[SPARK-32106][SQL]Implement SparkScriptTransformationExec in sql/core

sql/catalyst/src/main/scala/org/apache/spark/sql/catalyst/parser/AstBuilder.scala

@@ -744,8 +744,29 @@ class AstBuilder(conf: SQLConf) extends SqlBaseBaseVisitor[AnyRef] with Logging
     selectClause.hints.asScala.foldRight(withWindow)(withHints)
   }
 
+  // Decode and input/output format.
+  type Format = (Seq[(String, String)], Option[String], Seq[(String, String)], Option[String])
+
+  protected def getRowFormatDelimited(ctx: RowFormatDelimitedContext): Format = {
+    // TODO we should use the visitRowFormatDelimited function here. However HiveScriptIOSchema
+    // expects a seq of pairs in which the old parsers' token names are used as keys.
+    // Transforming the result of visitRowFormatDelimited would be quite a bit messier than
+    // retrieving the key value pairs ourselves.
+    def entry(key: String, value: Token): Seq[(String, String)] = {
+      Option(value).map(t => key -> t.getText).toSeq
+    }
+
+    val entries = entry("TOK_TABLEROWFORMATFIELD", ctx.fieldsTerminatedBy) ++
+      entry("TOK_TABLEROWFORMATCOLLITEMS", ctx.collectionItemsTerminatedBy) ++
+      entry("TOK_TABLEROWFORMATMAPKEYS", ctx.keysTerminatedBy) ++
+      entry("TOK_TABLEROWFORMATLINES", ctx.linesSeparatedBy) ++
+      entry("TOK_TABLEROWFORMATNULL", ctx.nullDefinedAs)
+
+    (entries, None, Seq.empty, None)
+  }
+
   /**
-   * Create a (Hive based) [[ScriptInputOutputSchema]].
+   * Create a [[ScriptInputOutputSchema]].
    */
   protected def withScriptIOSchema(
       ctx: ParserRuleContext,
@@ -754,7 +775,30 @@ class AstBuilder(conf: SQLConf) extends SqlBaseBaseVisitor[AnyRef] with Logging
       outRowFormat: RowFormatContext,
       recordReader: Token,
       schemaLess: Boolean): ScriptInputOutputSchema = {
-    throw new ParseException("Script Transform is not supported", ctx)
+
+    def format(fmt: RowFormatContext): Format = fmt match {
+      case c: RowFormatDelimitedContext =>
+        getRowFormatDelimited(c)
+
+      case c: RowFormatSerdeContext =>
+        throw new ParseException("TRANSFORM with serde is only supported in hive mode", ctx)
+
+      // SPARK-32106: When there is no definition about format, we return empty result
+      // to use a built-in default Serde in SparkScriptTransformationExec.
+      case null =>
+        (Nil, None, Seq.empty, None)
+    }
+
+    val (inFormat, inSerdeClass, inSerdeProps, reader) = format(inRowFormat)
+
+    val (outFormat, outSerdeClass, outSerdeProps, writer) = format(outRowFormat)
+
+    ScriptInputOutputSchema(
+      inFormat, outFormat,
+      inSerdeClass, outSerdeClass,
+      inSerdeProps, outSerdeProps,
+      reader, writer,
+      schemaLess)
   }
 
   /**

sql/catalyst/src/test/scala/org/apache/spark/sql/catalyst/parser/PlanParserSuite.scala

@@ -23,7 +23,7 @@ import org.apache.spark.sql.catalyst.expressions._
 import org.apache.spark.sql.catalyst.plans._
 import org.apache.spark.sql.catalyst.plans.logical._
 import org.apache.spark.sql.internal.SQLConf
-import org.apache.spark.sql.types.IntegerType
+import org.apache.spark.sql.types.{IntegerType, LongType, StringType}
 
 /**
  * Parser test cases for rules defined in [[CatalystSqlParser]] / [[AstBuilder]].
@@ -1031,4 +1031,96 @@ class PlanParserSuite extends AnalysisTest {
     assertEqual("select a, b from db.c;;;", table("db", "c").select('a, 'b))
     assertEqual("select a, b from db.c; ;;  ;", table("db", "c").select('a, 'b))
   }
+
+  test("SPARK-32106: TRANSFORM without serde") {
+    // verify schema less
+      assertEqual(
+        """
+          |SELECT TRANSFORM(a, b, c)
+          |USING 'cat'
+          |FROM testData
+        """.stripMargin,
+        ScriptTransformation(
+          Seq('a, 'b, 'c),
+          "cat",
+          Seq(AttributeReference("key", StringType)(),
+            AttributeReference("value", StringType)()),
+          UnresolvedRelation(TableIdentifier("testData")),
+          ScriptInputOutputSchema(List.empty, List.empty, None, None,
+            List.empty, List.empty, None, None, true))
+      )
+
+    // verify without output schema
+    assertEqual(
+      """
+        |SELECT TRANSFORM(a, b, c)
+        |USING 'cat' AS (a, b, c)
+        |FROM testData
+      """.stripMargin,
+      ScriptTransformation(
+        Seq('a, 'b, 'c),
+        "cat",
+        Seq(AttributeReference("a", StringType)(),
+          AttributeReference("b", StringType)(),
+          AttributeReference("c", StringType)()),
+        UnresolvedRelation(TableIdentifier("testData")),
+        ScriptInputOutputSchema(List.empty, List.empty, None, None,
+          List.empty, List.empty, None, None, false)))
+
+    // verify with output schema
+    assertEqual(
+      """
+        |SELECT TRANSFORM(a, b, c)
+        |USING 'cat' AS (a int, b string, c long)
+        |FROM testData
+      """.stripMargin,
+      ScriptTransformation(
+        Seq('a, 'b, 'c),
+        "cat",
+        Seq(AttributeReference("a", IntegerType)(),
+          AttributeReference("b", StringType)(),
+          AttributeReference("c", LongType)()),
+        UnresolvedRelation(TableIdentifier("testData")),
+        ScriptInputOutputSchema(List.empty, List.empty, None, None,
+          List.empty, List.empty, None, None, false)))
+
+    // verify with ROW FORMAT DELIMETED
+    assertEqual(
+      """
+        |SELECT TRANSFORM(a, b, c)
+        |ROW FORMAT DELIMITED
+        |FIELDS TERMINATED BY '\t'
+        |COLLECTION ITEMS TERMINATED BY '\u0002'
+        |MAP KEYS TERMINATED BY '\u0003'
+        |LINES TERMINATED BY '\n'
+        |NULL DEFINED AS 'null'
+        |USING 'cat' AS (a, b, c)
+        |ROW FORMAT DELIMITED
+        |FIELDS TERMINATED BY '\t'
+        |COLLECTION ITEMS TERMINATED BY '\u0004'
+        |MAP KEYS TERMINATED BY '\u0005'
+        |LINES TERMINATED BY '\n'
+        |NULL DEFINED AS 'NULL'
+        |FROM testData
+      """.stripMargin,
+      ScriptTransformation(
+        Seq('a, 'b, 'c),
+        "cat",
+        Seq(AttributeReference("a", StringType)(),
+          AttributeReference("b", StringType)(),
+          AttributeReference("c", StringType)()),
+        UnresolvedRelation(TableIdentifier("testData")),
+        ScriptInputOutputSchema(
+          Seq(("TOK_TABLEROWFORMATFIELD", "'\\t'"),
+            ("TOK_TABLEROWFORMATCOLLITEMS", "'\u0002'"),
+            ("TOK_TABLEROWFORMATMAPKEYS", "'\u0003'"),
+            ("TOK_TABLEROWFORMATLINES", "'\\n'"),
+            ("TOK_TABLEROWFORMATNULL", "'null'")),
+          Seq(("TOK_TABLEROWFORMATFIELD", "'\\t'"),
+            ("TOK_TABLEROWFORMATCOLLITEMS", "'\u0004'"),
+            ("TOK_TABLEROWFORMATMAPKEYS", "'\u0005'"),
+            ("TOK_TABLEROWFORMATLINES", "'\\n'"),
+            ("TOK_TABLEROWFORMATNULL", "'NULL'")), None, None,
+          List.empty, List.empty, None, None, false)))
+  }
 }