update as feedback

Author: chenghao-intel

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

@@ -567,7 +567,6 @@ case class Length(child: Expression) extends UnaryExpression with ExpectsInputTy
     child.dataType match {
       case StringType => defineCodeGen(ctx, ev, c => s"($c).numChars()")
       case BinaryType => defineCodeGen(ctx, ev, c => s"($c).length")
-      case NullType => defineCodeGen(ctx, ev, c => s"-1")
     }
   }
 
@@ -685,8 +684,6 @@ case class FormatNumber(x: Expression, d: Expression)
   override def right: Expression = d
   override def dataType: DataType = StringType
   override def inputTypes: Seq[AbstractDataType] = Seq(NumericType, IntegerType)
-  override def foldable: Boolean = x.foldable && d.foldable
-  override def nullable: Boolean = x.nullable || d.nullable
 
   @transient
   private var lastDValue: Int = -100
@@ -706,8 +703,7 @@ case class FormatNumber(x: Expression, d: Expression)
     val dObject = d.eval(input)
 
     if (dObject == null || dObject.asInstanceOf[Int] < 0) {
-      throw new IllegalArgumentException(
-        s"Argument 2 of function FORMAT_NUMBER must be >= 0, but $dObject was found")
+      return null
     }
     val dValue = dObject.asInstanceOf[Int]
 
@@ -742,5 +738,7 @@ case class FormatNumber(x: Expression, d: Expression)
         UTF8String.fromString(numberFormat.format(xObject.asInstanceOf[Decimal].toJavaBigDecimal))
     }
   }
+
+  override def prettyName: String = "format_number"
 }
 

sql/catalyst/src/test/scala/org/apache/spark/sql/catalyst/expressions/StringFunctionsSuite.scala

@@ -441,8 +441,6 @@ class StringFunctionsSuite extends SparkFunSuite with ExpressionEvalHelper {
 
     checkEvaluation(Length(Literal.create(null, StringType)), null, create_row(string))
     checkEvaluation(Length(Literal.create(null, BinaryType)), null, create_row(bytes))
-
-    checkEvaluation(Length(Literal.create(null, NullType)), null, create_row(null))
   }
 
   test("number format") {
@@ -453,6 +451,7 @@ class StringFunctionsSuite extends SparkFunSuite with ExpressionEvalHelper {
     checkEvaluation(FormatNumber(Literal(12831273.23481d), Literal(3)), "12,831,273.235")
     checkEvaluation(FormatNumber(Literal(12831273.83421d), Literal(0)), "12,831,274")
     checkEvaluation(FormatNumber(Literal(123123324123L), Literal(3)), "123,123,324,123.000")
+    checkEvaluation(FormatNumber(Literal(123123324123L), Literal(-1)), null)
     checkEvaluation(
       FormatNumber(
         Literal(Decimal(123123324123L) * Decimal(123123.21234d)), Literal(4)),

sql/core/src/main/scala/org/apache/spark/sql/functions.scala

@@ -1685,41 +1685,43 @@ object functions {
   //////////////////////////////////////////////////////////////////////////////////////////////
 
   /**
-   * Computes the length of a given string / binary value
+   * Computes the length of a given string / binary value.
    *
    * @group string_funcs
    * @since 1.5.0
    */
   def length(e: Column): Column = Length(e.expr)
 
   /**
-   * Computes the length of a given string / binary  column
+   * Computes the length of a given string / binary column.
    *
    * @group string_funcs
    * @since 1.5.0
    */
   def length(columnName: String): Column = length(Column(columnName))
 
   /**
-   * Formats the number X to a format like '#,###,###.##', rounded to D decimal places,
-   * and returns the result as a string. If D is 0, the result has no decimal point or
-   * fractional part.
+   * Formats the number X to a format like '#,###,###.##', rounded to d decimal places,
+   * and returns the result as a string.
+   * If d is 0, the result has no decimal point or fractional part.
+   * If d < 0, the result will be null.
    *
    * @group string_funcs
    * @since 1.5.0
    */
-  def formatNumber(x: Column, d: Column): Column = FormatNumber(x.expr, d.expr)
+  def format_number(x: Column, d: Int): Column = FormatNumber(x.expr, lit(d).expr)
 
   /**
-   * Formats the number X to a format like '#,###,###.##', rounded to D decimal places,
-   * and returns the result as a string. If D is 0, the result has no decimal point or
-   * fractional part.
+   * Formats the number X to a format like '#,###,###.##', rounded to d decimal places,
+   * and returns the result as a string.
+   * If d is 0, the result has no decimal point or fractional part.
+   * If d < 0, the result will be null.
    *
    * @group string_funcs
    * @since 1.5.0
    */
-  def formatNumber(columnXName: String, columnDName: String): Column = {
-    formatNumber(Column(columnXName), Column(columnDName))
+  def format_number(columnXName: String, d: Int): Column = {
+    format_number(Column(columnXName), d)
   }
 
   /**

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

@@ -465,8 +465,8 @@ class DataFrameFunctionsSuite extends QueryTest {
 
     checkAnswer(
       df.select(
-        formatNumber($"f", $"e"),
-        formatNumber("f", "e")),
+        format_number($"f", 4),
+        format_number("f", 4)),
       Row("5.0000", "5.0000"))
 
     checkAnswer(