fqaiser94 commented on a change in pull request #27066:
URL: https://github.com/apache/spark/pull/27066#discussion_r448598639
##########
File path:
sql/catalyst/src/main/scala/org/apache/spark/sql/catalyst/expressions/complexTypeCreator.scala
##########
@@ -539,3 +539,79 @@ case class StringToMap(text: Expression, pairDelim:
Expression, keyValueDelim: E
override def prettyName: String = "str_to_map"
}
+
+/**
+ * Adds/replaces field in struct by name.
+ */
+case class WithFields(
+ structExpr: Expression,
+ names: Seq[String],
+ valExprs: Seq[Expression]) extends Expression {
+
+ assert(names.length == valExprs.length)
+
+ override def checkInputDataTypes(): TypeCheckResult = {
+ if (!structExpr.dataType.isInstanceOf[StructType]) {
+ TypeCheckResult.TypeCheckFailure(
+ "struct argument should be struct type, got: " +
structExpr.dataType.catalogString)
+ } else {
+ TypeCheckResult.TypeCheckSuccess
+ }
+ }
+
+ override def children: Seq[Expression] = structExpr +: valExprs
+
+ private lazy val addOrReplaceExprs = names.zip(valExprs)
+
+ override def dataType: StructType = {
Review comment:
Just to be clear, even after reverting the codegen changes, this is
still not possible.
The main issue is with nullable structs.
When we call `WithFields` on a nullable struct, we want the **existing**
fields inside the struct to maintain the same nullability as before. Because we
use `GetStructField` to extract the existing fields we won't get this behaviour
as `GetStructField` returns a nullable dataType if the struct that it's being
called upon is nullable.
The `withField should add field to null struct` test is a good example of a
test that will fail with your suggested change:
```
StructType(StructField(a,StructType(StructField(a,IntegerType,true),
StructField(b,IntegerType,true), StructField(c,IntegerType,true),
StructField(d,IntegerType,false)),true))
did not equal
StructType(StructField(a,StructType(StructField(a,IntegerType,false),
StructField(b,IntegerType,true), StructField(c,IntegerType,false),
StructField(d,IntegerType,false)),true))
```
----------------------------------------------------------------
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.
For queries about this service, please contact Infrastructure at:
[email protected]
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
