[SPARK-12848][SQL] Change parsed decimal literal datatype from Double to Decimal

The current parser turns a decimal literal, for example ```12.1```, into a Double. The problem with this approach is that we convert an exact literal into a non-exact ```Double```. The PR changes this behavior, a Decimal literal is now converted into an extact ```BigDecimal```.

The behavior for scientific decimals, for example ```12.1e01```, is unchanged. This will be converted into a Double.

This PR replaces the ```BigDecimal``` literal by a ```Double``` literal, because the ```BigDecimal``` is the default now. You can use the double literal by appending a 'D' to the value, for instance: ```3.141527D```

cc davies rxin

Author: Herman van Hovell <hvanhovell@questtec.nl>

Closes #10796 from hvanhovell/SPARK-12848.

R/pkg/inst/tests/testthat/test_sparkSQL.R

@@ -1812,7 +1812,7 @@ test_that("Method coltypes() to get and set R's data types of a DataFrame", {
   expect_equal(coltypes(x), "map<string,string>")
 
   df <- selectExpr(read.json(sqlContext, jsonPath), "name", "(age * 1.21) as age")
-  expect_equal(dtypes(df), list(c("name", "string"), c("age", "double")))
+  expect_equal(dtypes(df), list(c("name", "string"), c("age", "decimal(24,2)")))
 
   df1 <- select(df, cast(df$age, "integer"))
   coltypes(df) <- c("character", "integer")

sql/catalyst/src/main/antlr3/org/apache/spark/sql/catalyst/parser/ExpressionParser.g

@@ -122,7 +122,7 @@ constant
     | BigintLiteral
     | SmallintLiteral
     | TinyintLiteral
-    | DecimalLiteral
+    | DoubleLiteral
     | booleanValue
     ;
 

sql/catalyst/src/main/antlr3/org/apache/spark/sql/catalyst/parser/SparkSqlLexer.g

@@ -418,9 +418,9 @@ TinyintLiteral
     (Digit)+ 'Y'
     ;
 
-DecimalLiteral
+DoubleLiteral
     :
-    Number 'B' 'D'
+    Number 'D'
     ;
 
 ByteLengthLiteral

sql/catalyst/src/main/scala/org/apache/spark/sql/catalyst/CatalystQl.scala

@@ -623,6 +623,7 @@ https://cwiki.apache.org/confluence/display/Hive/Enhanced+Aggregation%2C+Cube%2C
   val CASE = "(?i)CASE".r
 
   val INTEGRAL = "[+-]?\\d+".r
+  val DECIMAL = "[+-]?((\\d+(\\.\\d*)?)|(\\.\\d+))".r
 
   protected def nodeToExpr(node: ASTNode): Expression = node match {
     /* Attribute References */
@@ -785,8 +786,8 @@ https://cwiki.apache.org/confluence/display/Hive/Enhanced+Aggregation%2C+Cube%2C
     case ast if ast.tokenType == SparkSqlParser.BigintLiteral =>
       Literal.create(ast.text.substring(0, ast.text.length() - 1).toLong, LongType)
 
-    case ast if ast.tokenType == SparkSqlParser.DecimalLiteral =>
-      Literal(Decimal(ast.text.substring(0, ast.text.length() - 2)))
+    case ast if ast.tokenType == SparkSqlParser.DoubleLiteral =>
+      Literal(ast.text.toDouble)
 
     case ast if ast.tokenType == SparkSqlParser.Number =>
       val text = ast.text
@@ -799,7 +800,10 @@ https://cwiki.apache.org/confluence/display/Hive/Enhanced+Aggregation%2C+Cube%2C
               Literal(v.longValue())
             case v => Literal(v.underlying())
           }
+        case DECIMAL(_*) =>
+          Literal(BigDecimal(text).underlying())
         case _ =>
+          // Convert a scientifically notated decimal into a double.
           Literal(text.toDouble)
       }
     case ast if ast.tokenType == SparkSqlParser.StringLiteral =>

sql/catalyst/src/main/scala/org/apache/spark/sql/catalyst/analysis/HiveTypeCoercion.scala

@@ -692,12 +692,11 @@ object HiveTypeCoercion {
       case e if !e.childrenResolved => e
       // Find tightest common type for If, if the true value and false value have different types.
       case i @ If(pred, left, right) if left.dataType != right.dataType =>
-        findTightestCommonTypeToString(left.dataType, right.dataType).map { widestType =>
+        findWiderTypeForTwo(left.dataType, right.dataType).map { widestType =>
           val newLeft = if (left.dataType == widestType) left else Cast(left, widestType)
           val newRight = if (right.dataType == widestType) right else Cast(right, widestType)
           If(pred, newLeft, newRight)
         }.getOrElse(i)  // If there is no applicable conversion, leave expression unchanged.
-
       // Convert If(null literal, _, _) into boolean type.
       // In the optimizer, we should short-circuit this directly into false value.
       case If(pred, left, right) if pred.dataType == NullType =>

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

@@ -212,7 +212,7 @@ class MathExpressionsSuite extends QueryTest with SharedSQLContext {
       Seq(Row(5, 10, 0), Row(55, 60, 100), Row(555, 560, 600))
     )
 
-    val pi = "3.1415BD"
+    val pi = "3.1415"
     checkAnswer(
       sql(s"SELECT round($pi, -3), round($pi, -2), round($pi, -1), " +
         s"round($pi, 0), round($pi, 1), round($pi, 2), round($pi, 3)"),
@@ -367,6 +367,16 @@ class MathExpressionsSuite extends QueryTest with SharedSQLContext {
     checkAnswer(
       input.toDF("key", "value").selectExpr("abs(key) a").sort("a"),
       input.map(pair => Row(pair._2)))
+
+    checkAnswer(
+      sql("select abs(0), abs(-1), abs(123), abs(-9223372036854775807), abs(9223372036854775807)"),
+      Row(0, 1, 123, 9223372036854775807L, 9223372036854775807L)
+    )
+
+    checkAnswer(
+      sql("select abs(0.0), abs(-3.14159265), abs(3.14159265)"),
+      Row(BigDecimal("0.0"), BigDecimal("3.14159265"), BigDecimal("3.14159265"))
+    )
   }
 
   test("log2") {

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

@@ -1174,19 +1174,19 @@ class SQLQuerySuite extends QueryTest with SharedSQLContext {
 
   test("Floating point number format") {
     checkAnswer(
-      sql("SELECT 0.3"), Row(0.3)
+      sql("SELECT 0.3"), Row(BigDecimal(0.3))
     )
 
     checkAnswer(
-      sql("SELECT -0.8"), Row(-0.8)
+      sql("SELECT -0.8"), Row(BigDecimal(-0.8))
     )
 
     checkAnswer(
-      sql("SELECT .5"), Row(0.5)
+      sql("SELECT .5"), Row(BigDecimal(0.5))
     )
 
     checkAnswer(
-      sql("SELECT -.18"), Row(-0.18)
+      sql("SELECT -.18"), Row(BigDecimal(-0.18))
     )
   }
 
@@ -1200,11 +1200,11 @@ class SQLQuerySuite extends QueryTest with SharedSQLContext {
     )
 
     checkAnswer(
-      sql("SELECT 9223372036854775808BD"), Row(new java.math.BigDecimal("9223372036854775808"))
+      sql("SELECT 9223372036854775808"), Row(new java.math.BigDecimal("9223372036854775808"))
     )
 
     checkAnswer(
-      sql("SELECT -9223372036854775809BD"), Row(new java.math.BigDecimal("-9223372036854775809"))
+      sql("SELECT -9223372036854775809"), Row(new java.math.BigDecimal("-9223372036854775809"))
     )
   }
 
@@ -1219,11 +1219,11 @@ class SQLQuerySuite extends QueryTest with SharedSQLContext {
     )
 
     checkAnswer(
-      sql("SELECT -5.2BD"), Row(BigDecimal(-5.2))
+      sql("SELECT -5.2"), Row(BigDecimal(-5.2))
     )
 
     checkAnswer(
-      sql("SELECT +6.8"), Row(6.8d)
+      sql("SELECT +6.8e0"), Row(6.8d)
     )
 
     checkAnswer(
@@ -1598,20 +1598,20 @@ class SQLQuerySuite extends QueryTest with SharedSQLContext {
   }
 
   test("decimal precision with multiply/division") {
-    checkAnswer(sql("select 10.3BD * 3.0BD"), Row(BigDecimal("30.90")))
-    checkAnswer(sql("select 10.3000BD * 3.0BD"), Row(BigDecimal("30.90000")))
-    checkAnswer(sql("select 10.30000BD * 30.0BD"), Row(BigDecimal("309.000000")))
-    checkAnswer(sql("select 10.300000000000000000BD * 3.000000000000000000BD"),
+    checkAnswer(sql("select 10.3 * 3.0"), Row(BigDecimal("30.90")))
+    checkAnswer(sql("select 10.3000 * 3.0"), Row(BigDecimal("30.90000")))
+    checkAnswer(sql("select 10.30000 * 30.0"), Row(BigDecimal("309.000000")))
+    checkAnswer(sql("select 10.300000000000000000 * 3.000000000000000000"),
       Row(BigDecimal("30.900000000000000000000000000000000000", new MathContext(38))))
-    checkAnswer(sql("select 10.300000000000000000BD * 3.0000000000000000000BD"),
+    checkAnswer(sql("select 10.300000000000000000 * 3.0000000000000000000"),
       Row(null))
 
-    checkAnswer(sql("select 10.3BD / 3.0BD"), Row(BigDecimal("3.433333")))
-    checkAnswer(sql("select 10.3000BD / 3.0BD"), Row(BigDecimal("3.4333333")))
-    checkAnswer(sql("select 10.30000BD / 30.0BD"), Row(BigDecimal("0.343333333")))
-    checkAnswer(sql("select 10.300000000000000000BD / 3.00000000000000000BD"),
+    checkAnswer(sql("select 10.3 / 3.0"), Row(BigDecimal("3.433333")))
+    checkAnswer(sql("select 10.3000 / 3.0"), Row(BigDecimal("3.4333333")))
+    checkAnswer(sql("select 10.30000 / 30.0"), Row(BigDecimal("0.343333333")))
+    checkAnswer(sql("select 10.300000000000000000 / 3.00000000000000000"),
       Row(BigDecimal("3.433333333333333333333333333", new MathContext(38))))
-    checkAnswer(sql("select 10.3000000000000000000BD / 3.00000000000000000BD"),
+    checkAnswer(sql("select 10.3000000000000000000 / 3.00000000000000000"),
       Row(BigDecimal("3.4333333333333333333333333333", new MathContext(38))))
   }
 
@@ -1637,13 +1637,13 @@ class SQLQuerySuite extends QueryTest with SharedSQLContext {
   }
 
   test("precision smaller than scale") {
-    checkAnswer(sql("select 10.00BD"), Row(BigDecimal("10.00")))
-    checkAnswer(sql("select 1.00BD"), Row(BigDecimal("1.00")))
-    checkAnswer(sql("select 0.10BD"), Row(BigDecimal("0.10")))
-    checkAnswer(sql("select 0.01BD"), Row(BigDecimal("0.01")))
-    checkAnswer(sql("select 0.001BD"), Row(BigDecimal("0.001")))
-    checkAnswer(sql("select -0.01BD"), Row(BigDecimal("-0.01")))
-    checkAnswer(sql("select -0.001BD"), Row(BigDecimal("-0.001")))
+    checkAnswer(sql("select 10.00"), Row(BigDecimal("10.00")))
+    checkAnswer(sql("select 1.00"), Row(BigDecimal("1.00")))
+    checkAnswer(sql("select 0.10"), Row(BigDecimal("0.10")))
+    checkAnswer(sql("select 0.01"), Row(BigDecimal("0.01")))
+    checkAnswer(sql("select 0.001"), Row(BigDecimal("0.001")))
+    checkAnswer(sql("select -0.01"), Row(BigDecimal("-0.01")))
+    checkAnswer(sql("select -0.001"), Row(BigDecimal("-0.001")))
   }
 
   test("external sorting updates peak execution memory") {

sql/core/src/test/scala/org/apache/spark/sql/execution/datasources/json/JsonSuite.scala

@@ -442,13 +442,13 @@ class JsonSuite extends QueryTest with SharedSQLContext with TestJsonData {
 
     // Number and String conflict: resolve the type as number in this query.
     checkAnswer(
-      sql("select num_str + 1.2BD from jsonTable where num_str > 14"),
+      sql("select num_str + 1.2 from jsonTable where num_str > 14"),
       Row(BigDecimal("92233720368547758071.2"))
     )
 
     // Number and String conflict: resolve the type as number in this query.
     checkAnswer(
-      sql("select num_str + 1.2BD from jsonTable where num_str >= 92233720368547758060BD"),
+      sql("select num_str + 1.2 from jsonTable where num_str >= 92233720368547758060"),
       Row(new java.math.BigDecimal("92233720368547758071.2"))
     )
 

sql/hive/compatibility/src/test/scala/org/apache/spark/sql/hive/execution/HiveCompatibilitySuite.scala

@@ -323,7 +323,14 @@ class HiveCompatibilitySuite extends HiveQueryFileTest with BeforeAndAfter {
     // Feature removed in HIVE-11145
     "alter_partition_protect_mode",
     "drop_partitions_ignore_protection",
-    "protectmode"
+    "protectmode",
+
+    // Spark parser treats numerical literals differently: it creates decimals instead of doubles.
+    "udf_abs",
+    "udf_format_number",
+    "udf_round",
+    "udf_round_3",
+    "view_cast"
   )
 
   /**
@@ -884,7 +891,6 @@ class HiveCompatibilitySuite extends HiveQueryFileTest with BeforeAndAfter {
     "udf_10_trims",
     "udf_E",
     "udf_PI",
-    "udf_abs",
     "udf_acos",
     "udf_add",
     "udf_array",
@@ -928,7 +934,6 @@ class HiveCompatibilitySuite extends HiveQueryFileTest with BeforeAndAfter {
     "udf_find_in_set",
     "udf_float",
     "udf_floor",
-    "udf_format_number",
     "udf_from_unixtime",
     "udf_greaterthan",
     "udf_greaterthanorequal",
@@ -976,8 +981,6 @@ class HiveCompatibilitySuite extends HiveQueryFileTest with BeforeAndAfter {
     "udf_regexp_replace",
     "udf_repeat",
     "udf_rlike",
-    "udf_round",
-    "udf_round_3",
     "udf_rpad",
     "udf_rtrim",
     "udf_sign",

sql/hive/compatibility/src/test/scala/org/apache/spark/sql/hive/execution/HiveWindowFunctionQuerySuite.scala

@@ -559,7 +559,7 @@ class HiveWindowFunctionQuerySuite extends HiveComparisonTest with BeforeAndAfte
     """
       |select  p_mfgr,p_name, p_size,
       |histogram_numeric(p_retailprice, 5) over w1 as hist,
-      |percentile(p_partkey, 0.5) over w1 as per,
+      |percentile(p_partkey, cast(0.5 as double)) over w1 as per,
       |row_number() over(distribute by p_mfgr sort by p_name) as rn
       |from part
       |window w1 as (distribute by p_mfgr sort by p_mfgr, p_name