Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

feat: Add support for TryCast expression in Spark 3.2 and 3.3 #416

Merged
merged 7 commits into from
May 15, 2024
Merged
Show file tree
Hide file tree
Changes from 4 commits
Commits
File filter

Filter by extension

Filter by extension

Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
95 changes: 54 additions & 41 deletions spark/src/main/scala/org/apache/comet/serde/QueryPlanSerde.scala
Original file line number Diff line number Diff line change
Expand Up @@ -604,6 +604,52 @@ object QueryPlanSerde extends Logging with ShimQueryPlanSerde with CometExprShim

def exprToProtoInternal(expr: Expression, inputs: Seq[Attribute]): Option[Expr] = {
SQLConf.get

def handleCast(
child: Expression,
inputs: Seq[Attribute],
dt: DataType,
timeZoneId: Option[String],
actualEvalModeStr: String): Option[Expr] = {

val childExpr = exprToProtoInternal(child, inputs)
if (childExpr.isDefined) {
val castSupport =
CometCast.isSupported(child.dataType, dt, timeZoneId, actualEvalModeStr)

def getIncompatMessage(reason: Option[String]): String =
"Comet does not guarantee correct results for cast " +
s"from ${child.dataType} to $dt " +
s"with timezone $timeZoneId and evalMode $actualEvalModeStr" +
reason.map(str => s" ($str)").getOrElse("")

castSupport match {
case Compatible(_) =>
castToProto(timeZoneId, dt, childExpr, actualEvalModeStr)
case Incompatible(reason) =>
if (CometConf.COMET_CAST_ALLOW_INCOMPATIBLE.get()) {
logWarning(getIncompatMessage(reason))
castToProto(timeZoneId, dt, childExpr, actualEvalModeStr)
} else {
withInfo(
expr,
s"${getIncompatMessage(reason)}. To enable all incompatible casts, set " +
s"${CometConf.COMET_CAST_ALLOW_INCOMPATIBLE.key}=true")
None
}
case Unsupported =>
withInfo(
expr,
s"Unsupported cast from ${child.dataType} to $dt " +
s"with timezone $timeZoneId and evalMode $actualEvalModeStr")
None
}
} else {
withInfo(expr, child)
None
}
}

expr match {
case a @ Alias(_, _) =>
val r = exprToProtoInternal(a.child, inputs)
Expand All @@ -617,50 +663,17 @@ object QueryPlanSerde extends Logging with ShimQueryPlanSerde with CometExprShim
val value = cast.eval()
exprToProtoInternal(Literal(value, dataType), inputs)

case UnaryExpression(child) if expr.prettyName == "trycast" =>
Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Just making sure whether it is trycast, not try_cast. Looks like Spark 3.4 is using try_cast, haven't been able to find the source for Spark 3.3

Copy link
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

yes, it is trycast.

val timeZoneId = SQLConf.get.sessionLocalTimeZone
handleCast(child, inputs, expr.dataType, Some(timeZoneId), "TRY")
Comment on lines +666 to +668
Copy link
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I like this approach 👍


case Cast(child, dt, timeZoneId, evalMode) =>
val childExpr = exprToProtoInternal(child, inputs)
if (childExpr.isDefined) {
val evalModeStr = if (evalMode.isInstanceOf[Boolean]) {
// Spark 3.2 & 3.3 has ansiEnabled boolean
if (evalMode.asInstanceOf[Boolean]) "ANSI" else "LEGACY"
} else {
// Spark 3.4+ has EvalMode enum with values LEGACY, ANSI, and TRY
evalMode.toString
}
val castSupport =
CometCast.isSupported(child.dataType, dt, timeZoneId, evalModeStr)

def getIncompatMessage(reason: Option[String]) =
"Comet does not guarantee correct results for cast " +
s"from ${child.dataType} to $dt " +
s"with timezone $timeZoneId and evalMode $evalModeStr" +
reason.map(str => s" ($str)").getOrElse("")

castSupport match {
case Compatible(_) =>
castToProto(timeZoneId, dt, childExpr, evalModeStr)
case Incompatible(reason) =>
if (CometConf.COMET_CAST_ALLOW_INCOMPATIBLE.get()) {
logWarning(getIncompatMessage(reason))
castToProto(timeZoneId, dt, childExpr, evalModeStr)
} else {
withInfo(
expr,
s"${getIncompatMessage(reason)}. To enable all incompatible casts, set " +
s"${CometConf.COMET_CAST_ALLOW_INCOMPATIBLE.key}=true")
None
}
case Unsupported =>
withInfo(
expr,
s"Unsupported cast from ${child.dataType} to $dt " +
s"with timezone $timeZoneId and evalMode $evalModeStr")
None
}
val evalModeStr = if (evalMode.isInstanceOf[Boolean]) {
if (evalMode.asInstanceOf[Boolean]) "ANSI" else "LEGACY"
vaibhawvipul marked this conversation as resolved.
Show resolved Hide resolved
} else {
withInfo(expr, child)
None
evalMode.toString
}
handleCast(child, inputs, dt, timeZoneId, evalModeStr)

case add @ Add(left, right, _) if supportedDataType(left.dataType) =>
val leftExpr = exprToProtoInternal(left, inputs)
Expand Down
3 changes: 2 additions & 1 deletion spark/src/test/scala/org/apache/comet/CometCastSuite.scala
Original file line number Diff line number Diff line change
Expand Up @@ -950,7 +950,8 @@ class CometCastSuite extends CometTestBase with AdaptiveSparkPlanHelper {

// we do not support the TryCast expression in Spark 3.2 and 3.3
// https://github.com/apache/datafusion-comet/issues/374
val testTryCast = CometSparkSessionExtensions.isSpark34Plus
// val testTryCast = CometSparkSessionExtensions.isSpark34Plus
val testTryCast = true
vaibhawvipul marked this conversation as resolved.
Show resolved Hide resolved

withTempPath { dir =>
val data = roundtripParquet(input, dir).coalesce(1)
Expand Down
Loading