[SPARK-1495][SQL]add support for left semi join

sql/catalyst/src/main/scala/org/apache/spark/sql/catalyst/SqlParser.scala

@@ -128,6 +128,7 @@ class SqlParser extends StandardTokenParsers with PackratParsers {
   protected val OUTER = Keyword("OUTER")
   protected val RIGHT = Keyword("RIGHT")
   protected val SELECT = Keyword("SELECT")
+  protected val SEMI = Keyword("SEMI")
   protected val STRING = Keyword("STRING")
   protected val SUM = Keyword("SUM")
   protected val TRUE = Keyword("TRUE")
@@ -238,6 +239,7 @@ class SqlParser extends StandardTokenParsers with PackratParsers {
 
    protected lazy val joinType: Parser[JoinType] =
      INNER ^^^ Inner |
+     LEFT ~ SEMI ^^^ LeftSemi |
      LEFT ~ opt(OUTER) ^^^ LeftOuter |
      RIGHT ~ opt(OUTER) ^^^ RightOuter |
      FULL ~ opt(OUTER) ^^^ FullOuter

sql/catalyst/src/main/scala/org/apache/spark/sql/catalyst/planning/patterns.scala

@@ -119,6 +119,11 @@ object HashFilteredJoin extends Logging with PredicateHelper {
     case FilteredOperation(predicates, join @ Join(left, right, Inner, condition)) =>
       logger.debug(s"Considering hash inner join on: ${predicates ++ condition}")
       splitPredicates(predicates ++ condition, join)
+    // All predicates can be evaluated for left semi join (those that are in the WHERE
+    // clause can only from left table, so they can all be pushed down.)
+    case FilteredOperation(predicates, join @ Join(left, right, LeftSemi, condition)) =>
+      logger.debug(s"Considering hash left semi join on: ${predicates ++ condition}")
+      splitPredicates(predicates ++ condition, join)
     case join @ Join(left, right, joinType, condition) =>
       logger.debug(s"Considering hash join on: $condition")
       splitPredicates(condition.toSeq, join)

sql/catalyst/src/main/scala/org/apache/spark/sql/catalyst/plans/joinTypes.scala

@@ -22,3 +22,4 @@ case object Inner extends JoinType
 case object LeftOuter extends JoinType
 case object RightOuter extends JoinType
 case object FullOuter extends JoinType
+case object LeftSemi extends JoinType

sql/catalyst/src/main/scala/org/apache/spark/sql/catalyst/plans/logical/basicOperators.scala

@@ -18,7 +18,7 @@
 package org.apache.spark.sql.catalyst.plans.logical
 
 import org.apache.spark.sql.catalyst.expressions._
-import org.apache.spark.sql.catalyst.plans.JoinType
+import org.apache.spark.sql.catalyst.plans.{LeftSemi, JoinType}
 import org.apache.spark.sql.catalyst.types._
 
 case class Project(projectList: Seq[NamedExpression], child: LogicalPlan) extends UnaryNode {
@@ -81,7 +81,12 @@ case class Join(
   condition: Option[Expression]) extends BinaryNode {
 
   def references = condition.map(_.references).getOrElse(Set.empty)
-  def output = left.output ++ right.output
+  def output = joinType match {
+    case LeftSemi =>
+      left.output
+    case _ =>
+      left.output ++ right.output
+  }
 }
 
 case class InsertIntoTable(

sql/core/src/main/scala/org/apache/spark/sql/SQLContext.scala

@@ -193,6 +193,7 @@ class SQLContext(@transient val sparkContext: SparkContext)
     val strategies: Seq[Strategy] =
       TakeOrdered ::
       PartialAggregation ::
+      LeftSemiJoin ::
       HashJoin ::
       ParquetOperations ::
       BasicOperators ::

sql/core/src/main/scala/org/apache/spark/sql/execution/SparkStrategies.scala

@@ -28,6 +28,22 @@ import org.apache.spark.sql.parquet._
 private[sql] abstract class SparkStrategies extends QueryPlanner[SparkPlan] {
   self: SQLContext#SparkPlanner =>
 
+  object LeftSemiJoin extends Strategy with PredicateHelper {
+    def apply(plan: LogicalPlan): Seq[SparkPlan] = plan match {
+      // Find left semi joins where at least some predicates can be evaluated by matching hash
+      // keys using the HashFilteredJoin pattern.
+      case HashFilteredJoin(LeftSemi, leftKeys, rightKeys, condition, left, right) =>
+        val semiJoin = execution.LeftSemiJoinHash(
+          leftKeys, rightKeys, planLater(left), planLater(right))
+        condition.map(Filter(_, semiJoin)).getOrElse(semiJoin) :: Nil
+      // no predicate can be evaluated by matching hash keys
+      case logical.Join(left, right, LeftSemi, condition) =>
+        execution.LeftSemiJoinBNL(
+          planLater(left), planLater(right), condition)(sparkContext) :: Nil
+      case _ => Nil
+    }
+  }
+
   object HashJoin extends Strategy with PredicateHelper {
     def apply(plan: LogicalPlan): Seq[SparkPlan] = plan match {
       // Find inner joins where at least some predicates can be evaluated by matching hash keys

sql/core/src/main/scala/org/apache/spark/sql/execution/joins.scala

@@ -140,6 +140,137 @@ case class HashJoin(
   }
 }
 
+/**
+ * :: DeveloperApi ::
+ * Build the right table's join keys into a HashSet, and iteratively go through the left
+ * table, to find the if join keys are in the Hash set.
+ */
+@DeveloperApi
+case class LeftSemiJoinHash(
+    leftKeys: Seq[Expression],
+    rightKeys: Seq[Expression],
+    left: SparkPlan,
+    right: SparkPlan) extends BinaryNode {
+
+  override def outputPartitioning: Partitioning = left.outputPartitioning
+
+  override def requiredChildDistribution =
+    ClusteredDistribution(leftKeys) :: ClusteredDistribution(rightKeys) :: Nil
+
+  val (buildPlan, streamedPlan) = (right, left)
+  val (buildKeys, streamedKeys) = (rightKeys, leftKeys)
+
+  def output = left.output
+
+  @transient lazy val buildSideKeyGenerator = new Projection(buildKeys, buildPlan.output)
+  @transient lazy val streamSideKeyGenerator =
+    () => new MutableProjection(streamedKeys, streamedPlan.output)
+
+  def execute() = {
+
+    buildPlan.execute().zipPartitions(streamedPlan.execute()) { (buildIter, streamIter) =>
+      val hashTable = new java.util.HashSet[Row]()
+      var currentRow: Row = null
+
+      // Create a Hash set of buildKeys
+      while (buildIter.hasNext) {
+        currentRow = buildIter.next()
+        val rowKey = buildSideKeyGenerator(currentRow)
+        if(!rowKey.anyNull) {
+          val keyExists = hashTable.contains(rowKey)
+          if (!keyExists) {
+            hashTable.add(rowKey)
+          }
+        }
+      }
+
+      new Iterator[Row] {
+        private[this] var currentStreamedRow: Row = _
+        private[this] var currentHashMatched: Boolean = false
+
+        private[this] val joinKeys = streamSideKeyGenerator()
+
+        override final def hasNext: Boolean =
+          streamIter.hasNext && fetchNext()
+
+        override final def next() = {
+          currentStreamedRow
+        }
+
+        /**
+         * Searches the streamed iterator for the next row that has at least one match in hashtable.
+         *
+         * @return true if the search is successful, and false the streamed iterator runs out of
+         *         tuples.
+         */
+        private final def fetchNext(): Boolean = {
+          currentHashMatched = false
+          while (!currentHashMatched && streamIter.hasNext) {
+            currentStreamedRow = streamIter.next()
+            if (!joinKeys(currentStreamedRow).anyNull) {
+              currentHashMatched = hashTable.contains(joinKeys.currentValue)
+            }
+          }
+          currentHashMatched
+        }
+      }
+    }
+  }
+}
+
+/**
+ * :: DeveloperApi ::
+ * Using BroadcastNestedLoopJoin to calculate left semi join result when there's no join keys
+ * for hash join.
+ */
+@DeveloperApi
+case class LeftSemiJoinBNL(
+    streamed: SparkPlan, broadcast: SparkPlan, condition: Option[Expression])
+    (@transient sc: SparkContext)
+  extends BinaryNode {
+  // TODO: Override requiredChildDistribution.
+
+  override def outputPartitioning: Partitioning = streamed.outputPartitioning
+
+  override def otherCopyArgs = sc :: Nil
+
+  def output = left.output
+
+  /** The Streamed Relation */
+  def left = streamed
+  /** The Broadcast relation */
+  def right = broadcast
+
+  @transient lazy val boundCondition =
+    InterpretedPredicate(
+      condition
+        .map(c => BindReferences.bindReference(c, left.output ++ right.output))
+        .getOrElse(Literal(true)))
+
+
+  def execute() = {
+    val broadcastedRelation = sc.broadcast(broadcast.execute().map(_.copy()).collect().toIndexedSeq)
+
+    streamed.execute().mapPartitions { streamedIter =>
+      val joinedRow = new JoinedRow
+
+      streamedIter.filter(streamedRow => {
+        var i = 0
+        var matched = false
+
+        while (i < broadcastedRelation.value.size && !matched) {
+          val broadcastedRow = broadcastedRelation.value(i)
+          if (boundCondition(joinedRow(streamedRow, broadcastedRow))) {
+            matched = true
+          }
+          i += 1
+        }
+        matched
+      })
+    }
+  }
+}
+
 /**
  * :: DeveloperApi ::
  */

sql/core/src/test/scala/org/apache/spark/sql/QueryTest.scala

@@ -44,7 +44,7 @@ class QueryTest extends FunSuite {
         fail(
           s"""
             |Exception thrown while executing query:
-            |${rdd.logicalPlan}
+            |${rdd.queryExecution}
             |== Exception ==
             |$e
           """.stripMargin)

sql/core/src/test/scala/org/apache/spark/sql/SQLQuerySuite.scala

@@ -34,6 +34,13 @@ class SQLQuerySuite extends QueryTest {
       arrayData.map(d => (d.data, d.data(0), d.data(0) + d.data(1), d.data(1))).collect().toSeq)
   }
 
+  test("left semi greater than predicate") {
+    checkAnswer(
+      sql("SELECT * FROM testData2 x LEFT SEMI JOIN testData2 y ON x.a >= y.a + 2"),
+      Seq((3,1), (3,2))
+    )
+  }
+
   test("index into array of arrays") {
     checkAnswer(
       sql(

sql/hive/src/main/scala/org/apache/spark/sql/hive/HiveContext.scala

@@ -224,6 +224,7 @@ class HiveContext(sc: SparkContext) extends SQLContext(sc) {
       DataSinks,
       Scripts,
       PartialAggregation,
+      LeftSemiJoin,
       HashJoin,
       BasicOperators,
       CartesianProduct,

sql/hive/src/main/scala/org/apache/spark/sql/hive/HiveQl.scala

@@ -680,6 +680,7 @@ private[hive] object HiveQl {
         case "TOK_RIGHTOUTERJOIN" => RightOuter
         case "TOK_LEFTOUTERJOIN" => LeftOuter
         case "TOK_FULLOUTERJOIN" => FullOuter
+        case "TOK_LEFTSEMIJOIN" => LeftSemi
       }
       assert(other.size <= 1, "Unhandled join clauses.")
       Join(nodeToRelation(relation1),

sql/hive/src/test/resources/golden/leftsemijoin-10-89737a8857b5b61cc909e0c797f86aea

@@ -0,0 +1,4 @@
+Hank	2
+Hank	2
+Joe	2
+Joe	2

sql/hive/src/test/resources/golden/leftsemijoin-2-43d53504df013e6b35f81811138a167a

@@ -0,0 +1 @@
+0

sql/hive/src/test/resources/golden/leftsemijoin-8-73cad58a10a1483ccb15e94a857013

@@ -0,0 +1,4 @@
+Hank	2
+Hank	2
+Joe	2
+Joe	2

sql/hive/src/test/resources/golden/leftsemijoin-9-c5efa6b8771a51610d655be461670e1e

@@ -0,0 +1,2 @@
+2	Tie
+2	Tie

sql/hive/src/test/resources/golden/leftsemijoin_mr-10-9914f44ecb6ae7587b62e5349ff60d04

@@ -0,0 +1 @@
+0

sql/hive/src/test/resources/golden/leftsemijoin_mr-11-2027ecb1495d5550c5d56abf6b95b0a7

@@ -0,0 +1,2 @@
+1
+1

sql/hive/src/test/resources/golden/leftsemijoin_mr-4-333895fe6abca27c8edb5c91bfe10d2f

@@ -0,0 +1,2 @@
+1
+1

sql/hive/src/test/resources/golden/leftsemijoin_mr-5-896d0948c1df849df9764a6d8ad8fff9

@@ -0,0 +1,20 @@
+1
+1
+1
+1
+1
+1
+1
+1
+1
+1
+1
+1
+1
+1
+1
+1
+1
+1
+1
+1

sql/hive/src/test/resources/golden/leftsemijoin_mr-6-b1e2ade89ae898650f0be4f796d8947b

@@ -0,0 +1 @@
+0

sql/hive/src/test/resources/golden/leftsemijoin_mr-7-8e9c2969b999557363e40f9ebb3f6d7c

@@ -0,0 +1 @@
+0

sql/hive/src/test/resources/golden/leftsemijoin_mr-8-c61b972d4409babe41d8963e841af45b

@@ -0,0 +1 @@
+0

sql/hive/src/test/resources/golden/leftsemijoin_mr-9-2027ecb1495d5550c5d56abf6b95b0a7

@@ -0,0 +1,2 @@
+1
+1

sql/hive/src/test/scala/org/apache/spark/sql/hive/execution/HiveCompatibilitySuite.scala

@@ -478,6 +478,8 @@ class HiveCompatibilitySuite extends HiveQueryFileTest with BeforeAndAfter {
     "lateral_view_cp",
     "lateral_view_outer",
     "lateral_view_ppd",
+    "leftsemijoin",
+    "leftsemijoin_mr",
     "lineage1",
     "literal_double",
     "literal_ints",