Data Types

Sail supports Arrow data types that is a superset of data types available in Spark SQL. For more background information, you can refer to the Data Types guide for the Spark DataFrame API.

The following table shows the SQL type syntax along with the corresponding Spark data types and Arrow data types. Many data types have aliases not supported in JVM Spark. These are extensions in Sail.

Many Arrow data types do not have a corresponding SQL type syntax, but they are still supported in Sail. You can work with these types in Python UDFs or data sources.

SQL Type Syntax	Spark Data Type	Arrow Data Type
NULL VOID	NullType	Null
BOOLEAN BOOL	BooleanType	Boolean
BYTE TINYINT INT8	ByteType	Int8
SHORT SMALLINT INT16	ShortType	Int16
INTEGER INT INT32	IntegerType	Int32
LONG BIGINT INT64	LongType	Int64
UNSIGNED BYTE UNSIGNED TINYINT UINT8	-	UInt8
UNSIGNED SHORT UNSIGNED SMALLINT UINT16	-	UInt16
UNSIGNED INTEGER UNSIGNED INT UINT32	-	UInt32
UNSIGNED LONG UNSIGNED BIGINT UINT64	-	UInt64
-	-	Float16
FLOAT REAL FLOAT32	FloatType	Float32
DOUBLE FLOAT64	DoubleType	Float64
DATE DATE32	DateType	Date32
DATE64	-	Date64
-	-	Time32(Second) Time32(Millisecond) Time64(Microsecond) Time64(Nanosecond)
TIMESTAMP[(p)]	TimestampType TimestampNTZType	Timestamp(_, _)
TIMESTAMP_LTZ[(p)] TIMESTAMP[(p)] WITH [LOCAL ]TIME ZONE	TimestampType	Timestamp(_, TimeZone(_))
TIMESTAMP_NTZ[(p)] TIMESTAMP[(p)] WITHOUT TIME ZONE	TimestampNTZType	Timestamp(_, NoTimeZone)
STRING	StringType	Utf8 LargeUtf8
TEXT	-	LargeUtf8
CHAR(n) CHARACTER(n)	CharType(n)	Utf8 LargeUtf8
VARCHAR(n)	VarcharType(n)	Utf8 LargeUtf8
-	-	Utf8View
BINARY BYTEA	BinaryType	Binary LargeBinary
-	-	FixedSizeBinary BinaryView
-	-	Decimal32 Decimal64
DECIMAL[(p[, s])] DEC[(p[, s])] NUMERIC[(p[, s])]	DecimalType	Decimal128 Decimal256
-	-	Duration(Second) Duration(Millisecond) Duration(Nanosecond)
INTERVAL YEAR INTERVAL YEAR TO MONTH INTERVAL MONTH	YearMonthIntervalType	Interval(YearMonth)
-	-	Interval(DayTime)
INTERVAL DAY INTERVAL DAY TO HOUR INTERVAL DAY TO MINUTE INTERVAL DAY TO SECOND INTERVAL HOUR INTERVAL HOUR TO MINUTE INTERVAL HOUR TO SECOND INTERVAL MINUTE INTERVAL MINUTE TO SECOND INTERVAL SECOND	DayTimeIntervalType	Duration(Microsecond)
INTERVAL	CalendarIntervalType	Interval(MonthDayNano)
ARRAY<type>	ArrayType	List
-	-	LargeList FixedSizeList ListView LargeListView
MAP<key-type, value-type>	MapType	Map
STRUCT<name[:] type(, name[:] type)*>	StructType	Struct
-	-	Union
-	-	Dictionary
-	-	RunEndEncoded

Notes

1. The SQL string types (except TEXT) are mapped to either the Utf8 or LargeUtf8 type in Arrow, depending on the spark.sql.execution.arrow.useLargeVarTypes configuration option.
2. The SQL binary types are mapped to either the Binary or LargeBinary type in Arrow, depending on the spark.sql.execution.arrow.useLargeVarTypes configuration option.
3. The SQL TIMESTAMP type can either represent timestamps with local time zone (TIMESTAMP_LTZ, the default) or timestamps without time zone (TIMESTAMP_NTZ), depending on the spark.sql.timestampType configuration option.
4. For the SQL timestamp types, the optional p parameter specifies the precision of the timestamp. A number of 0, 3, 6, or 9 represents second, millisecond, microsecond, or nanosecond precision respectively. The default value is 6 (microsecond precision). Note that only the microsecond precision timestamp is compatible with Spark.
5. For the SQL decimal types, the optional p and s parameters specify the precision and scale of the decimal number respectively. The default precision is 10 and the default scale is 0. The decimal type maps to either Decimal128 or Decimal256 type in Arrow depending on the specified precision.
6. The SQL INTERVAL type is mapped to the Interval(MonthDayNano) Arrow type which has nanosecond precision. CalendarIntervalType in Spark has microsecond precision so the supported data range is different.