match notation for decimal parsing (#1875)

Author: jycor

enginetest/queries/script_queries.go

@@ -2804,6 +2804,18 @@ var ScriptTests = []ScriptTest{
 			},
 		},
 	},
+	{
+		Name: "scientific notation for floats",
+		SetUpScript: []string{
+			"create table t (b bigint unsigned);",
+		},
+		Assertions: []ScriptTestAssertion{
+			{
+				Query:    "insert into t values (5.2443381514267e+18);",
+				Expected: []sql.Row{{types.NewOkResult(1)}},
+			},
+		},
+	},
 	{
 		Name: "INSERT IGNORE throws an error when json is badly formatted",
 		SetUpScript: []string{

sql/parse/parse.go

@@ -4346,7 +4346,11 @@ func convertVal(ctx *sql.Context, v *sqlparser.SQLVal) (sql.Expression, error) {
 		// use the value as string format to keep precision and scale as defined for DECIMAL data type to avoid rounded up float64 value
 		if ps := strings.Split(string(v.Val), "."); len(ps) == 2 {
 			ogVal := string(v.Val)
-			floatVal := strconv.FormatFloat(val, 'f', -1, 64)
+			var fmtStr byte = 'f'
+			if strings.Contains(ogVal, "e") {
+				fmtStr = 'e'
+			}
+			floatVal := strconv.FormatFloat(val, fmtStr, -1, 64)
 			if len(ogVal) >= len(floatVal) && ogVal != floatVal {
 				p, s := expression.GetDecimalPrecisionAndScale(ogVal)
 				dt, err := types.CreateDecimalType(p, s)