[SPARK-52583][SQL] Add an Developer API for stringifying values in UserDefinedType

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

This PR proposes to add a Developer API for stringifying values in UserDefinedType. When casting a Class X instance that UserDefinedType represents in the Catalyst layer, this API will be called to get the string representation. The default implementation of this API is to call `obj.toString` with the AS-IS behavior.

### Why are the changes needed?

The Class X's toString method doesn't always meet the needs of users, and it's difficult for them to override it in some circumstances. For example, it's a JVM implementation or from other dependent projects.

This stringifyValue API gives a user or developer an extra opportunity to do the override.

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

No, it's a developer API addition and the behavior remains

### How was this patch tested?
New test cases

### Was this patch authored or co-authored using generative AI tooling?
no

Closes #51289 from yaooqinn/SPARK-52583.

Authored-by: Kent Yao <[email protected]>
Signed-off-by: Dongjoon Hyun <[email protected]>

Author: yaooqinn

sql/api/src/main/scala/org/apache/spark/sql/types/UserDefinedType.scala

@@ -94,6 +94,22 @@ abstract class UserDefinedType[UserType >: Null] extends DataType with Serializa
   }
 
   override def catalogString: String = sqlType.simpleString
+
+  /**
+   * This method is used to convert the value of a UDT to a string representation.
+   *
+   * By default, it simply calls `toString` on the object.
+   *
+   * @param obj
+   *   The object to convert to a string.
+   * @return
+   *   A string representation of the object.
+   * @since 4.1.0
+   */
+  @Since("4.1.0")
+  def stringifyValue(obj: Any): String = {
+    obj.toString
+  }
 }
 
 private[spark] object UserDefinedType {

sql/catalyst/src/main/scala/org/apache/spark/sql/catalyst/expressions/ToStringBase.scala

@@ -165,7 +165,7 @@ trait ToStringBase { self: UnaryExpression with TimeZoneAwareExpression =>
       })
     case pudt: PythonUserDefinedType => castToString(pudt.sqlType)
     case udt: UserDefinedType[_] =>
-      o => UTF8String.fromString(udt.deserialize(o).toString)
+      o => UTF8String.fromString(udt.stringifyValue(udt.deserialize(o)))
     case YearMonthIntervalType(startField, endField) =>
       acceptAny[Int](i => UTF8String.fromString(
         IntervalUtils.toYearMonthIntervalString(i, ANSI_STYLE, startField, endField)))
@@ -274,7 +274,7 @@ trait ToStringBase { self: UnaryExpression with TimeZoneAwareExpression =>
       case udt: UserDefinedType[_] =>
         val udtRef = JavaCode.global(ctx.addReferenceObj("udt", udt), udt.sqlType)
         (c, evPrim) =>
-          code"$evPrim = UTF8String.fromString($udtRef.deserialize($c).toString());"
+          code"$evPrim = UTF8String.fromString($udtRef.stringifyValue($udtRef.deserialize($c)));"
       case i: YearMonthIntervalType =>
         val iu = IntervalUtils.getClass.getName.stripSuffix("$")
         val iss = IntervalStringStyles.getClass.getName.stripSuffix("$")

sql/core/src/test/scala/org/apache/spark/sql/UserDefinedTypeSuite.scala

@@ -245,6 +245,19 @@ class UserDefinedTypeSuite extends QueryTest with SharedSparkSession with Parque
     checkEvaluation(ret, "(1.0, 3.0, 5.0, 7.0, 9.0)")
   }
 
+  test("SPARK-52583: Cast UserDefinedType to string with custom stringifyValue") {
+    val udt = new TestUDT.MyDenseVectorUDT() {
+      override def stringifyValue(obj: Any): String = {
+        val v = obj.asInstanceOf[TestUDT.MyDenseVector]
+        v.toString.stripPrefix("(").stripSuffix(")")
+      }
+    }
+    val vector = new TestUDT.MyDenseVector(Array(1.0, 3.0, 5.0, 7.0, 9.0))
+    val data = udt.serialize(vector)
+    val ret = Cast(Literal(data, udt), StringType, None)
+    checkEvaluation(ret, "1.0, 3.0, 5.0, 7.0, 9.0")
+  }
+
   test("SPARK-28497 Can't up cast UserDefinedType to string") {
     val udt = new TestUDT.MyDenseVectorUDT()
     assert(!Cast.canUpCast(udt, StringType))