milastdbx commented on code in PR #48621:
URL: https://github.com/apache/spark/pull/48621#discussion_r1816922884
##########
sql/core/src/main/scala/org/apache/spark/sql/jdbc/MsSqlServerDialect.scala:
##########
@@ -88,11 +88,31 @@ private case class MsSqlServerDialect() extends JdbcDialect
with NoLegacyJDBCErr
case e: Predicate => e.name() match {
case "=" | "<>" | "<=>" | "<" | "<=" | ">" | ">=" =>
val Array(l, r) = e.children().map {
- case p: Predicate => s"CASE WHEN ${inputToSQL(p)} THEN 1 ELSE 0
END"
+ case p: Predicate => inputToCaseWhenSQL(p)
case o => inputToSQL(o)
}
visitBinaryComparison(e.name(), l, r)
- case "CASE_WHEN" =>
visitCaseWhen(expressionsToStringArray(e.children())) + " = 1"
+ case "CASE_WHEN" =>
+ // Since MsSqlServer cannot handle boolean expressions inside
+ // a CASE WHEN, it is necessary to convert those to another
+ // CASE WHEN expression that will return 1 or 0 depending on
+ // the result. Exceptions are TRUE and FALSE, which already
+ // get translated to 1 and 0.
+ // Example:
+ // In: ... CASE WHEN a = b THEN c = d ... END
+ // Out: ... CASE WHEN a = b THEN CASE WHEN c = d THEN 1 ELSE 0 END
... END = 1
+
+ // grouped turns Array[Expression] to Array[Array[Expression]]
+ // with a len of max 2 (final one will have only one)
+ val stringArray = e.children().grouped(2).flatMap { arr =>
+ arr.dropRight(1).map(inputToSQL) :+
Review Comment:
this looks wierd.
Lets do something like
```
.flatMap{
case List(left, right) => // put more meaningufl name than left, right
....
case List(elseBranch) =>
....
}
##########
sql/core/src/main/scala/org/apache/spark/sql/catalyst/util/V2ExpressionBuilder.scala:
##########
@@ -221,8 +221,18 @@ class V2ExpressionBuilder(e: Expression, isPredicate:
Boolean = false) extends L
case _: BitwiseNot => generateExpressionWithName("~", expr, isPredicate)
case caseWhen @ CaseWhen(branches, elseValue) =>
val conditions = branches.map(_._1).flatMap(generateExpression(_, true))
- val values = branches.map(_._2).flatMap(generateExpression(_))
- val elseExprOpt = elseValue.flatMap(generateExpression(_))
+ val values = branches.map(_._2).flatMap(child =>
+ generateExpression(
+ child,
+ isPredicate && child.dataType.isInstanceOf[BooleanType]
Review Comment:
I agree with @cloud-fan, if there is a design problem of this class (not
being safe against data type check) lets fix it at class level and not per
expression level
##########
sql/catalyst/src/main/java/org/apache/spark/sql/connector/util/V2ExpressionSQLBuilder.java:
##########
@@ -219,6 +219,10 @@ protected String inputToSQL(Expression input) {
}
}
+ protected String inputToCaseWhenSQL(Expression input) {
Review Comment:
Lets add it in JdbcDialect
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: reviews-unsubscr...@spark.apache.org
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org
---------------------------------------------------------------------
To unsubscribe, e-mail: reviews-unsubscr...@spark.apache.org
For additional commands, e-mail: reviews-h...@spark.apache.org
