[SPARK-56976][SQL] Support Two-Argument IF(condition, value) Returning NULL

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

@@ -32,18 +32,23 @@ import org.apache.spark.util.ArrayImplicits._
 
 // scalastyle:off line.size.limit
 @ExpressionDescription(
-  usage = "_FUNC_(expr1, expr2, expr3) - If `expr1` evaluates to true, then returns `expr2`; otherwise returns `expr3`.",
+  usage = "_FUNC_(expr1, expr2, expr3) - If `expr1` evaluates to true, then returns `expr2`; otherwise returns `expr3`. If `expr3` is not provided, returns NULL when `expr1` is false.",
   examples = """
     Examples:
       > SELECT _FUNC_(1 < 2, 'a', 'b');
        a
+      > SELECT _FUNC_(1 > 2, 'a');
+       NULL
   """,
   since = "1.0.0",
   group = "conditional_funcs")
 // scalastyle:on line.size.limit
 case class If(predicate: Expression, trueValue: Expression, falseValue: Expression)
   extends ComplexTypeMergingExpression with ConditionalExpression with TernaryLike[Expression] {
 
+  def this(predicate: Expression, trueValue: Expression) =
+    this(predicate, trueValue, Literal(null))
+
   @transient
   override lazy val inputTypesForMerging: Seq[DataType] = {
     Seq(trueValue.dataType, falseValue.dataType)

sql/core/src/test/resources/sql-tests/analyzer-results/if-two-arg.sql.out

@@ -0,0 +1,81 @@
+-- Automatically generated by SQLQueryTestSuite
+-- !query
+SELECT if(1 < 2, 'yes')
+-- !query analysis
+Project [if ((1 < 2)) yes else cast(null as string) AS (IF((1 < 2), yes, NULL))#x]
++- OneRowRelation
+
+
+-- !query
+SELECT if(1 > 2, 'yes')
+-- !query analysis
+Project [if ((1 > 2)) yes else cast(null as string) AS (IF((1 > 2), yes, NULL))#x]
++- OneRowRelation
+
+
+-- !query
+SELECT if(id > 1, id * 10) FROM VALUES (1), (2), (3) AS t(id)
+-- !query analysis
+Project [if ((id#x > 1)) (id#x * 10) else cast(null as int) AS (IF((id > 1), (id * 10), NULL))#x]
++- SubqueryAlias t
+   +- LocalRelation [id#x]
+
+
+-- !query
+SELECT if(1 < 2, 'a', 'b')
+-- !query analysis
+Project [if ((1 < 2)) a else b AS (IF((1 < 2), a, b))#x]
++- OneRowRelation
+
+
+-- !query
+SELECT if(length('hello') > 3, upper('hello'))
+-- !query analysis
+Project [if ((length(hello) > 3)) upper(hello) else cast(null as string) AS (IF((length(hello) > 3), upper(hello), NULL))#x]
++- OneRowRelation
+
+
+-- !query
+SELECT if(true)
+-- !query analysis
+org.apache.spark.sql.AnalysisException
+{
+  "errorClass" : "WRONG_NUM_ARGS.WITHOUT_SUGGESTION",
+  "sqlState" : "42605",
+  "messageParameters" : {
+    "actualNum" : "1",
+    "docroot" : "https://spark.apache.org/docs/latest",
+    "expectedNum" : "[2, 3]",
+    "functionName" : "`if`"
+  },
+  "queryContext" : [ {
+    "objectType" : "",
+    "objectName" : "",
+    "startIndex" : 8,
+    "stopIndex" : 15,
+    "fragment" : "if(true)"
+  } ]
+}
+
+
+-- !query
+SELECT if(true, 1, 2, 3)
+-- !query analysis
+org.apache.spark.sql.AnalysisException
+{
+  "errorClass" : "WRONG_NUM_ARGS.WITHOUT_SUGGESTION",
+  "sqlState" : "42605",
+  "messageParameters" : {
+    "actualNum" : "4",
+    "docroot" : "https://spark.apache.org/docs/latest",
+    "expectedNum" : "[2, 3]",
+    "functionName" : "`if`"
+  },
+  "queryContext" : [ {
+    "objectType" : "",
+    "objectName" : "",
+    "startIndex" : 8,
+    "stopIndex" : 24,
+    "fragment" : "if(true, 1, 2, 3)"
+  } ]
+}

sql/core/src/test/resources/sql-tests/inputs/if-two-arg.sql

@@ -0,0 +1,22 @@
+-- Tests for two-argument IF (SQL standard: IF(cond, value) returns NULL when false)
+
+-- Two-argument IF: returns value when true
+SELECT if(1 < 2, 'yes');
+
+-- Two-argument IF: returns NULL when false
+SELECT if(1 > 2, 'yes');
+
+-- Two-argument IF: with column reference
+SELECT if(id > 1, id * 10) FROM VALUES (1), (2), (3) AS t(id);
+
+-- Three-argument IF still works (regression check)
+SELECT if(1 < 2, 'a', 'b');
+
+-- Two-argument IF with complex expression
+SELECT if(length('hello') > 3, upper('hello'));
+
+-- Error: one argument
+SELECT if(true);
+
+-- Error: four arguments
+SELECT if(true, 1, 2, 3);

sql/core/src/test/resources/sql-tests/results/if-two-arg.sql.out

@@ -0,0 +1,91 @@
+-- Automatically generated by SQLQueryTestSuite
+-- !query
+SELECT if(1 < 2, 'yes')
+-- !query schema
+struct<(IF((1 < 2), yes, NULL)):string>
+-- !query output
+yes
+
+
+-- !query
+SELECT if(1 > 2, 'yes')
+-- !query schema
+struct<(IF((1 > 2), yes, NULL)):string>
+-- !query output
+NULL
+
+
+-- !query
+SELECT if(id > 1, id * 10) FROM VALUES (1), (2), (3) AS t(id)
+-- !query schema
+struct<(IF((id > 1), (id * 10), NULL)):int>
+-- !query output
+20
+30
+NULL
+
+
+-- !query
+SELECT if(1 < 2, 'a', 'b')
+-- !query schema
+struct<(IF((1 < 2), a, b)):string>
+-- !query output
+a
+
+
+-- !query
+SELECT if(length('hello') > 3, upper('hello'))
+-- !query schema
+struct<(IF((length(hello) > 3), upper(hello), NULL)):string>
+-- !query output
+HELLO
+
+
+-- !query
+SELECT if(true)
+-- !query schema
+struct<>
+-- !query output
+org.apache.spark.sql.AnalysisException
+{
+  "errorClass" : "WRONG_NUM_ARGS.WITHOUT_SUGGESTION",
+  "sqlState" : "42605",
+  "messageParameters" : {
+    "actualNum" : "1",
+    "docroot" : "https://spark.apache.org/docs/latest",
+    "expectedNum" : "[2, 3]",
+    "functionName" : "`if`"
+  },
+  "queryContext" : [ {
+    "objectType" : "",
+    "objectName" : "",
+    "startIndex" : 8,
+    "stopIndex" : 15,
+    "fragment" : "if(true)"
+  } ]
+}
+
+
+-- !query
+SELECT if(true, 1, 2, 3)
+-- !query schema
+struct<>
+-- !query output
+org.apache.spark.sql.AnalysisException
+{
+  "errorClass" : "WRONG_NUM_ARGS.WITHOUT_SUGGESTION",
+  "sqlState" : "42605",
+  "messageParameters" : {
+    "actualNum" : "4",
+    "docroot" : "https://spark.apache.org/docs/latest",
+    "expectedNum" : "[2, 3]",
+    "functionName" : "`if`"
+  },
+  "queryContext" : [ {
+    "objectType" : "",
+    "objectName" : "",
+    "startIndex" : 8,
+    "stopIndex" : 24,
+    "fragment" : "if(true, 1, 2, 3)"
+  } ]
+}