;
import cascading.tap.SinkMode;
import cascading.tap.Tap;
import cascading.tuple.Fields;
import cascading.tuple.Hasher;
import cascading.tuple.Tuple;
import org.junit.Test;

import static data.InputData.*;

public class JoinFieldedPipesPlatformTest extends PlatformTestCase
  {
  public JoinFieldedPipesPlatformTest()
    {
    super( true, 4, 1 ); // leave cluster testing enabled
    }

  @Test
  public void testCross() throws Exception
    {
    getPlatform().copyFromLocal( inputFileLhs );
    getPlatform().copyFromLocal( inputFileRhs );

    Map sources = new HashMap();

    sources.put( "lhs", getPlatform().getTextFile( inputFileLhs ) );
    sources.put( "rhs", getPlatform().getTextFile( inputFileRhs ) );

    Tap sink = getPlatform().getTextFile( new Fields( "line" ), getOutputPath( "cross" ), SinkMode.REPLACE );

    Pipe pipeLower = new Each( "lhs", new Fields( "line" ), new RegexSplitter( new Fields( "numLHS", "charLHS" ), " " ) );
    Pipe pipeUpper = new Each( "rhs", new Fields( "line" ), new RegexSplitter( new Fields( "numRHS", "charRHS" ), " " ) );

    Pipe cross = new HashJoin( pipeLower, new Fields( "numLHS" ), pipeUpper, new Fields( "numRHS" ), new InnerJoin() );

    Flow flow = getPlatform().getFlowConnector().connect( sources, sink, cross );

    flow.complete();

    validateLength( flow, 37, null );

    List<Tuple> values = getSinkAsList( flow );

    assertTrue( values.contains( new Tuple( "1\ta\t1\tA" ) ) );
    assertTrue( values.contains( new Tuple( "1\ta\t1\tB" ) ) );
    }

  @Test
  public void testJoin() throws Exception
    {
    getPlatform().copyFromLocal( inputFileLower );
    getPlatform().copyFromLocal( inputFileUpper );

    Tap sourceLower = getPlatform().getTextFile( new Fields( "offset", "line" ), inputFileLower );
    Tap sourceUpper = getPlatform().getTextFile( new Fields( "offset", "line" ), inputFileUpper );

    Map sources = new HashMap();

    sources.put( "lower", sourceLower );
    sources.put( "upper", sourceUpper );

    Tap sink = getPlatform().getTextFile( new Fields( "line" ), getOutputPath( "join" ), SinkMode.REPLACE );

    Function splitter = new RegexSplitter( new Fields( "num", "char" ), " " );

    Pipe pipeLower = new Each( new Pipe( "lower" ), new Fields( "line" ), splitter );
    Pipe pipeUpper = new Each( new Pipe( "upper" ), new Fields( "line" ), splitter );

    Pipe splice = new HashJoin( pipeLower, new Fields( "num" ), pipeUpper, new Fields( "num" ), Fields.size( 4 ) );

    Map<Object, Object> properties = getProperties();

    Flow flow = getPlatform().getFlowConnector( properties ).connect( sources, sink, splice );

    flow.complete();

    validateLength( flow, 5 );

    List<Tuple> values = getSinkAsList( flow );

    assertTrue( values.contains( new Tuple( "1\ta\t1\tA" ) ) );
    assertTrue( values.contains( new Tuple( "2\tb\t2\tB" ) ) );
    }

  @Test
  public void testJoinSamePipeName() throws Exception
    {
    getPlatform().copyFromLocal( inputFileLower );
    getPlatform().copyFromLocal( inputFileUpper );

    Tap sourceLower = getPlatform().getTextFile( new Fields( "offset", "line" ), inputFileLower );
    Tap sourceUpper = getPlatform().getTextFile( new Fields( "offset", "line" ), inputFileUpper );

    Map sources = new HashMap();

    sources.put( "lower", sourceLower );
    sources.put( "upper", sourceUpper );

    Tap sink = getPlatform().getTextFile( new Fields( "line" ), getOutputPath( "renamedpipes" ), SinkMode.REPLACE );

    Function splitter = new RegexSplitter( new Fields( "num", "char" ), " " );

    Pipe pipeLower = new Pipe( "lower" );
    Pipe pipeUpper = new Pipe( "upper" );

    // these pipes will hide the source name, and could cause one to be lost
    pipeLower = new Pipe( "same", pipeLower );
    pipeUpper = new Pipe( "same", pipeUpper );

    pipeLower = new Each( pipeLower, new Fields( "line" ), splitter );
    pipeUpper = new Each( pipeUpper, new Fields( "line" ), splitter );

//    pipeLower = new Each( pipeLower, new Fields( "num", "char" ), new Identity( new Fields( "num", "char" ) ) );
//    pipeUpper = new Each( pipeUpper, new Fields( "num", "char" ), new Identity( new Fields( "num", "char" ) ) );

    pipeLower = new Pipe( "left", pipeLower );
    pipeUpper = new Pipe( "right", pipeUpper );

//    pipeLower = new Each( pipeLower, new Debug( true ) );
//    pipeUpper = new Each( pipeUpper, new Debug( true ) );

    Pipe splice = new HashJoin( pipeLower, new Fields( "num" ), pipeUpper, new Fields( "num" ), Fields.size( 4 ) );

//    splice = new Each( splice, new Debug( true ) );
    splice = new Pipe( "splice", splice );
    splice = new Pipe( "tail", splice );

    Flow flow = getPlatform().getFlowConnector().connect( sources, sink, splice );

    flow.complete();

    validateLength( flow, 5 );

    List<Tuple> values = getSinkAsList( flow );

    assertTrue( values.contains( new Tuple( "1\ta\t1\tA" ) ) );
    assertTrue( values.contains( new Tuple( "2\tb\t2\tB" ) ) );
    }

  @Test
  public void testJoinWithUnknowns() throws Exception
    {
    getPlatform().copyFromLocal( inputFileLower );
    getPlatform().copyFromLocal( inputFileUpper );

    Tap sourceLower = getPlatform().getTextFile( new Fields( "offset", "line" ), inputFileLower );
    Tap sourceUpper = getPlatform().getTextFile( new Fields( "offset", "line" ), inputFileUpper );

    Map sources = new HashMap();

    sources.put( "lower", sourceLower );
    sources.put( "upper", sourceUpper );

    Tap sink = getPlatform().getTextFile( new Fields( "line" ), getOutputPath( "unknown" ), SinkMode.REPLACE );

    Function splitter = new RegexSplitter( Fields.UNKNOWN, " " );

    Pipe pipeLower = new Each( new Pipe( "lower" ), new Fields( "line" ), splitter );
    Pipe pipeUpper = new Each( new Pipe( "upper" ), new Fields( "line" ), splitter );

    Pipe splice = new HashJoin( pipeLower, new Fields( 0 ), pipeUpper, new Fields( 0 ), Fields.size( 4 ) );

    Flow flow = getPlatform().getFlowConnector().connect( sources, sink, splice );

    flow.complete();

    validateLength( flow, 5 );

    List<Tuple> values = getSinkAsList( flow );

    assertTrue( values.contains( new Tuple( "1\ta\t1\tA" ) ) );
    assertTrue( values.contains( new Tuple( "2\tb\t2\tB" ) ) );
    }

  /**
   * this test intentionally filters out all values so the intermediate tap is empty. this tap is cogrouped with
   * a new stream using an outerjoin.
   *
   * @throws Exception
   */
  @Test
  public void testJoinFilteredBranch() throws Exception
    {
    getPlatform().copyFromLocal( inputFileLower );
    getPlatform().copyFromLocal( inputFileUpper );

    Tap sourceLower = getPlatform().getTextFile( new Fields( "offset", "line" ), inputFileLower );
    Tap sourceUpper = getPlatform().getTextFile( new Fields( "offset", "line" ), inputFileUpper );

    Map sources = new HashMap();

    sources.put( "lower", sourceLower );
    sources.put( "upper", sourceUpper );

    Tap sink = getPlatform().getTextFile( new Fields( "line" ), getOutputPath( "joinfilteredbranch" ), SinkMode.REPLACE );

    Function splitter = new RegexSplitter( new Fields( "num", "char" ), " " );

    Pipe pipeLower = new Each( new Pipe( "lower" ), new Fields( "line" ), splitter );
    Pipe pipeUpper = new Each( new Pipe( "upper" ), new Fields( "line" ), splitter );
    pipeUpper = new Each( pipeUpper, new Fields( "num" ), new RegexFilter( "^fobar" ) ); // intentionally filtering all
    pipeUpper = new GroupBy( pipeUpper, new Fields( "num" ) );

    Pipe splice = new HashJoin( pipeLower, new Fields( "num" ), pipeUpper, new Fields( "num" ), Fields.size( 4 ), new OuterJoin() );

    Flow flow = getPlatform().getFlowConnector().connect( sources, sink, splice );

    flow.complete();

    validateLength( flow, 5 );

    List<Tuple> values = getSinkAsList( flow );

    assertTrue( values.contains( new Tuple( "1\ta\tnull\tnull" ) ) );
    assertTrue( values.contains( new Tuple( "2\tb\tnull\tnull" ) ) );
    }

  @Test
  public void testJoinSelf() throws Exception
    {
    getPlatform().copyFromLocal( inputFileLhs );

    Tap sourceLhs = getPlatform().getTextFile( new Fields( "offset", "line" ), inputFileLhs );
    Tap sourceRhs = getPlatform().getTextFile( new Fields( "offset", "line" ), inputFileLhs );

    Map sources = new HashMap();

    sources.put( "lhs", sourceLhs );
    sources.put( "rhs", sourceRhs );

    Tap sink = getPlatform().getTextFile( new Fields( "line" ), getOutputPath( "joinself" ), SinkMode.REPLACE );

    Function splitter = new RegexSplitter( new Fields( "num", "char" ), " " );

    Pipe pipeLower = new Each( new Pipe( "lhs" ), new Fields( "line" ), splitter );
    Pipe pipeUpper = new Each( new Pipe( "rhs" ), new Fields( "line" ), splitter );

    Pipe splice = new HashJoin( pipeLower, new Fields( "num" ), pipeUpper, new Fields( "num" ), Fields.size( 4 ) );

    Flow flow = getPlatform().getFlowConnector().connect( sources, sink, splice );

    flow.complete();

    validateLength( flow, 37 );

    List<Tuple> values = getSinkAsList( flow );

    assertTrue( values.contains( new Tuple( "1\ta\t1\ta" ) ) );
    assertTrue( values.contains( new Tuple( "2\tb\t2\tb" ) ) );
    }

  @Test
  public void testSameSourceJoin() throws Exception
    {
    getPlatform().copyFromLocal( inputFileLhs );

    Tap source = getPlatform().getDelimitedFile( new Fields( "num", "char" ), " ", inputFileLhs );

    Map sources = new HashMap();

    sources.put( "lhs", source );
    sources.put( "rhs", source );

    Tap sink = getPlatform().getTextFile( new Fields( "line" ), getOutputPath(), SinkMode.REPLACE );

    Pipe pipeLower = new Pipe( "lhs" );
    Pipe pipeUpper = new Pipe( "rhs" );

    Pipe splice = new HashJoin( pipeLower, new Fields( "num" ), pipeUpper, new Fields( "num" ), Fields.size( 4 ) );

    Flow flow = getPlatform().getFlowConnector().connect( sources, sink, splice );

    flow.complete();

    validateLength( flow, 37 );

    List<Tuple> values = getSinkAsList( flow );

    assertTrue( values.contains( new Tuple( "1\ta\t1\ta" ) ) );
    assertTrue( values.contains( new Tuple( "2\tb\t2\tb" ) ) );
    }

  /**
   * Method testCoGroupAfterEvery tests that a tmp tap is inserted after the Every in the cogroup join
   *
   * @throws Exception when
   */
  @Test
  public void testJoinAfterEvery() throws Exception
    {
    getPlatform().copyFromLocal( inputFileLower );
    getPlatform().copyFromLocal( inputFileUpper );

    Tap sourceLower = getPlatform().getTextFile( new Fields( "offset", "line" ), inputFileLower );
    Tap sourceUpper = getPlatform().getTextFile( new Fields( "offset", "line" ), inputFileUpper );

    Map sources = new HashMap();

    sources.put( "lower", sourceLower );
    sources.put( "upper", sourceUpper );

    Tap sink = getPlatform().getTextFile( new Fields( "line" ), getOutputPath( "afterevery" ), SinkMode.REPLACE );

    Function splitter = new RegexSplitter( new Fields( "num", "char" ), " " );

    Pipe pipeLower = new Each( new Pipe( "lower" ), new Fields( "line" ), splitter );
    pipeLower = new GroupBy( pipeLower, new Fields( "num" ) );
    pipeLower = new Every( pipeLower, new Fields( "char" ), new First(), Fields.ALL );

    Pipe pipeUpper = new Each( new Pipe( "upper" ), new Fields( "line" ), splitter );
    pipeUpper = new GroupBy( pipeUpper, new Fields( "num" ) );
    pipeUpper = new Every( pipeUpper, new Fields( "char" ), new First(), Fields.ALL );

    Pipe splice = new HashJoin( pipeLower, new Fields( "num" ), pipeUpper, new Fields( "num" ), Fields.size( 4 ) );

    Flow flow = getPlatform().getFlowConnector().connect( sources, sink, splice );

    flow.complete();

    validateLength( flow, 5, null );

    List<Tuple> values = getSinkAsList( flow );

    assertTrue( values.contains( new Tuple( "1\ta\t1\tA" ) ) );
    assertTrue( values.contains( new Tuple( "2\tb\t2\tB" ) ) );
    }

  @Test
  public void testJoinInnerSingleField() throws Exception
    {
    getPlatform().copyFromLocal( inputFileLowerOffset );
    getPlatform().copyFromLocal( inputFileUpper );

    Tap sourceLower = getPlatform().getTextFile( new Fields( "offset", "line" ), inputFileLowerOffset );
    Tap sourceUpper = getPlatform().getTextFile( new Fields( "offset", "line" ), inputFileUpper );

    Map sources = new HashMap();

    sources.put( "lower", sourceLower );
    sources.put( "upper", sourceUpper );

    Tap sink = getPlatform().getTextFile( new Fields( "line" ), getOutputPath( "joininnersingle" ), SinkMode.REPLACE );

    Pipe pipeLower = new Each( new Pipe( "lower" ), new Fields( "line" ), new RegexSplitter( new Fields( "num1", "char" ), " " ), new Fields( "num1" ) );
    Pipe pipeUpper = new Each( new Pipe( "upper" ), new Fields( "line" ), new RegexSplitter( new Fields( "num2", "char" ), " " ), new Fields( "num2" ) );

    Pipe join = new HashJoin( pipeLower, new Fields( "num1" ), pipeUpper, new Fields( "num2" ) );

    Flow flow = getPlatform().getFlowConnector().connect( sources, sink, join );

    flow.complete();

    validateLength( flow, 3, null );

    Set<Tuple> results = new HashSet<Tuple>();

    results.add( new Tuple( "1\t1" ) );
    results.add( new Tuple( "5\t5" ) );

    List<Tuple> actual = getSinkAsList( flow );

    results.removeAll( actual );

    assertEquals( 0, results.size() );
    }

  /**
   * 1 a1
   * 1 a2
   * 1 a3
   * 2 b1
   * 3 c1
   * 4 d1
   * 4 d2
   * 4 d3
   * 5 e1
   * 5 e2
   * 5 e3
   * 7 g1
   * 7 g2
   * 7 g3
   * 7 g4
   * 7 g5
   * null h1
   * <p/>
   * 1 A1
   * 1 A2
   * 1 A3
   * 2 B1
   * 2 B2
   * 2 B3
   * 4 D1
   * 6 F1
   * 6 F2
   * null H1
   * <p/>
   * 1  a1      1       A1
   * 1  a1      1       A2
   * 1  a1      1       A3
   * 1  a2      1       A1
   * 1  a2      1       A2
   * 1  a2      1       A3
   * 1  a3      1       A1
   * 1  a3      1       A2
   * 1  a3      1       A3
   * 2  b1      2       B1
   * 2  b1      2       B2
   * 2  b1      2       B3
   * 4  d1      4       D1
   * 4  d2      4       D1
   * 4  d3      4       D1
   * null h1  null  H1
   *
   * @throws Exception
   */
  @Test
  public void testJoinInner() throws Exception
    {
    HashSet<Tuple> results = new HashSet<Tuple>();

    results.add( new Tuple( "1", "a1", "1", "A1" ) );
    results.add( new Tuple( "1", "a1", "1", "A2" ) );
    results.add( new Tuple( "1", "a1", "1", "A3" ) );
    results.add( new Tuple( "1", "a2", "1", "A1" ) );
    results.add( new Tuple( "1", "a2", "1", "A2" ) );
    results.add( new Tuple( "1", "a2", "1", "A3" ) );
    results.add( new Tuple( "1", "a3", "1", "A1" ) );
    results.add( new Tuple( "1", "a3", "1", "A2" ) );
    results.add( new Tuple( "1", "a3", "1", "A3" ) );
    results.add( new Tuple( "2", "b1", "2", "B1" ) );
    results.add( new Tuple( "2", "b1", "2", "B2" ) );
    results.add( new Tuple( "2", "b1", "2", "B3" ) );
    results.add( new Tuple( "4", "d1", "4", "D1" ) );
    results.add( new Tuple( "4", "d2", "4", "D1" ) );
    results.add( new Tuple( "4", "d3", "4", "D1" ) );
    results.add( new Tuple( null, "h1", null, "H1" ) );

    handleJoins( "joininner", new InnerJoin(), results );
    }

  /**
   * /**
   * 1 a1
   * 1 a2
   * 1 a3
   * 2 b1
   * 3 c1
   * 4 d1
   * 4 d2
   * 4 d3
   * 5 e1
   * 5 e2
   * 5 e3
   * 7 g1
   * 7 g2
   * 7 g3
   * 7 g4
   * 7 g5
   * null h1
   * <p/>
   * 1 A1
   * 1 A2
   * 1 A3
   * 2 B1
   * 2 B2
   * 2 B3
   * 4 D1
   * 6 F1
   * 6 F2
   * null H1
   * <p/>
   * 1  a1      1       A1
   * 1  a1      1       A2
   * 1  a1      1       A3
   * 1  a2      1       A1
   * 1  a2      1       A2
   * 1  a2      1       A3
   * 1  a3      1       A1
   * 1  a3      1       A2
   * 1  a3      1       A3
   * 2  b1      2       B1
   * 2  b1      2       B2
   * 2  b1      2       B3
   * 3  c1      null    null
   * 4  d1      4       D1
   * 4  d2      4       D1
   * 4  d3      4       D1
   * 5  e1      null    null
   * 5  e2      null    null
   * 5  e3      null    null
   * null       null    6       F1
   * null       null    6       F2
   * 7  g1      null    null
   * 7  g2      null    null
   * 7  g3      null    null
   * 7  g4      null    null
   * 7  g5      null    null
   * null h1  null  H1
   *
   * @throws Exception
   */
  @Test
  public void testJoinOuter() throws Exception
    {
    // skip if hadoop cluster mode, outer joins don't behave the same
    if( getPlatform().isMapReduce() && getPlatform().isUseCluster() )
      return;

    Set<Tuple> results = new HashSet<Tuple>();

    results.add( new Tuple( "1", "a1", "1", "A1" ) );
    results.add( new Tuple( "1", "a1", "1", "A2" ) );
    results.add( new Tuple( "1", "a1", "1", "A3" ) );
    results.add( new Tuple( "1", "a2", "1", "A1" ) );
    results.add( new Tuple( "1", "a2", "1", "A2" ) );
    results.add( new Tuple( "1", "a2", "1", "A3" ) );
    results.add( new Tuple( "1", "a3", "1", "A1" ) );
    results.add( new Tuple( "1", "a3", "1", "A2" ) );
    results.add( new Tuple( "1", "a3", "1", "A3" ) );
    results.add( new Tuple( "2", "b1", "2", "B1" ) );
    results.add( new Tuple( "2", "b1", "2", "B2" ) );
    results.add( new Tuple( "2", "b1", "2", "B3" ) );
    results.add( new Tuple( "3", "c1", null, null ) );
    results.add( new Tuple( "4", "d1", "4", "D1" ) );
    results.add( new Tuple( "4", "d2", "4", "D1" ) );
    results.add( new Tuple( "4", "d3", "4", "D1" ) );
    results.add( new Tuple( "5", "e1", null, null ) );
    results.add( new Tuple( "5", "e2", null, null ) );
    results.add( new Tuple( "5", "e3", null, null ) );
    results.add( new Tuple( null, null, "6", "F1" ) );
    results.add( new Tuple( null, null, "6", "F2" ) );
    results.add( new Tuple( "7", "g1", null, null ) );
    results.add( new Tuple( "7", "g2", null, null ) );
    results.add( new Tuple( "7", "g3", null, null ) );
    results.add( new Tuple( "7", "g4", null, null ) );
    results.add( new Tuple( "7", "g5", null, null ) );
    results.add( new Tuple( null, "h1", null, "H1" ) );

    handleJoins( "joinouter", new OuterJoin(), results );
    }

  /**
   * 1 a1
   * 1 a2
   * 1 a3
   * 2 b1
   * 3 c1
   * 4 d1
   * 4 d2
   * 4 d3
   * 5 e1
   * 5 e2
   * 5 e3
   * 7 g1
   * 7 g2
   * 7 g3
   * 7 g4
   * 7 g5
   * null h1
   * <p/>
   * 1 A1
   * 1 A2
   * 1 A3
   * 2 B1
   * 2 B2
   * 2 B3
   * 4 D1
   * 6 F1
   * 6 F2
   * null H1
   * <p/>
   * 1  a1      1       A1
   * 1  a1      1       A2
   * 1  a1      1       A3
   * 1  a2      1       A1
   * 1  a2      1       A2
   * 1  a2      1       A3
   * 1  a3      1       A1
   * 1  a3      1       A2
   * 1  a3      1       A3
   * 2  b1      2       B1
   * 2  b1      2       B2
   * 2  b1      2       B3
   * 3  c1      null    null
   * 4  d1      4       D1
   * 4  d2      4       D1
   * 4  d3      4       D1
   * 5  e1      null    null
   * 5  e2      null    null
   * 5  e3      null    null
   * 7  g1      null    null
   * 7  g2      null    null
   * 7  g3      null    null
   * 7  g4      null    null
   * 7  g5      null    null
   * null h1    null    H1
   *
   * @throws Exception
   */
  @Test
  public void testJoinInnerOuter() throws Exception
    {
    Set<Tuple> results = new HashSet<Tuple>();

    results.add( new Tuple( "1", "a1", "1", "A1" ) );
    results.add( new Tuple( "1", "a1", "1", "A2" ) );
    results.add( new Tuple( "1", "a1", "1", "A3" ) );
    results.add( new Tuple( "1", "a2", "1", "A1" ) );
    results.add( new Tuple( "1", "a2", "1", "A2" ) );
    results.add( new Tuple( "1", "a2", "1", "A3" ) );
    results.add( new Tuple( "1", "a3", "1", "A1" ) );
    results.add( new Tuple( "1", "a3", "1", "A2" ) );
    results.add( new Tuple( "1", "a3", "1", "A3" ) );
    results.add( new Tuple( "2", "b1", "2", "B1" ) );
    results.add( new Tuple( "2", "b1", "2", "B2" ) );
    results.add( new Tuple( "2", "b1", "2", "B3" ) );
    results.add( new Tuple( "3", "c1", null, null ) );
    results.add( new Tuple( "4", "d1", "4", "D1" ) );
    results.add( new Tuple( "4", "d2", "4", "D1" ) );
    results.add( new Tuple( "4", "d3", "4", "D1" ) );
    results.add( new Tuple( "5", "e1", null, null ) );
    results.add( new Tuple( "5", "e2", null, null ) );
    results.add( new Tuple( "5", "e3", null, null ) );
    results.add( new Tuple( "7", "g1", null, null ) );
    results.add( new Tuple( "7", "g2", null, null ) );
    results.add( new Tuple( "7", "g3", null, null ) );
    results.add( new Tuple( "7", "g4", null, null ) );
    results.add( new Tuple( "7", "g5", null, null ) );
    results.add( new Tuple( null, "h1", null, "H1" ) );

    handleJoins( "joininnerouter", new LeftJoin(), results );
    }

  /**
   * 1 a1
   * 1 a2
   * 1 a3
   * 2 b1
   * 3 c1
   * 4 d1
   * 4 d2
   * 4 d3
   * 5 e1
   * 5 e2
   * 5 e3
   * 7 g1
   * 7 g2
   * 7 g3
   * 7 g4
   * 7 g5
   * null h1
   * <p/>
   * 1 A1
   * 1 A2
   * 1 A3
   * 2 B1
   * 2 B2
   * 2 B3
   * 4 D1
   * 6 F1
   * 6 F2
   * null H1
   * <p/>
   * 1  a1      1       A1
   * 1  a1      1       A2
   * 1  a1      1       A3
   * 1  a2      1       A1
   * 1  a2      1       A2
   * 1  a2      1       A3
   * 1  a3      1       A1
   * 1  a3      1       A2
   * 1  a3      1       A3
   * 2  b1      2       B1
   * 2  b1      2       B2
   * 2  b1      2       B3
   * 4  d1      4       D1
   * 4  d2      4       D1
   * 4  d3      4       D1
   * null       null    6       F1
   * null       null    6       F2
   * null h1    null    H1
   *
   * @throws Exception
   */
  @Test
  public void testJoinOuterInner() throws Exception
    {
    // skip if hadoop cluster mode, outer joins don't behave the same
    if( getPlatform().isMapReduce() && getPlatform().isUseCluster() )
      return;

    Set<Tuple> results = new HashSet<Tuple>();

    results.add( new Tuple( "1", "a1", "1", "A1" ) );
    results.add( new Tuple( "1", "a1", "1", "A2" ) );
    results.add( new Tuple( "1", "a1", "1", "A3" ) );
    results.add( new Tuple( "1", "a2", "1", "A1" ) );
    results.add( new Tuple( "1", "a2", "1", "A2" ) );
    results.add( new Tuple( "1", "a2", "1", "A3" ) );
    results.add( new Tuple( "1", "a3", "1", "A1" ) );
    results.add( new Tuple( "1", "a3", "1", "A2" ) );
    results.add( new Tuple( "1", "a3", "1", "A3" ) );
    results.add( new Tuple( "2", "b1", "2", "B1" ) );
    results.add( new Tuple( "2", "b1", "2", "B2" ) );
    results.add( new Tuple( "2", "b1", "2", "B3" ) );
    results.add( new Tuple( "4", "d1", "4", "D1" ) );
    results.add( new Tuple( "4", "d2", "4", "D1" ) );
    results.add( new Tuple( "4", "d3", "4", "D1" ) );
    results.add( new Tuple( null, null, "6", "F1" ) );
    results.add( new Tuple( null, null, "6", "F2" ) );
    results.add( new Tuple( null, "h1", null, "H1" ) );

    handleJoins( "joinouterinner", new RightJoin(), results );
    }

  private void handleJoins( String path, Joiner joiner, Set<Tuple> results ) throws Exception
    {
    getPlatform().copyFromLocal( inputFileLhsSparse );
    getPlatform().copyFromLocal( inputFileRhsSparse );

    Fields fields = new Fields( "num", "char" ).applyTypes( Integer.class, String.class );
    Tap sourceLower = getPlatform().getDelimitedFile( fields, " ", inputFileLhsSparse );
    Tap sourceUpper = getPlatform().getDelimitedFile( fields, " ", inputFileRhsSparse );

    Map sources = new HashMap();

    sources.put( "lower", sourceLower );
    sources.put( "upper", sourceUpper );

    Tap sink = getPlatform().getDelimitedFile( Fields.size( 4, String.class ), "\t", getOutputPath( path ), SinkMode.REPLACE );

    Pipe pipeLower = new Pipe( "lower" );
    Pipe pipeUpper = new Pipe( "upper" );

    Fields declaredFields = new Fields( "num", "char", "num2", "char2" );
    Fields groupingFields = new Fields( "num" );

    Pipe splice = new HashJoin( pipeLower, groupingFields, pipeUpper, groupingFields, declaredFields, joiner );

    splice = new Each( splice, Fields.ALL, new Identity(), Fields.RESULTS );

    Flow flow = getPlatform().getFlowConnector().connect( sources, sink, splice );

    flow.complete();

    validateLength( flow, results.size() );

    List<Tuple> actual = getSinkAsList( flow );

    results.removeAll( actual );

    assertEquals( 0, results.size() );
    }

  /**
   * 1 a
   * 5 b
   * 6 c
   * 5 b
   * 5 e
   * <p/>
   * 1 A
   * 2 B
   * 3 C
   * 4 D
   * 5 E
   * <p/>
   * 1 a
   * 2 b
   * 3 c
   * 4 d
   * 5 e
   * <p/>
   * 1  a       1       A  1  a
   * -  -   2   B  2  b
   * -  -   3   C  3  c
   * -  -   4   D  4  d
   * 5  b       5   E  5  e
   * 5  e       5   E  5  e
   *
   * @throws Exception
   */
  @Test
  public void testJoinMixed() throws Exception
    {
    // skip if hadoop cluster mode, outer joins don't behave the same
    if( getPlatform().isMapReduce() && getPlatform().isUseCluster() )
      return;

    getPlatform().copyFromLocal( inputFileLowerOffset );
    getPlatform().copyFromLocal( inputFileLower );
    getPlatform().copyFromLocal( inputFileUpper );

    Tap sourceLowerOffset = getPlatform().getTextFile( new Fields( "offset", "line" ), inputFileLowerOffset );
    Tap sourceUpper = getPlatform().getTextFile( new Fields( "offset", "line" ), inputFileUpper );
    Tap sourceLower = getPlatform().getTextFile( new Fields( "offset", "line" ), inputFileLower );

    Map sources = new HashMap();

    sources.put( "loweroffset", sourceLowerOffset );
    sources.put( "lower", sourceLower );
    sources.put( "upper", sourceUpper );

    Tap sink = getPlatform().getTextFile( new Fields( "line" ), getOutputPath( "joinmixed" ), SinkMode.REPLACE );

    Function splitter = new RegexSplitter( new Fields( "num", "char" ), " " );

    Pipe pipeLowerOffset = new Each( new Pipe( "loweroffset" ), new Fields( "line" ), splitter );
    Pipe pipeLower = new Each( new Pipe( "lower" ), new Fields( "line" ), splitter );
    Pipe pipeUpper = new Each( new Pipe( "upper" ), new Fields( "line" ), splitter );

    Pipe[] pipes = Pipe.pipes( pipeLowerOffset, pipeUpper, pipeLower );
    Fields[] fields = Fields.fields( new Fields( "num" ), new Fields( "num" ), new Fields( "num" ) );

    MixedJoin join = new MixedJoin( new boolean[]{MixedJoin.OUTER, MixedJoin.INNER, MixedJoin.OUTER} );
    Pipe splice = new HashJoin( pipes, fields, Fields.size( 6 ), join );

    Flow flow = getPlatform().getFlowConnector().connect( sources, sink, splice );

    flow.complete();

    validateLength( flow, 6 );

    Set<Tuple> results = new HashSet<Tuple>();

    results.add( new Tuple( "1\ta\t1\tA\t1\ta" ) );
    results.add( new Tuple( "null\tnull\t2\tB\t2\tb" ) );
    results.add( new Tuple( "null\tnull\t3\tC\t3\tc" ) );
    results.add( new Tuple( "null\tnull\t4\tD\t4\td" ) );
    results.add( new Tuple( "5\tb\t5\tE\t5\te" ) );
    results.add( new Tuple( "5\te\t5\tE\t5\te" ) );

    List<Tuple> actual = getSinkAsList( flow );

    results.removeAll( actual );

    assertEquals( 0, results.size() );
    }

  @Test
  public void testJoinDiffFields() throws Exception
    {
    getPlatform().copyFromLocal( inputFileLower );
    getPlatform().copyFromLocal( inputFileUpper );

    Tap sourceLower = getPlatform().getTextFile( new Fields( "offset", "line" ), inputFileLower );
    Tap sourceUpper = getPlatform().getTextFile( new Fields( "offset", "line" ), inputFileUpper );

    Map sources = new HashMap();

    sources.put( "lower", sourceLower );
    sources.put( "upper", sourceUpper );

    Tap sink = getPlatform().getTextFile( new Fields( "line" ), getOutputPath( "difffields" ), SinkMode.REPLACE );

    Function splitterLower = new RegexSplitter( new Fields( "numA", "lower" ), " " );
    Function splitterUpper = new RegexSplitter( new Fields( "numB", "upper" ), " " );

    Pipe pipeLower = new Each( new Pipe( "lower" ), new Fields( "line" ), splitterLower );
    Pipe pipeUpper = new Each( new Pipe( "upper" ), new Fields( "line" ), splitterUpper );

    Pipe pipe = new HashJoin( pipeLower, new Fields( "numA" ), pipeUpper, new Fields( "numB" ) );

    Flow flow = getPlatform().getFlowConnector().connect( sources, sink, pipe );

    flow.complete();

    validateLength( flow, 5 );

    List<Tuple> actual = getSinkAsList( flow );

    assertTrue( actual.contains( new Tuple( "1\ta\t1\tA" ) ) );
    assertTrue( actual.contains( new Tuple( "2\tb\t2\tB" ) ) );
    }

  @Test
  public void testJoinGroupBy() throws Exception
    {
    getPlatform().copyFromLocal( inputFileLower );
    getPlatform().copyFromLocal( inputFileUpper );

    Tap sourceLower = getPlatform().getTextFile( new Fields( "offset", "line" ), inputFileLower );
    Tap sourceUpper = getPlatform().getTextFile( new Fields( "offset", "line" ), inputFileUpper );

    Map sources = new HashMap();

    sources.put( "lower", sourceLower );
    sources.put( "upper", sourceUpper );

    Tap sink = getPlatform().getTextFile( new Fields( "line" ), getOutputPath( "joingroupby" ), SinkMode.REPLACE );

    Function splitterLower = new RegexSplitter( new Fields( "numA", "lower" ), " " );
    Function splitterUpper = new RegexSplitter( new Fields( "numB", "upper" ), " " );

    Pipe pipeLower = new Each( new Pipe( "lower" ), new Fields( "line" ), splitterLower );
    Pipe pipeUpper = new Each( new Pipe( "upper" ), new Fields( "line" ), splitterUpper );

    Pipe pipe = new HashJoin( pipeLower, new Fields( "numA" ), pipeUpper, new Fields( "numB" ) );

    Pipe groupby = new GroupBy( pipe, new Fields( "numA" ) );

    Flow flow = getPlatform().getFlowConnector().connect( sources, sink, groupby );

    flow.complete();

    validateLength( flow, 5, null );

    List<Tuple> actual = getSinkAsList( flow );

    assertTrue( actual.contains( new Tuple( "1\ta\t1\tA" ) ) );
    assertTrue( actual.contains( new Tuple( "2\tb\t2\tB" ) ) );
    }

  @Test
  public void testJoinSamePipe() throws Exception
    {
    getPlatform().copyFromLocal( inputFileLower );

    Tap source = getPlatform().getTextFile( new Fields( "offset", "line" ), inputFileLower );

    Map sources = new HashMap();

    sources.put( "lower", source );

    Tap sink = getPlatform().getTextFile( new Fields( "line" ), getOutputPath( "samepipe" ), SinkMode.REPLACE );

    Function splitter = new RegexSplitter( new Fields( "num", "char" ), " " );

    Pipe pipeLower = new Each( new Pipe( "lower" ), new Fields( "line" ), splitter );

    Pipe pipe = new HashJoin( pipeLower, new Fields( "num" ), 1, new Fields( "num1", "char1", "num2", "char2" ) );

    Flow flow = getPlatform().getFlowConnector().connect( sources, sink, pipe );

    flow.complete();

    validateLength( flow, 5, null );

    List<Tuple> actual = getSinkAsList( flow );

    assertTrue( actual.contains( new Tuple( "1\ta\t1\ta" ) ) );
    assertTrue( actual.contains( new Tuple( "2\tb\t2\tb" ) ) );
    }

  @Test
  public void testJoinSamePipe2() throws Exception
    {
    getPlatform().copyFromLocal( inputFileLower );

    Tap source = getPlatform().getTextFile( new Fields( "offset", "line" ), inputFileLower );

    Map sources = new HashMap();

    sources.put( "lower", source );

    Tap sink = getPlatform().getTextFile( new Fields( "line" ), getOutputPath( "samepipe2" ), SinkMode.REPLACE );

    Function splitter = new RegexSplitter( new Fields( "num", "char" ), " " );

    Pipe pipeLower = new Each( new Pipe( "lower" ), new Fields( "line" ), splitter );

    Pipe join = new HashJoin( pipeLower, new Fields( "num" ), pipeLower, new Fields( "num" ), new Fields( "num1", "char1", "num2", "char2" ) );

    Flow flow = getPlatform().getFlowConnector().connect( sources, sink, join );

    flow.complete();

    validateLength( flow, 5, null );

    List<Tuple> actual = getSinkAsList( flow );

    assertTrue( actual.contains( new Tuple( "1\ta\t1\ta" ) ) );
    assertTrue( actual.contains( new Tuple( "2\tb\t2\tb" ) ) );
    }

  @Test
  public void testJoinSamePipe3() throws Exception
    {
    getPlatform().copyFromLocal( inputFileLower );

    Tap source = getPlatform().getDelimitedFile( new Fields( "num", "char" ), " ", inputFileLower );

    Map sources = new HashMap();

    sources.put( "lower", source );

    Tap sink = getPlatform().getTextFile( new Fields( "line" ), getOutputPath( "samepipe3" ), SinkMode.REPLACE );

    Pipe pipe = new Pipe( "lower" );

    Pipe lhs = new Pipe( "lhs", pipe );
    Pipe rhs = new Pipe( "rhs", pipe );

    Pipe join = new HashJoin( lhs, new Fields( "num" ), rhs, new Fields( "num" ), new Fields( "num1", "char1", "num2", "char2" ) );

    Flow flow = getPlatform().getFlowConnector().connect( sources, sink, join );

    flow.complete();

    validateLength( flow, 5, null );

    List<Tuple> actual = getSinkAsList( flow );

    assertTrue( actual.contains( new Tuple( "1\ta\t1\ta" ) ) );
    assertTrue( actual.contains( new Tuple( "2\tb\t2\tb" ) ) );
    }

  /**
   * Same source as rightmost
   * <p/>
   * should be a single job as the same file accumulates into the joins
   *
   * @throws Exception
   */
  @Test
  public void testJoinAroundJoinRightMost() throws Exception
    {
    getPlatform().copyFromLocal( inputFileLower );
    getPlatform().copyFromLocal( inputFileUpper );

    Tap sourceLower = getPlatform().getTextFile( new Fields( "offset", "line" ), inputFileLower );
    Tap sourceUpper = getPlatform().getTextFile( new Fields( "offset", "line" ), inputFileUpper );

    Map sources = new HashMap();

    sources.put( "lower", sourceLower );
    sources.put( "upper1", sourceUpper );
    sources.put( "upper2", sourceUpper );

    Tap sink = getPlatform().getTextFile( new Fields( "line" ), getOutputPath( "joinaroundjoinrightmost" ), SinkMode.REPLACE );

    Function splitter = new RegexSplitter( new Fields( "num", "char" ), " " );

    Pipe pipeLower = new Each( new Pipe( "lower" ), new Fields( "line" ), splitter );
    Pipe pipeUpper1 = new Each( new Pipe( "upper1" ), new Fields( "line" ), splitter );
    Pipe pipeUpper2 = new Each( new Pipe( "upper2" ), new Fields( "line" ), splitter );

    Pipe splice1 = new HashJoin( pipeLower, new Fields( "num" ), pipeUpper1, new Fields( "num" ), new Fields( "num1", "char1", "num2", "char2" ) );

    splice1 = new Each( splice1, new Identity() );

    Pipe splice2 = new HashJoin( splice1, new Fields( "num1" ), pipeUpper2, new Fields( "num" ), new Fields( "num1", "char1", "num2", "char2", "num3", "char3" ) );

    Flow flow = getPlatform().getFlowConnector().connect( sources, sink, splice2 );

//    flow.writeDOT( "joinaroundrightmost.dot" );

    if( getPlatform().isMapReduce() )
      assertEquals( "wrong number of steps", 1, flow.getFlowSteps().size() );

    flow.complete();

    validateLength( flow, 5, null );

    List<Tuple> actual = getSinkAsList( flow );

    assertTrue( actual.contains( new Tuple( "1\ta\t1\tA\t1\tA" ) ) );
    assertTrue( actual.contains( new Tuple( "2\tb\t2\tB\t2\tB" ) ) );
    }

  /**
   * Same source as leftmost
   *
   * @throws Exception
   */
  @Test
  public void testJoinAroundJoinLeftMost() throws Exception
    {
    getPlatform().copyFromLocal( inputFileLower );
    getPlatform().copyFromLocal( inputFileUpper );

    Tap sourceLower = getPlatform().getTextFile( new Fields( "offset", "line" ), inputFileLower );
    Tap sourceUpper = getPlatform().getTextFile( new Fields( "offset", "line" ), inputFileUpper );

    Map sources = new HashMap();

    sources.put( "lower", sourceLower );
    sources.put( "upper1", sourceUpper );
    sources.put( "upper2", sourceUpper );

    Tap sink = getPlatform().getTextFile( new Fields( "line" ), getOutputPath( "joinaroundjoinleftmost" ), SinkMode.REPLACE );

    Function splitter = new RegexSplitter( new Fields( "num", "char" ), " " );

    Pipe pipeLower = new Each( new Pipe( "lower" ), new Fields( "line" ), splitter );
    Pipe pipeUpper1 = new Each( new Pipe( "upper1" ), new Fields( "line" ), splitter );
    Pipe pipeUpper2 = new Each( new Pipe( "upper2" ), new Fields( "line" ), splitter );

    Pipe splice1 = new HashJoin( pipeUpper1, new Fields( "num" ), pipeUpper2, new Fields( "num" ), new Fields( "num1", "char1", "num2", "char2" ) );

    splice1 = new Each( splice1, new Identity() );

    Pipe splice2 = new HashJoin( splice1, new Fields( "num1" ), pipeLower, new Fields( "num" ), new Fields( "num1", "char1", "num2", "char2", "num3", "char3" ) );

    Flow flow = getPlatform().getFlowConnector().connect( sources, sink, splice2 );

//    flow.writeDOT( "joinaroundleftmost.dot" );

    if( getPlatform().isMapReduce() )
      assertEquals( "wrong number of steps", 2, flow.getFlowSteps().size() );

    flow.complete();

    validateLength( flow, 5, null );

    List<Tuple> actual = getSinkAsList( flow );

    assertTrue( actual.contains( new Tuple( "1\tA\t1\tA\t1\ta" ) ) );
    assertTrue( actual.contains( new Tuple( "2\tB\t2\tB\t2\tb" ) ) );
    }

  /**
   * Upper as leftmost and rightmost forcing two jobs
   *
   * @throws Exception
   */
  @Test
  public void testJoinAroundJoinRightMostSwapped() throws Exception
    {
    getPlatform().copyFromLocal( inputFileLower );
    getPlatform().copyFromLocal( inputFileUpper );

    Tap sourceLower = getPlatform().getTextFile( new Fields( "offset", "line" ), inputFileLower );
    Tap sourceUpper = getPlatform().getTextFile( new Fields( "offset", "line" ), inputFileUpper );

    Map sources = new HashMap();

    sources.put( "lower", sourceLower );
    sources.put( "upper1", sourceUpper );
    sources.put( "upper2", sourceUpper );

    Tap sink = getPlatform().getTextFile( new Fields( "line" ), getOutputPath( "joinaroundjoinswapped" ), SinkMode.REPLACE );

    Function splitter = new RegexSplitter( new Fields( "num", "char" ), " " );

    Pipe pipeLower = new Each( new Pipe( "lower" ), new Fields( "line" ), splitter );
    Pipe pipeUpper1 = new Each( new Pipe( "upper1" ), new Fields( "line" ), splitter );
    Pipe pipeUpper2 = new Each( new Pipe( "upper2" ), new Fields( "line" ), splitter );

    Pipe splice1 = new HashJoin( pipeLower, new Fields( "num" ), pipeUpper1, new Fields( "num" ), new Fields( "num1", "char1", "num2", "char2" ) );

    splice1 = new Each( splice1, new Identity() );

    // upper2 becomes leftmost, forcing a tap between the joins
    Pipe splice2 = new HashJoin( pipeUpper2, new Fields( "num" ), splice1, new Fields( "num1" ), new Fields( "num1", "char1", "num2", "char2", "num3", "char3" ) );

    Flow flow = getPlatform().getFlowConnector().connect( sources, sink, splice2 );

    if( getPlatform().isMapReduce() )
      assertEquals( "wrong number of steps", 2, flow.getFlowSteps().size() );

    flow.complete();

    validateLength( flow, 5, null );

    List<Tuple> actual = getSinkAsList( flow );

    assertTrue( actual.contains( new Tuple( "1\tA\t1\ta\t1\tA" ) ) );
    assertTrue( actual.contains( new Tuple( "2\tB\t2\tb\t2\tB" ) ) );
    }

  @Test
  public void testJoinGroupByJoin() throws Exception
    {
    getPlatform().copyFromLocal( inputFileLower );
    getPlatform().copyFromLocal( inputFileUpper );
    getPlatform().copyFromLocal( inputFileJoined );

    Tap sourceLower = getPlatform().getTextFile( new Fields( "offset", "line" ), inputFileLower );
    Tap sourceUpper = getPlatform().getTextFile( new Fields( "offset", "line" ), inputFileUpper );
    Tap sourceJoined = getPlatform().getTextFile( new Fields( "offset", "line" ), inputFileJoined );

    Map sources = new HashMap();

    sources.put( "lower", sourceLower );
    sources.put( "upper", sourceUpper );
    sources.put( "joined", sourceJoined );

    Tap sink = getPlatform().getTextFile( new Fields( "line" ), getOutputPath( "joingroupbyjoin" ), SinkMode.REPLACE );

    Function splitterLower = new RegexSplitter( new Fields( "numA", "lower" ), " " );
    Function splitterUpper = new RegexSplitter( new Fields( "numB", "upper" ), " " );
    Function splitterJoined = new RegexSplitter( new Fields( "numC", "lowerC", "upperC" ), "\t" );

    Pipe pipeLower = new Each( new Pipe( "lower" ), new Fields( "line" ), splitterLower );
    Pipe pipeUpper = new Each( new Pipe( "upper" ), new Fields( "line" ), splitterUpper );
    Pipe pipeJoined = new Each( new Pipe( "joined" ), new Fields( "line" ), splitterJoined );

    Pipe pipe = new HashJoin( pipeLower, new Fields( "numA" ), pipeUpper, new Fields( "numB" ) );

    pipe = new GroupBy( pipe, new Fields( "numA" ) );

    pipe = new HashJoin( pipe, new Fields( "numA" ), pipeJoined, new Fields( "numC" ) );

    Flow flow = getPlatform().getFlowConnector().connect( sources, sink, pipe );

    if( getPlatform().isMapReduce() )
      assertEquals( "wrong number of steps", 2, flow.getFlowSteps().size() );

    flow.complete();

    validateLength( flow, 5, null );

    List<Tuple> actual = getSinkAsList( flow );

    assertTrue( actual.contains( new Tuple( "1\ta\t1\tA\t1\ta\tA" ) ) );
    assertTrue( actual.contains( new Tuple( "2\tb\t2\tB\t2\tb\tB" ) ) );
    }

  /**
   * here the same file is fed into the same HashJoin.
   * <p/>
   * This is three jobs.
   * <p/>
   * a temp tap is inserted before the accumulated branch for two reasons on the common HashJoin
   * <p/>
   * it is assumed the accumulated side is filtered down, so pushing to disk will preserve io
   * if accumulated side was streamed instead via a fork, only part of the file will accumulate into the HashJoin
   * <p/>
   * /-T-\ <-- accumulated
   * T      HJ
   * \---/ <-- streamed
   *
   * @throws Exception
   */
  @Test
  public void testJoinSameSourceIntoJoin() throws Exception
    {
    getPlatform().copyFromLocal( inputFileLower );
    getPlatform().copyFromLocal( inputFileUpper );

    Tap sourceLower = getPlatform().getTextFile( new Fields( "offset", "line" ), inputFileLower );
    Tap sourceUpper = getPlatform().getTextFile( new Fields( "offset", "line" ), inputFileUpper );

    Map sources = new HashMap();

    sources.put( "lower", sourceLower );
    sources.put( "upper1", sourceUpper );
    sources.put( "upper2", sourceUpper );

    Tap sink = getPlatform().getTextFile( new Fields( "line" ), getOutputPath( "joinsamesourceintojoin" ), SinkMode.REPLACE );

    Function splitter = new RegexSplitter( new Fields( "num", "char" ), " " );

    Pipe pipeLower = new Each( new Pipe( "lower" ), new Fields( "line" ), splitter );
    Pipe pipeUpper1 = new Each( new Pipe( "upper1" ), new Fields( "line" ), splitter );
    Pipe pipeUpper2 = new Each( new Pipe( "upper2" ), new Fields( "line" ), splitter );

    Pipe splice1 = new HashJoin( pipeUpper1, new Fields( "num" ), pipeUpper2, new Fields( "num" ), new Fields( "num1", "char1", "num2", "char2" ) );

    splice1 = new Each( splice1, new Identity() );

    Pipe splice2 = new HashJoin( pipeLower, new Fields( "num" ), splice1, new Fields( "num1" ), new Fields( "num1", "char1", "num2", "char2", "num3", "char3" ) );

    Flow flow = getPlatform().getFlowConnector().connect( sources, sink, splice2 );

//    flow.writeDOT( "joinsamesourceintojoin.dot" );

    if( getPlatform().isMapReduce() )
      assertEquals( "wrong number of steps", 3, flow.getFlowSteps().size() );

    flow.complete();

    validateLength( flow, 5, null );

    List<Tuple> actual = getSinkAsList( flow );

    assertTrue( actual.contains( new Tuple( "1\ta\t1\tA\t1\tA" ) ) );
    assertTrue( actual.contains( new Tuple( "2\tb\t2\tB\t2\tB" ) ) );
    }

  @Test
  public void testJoinSameSourceIntoJoinSimple() throws Exception
    {
    getPlatform().copyFromLocal( inputFileUpper );

    Tap sourceUpper = getPlatform().getTextFile( new Fields( "offset", "line" ), inputFileUpper );

    Map sources = new HashMap();

    sources.put( "upper1", sourceUpper );
    sources.put( "upper2", sourceUpper );

    Tap sink = getPlatform().getTextFile( new Fields( "line" ), getOutputPath( "joinsamesourceintojoinsimple" ), SinkMode.REPLACE );

    Function splitter = new RegexSplitter( new Fields( "num", "char" ), " " );

    Pipe pipeUpper1 = new Each( new Pipe( "upper1" ), new Fields( "line" ), splitter );
    Pipe pipeUpper2 = new Each( new Pipe( "upper2" ), new Fields( "line" ), splitter );

    Pipe splice1 = new HashJoin( pipeUpper1, new Fields( "num" ), pipeUpper2, new Fields( "num" ), new Fields( "num1", "char1", "num2", "char2" ) );

    splice1 = new Each( splice1, new Identity() );

    Flow flow = getPlatform().getFlowConnector().connect( sources, sink, splice1 );

//    flow.writeDOT( "joinsamesourceintojoin.dot" );

    if( getPlatform().isMapReduce() )
      assertEquals( "wrong number of steps", 2, flow.getFlowSteps().size() );

    flow.complete();

    validateLength( flow, 5, null );

    List<Tuple> actual = getSinkAsList( flow );

    assertTrue( actual.contains( new Tuple( "1\tA\t1\tA" ) ) );
    assertTrue( actual.contains( new Tuple( "2\tB\t2\tB" ) ) );
    }

  /**
   * Loosely tests for a deadlock when BlockingHashJoinAnnotator rule doesn't excluce the GroupBy from the blocking
   * annotation.
   * <p/>
   * the deadlock is random on the order of the paths traversed from the Source Tap + fork.
   *
   * @throws Exception
   */
  @Test
  public void testJoinSameSourceOverGroupByIntoJoinSimple() throws Exception
    {
    getPlatform().copyFromLocal( inputFileLower );
    getPlatform().copyFromLocal( inputFileUpper );

    Tap sourceUpper = getPlatform().getTextFile( new Fields( "offset", "line" ), inputFileUpper );

    Map sources = new HashMap();

    sources.put( "upper1", sourceUpper );
    sources.put( "upper2", sourceUpper );

    Tap sink = getPlatform().getTextFile( new Fields( "line" ), getOutputPath( "joinsamesourceovergroupbyintojoinsimple" ), SinkMode.REPLACE );

    Function splitter = new RegexSplitter( new Fields( "num", "char" ), " " );

    Pipe pipeUpper1 = new Each( new Pipe( "upper1" ), new Fields( "line" ), splitter );
    Pipe pipeUpper2 = new Each( new Pipe( "upper2" ), new Fields( "line" ), splitter );

    pipeUpper1 = new GroupBy( pipeUpper1, new Fields( "num" ) );
    pipeUpper2 = new GroupBy( pipeUpper2, new Fields( "num" ) );

    Pipe splice1 = new HashJoin( pipeUpper1, new Fields( "num" ), pipeUpper2, new Fields( "num" ), new Fields( "num1", "char1", "num2", "char2" ) );

    splice1 = new Each( splice1, new Identity() );

    Flow flow = getPlatform().getFlowConnector().connect( sources, sink, splice1 );

    if( getPlatform().isMapReduce() )
      assertEquals( "wrong number of steps", 3, flow.getFlowSteps().size() );

    flow.complete();

    validateLength( flow, 5, null );

    List<Tuple> actual = getSinkAsList( flow );

    assertTrue( actual.contains( new Tuple( "1\tA\t1\tA" ) ) );
    assertTrue( actual.contains( new Tuple( "2\tB\t2\tB" ) ) );
    }

  /**
   * Tests that two independent streamed sources with loadable tributaries properly plan into a GroupBy
   * without loading unused sources
   *
   * @throws Exception
   */
  @Test
  public void testJoinsIntoGroupBy() throws Exception
    {
    getPlatform().copyFromLocal( inputFileLower );
    getPlatform().copyFromLocal( inputFileUpper );

    getPlatform().copyFromLocal( inputFileLhs );
    getPlatform().copyFromLocal( inputFileRhs );

    Tap sourceLower = getPlatform().getTextFile( new Fields( "offset", "line" ), inputFileLower );
    Tap sourceUpper = getPlatform().getTextFile( new Fields( "offset", "line" ), inputFileUpper );

    Tap sourceLhs = getPlatform().getTextFile( new Fields( "offset", "line" ), inputFileLhs );
    Tap sourceRhs = getPlatform().getTextFile( new Fields( "offset", "line" ), inputFileRhs );

    Map sources = new HashMap();

    sources.put( "lower", sourceLower );
    sources.put( "upper", sourceUpper );
    sources.put( "lhs", sourceLhs );
    sources.put( "rhs", sourceRhs );

    Tap sink = getPlatform().getTextFile( new Fields( "line" ), getOutputPath( "joinsintogroupby" ), SinkMode.REPLACE );

    Function splitter = new RegexSplitter( new Fields( "num", "char" ), " " );

    Pipe pipeLower = new Each( new Pipe( "lower" ), new Fields( "line" ), splitter );
    Pipe pipeUpper = new Each( new Pipe( "upper" ), new Fields( "line" ), splitter );

    Pipe pipeLhs = new Each( new Pipe( "lhs" ), new Fields( "line" ), splitter );
    Pipe pipeRhs = new Each( new Pipe( "rhs" ), new Fields( "line" ), splitter );

    Pipe upperLower = new HashJoin( pipeLower, new Fields( "num" ), pipeUpper, new Fields( "num" ), new Fields( "num1", "char1", "num2", "char2" ) );

    upperLower = new Each( upperLower, new Identity() );

    Pipe lhsRhs = new HashJoin( pipeLhs, new Fields( "num" ), pipeRhs, new Fields( "num" ), new Fields( "num1", "char1", "num2", "char2" ) );

    lhsRhs = new Each( lhsRhs, new Identity() );

    Pipe grouped = new GroupBy( "merging", Pipe.pipes( upperLower, lhsRhs ), new Fields( "num1" ) );

    Flow flow = getPlatform().getFlowConnector().connect( sources, sink, grouped );

    if( getPlatform().isMapReduce() )
      assertEquals( "wrong number of steps", 1, flow.getFlowSteps().size() );

    flow.complete();

    validateLength( flow, 42, null );

    List<Tuple> actual = getSinkAsList( flow );

    assertTrue( actual.contains( new Tuple( "1\ta\t1\tA" ) ) );
    assertTrue( actual.contains( new Tuple( "5\te\t5\tE" ) ) );
    }

  @Test
  public void testJoinSamePipeAroundGroupBy() throws Exception
    {
    getPlatform().copyFromLocal( inputFileLower );

    Tap source = getPlatform().getTextFile( new Fields( "offset", "line" ), inputFileLower );
    Tap sink = getPlatform().getTextFile( new Fields( "line" ), getOutputPath( "samepipearoundgroupby" ), SinkMode.REPLACE );

    Function splitter = new RegexSplitter( new Fields( "num", "char" ), " " );

    Pipe pipeLower = new Each( new Pipe( "lower" ), new Fields( "line" ), splitter );

    Pipe lhsPipe = new Each( new Pipe( "lhs", pipeLower ), new Identity() );

    Pipe rhsPipe = new Each( new Pipe( "rhs", pipeLower ), new Identity() );

    rhsPipe = new GroupBy( rhsPipe, new Fields( "num" ) );

    rhsPipe = new Each( rhsPipe, new Identity() );

    Pipe pipe = new HashJoin( lhsPipe, new Fields( "num" ), rhsPipe, new Fields( "num" ), new Fields( "num1", "char1", "num2", "char2" ) );

    Flow flow = getPlatform().getFlowConnector().connect( source, sink, pipe );

    flow.complete();

    validateLength( flow, 5, null );

    List<Tuple> actual = getSinkAsList( flow );

    assertTrue( actual.contains( new Tuple( "1\ta\t1\ta" ) ) );
    assertTrue( actual.contains( new Tuple( "2\tb\t2\tb" ) ) );
    }

  /**
   * This test results in two MR jobs because one join feeds into the accumulated side of the second. A mapper
   * can only stream on branch at a time forcing a temp file between the mappers. see next test for swapped join
   *
   * @throws Exception
   */
  @Test
  public void testJoinsIntoCoGroupLhs() throws Exception
    {
    getPlatform().copyFromLocal( inputFileLower );
    getPlatform().copyFromLocal( inputFileUpper );

    getPlatform().copyFromLocal( inputFileLhs );
    getPlatform().copyFromLocal( inputFileRhs );

    Tap sourceLower = getPlatform().getTextFile( new Fields( "offset", "line" ), inputFileLower );
    Tap sourceUpper = getPlatform().getTextFile( new Fields( "offset", "line" ), inputFileUpper );

    Tap sourceLhs = getPlatform().getTextFile( new Fields( "offset", "line" ), inputFileLhs );
    Tap sourceRhs = getPlatform().getTextFile( new Fields( "offset", "line" ), inputFileRhs );

    Map sources = new HashMap();

    sources.put( "lower", sourceLower );
    sources.put( "upper", sourceUpper );
    sources.put( "lhs", sourceLhs );
    sources.put( "rhs", sourceRhs );

    Tap sink = getPlatform().getTextFile( new Fields( "line" ), getOutputPath( "joinsintocogrouplhs" ), SinkMode.REPLACE );

    Function splitter = new RegexSplitter( new Fields( "num", "char" ), " " );

    Pipe pipeLower = new Each( new Pipe( "lower" ), new Fields( "line" ), splitter );
    Pipe pipeUpper = new Each( new Pipe( "upper" ), new Fields( "line" ), splitter );

    Pipe pipeLhs = new Each( new Pipe( "lhs" ), new Fields( "line" ), splitter );
    Pipe pipeRhs = new Each( new Pipe( "rhs" ), new Fields( "line" ), splitter );

    Pipe upperLower = new HashJoin( pipeLower, new Fields( "num" ), pipeUpper, new Fields( "num" ), new Fields( "numUpperLower", "charUpperLower", "num2UpperLower", "char2UpperLower" ) );

    upperLower = new Each( upperLower, new Identity() );

    Pipe lhsUpperLower = new HashJoin( pipeLhs, new Fields( "num" ), upperLower, new Fields( "numUpperLower" ), new Fields( "numLhs", "charLhs", "numUpperLower", "charUpperLower", "num2UpperLower", "char2UpperLower" ) );

    lhsUpperLower = new Each( lhsUpperLower, new Identity() );

    Pipe grouped = new CoGroup( "cogrouping", lhsUpperLower, new Fields( "numLhs" ), pipeRhs, new Fields( "num" ) );

    Flow flow = getPlatform().getFlowConnector().connect( sources, sink, grouped );

    if( getPlatform().isMapReduce() )
      assertEquals( "wrong number of steps", 2, flow.getFlowSteps().size() );

    flow.complete();

    validateLength( flow, 37, null );

    List<Tuple> actual = getSinkAsList( flow );

    assertTrue( actual.contains( new Tuple( "1\ta\t1\ta\t1\tA\t1\tA" ) ) );
    assertTrue( actual.contains( new Tuple( "5\ta\t5\te\t5\tE\t5\tA" ) ) );
    }

  /**
   * This test results in one MR jobs because one join feeds into the streamed side of the second.
   *
   * @throws Exception
   */
  @Test
  public void testJoinsIntoCoGroupLhsSwappedJoin() throws Exception
    {
    getPlatform().copyFromLocal( inputFileLower );
    getPlatform().copyFromLocal( inputFileUpper );

    getPlatform().copyFromLocal( inputFileLhs );
    getPlatform().copyFromLocal( inputFileRhs );

    Tap sourceLower = getPlatform().getTextFile( new Fields( "offset", "line" ), inputFileLower );
    Tap sourceUpper = getPlatform().getTextFile( new Fields( "offset", "line" ), inputFileUpper );

    Tap sourceLhs = getPlatform().getTextFile( new Fields( "offset", "line" ), inputFileLhs );
    Tap sourceRhs = getPlatform().getTextFile( new Fields( "offset", "line" ), inputFileRhs );

    Map sources = new HashMap();

    sources.put( "lower", sourceLower );
    sources.put( "upper", sourceUpper );
    sources.put( "lhs", sourceLhs );
    sources.put( "rhs", sourceRhs );

    Tap sink = getPlatform().getTextFile( new Fields( "line" ), getOutputPath( "joinsintocogrouplhsswappedjoin" ), SinkMode.REPLACE );

    Function splitter = new RegexSplitter( new Fields( "num", "char" ), " " );

    Pipe pipeLower = new Each( new Pipe( "lower" ), new Fields( "line" ), splitter );
    Pipe pipeUpper = new Each( new Pipe( "upper" ), new Fields( "line" ), splitter );

    Pipe pipeLhs = new Each( new Pipe( "lhs" ), new Fields( "line" ), splitter );
    Pipe pipeRhs = new Each( new Pipe( "rhs" ), new Fields( "line" ), splitter );

    Pipe upperLower = new HashJoin( pipeLower, new Fields( "num" ), pipeUpper, new Fields( "num" ), new Fields( "numUpperLower", "charUpperLower", "num2UpperLower", "char2UpperLower" ) );

    upperLower = new Each( upperLower, new Identity() );

    Pipe lhsUpperLower = new HashJoin( upperLower, new Fields( "numUpperLower" ), pipeLhs, new Fields( "num" ), new Fields( "numUpperLower", "charUpperLower", "num2UpperLower", "char2UpperLower", "numLhs", "charLhs" ) );

    lhsUpperLower = new Each( lhsUpperLower, new Identity() );

    Pipe grouped = new CoGroup( "cogrouping", lhsUpperLower, new Fields( "numLhs" ), pipeRhs, new Fields( "num" ) );

    Flow flow = getPlatform().getFlowConnector().connect( sources, sink, grouped );

    if( getPlatform().isMapReduce() )
      assertEquals( "wrong number of steps", 1, flow.getFlowSteps().size() );

    flow.complete();

    validateLength( flow, 37, null );

    List<Tuple> actual = getSinkAsList( flow );

    assertTrue( actual.contains( new Tuple( "1\ta\t1\tA\t1\ta\t1\tA" ) ) );
    assertTrue( actual.contains( new Tuple( "5\te\t5\tE\t5\te\t5\tE" ) ) );
    }

  @Test
  public void testJoinsIntoCoGroupRhs() throws Exception
    {
    getPlatform().copyFromLocal( inputFileLower );
    getPlatform().copyFromLocal( inputFileUpper );

    getPlatform().copyFromLocal( inputFileLhs );
    getPlatform().copyFromLocal( inputFileRhs );

    Tap sourceLower = getPlatform().getTextFile( new Fields( "offset", "line" ), inputFileLower );
    Tap sourceUpper = getPlatform().getTextFile( new Fields( "offset", "line" ), inputFileUpper );

    Tap sourceLhs = getPlatform().getTextFile( new Fields( "offset", "line" ), inputFileLhs );
    Tap sourceRhs = getPlatform().getTextFile( new Fields( "offset", "line" ), inputFileRhs );

    Map sources = new HashMap();

    sources.put( "lower", sourceLower );
    sources.put( "upper", sourceUpper );
    sources.put( "lhs", sourceLhs );
    sources.put( "rhs", sourceRhs );

    Tap sink = getPlatform().getTextFile( new Fields( "line" ), getOutputPath( "joinsintocogrouprhs" ), SinkMode.REPLACE );

    Function splitter = new RegexSplitter( new Fields( "num", "char" ), " " );

    Pipe pipeLower = new Each( new Pipe( "lower" ), new Fields( "line" ), splitter );
    Pipe pipeUpper = new Each( new Pipe( "upper" ), new Fields( "line" ), splitter );

    Pipe pipeLhs = new Each( new Pipe( "lhs" ), new Fields( "line" ), splitter );
    Pipe pipeRhs = new Each( new Pipe( "rhs" ), new Fields( "line" ), splitter );

    Pipe upperLower = new HashJoin( pipeLower, new Fields( "num" ), pipeUpper, new Fields( "num" ), new Fields( "numUpperLower", "charUpperLower", "num2UpperLower", "char2UpperLower" ) );

    upperLower = new Each( upperLower, new Identity() );

    Pipe lhsUpperLower = new HashJoin( pipeLhs, new Fields( "num" ), upperLower, new Fields( "numUpperLower" ), new Fields( "numLhs", "charLhs", "numUpperLower", "charUpperLower", "num2UpperLower", "char2UpperLower" ) );

    lhsUpperLower = new Each( lhsUpperLower, new Identity() );

    Pipe grouped = new CoGroup( "cogrouping", pipeRhs, new Fields( "num" ), lhsUpperLower, new Fields( "numLhs" ) );

    Flow flow = getPlatform().getFlowConnector().connect( sources, sink, grouped );

    if( getPlatform().isMapReduce() )
      assertEquals( "wrong number of steps", 2, flow.getFlowSteps().size() );

    flow.complete();

    validateLength( flow, 37, null );

    List<Tuple> actual = getSinkAsList( flow );

    assertTrue( actual.contains( new Tuple( "1\tA\t1\ta\t1\ta\t1\tA" ) ) );
    assertTrue( actual.contains( new Tuple( "5\tE\t5\te\t5\te\t5\tE" ) ) );
    }

  @Test
  public void testJoinsIntoCoGroup() throws Exception
    {
    getPlatform().copyFromLocal( inputFileLower );
    getPlatform().copyFromLocal( inputFileUpper );

    getPlatform().copyFromLocal( inputFileLhs );
    getPlatform().copyFromLocal( inputFileRhs );

    Tap sourceLower = getPlatform().getTextFile( new Fields( "offset", "line" ), inputFileLower );
    Tap sourceUpper = getPlatform().getTextFile( new Fields( "offset", "line" ), inputFileUpper );

    Tap sourceLhs = getPlatform().getTextFile( new Fields( "offset", "line" ), inputFileLhs );
    Tap sourceRhs = getPlatform().getTextFile( new Fields( "offset", "line" ), inputFileRhs );

    Map sources = new HashMap();

    sources.put( "lower", sourceLower );
    sources.put( "upper", sourceUpper );
    sources.put( "lhs", sourceLhs );
    sources.put( "rhs", sourceRhs );

    Tap sink = getPlatform().getTextFile( new Fields( "line" ), getOutputPath( "joinsintocogroup" ), SinkMode.REPLACE );

    Function splitter = new RegexSplitter( new Fields( "num", "char" ), " " );

    Pipe pipeLower = new Each( new Pipe( "lower" ), new Fields( "line" ), splitter );
    Pipe pipeUpper = new Each( new Pipe( "upper" ), new Fields( "line" ), splitter );

    Pipe pipeLhs = new Each( new Pipe( "lhs" ), new Fields( "line" ), splitter );
    Pipe pipeRhs = new Each( new Pipe( "rhs" ), new Fields( "line" ), splitter );

    Pipe upperLower = new HashJoin( pipeLower, new Fields( "num" ), pipeUpper, new Fields( "num" ), new Fields( "numUpperLower1", "charUpperLower1", "numUpperLower2", "charUpperLower2" ) );

    upperLower = new Each( upperLower, new Identity() );

    Pipe lhsRhs = new HashJoin( pipeLhs, new Fields( "num" ), pipeRhs, new Fields( "num" ), new Fields( "numLhsRhs1", "charLhsRhs1", "numLhsRhs2", "charLhsRhs2" ) );

    lhsRhs = new Each( lhsRhs, new Identity() );

    Pipe grouped = new CoGroup( "cogrouping", upperLower, new Fields( "numUpperLower1" ), lhsRhs, new Fields( "numLhsRhs1" ) );

    Flow flow = getPlatform().getFlowConnector().connect( sources, sink, grouped );

    if( getPlatform().isMapReduce() )
      assertEquals( "wrong number of steps", 1, flow.getFlowSteps().size() );

    flow.complete();

    validateLength( flow, 37, null );

    List<Tuple> actual = getSinkAsList( flow );

    assertTrue( actual.contains( new Tuple( "1\ta\t1\tA\t1\ta\t1\tA" ) ) );
    assertTrue( actual.contains( new Tuple( "5\te\t5\tE\t5\te\t5\tE" ) ) );
    }

  public static class AllComparator implements Comparator<Comparable>, Hasher<Comparable>, Serializable
    {

    @Override
    public int compare( Comparable lhs, Comparable rhs )
      {
      return lhs.toString().compareTo( rhs.toString() );
      }

    @Override
    public int hashCode( Comparable value )
      {
      if( value == null )
        return 0;

      return value.toString().hashCode();
      }
    }

  /**
   * Tests Hasher being honored even if default comparator is null.
   *
   * @throws Exception
   */
  @Test
  public void testJoinWithHasher() throws Exception
    {
    getPlatform().copyFromLocal( inputFileLower );
    getPlatform().copyFromLocal( inputFileUpper );

    Tap sourceLower = getPlatform().getTextFile( new Fields( "offset", "line" ), inputFileLower );
    Tap sourceUpper = getPlatform().getTextFile( new Fields( "offset", "line" ), inputFileUpper );

    Map sources = new HashMap();

    sources.put( "lower", sourceLower );
    sources.put( "upper", sourceUpper );

    Tap sink = getPlatform().getTextFile( new Fields( "line" ), getOutputPath( "joinhasher" ), SinkMode.REPLACE );

    Function splitter = new RegexSplitter( new Fields( "num", "char" ), " " );

    Pipe pipeLower = new Each( new Pipe( "lower" ), new Fields( "line" ), splitter );

    pipeLower = new Each( pipeLower, new Fields( "num" ), new ExpressionFunction( Fields.ARGS, "Integer.parseInt( num )", String.class ), Fields.REPLACE );

    Pipe pipeUpper = new Each( new Pipe( "upper" ), new Fields( "line" ), splitter );

    Fields num = new Fields( "num" );
    num.setComparator( "num", new AllComparator() );

    Pipe splice = new HashJoin( pipeLower, num, pipeUpper, new Fields( "num" ), Fields.size( 4 ) );

    Map<Object, Object> properties = getProperties();

    Flow flow = getPlatform().getFlowConnector( properties ).connect( sources, sink, splice );

    flow.complete();

    validateLength( flow, 5 );

    List<Tuple> values = getSinkAsList( flow );

    assertTrue( values.contains( new Tuple( "1\ta\t1\tA" ) ) );
    assertTrue( values.contains( new Tuple( "2\tb\t2\tB" ) ) );
    }

  @Test
  public void testJoinNone() throws Exception
    {
    getPlatform().copyFromLocal( inputFileLower );
    getPlatform().copyFromLocal( inputFileUpper );

    Tap sourceLower = getPlatform().getTextFile( new Fields( "offset", "line" ), inputFileLower );
    Tap sourceUpper = getPlatform().getTextFile( new Fields( "offset", "line" ), inputFileUpper );

    Map sources = new HashMap();

    sources.put( "lower", sourceLower );
    sources.put( "upper", sourceUpper );

    Tap sink = getPlatform().getTextFile( new Fields( "line" ), getOutputPath( "joinnone" ), SinkMode.REPLACE );

    Function splitter = new RegexSplitter( new Fields( "num", "char" ), " " );

    Pipe pipeLower = new Each( new Pipe( "lower" ), new Fields( "line" ), splitter );
    Pipe pipeUpper = new Each( new Pipe( "upper" ), new Fields( "line" ), splitter );

    Pipe splice = new HashJoin( pipeLower, Fields.NONE, pipeUpper, Fields.NONE, Fields.size( 4 ) );

    Map<Object, Object> properties = getProperties();

    Flow flow = getPlatform().getFlowConnector( properties ).connect( sources, sink, splice );

    flow.complete();

    validateLength( flow, 25 );

    List<Tuple> values = getSinkAsList( flow );

    assertTrue( values.contains( new Tuple( "1\ta\t1\tA" ) ) );
    assertTrue( values.contains( new Tuple( "1\ta\t2\tB" ) ) );
    assertTrue( values.contains( new Tuple( "2\tb\t2\tB" ) ) );
    }

  @Test
  public void testGroupBySplitJoins() throws Exception
    {
    getPlatform().copyFromLocal( inputFileLower );
    getPlatform().copyFromLocal( inputFileUpper );
    getPlatform().copyFromLocal( inputFileJoined );

    Tap sourceLower = getPlatform().getTextFile( new Fields( "offset", "line" ), inputFileLower );
    Tap sourceUpper = getPlatform().getTextFile( new Fields( "offset", "line" ), inputFileUpper );
    Tap sourceJoined = getPlatform().getTextFile( new Fields( "offset", "line" ), inputFileJoined );

    Map sources = new HashMap();

    sources.put( "lower", sourceLower );
    sources.put( "upper", sourceUpper );
    sources.put( "joined", sourceJoined );

    Tap lhsSink = getPlatform().getTextFile( new Fields( "line" ), getOutputPath( "lhs" ), SinkMode.REPLACE );
    Tap rhsSink = getPlatform().getTextFile( new Fields( "line" ), getOutputPath( "rhs" ), SinkMode.REPLACE );

    Map sinks = new HashMap();

    sinks.put( "lhs", lhsSink );
    sinks.put( "rhs", rhsSink );

    Function splitterLower = new RegexSplitter( new Fields( "numA", "lower" ), " " );
    Function splitterUpper = new RegexSplitter( new Fields( "numB", "upper" ), " " );
    Function splitterJoined = new RegexSplitter( new Fields( "numC", "lowerC", "upperC" ), "\t" );

    Pipe pipeLower = new Each( new Pipe( "lower" ), new Fields( "line" ), splitterLower );
    Pipe pipeUpper = new Each( new Pipe( "upper" ), new Fields( "line" ), splitterUpper );
    Pipe pipeJoined = new Each( new Pipe( "joined" ), new Fields( "line" ), splitterJoined );

    Pipe pipe = new GroupBy( pipeLower, new Fields( "numA" ) );

    pipe = new Every( pipe, Fields.ALL, new TestIdentityBuffer( new Fields( "numA" ), 5, false ), Fields.RESULTS );

    Pipe lhsPipe = new Each( pipe, new Identity() );
    lhsPipe = new HashJoin( "lhs", lhsPipe, new Fields( "numA" ), pipeUpper, new Fields( "numB" ) );

    Pipe rhsPipe = new Each( pipe, new Identity() );
    rhsPipe = new HashJoin( "rhs", rhsPipe, new Fields( "numA" ), pipeJoined, new Fields( "numC" ) );

    Flow flow = getPlatform().getFlowConnector().connect( sources, sinks, lhsPipe, rhsPipe );

    if( getPlatform().isMapReduce() )
      assertEquals( "wrong number of steps", 3, flow.getFlowSteps().size() );

    flow.complete();

    validateLength( flow.openSink( "lhs" ), 5, null );
    validateLength( flow.openSink( "rhs" ), 5, null );

    List<Tuple> lhsActual = asList( flow, lhsSink );

    assertTrue( lhsActual.contains( new Tuple( "1\ta\t1\tA" ) ) );
    assertTrue( lhsActual.contains( new Tuple( "2\tb\t2\tB" ) ) );

    List<Tuple> rhsActual = asList( flow, rhsSink );

    assertTrue( rhsActual.contains( new Tuple( "1\ta\t1\ta\tA" ) ) );
    assertTrue( rhsActual.contains( new Tuple( "2\tb\t2\tb\tB" ) ) );
    }

  /**
   * When run against a cluster a Merge before a GroupBy can hide the streamed/accumulated nature of a branch.
   * <p>
   * The planner nw
   * <p>
   * <p/>
   * commented code is for troubleshooting.
   *
   * @throws Exception
   */
  @Test
  public void testJoinMergeGroupBy() throws Exception
    {
    getPlatform().copyFromLocal( inputFileNums10 );
    getPlatform().copyFromLocal( inputFileNums20 );

    Tap lhsTap = getPlatform().getTextFile( new Fields( "id" ), inputFileNums10 );
    Tap rhsTap = getPlatform().getTextFile( new Fields( "id2" ), inputFileNums20 );

    Pipe lhs = new Pipe( "lhs" );
    Pipe rhs = new Pipe( "rhs" );

//    Pipe joined = new CoGroup( messages, new Fields( "id" ), people, new Fields( "id2" ) );
    Pipe joined = new HashJoin( lhs, new Fields( "id" ), rhs, new Fields( "id2" ) );

    Pipe pruned = new Each( joined, new Fields( "id2" ), new Identity(), Fields.RESULTS );
//    pruned = new Checkpoint( pruned );
    Pipe merged = new Merge( pruned, rhs );
    Pipe grouped = new GroupBy( merged, new Fields( "id2" ) );
//    Pipe grouped = new GroupBy( Pipe.pipes(  pruned, people  ), new Fields( "id2" ) );
    Aggregator count = new Count( new Fields( "count" ) );
    Pipe counted = new Every( grouped, count );

    String testJoinMerge = "testJoinMergeGroupBy/" + ( ( joined instanceof CoGroup ) ? "cogroup" : "hashjoin" );
    Tap sink = getPlatform().getDelimitedFile( Fields.ALL, true, "\t", null, getOutputPath( testJoinMerge ), SinkMode.REPLACE );

    FlowDef flowDef = FlowDef.flowDef()
      .setName( "join-merge" )
      .addSource( rhs, rhsTap )
      .addSource( lhs, lhsTap )
      .addTailSink( counted, sink );

    Flow flow = getPlatform().getFlowConnector().connect( flowDef );

//    flow.writeDOT( "joinmerge.dot" );
//    flow.writeStepsDOT( "joinmerge-steps.dot" );

    flow.complete();

    validateLength( flow, 20 );

    List<Tuple> values = getSinkAsList( flow );
    List<Tuple> expected = new ArrayList<Tuple>();

    expected.add( new Tuple( "1", "2" ) );
    expected.add( new Tuple( "10", "2" ) );
    expected.add( new Tuple( "11", "1" ) );
    expected.add( new Tuple( "12", "1" ) );
    expected.add( new Tuple( "13", "1" ) );
    expected.add( new Tuple( "14", "1" ) );
    expected.add( new Tuple( "15", "1" ) );
    expected.add( new Tuple( "16", "1" ) );
    expected.add( new Tuple( "17", "1" ) );
    expected.add( new Tuple( "18", "1" ) );
    expected.add( new Tuple( "19", "1" ) );
    expected.add( new Tuple( "2", "2" ) );
    expected.add( new Tuple( "20", "1" ) );
    expected.add( new Tuple( "3", "2" ) );
    expected.add( new Tuple( "4", "2" ) );
    expected.add( new Tuple( "5", "2" ) );
    expected.add( new Tuple( "6", "2" ) );
    expected.add( new Tuple( "7", "2" ) );
    expected.add( new Tuple( "8", "2" ) );
    expected.add( new Tuple( "9", "2" ) );

    Collections.sort( values );
    Collections.sort( expected );

    assertEquals( expected, values );
    }

  /**
   * Under tez, this can result in the HashJoin being duplicated across nodes for each split after the HashJoin
   * BoundaryBalanceJoinSplitTransformer inserts a Boundary at the split, preventing duplication of the path
   *
   * @throws Exception
   */
  @Test
  public void testJoinSplit() throws Exception
    {
    getPlatform().copyFromLocal( inputFileLhs );
    getPlatform().copyFromLocal( inputFileRhs );

    FlowDef flowDef = FlowDef.flowDef()
      .addSource( "lhs", getPlatform().getTextFile( inputFileLhs ) )
      .addSource( "rhs", getPlatform().getTextFile( inputFileRhs ) )
      .addSink( "lhsSink", getPlatform().getTextFile( new Fields( "line" ), getOutputPath( "lhs" ), SinkMode.REPLACE ) )
      .addSink( "rhsSink", getPlatform().getTextFile( new Fields( "line" ), getOutputPath( "rhs" ), SinkMode.REPLACE ) );

    Pipe pipeLower = new Each( "lhs", new Fields( "line" ), new RegexSplitter( new Fields( "numLHS", "charLHS" ), " " ) );
    Pipe pipeUpper = new Each( "rhs", new Fields( "line" ), new RegexSplitter( new Fields( "numRHS", "charRHS" ), " " ) );

    Pipe join = new HashJoin( pipeLower, new Fields( "numLHS" ), pipeUpper, new Fields( "numRHS" ), new InnerJoin() );

    Pipe pipeLhs = new Each( new Pipe( "lhsSink", join ), new Identity() );
    Pipe pipeRhs = new Each( new Pipe( "rhsSink", join ), new Identity() );

    flowDef
      .addTail( pipeLhs )
      .addTail( pipeRhs );

    Flow flow = getPlatform().getFlowConnector().connect( flowDef );

    flow.complete();

    validateLength( flow, 37, null );

    List<Tuple> values = asList( flow, flowDef.getSinks().get( "lhsSink" ) );

    assertTrue( values.contains( new Tuple( "1\ta\t1\tA" ) ) );
    assertTrue( values.contains( new Tuple( "1\ta\t1\tB" ) ) );

    values = asList( flow, flowDef.getSinks().get( "rhsSink" ) );

    assertTrue( values.contains( new Tuple( "1\ta\t1\tA" ) ) );
    assertTrue( values.contains( new Tuple( "1\ta\t1\tB" ) ) );
    }

  /**
   * catches a situation where BottomUpJoinedBoundariesNodePartitioner may capture an invalid HashJoin sub-graph
   * if the in-bound Boundary is split upon.
   */
  @Test
  public void testSameSourceJoinSplitIntoJoin() throws Exception
    {
    getPlatform().copyFromLocal( inputFileLhs );
    getPlatform().copyFromLocal( inputFileRhs );

    FlowDef flowDef = FlowDef.flowDef()
      .addSource( "lhs", getPlatform().getTextFile( inputFileLhs ) )
      .addSource( "rhs", getPlatform().getTextFile( inputFileLhs ) )
      .addSource( "joinSecond", getPlatform().getTextFile( inputFileRhs ) )
      .addSink( "lhsSink", getPlatform().getTextFile( new Fields( "line" ), getOutputPath( "lhs" ), SinkMode.REPLACE ) )
      .addSink( "rhsSink", getPlatform().getTextFile( new Fields( "line" ), getOutputPath( "rhs" ), SinkMode.REPLACE ) );

    Pipe pipeLower = new Each( "lhs", new Fields( "line" ), new RegexSplitter( new Fields( "numLHS", "charLHS" ), " " ) );
    Pipe pipeUpper = new Each( "rhs", new Fields( "line" ), new RegexSplitter( new Fields( "numRHS", "charRHS" ), " " ) );

    Pipe joinFirst = new HashJoin( pipeLower, new Fields( "numLHS" ), pipeUpper, new Fields( "numRHS" ), new InnerJoin() );

    Pipe pipeLhs = new Each( new Pipe( "lhsSink", joinFirst ), new Identity() );

    Pipe joinSecond = new Each( "joinSecond", new Fields( "line" ), new RegexSplitter( new Fields( "numRHSSecond", "charRHSSecond" ), " " ) );

    joinSecond = new HashJoin( joinFirst, new Fields( "numLHS" ), joinSecond, new Fields( "numRHSSecond" ) );

    Pipe pipeRhs = new Each( new Pipe( "rhsSink", joinSecond ), new Identity() );

    flowDef
      .addTail( pipeLhs )
      .addTail( pipeRhs );

    Flow flow = getPlatform().getFlowConnector().connect( flowDef );

    flow.complete();

    List<Tuple> values = asList( flow, flowDef.getSinks().get( "lhsSink" ) );

    assertEquals( 37, values.size() );
    assertTrue( values.contains( new Tuple( "1\ta\t1\ta" ) ) );
    assertTrue( values.contains( new Tuple( "1\ta\t1\tb" ) ) );

    values = asList( flow, flowDef.getSinks().get( "rhsSink" ) );

    assertEquals( 109, values.size() );
    assertTrue( values.contains( new Tuple( "1\ta\t1\ta\t1\tA" ) ) );
    assertTrue( values.contains( new Tuple( "1\ta\t1\tb\t1\tB" ) ) );
    }

  /**
   * checks that a split after a HashJoin does not result in the HashJoin execution being duplicated across
   * multiple nodes, one for each branch in the split.
   */
  @Test
  public void testJoinSplitBeforeJoin() throws Exception
    {
    getPlatform().copyFromLocal( inputFileLhs );
    getPlatform().copyFromLocal( inputFileRhs );

    FlowDef flowDef = FlowDef.flowDef()
      .addSource( "lhs", getPlatform().getTextFile( inputFileLhs ) )
      .addSource( "rhs", getPlatform().getTextFile( inputFileRhs ) )
      .addSource( "joinSecond", getPlatform().getTextFile( inputFileRhs ) )
      .addSink( "lhsSink", getPlatform().getTextFile( new Fields( "line" ), getOutputPath( "lhs" ), SinkMode.REPLACE ) )
      .addSink( "rhsSink", getPlatform().getTextFile( new Fields( "line" ), getOutputPath( "rhs" ), SinkMode.REPLACE ) );

    Pipe pipeLower = new Each( "lhs", new Fields( "line" ), new RegexSplitter( new Fields( "numLHS", "charLHS" ), " " ) );
    Pipe pipeUpper = new Each( "rhs", new Fields( "line" ), new RegexSplitter( new Fields( "numRHS", "charRHS" ), " " ) );

    pipeUpper = new Checkpoint( pipeUpper );

    HashJoin hashJoin = new HashJoin( pipeLower, new Fields( "numLHS" ), pipeUpper, new Fields( "numRHS" ), new InnerJoin() );

    Pipe joinFirst = hashJoin;

    joinFirst = new Each( joinFirst, new Identity() );

    Pipe pipeLhs = new Each( new Pipe( "lhsSink", joinFirst ), new Identity() );

    pipeLhs = new GroupBy( pipeLhs, new Fields( "numLHS" ) );

    joinFirst = new Each( new Pipe( "lhsSplit", joinFirst ), new Identity() );

    Pipe joinSecond = new Each( "joinSecond", new Fields( "line" ), new RegexSplitter( new Fields( "numRHSSecond", "charRHSSecond" ), " " ) );

    joinSecond = new CoGroup( joinFirst, new Fields( "numLHS" ), joinSecond, new Fields( "numRHSSecond" ) );

    Pipe pipeRhs = new Each( new Pipe( "rhsSink", joinSecond ), new Identity() );

    flowDef
      .addTail( pipeLhs )
      .addTail( pipeRhs );

    Flow flow = getPlatform().getFlowConnector().connect( flowDef );

    if( getPlatform().isDAG() )
      {
      FlowStep flowStep = (FlowStep) flow.getFlowSteps().get( 0 );
      List<ElementGraph> elementGraphs = flowStep.getFlowNodeGraph().getElementGraphs( hashJoin );

      assertEquals( 1, elementGraphs.size() );
      }

    flow.complete();

    List<Tuple> values = asList( flow, flowDef.getSinks().get( "lhsSink" ) );

    assertEquals( 37, values.size() );
    assertTrue( values.contains( new Tuple( "1\ta\t1\tA" ) ) );
    assertTrue( values.contains( new Tuple( "1\ta\t1\tB" ) ) );

    values = asList( flow, flowDef.getSinks().get( "rhsSink" ) );

    assertEquals( 109, values.size() );
    assertTrue( values.contains( new Tuple( "1\ta\t1\tA\t1\tA" ) ) );
    assertTrue( values.contains( new Tuple( "1\ta\t1\tB\t1\tB" ) ) );
    }

  @Test
  public void testGroupBySplitGroupByJoin() throws Exception
    {
    getPlatform().copyFromLocal( inputFileLower );

    Tap source = getPlatform().getTextFile( new Fields( "offset", "line" ), inputFileLower );

    Tap sink = getPlatform().getTextFile( new Fields( "line" ), getOutputPath( "sink" ), SinkMode.REPLACE );

    Function splitter = new RegexSplitter( new Fields( "num", "char" ), " " );

    Pipe pipeFirst = new Pipe( "first" );
    pipeFirst = new Each( pipeFirst, new Fields( "line" ), splitter );
    pipeFirst = new GroupBy( pipeFirst, new Fields( "num" ) );
    pipeFirst = new Every( pipeFirst, new Fields( "char" ), new First( new Fields( "firstFirst" ) ), Fields.ALL );

    Pipe pipeSecond = new Pipe( "second", pipeFirst );
    pipeSecond = new Each( pipeSecond, new Identity() );
    pipeSecond = new GroupBy( pipeSecond, new Fields( "num" ) );
    pipeSecond = new Every( pipeSecond, new Fields( "firstFirst" ), new First( new Fields( "secondFirst" ) ), Fields.ALL );
    pipeSecond = new GroupBy( pipeSecond, new Fields( "num" ) );
    pipeSecond = new Every( pipeSecond, new Fields( "secondFirst" ), new First( new Fields( "thirdFirst" ) ), Fields.ALL );

    Pipe splice = new HashJoin( pipeFirst, new Fields( "num" ), pipeSecond, new Fields( "num" ), Fields.size( 4 ) );

    Flow flow = getPlatform().getFlowConnector().connect( source, sink, splice );

    flow.complete();

    validateLength( flow, 5, null );

    List<Tuple> values = getSinkAsList( flow );

    assertTrue( values.contains( new Tuple( "1\ta\t1\ta" ) ) );
    assertTrue( values.contains( new Tuple( "2\tb\t2\tb" ) ) );
    assertTrue( values.contains( new Tuple( "3\tc\t3\tc" ) ) );
    assertTrue( values.contains( new Tuple( "4\td\t4\td" ) ) );
    assertTrue( values.contains( new Tuple( "5\te\t5\te" ) ) );
    }

  @Test
  public void testGroupBySplitSplitGroupByJoin() throws Exception
    {
    getPlatform().copyFromLocal( inputFileLower );

    Tap source = getPlatform().getTextFile( new Fields( "offset", "line" ), inputFileLower );

    Tap sink = getPlatform().getTextFile( new Fields( "line" ), getOutputPath( "sink" ), SinkMode.REPLACE );

    Function splitter = new RegexSplitter( new Fields( "num", "char" ), " " );

    Pipe pipeFirst = new Pipe( "first" );
    pipeFirst = new Each( pipeFirst, new Fields( "line" ), splitter );
    pipeFirst = new GroupBy( pipeFirst, new Fields( "num" ) );
    pipeFirst = new Every( pipeFirst, new Fields( "char" ), new First( new Fields( "firstFirst" ) ), Fields.ALL );

    Pipe pipeSecond = new Pipe( "second", pipeFirst );
    pipeSecond = new Each( pipeSecond, new Identity() );
    pipeSecond = new GroupBy( pipeSecond, new Fields( "num" ) );
    pipeSecond = new Every( pipeSecond, new Fields( "firstFirst" ), new First( new Fields( "secondFirst" ) ), Fields.ALL );

    Pipe splice = new HashJoin( pipeFirst, new Fields( "num" ), pipeSecond, new Fields( "num" ), Fields.size( 4 ) );
//    Pipe splice = new HashJoin( pipeSecond, new Fields( "num" ), pipeFirst, new Fields( "num" ), Fields.size( 4 ) );

    splice = new HashJoin( splice, new Fields( 0 ), pipeSecond, new Fields( "num" ), Fields.size( 6 ) );

    Flow flow = getPlatform().getFlowConnector().connect( source, sink, splice );

    flow.complete();

    validateLength( flow, 5, null );

    List<Tuple> values = getSinkAsList( flow );

    assertTrue( values.contains( new Tuple( "1\ta\t1\ta\t1\ta" ) ) );
    assertTrue( values.contains( new Tuple( "2\tb\t2\tb\t2\tb" ) ) );
    assertTrue( values.contains( new Tuple( "3\tc\t3\tc\t3\tc" ) ) );
    assertTrue( values.contains( new Tuple( "4\td\t4\td\t4\td" ) ) );
    assertTrue( values.contains( new Tuple( "5\te\t5\te\t5\te" ) ) );
    }

  @Test
  public void testGroupBySplitAroundSplitGroupByJoin() throws Exception
    {
    getPlatform().copyFromLocal( inputFileLower );

    Tap source = getPlatform().getTextFile( new Fields( "offset", "line" ), inputFileLower );

    Tap sink = getPlatform().getTextFile( new Fields( "line" ), getOutputPath( "sink" ), SinkMode.REPLACE );
    Tap sink2 = getPlatform().getTextFile( new Fields( "line" ), getOutputPath( "sink2" ), SinkMode.REPLACE );

    Function splitter = new RegexSplitter( new Fields( "num", "char" ), " " );

    Pipe pipeInit = new Pipe( "init" );
    Pipe pipeFirst = new Pipe( "first", pipeInit );
    pipeFirst = new Each( pipeFirst, new Fields( "line" ), splitter );
    pipeFirst = new GroupBy( pipeFirst, new Fields( "num" ) );
    pipeFirst = new Every( pipeFirst, new Fields( "char" ), new First( new Fields( "firstFirst" ) ), Fields.ALL );

    Pipe sink2Pipe = new Pipe( "sink2", pipeFirst );

    Pipe pipeSecond = new Pipe( "second", pipeInit );
    pipeSecond = new Each( pipeSecond, new Fields( "line" ), splitter );
    pipeSecond = new GroupBy( pipeSecond, new Fields( "num" ) );
    pipeSecond = new Every( pipeSecond, new Fields( "char" ), new First( new Fields( "secondFirst" ) ), Fields.ALL );

//    Pipe splice = new HashJoin( pipeFirst, new Fields( "num" ), pipeSecond, new Fields( "num" ), Fields.size( 4 ) );
    Pipe splice = new HashJoin( pipeSecond, new Fields( "num" ), pipeFirst, new Fields( "num" ), Fields.size( 4 ) );

    Pipe pipeThird = new Pipe( "third", pipeSecond );
    pipeThird = new Each( pipeThird, new Identity() );
    pipeThird = new GroupBy( pipeThird, new Fields( "num" ) );
    pipeThird = new Every( pipeThird, new Fields( "secondFirst" ), new First( new Fields( "thirdFirst" ) ), Fields.ALL );

    splice = new HashJoin( splice, new Fields( 0 ), pipeThird, new Fields( "num" ), Fields.size( 6 ) );

    FlowDef flowDef = FlowDef.flowDef()
      .setName( splice.getName() )
      .addSource( "init", source )
      .addTailSink( splice, sink )
      .addTailSink( sink2Pipe, sink2 );

    Flow flow = getPlatform().getFlowConnector().connect( flowDef );

    flow.complete();

    validateLength( flow, 5, null );

    List<Tuple> values = getSinkAsList( flow );

    assertTrue( values.contains( new Tuple( "1\ta\t1\ta\t1\ta" ) ) );
    assertTrue( values.contains( new Tuple( "2\tb\t2\tb\t2\tb" ) ) );
    assertTrue( values.contains( new Tuple( "3\tc\t3\tc\t3\tc" ) ) );
    assertTrue( values.contains( new Tuple( "4\td\t4\td\t4\td" ) ) );
    assertTrue( values.contains( new Tuple( "5\te\t5\te\t5\te" ) ) );
    }

  /**
   * This test checks for a deadlock when the same input is forked, adapted on one edge, then hashjoined back together.
   *
   * @throws Exception
   */
  @Test
  public void testForkThenJoin() throws Exception
    {
    getPlatform().copyFromLocal( inputFileLower );
    Tap sourceLower = getPlatform().getTextFile( new Fields( "offset", "line" ), inputFileLower );

    Map sources = new HashMap();

    sources.put( "lower", sourceLower );

    Tap sink = getPlatform().getTextFile( new Fields( "line" ), getOutputPath( "join" ), SinkMode.REPLACE );

    Function splitter = new RegexSplitter( new Fields( "num", "text" ), " " );

    Pipe pipeLower = new Each( new Pipe( "lower" ), new Fields( "line" ), splitter );
    Pipe pipeUpper = new Each( new Pipe( "upper", pipeLower ), new Fields( "text" ),
      new ExpressionFunction( Fields.ARGS, "text.toUpperCase(java.util.Locale.ROOT)", String.class ),
      Fields.REPLACE );

    Pipe splice = new HashJoin( pipeLower, new Fields( "num" ), pipeUpper, new Fields( "num" ), Fields.size( 4 ) );

    Map<Object, Object> properties = getProperties();

    Flow flow = getPlatform().getFlowConnector( properties ).connect( sources, sink, splice );

    flow.complete();

    validateLength( flow, 5 );

    List<Tuple> values = getSinkAsList( flow );

    assertTrue( values.contains( new Tuple( "1\ta\t1\tA" ) ) );
    assertTrue( values.contains( new Tuple( "2\tb\t2\tB" ) ) );
    }

  /**
   * This test checks for a deadlock when the same input is forked, adapted on one edge, then hashjoined back together.
   *
   * @throws Exception
   */
  @Test
  public void testForkCoGroupThenHashJoin() throws Exception
    {
    getPlatform().copyFromLocal( inputFileLower );
    Tap sourceLower = getPlatform().getTextFile( new Fields( "offset", "line" ), inputFileLower );
    Tap sourceUpper = getPlatform().getTextFile( new Fields( "offset", "line" ), inputFileUpper );

    Map sources = new HashMap();

    sources.put( "sourceLower", sourceLower );
    sources.put( "sourceUpper", sourceUpper );

    Tap sink = getPlatform().getTextFile( new Fields( "line" ), getOutputPath( "join" ), SinkMode.REPLACE );

    Function splitter = new RegexSplitter( new Fields( "num", "text" ), " " );

    Pipe leftPipeLower = new Each( new Pipe( "sourceLower" ), new Fields( "line" ), splitter );
    Pipe rightPipeUpper = new Each( new Pipe( "sourceUpper" ), new Fields( "line" ), splitter );

    Pipe leftPipeUpper = new Each( new Pipe( "leftUpper", leftPipeLower ), new Fields( "text" ),
      new ExpressionFunction( Fields.ARGS, "text.toUpperCase(java.util.Locale.ROOT)", String.class ),
      Fields.REPLACE );
    Pipe rightPipeLower = new Each( new Pipe( "rightLower", rightPipeUpper ), new Fields( "text" ),
      new ExpressionFunction( Fields.ARGS, "text.toLowerCase(java.util.Locale.ROOT)", String.class ),
      Fields.REPLACE );

    leftPipeUpper = new GroupBy( leftPipeUpper, new Fields( "num" ) );
    rightPipeLower = new GroupBy( rightPipeLower, new Fields( "num" ) );

    Pipe middleSplice = new CoGroup( "middleCoGroup", leftPipeUpper, new Fields( "num" ), rightPipeLower, new Fields( "num" ), new Fields( "numM1", "charM1", "numM2", "charM2" ) );

    Pipe leftSplice = new HashJoin( leftPipeLower, new Fields( "num" ), middleSplice, new Fields( "numM1" ) );

    Map<Object, Object> properties = getProperties();

    Flow flow = getPlatform().getFlowConnector( properties ).connect( sources, sink, leftSplice );

    flow.complete();

    validateLength( flow, 5 );

    List<Tuple> values = getSinkAsList( flow );
    // that the flow completes at all is already success.
    assertTrue( values.contains( new Tuple( "1\ta\t1\tA\t1\ta" ) ) );
    assertTrue( values.contains( new Tuple( "2\tb\t2\tB\t2\tb" ) ) );
    }

  /**
   * This test checks for a deadlock when the same input is forked, adapted on one edge, cogroup with something,
   * then hashjoined back together.
   *
   * @throws Exception
   */
  @Test
  public void testForkCoGroupThenHashJoinCoGroupAgain() throws Exception
    {
    getPlatform().copyFromLocal( inputFileLower );
    getPlatform().copyFromLocal( inputFileUpper );

    Tap sourceLower = getPlatform().getTextFile( new Fields( "offset", "line" ), inputFileLower );
    Tap sourceUpper = getPlatform().getTextFile( new Fields( "offset", "line" ), inputFileUpper );

    Map sources = new HashMap();

    sources.put( "sourceLower", sourceLower );
    sources.put( "sourceUpper", sourceUpper );

    Tap sink = getPlatform().getTextFile( new Fields( "line" ), getOutputPath( "join" ), SinkMode.REPLACE );

    Function splitter = new RegexSplitter( new Fields( "num", "text" ), " " );

    Pipe leftPipeLower = new Each( new Pipe( "sourceLower" ), new Fields( "line" ), splitter );
    Pipe rightPipeUpper = new Each( new Pipe( "sourceUpper" ), new Fields( "line" ), splitter );

    Pipe leftPipeUpper = new Each( new Pipe( "leftUpper", leftPipeLower ), new Fields( "text" ),
      new ExpressionFunction( Fields.ARGS, "text.toUpperCase(java.util.Locale.ROOT)", String.class ),
      Fields.REPLACE );
    Pipe rightPipeLower = new Each( new Pipe( "rightLower", rightPipeUpper ), new Fields( "text" ),
      new ExpressionFunction( Fields.ARGS, "text.toLowerCase(java.util.Locale.ROOT)", String.class ),
      Fields.REPLACE );

    leftPipeUpper = new GroupBy( leftPipeUpper, new Fields( "num" ) );
    rightPipeLower = new GroupBy( rightPipeLower, new Fields( "num" ) );

    Pipe middleSplice = new CoGroup( "middleCoGroup", leftPipeUpper, new Fields( "num" ), rightPipeLower, new Fields( "num" ), new Fields( "numM1", "charM1", "numM2", "charM2" ) );

    Pipe leftSplice = new HashJoin( leftPipeLower, new Fields( "num" ), middleSplice, new Fields( "numM1" ) );
    Pipe rightSplice = new HashJoin( rightPipeUpper, new Fields( "num" ), middleSplice, new Fields( "numM2" ) );

    leftSplice = new Rename( leftSplice, new Fields( "num", "text", "numM1", "charM1", "numM2", "charM2" ), new Fields( "numL1", "charL1", "numM1L", "charM1L", "numM2L", "charM2L" ) );
    rightSplice = new Rename( rightSplice, new Fields( "num", "text", "numM1", "charM1", "numM2", "charM2" ), new Fields( "numR1", "charR1", "numM1R", "charM1R", "numM2R", "charM2R" ) );

    leftSplice = new GroupBy( leftSplice, new Fields( "numM1L" ) );
    rightSplice = new GroupBy( rightSplice, new Fields( "numM2R" ) );

    Pipe splice = new CoGroup( "cogrouping", leftSplice, new Fields( "numM1L" ), rightSplice, new Fields( "numM2R" ) );

    Map<Object, Object> properties = getProperties();

    Flow flow = getPlatform().getFlowConnector( properties ).connect( sources, sink, splice );

    flow.complete();

    validateLength( flow, 5 );

    List<Tuple> values = getSinkAsList( flow );

    // getting this far is a success already (past old deadlocks)
    assertTrue( values.contains( new Tuple( "1\ta\t1\tA\t1\ta\t1\tA\t1\tA\t1\ta" ) ) );
    assertTrue( values.contains( new Tuple( "2\tb\t2\tB\t2\tb\t2\tB\t2\tB\t2\tb" ) ) );
    }
  }