[SPARK-46064][SQL][SS] Move out EliminateEventTimeWatermark to the analyzer and change to only take effect on resolved child

### What changes were proposed in this pull request?

This PR proposes to move out EliminateEventTimeWatermark to the analyzer (one of the analysis rule), and also make a change to eliminate EventTimeWatermark node only when the child of EventTimeWatermark is "resolved".

### Why are the changes needed?

Currently, we apply EliminateEventTimeWatermark immediately when withWatermark is called, which means the rule is applied immediately against the child, regardless whether child is resolved or not.

It is not an issue for the usage of DataFrame API initiated by read / readStream, because streaming sources have the flag isStreaming set to true even it is yet resolved. But mix-up of SQL and DataFrame API would expose the issue; we may not know the exact value of isStreaming flag on unresolved node and it is subject to change upon resolution.

### Does this PR introduce _any_ user-facing change?

No.

### How was this patch tested?

New UTs.

### Was this patch authored or co-authored using generative AI tooling?

No.

Closes #43971 from HeartSaVioR/SPARK-46064.

Authored-by: Jungtaek Lim <[email protected]>
Signed-off-by: Jungtaek Lim <[email protected]>

Author: HeartSaVioR

sql/catalyst/src/main/scala/org/apache/spark/sql/catalyst/analysis/Analyzer.scala

@@ -347,7 +347,9 @@ class Analyzer(override val catalogManager: CatalogManager) extends RuleExecutor
     Batch("Cleanup", fixedPoint,
       CleanupAliases),
     Batch("HandleSpecialCommand", Once,
-      HandleSpecialCommand)
+      HandleSpecialCommand),
+    Batch("Remove watermark for batch query", Once,
+      EliminateEventTimeWatermark)
   )
 
   /**
@@ -3938,7 +3940,7 @@ object CleanupAliases extends Rule[LogicalPlan] with AliasHelper {
 object EliminateEventTimeWatermark extends Rule[LogicalPlan] {
   override def apply(plan: LogicalPlan): LogicalPlan = plan.resolveOperatorsWithPruning(
     _.containsPattern(EVENT_TIME_WATERMARK)) {
-    case EventTimeWatermark(_, _, child) if !child.isStreaming => child
+    case EventTimeWatermark(_, _, child) if child.resolved && !child.isStreaming => child
   }
 }
 

sql/catalyst/src/test/scala/org/apache/spark/sql/catalyst/analysis/AnalysisSuite.scala

@@ -1724,4 +1724,27 @@ class AnalysisSuite extends AnalysisTest with Matchers {
       checkAnalysis(ident2.select($"a"), testRelation.select($"a").analyze)
     }
   }
+
+  test("SPARK-46064 Basic functionality of elimination for watermark node in batch query") {
+    val dfWithEventTimeWatermark = EventTimeWatermark($"ts",
+      IntervalUtils.fromIntervalString("10 seconds"), batchRelationWithTs)
+
+    val analyzed = getAnalyzer.executeAndCheck(dfWithEventTimeWatermark, new QueryPlanningTracker)
+
+    // EventTimeWatermark node is eliminated via EliminateEventTimeWatermark.
+    assert(!analyzed.exists(_.isInstanceOf[EventTimeWatermark]))
+  }
+
+  test("SPARK-46064 EliminateEventTimeWatermark properly handles the case where the child of " +
+    "EventTimeWatermark changes the isStreaming flag during resolution") {
+    // UnresolvedRelation which is batch initially and will be resolved as streaming
+    val dfWithTempView = UnresolvedRelation(TableIdentifier("streamingTable"))
+    val dfWithEventTimeWatermark = EventTimeWatermark($"ts",
+      IntervalUtils.fromIntervalString("10 seconds"), dfWithTempView)
+
+    val analyzed = getAnalyzer.executeAndCheck(dfWithEventTimeWatermark, new QueryPlanningTracker)
+
+    // EventTimeWatermark node is NOT eliminated.
+    assert(analyzed.exists(_.isInstanceOf[EventTimeWatermark]))
+  }
 }

sql/catalyst/src/test/scala/org/apache/spark/sql/catalyst/analysis/AnalysisTest.scala

@@ -82,6 +82,8 @@ trait AnalysisTest extends PlanTest {
     createTempView(catalog, "TaBlE3", TestRelations.testRelation3, overrideIfExists = true)
     createGlobalTempView(catalog, "TaBlE4", TestRelations.testRelation4, overrideIfExists = true)
     createGlobalTempView(catalog, "TaBlE5", TestRelations.testRelation5, overrideIfExists = true)
+    createTempView(catalog, "streamingTable", TestRelations.streamingRelation,
+      overrideIfExists = true)
     new Analyzer(catalog) {
       override val extendedResolutionRules = extendedAnalysisRules
     }

sql/catalyst/src/test/scala/org/apache/spark/sql/catalyst/analysis/TestRelations.scala

@@ -68,4 +68,18 @@ object TestRelations {
 
   val mapRelation = LocalRelation(
     AttributeReference("map", MapType(IntegerType, IntegerType))())
+
+  val streamingRelation = LocalRelation(
+    Seq(
+      AttributeReference("a", IntegerType)(),
+      AttributeReference("ts", TimestampType)()
+    ),
+    isStreaming = true)
+
+  val batchRelationWithTs = LocalRelation(
+    Seq(
+      AttributeReference("a", IntegerType)(),
+      AttributeReference("ts", TimestampType)()
+    ),
+    isStreaming = false)
 }

sql/catalyst/src/test/scala/org/apache/spark/sql/catalyst/optimizer/FilterPushdownSuite.scala

@@ -1190,7 +1190,7 @@ class FilterPushdownSuite extends PlanTest {
 
   test("watermark pushdown: no pushdown on watermark attribute #1") {
     val interval = new CalendarInterval(2, 2, 2000L)
-    val relation = LocalRelation(attrA, $"b".timestamp, attrC)
+    val relation = LocalRelation(Seq(attrA, $"b".timestamp, attrC), Nil, isStreaming = true)
 
     // Verify that all conditions except the watermark touching condition are pushed down
     // by the optimizer and others are not.
@@ -1205,7 +1205,7 @@ class FilterPushdownSuite extends PlanTest {
 
   test("watermark pushdown: no pushdown for nondeterministic filter") {
     val interval = new CalendarInterval(2, 2, 2000L)
-    val relation = LocalRelation(attrA, attrB, $"c".timestamp)
+    val relation = LocalRelation(Seq(attrA, attrB, $"c".timestamp), Nil, isStreaming = true)
 
     // Verify that all conditions except the watermark touching condition are pushed down
     // by the optimizer and others are not.
@@ -1221,7 +1221,7 @@ class FilterPushdownSuite extends PlanTest {
 
   test("watermark pushdown: full pushdown") {
     val interval = new CalendarInterval(2, 2, 2000L)
-    val relation = LocalRelation(attrA, attrB, $"c".timestamp)
+    val relation = LocalRelation(Seq(attrA, attrB, $"c".timestamp), Nil, isStreaming = true)
 
     // Verify that all conditions except the watermark touching condition are pushed down
     // by the optimizer and others are not.
@@ -1236,7 +1236,7 @@ class FilterPushdownSuite extends PlanTest {
 
   test("watermark pushdown: no pushdown on watermark attribute #2") {
     val interval = new CalendarInterval(2, 2, 2000L)
-    val relation = LocalRelation($"a".timestamp, attrB, attrC)
+    val relation = LocalRelation(Seq($"a".timestamp, attrB, attrC), Nil, isStreaming = true)
 
     val originalQuery = EventTimeWatermark($"a", interval, relation)
       .where($"a" === new java.sql.Timestamp(0) && $"b" === 10)

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

@@ -778,8 +778,7 @@ class Dataset[T] private[sql](
       val parsedDelay = IntervalUtils.fromIntervalString(delayThreshold)
       require(!IntervalUtils.isNegative(parsedDelay),
         s"delay threshold ($delayThreshold) should not be negative.")
-      EliminateEventTimeWatermark(
-        EventTimeWatermark(UnresolvedAttribute(eventTime), parsedDelay, logicalPlan))
+      EventTimeWatermark(UnresolvedAttribute(eventTime), parsedDelay, logicalPlan)
     }
   }