cloud-fan commented on a change in pull request #35352:
URL: https://github.com/apache/spark/pull/35352#discussion_r832166451
##########
File path:
sql/catalyst/src/main/scala/org/apache/spark/sql/catalyst/expressions/stringExpressions.scala
##########
@@ -2943,3 +2943,87 @@ case class Sentences(
copy(str = newFirst, language = newSecond, country = newThird)
}
+
+/**
+ * Splits a given string by a specified delimiter.
+ */
+case class SplitByDelimiter(
+ str: Expression,
+ delimiter: Expression)
+ extends BinaryExpression with NullIntolerant {
+ override def dataType: DataType = ArrayType(StringType, containsNull = false)
+ override def left: Expression = str
+ override def right: Expression = delimiter
+
+ override def nullSafeEval(string: Any, delimiter: Any): Any = {
+ val strings = {
+ // if delimiter is empty string, skip the regex based splitting directly
as regex
+ // treats empty string as matching anything, thus use the input directly.
+ if (delimiter.asInstanceOf[UTF8String].numBytes() == 0) {
+ Array(string)
+ } else {
+ string.asInstanceOf[UTF8String].splitSQL(
+ delimiter.asInstanceOf[UTF8String], -1)
+ }
+ }
+ new GenericArrayData(strings.asInstanceOf[Array[Any]])
+ }
+
+ override def doGenCode(ctx: CodegenContext, ev: ExprCode): ExprCode = {
+ val arrayClass = classOf[GenericArrayData].getName
+ nullSafeCodeGen(ctx, ev, (str, delimiter) => {
+ if (delimiter.asInstanceOf[UTF8String].numBytes() == 0) {
Review comment:
Actually, can we put this empty delimiter handling in
`UTF8String.splitSQL`? then we can make this expression extends
`RuntimeReplaceable` with `Invoke`, and we don't need to implement the
complicated codegen part.
--
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: [email protected]
For queries about this service, please contact Infrastructure at:
[email protected]
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
