Supported Data Types

The following data types are supported in the Fivetran Connector SDK:

BOOLEAN
SHORT
INT
LONG
DECIMAL
FLOAT
DOUBLE
NAIVE_DATE
NAIVE_DATETIME
UTC_DATETIME
BINARY
XML
STRING
JSON

If you don't specify a type, we infer the data type automatically based on the values. The Connector SDK doesn't accept every Python object for every column type. It validates values, coerces compatible inputs, and rejects values that don't fit the declared type. None values are interpreted as NULL, and NaN values as FLOAT.

Explicitly defined column types

Data type	Accepted Python types
`BOOLEAN`	Boolean and integer values
`SHORT`	Whole-number integers
`INT`	Whole-number integers
`LONG`	Whole-number integers
`DECIMAL`	Decimal string such as `"123.45"`
`FLOAT`	Numeric values
`DOUBLE`	Numeric values
`NAIVE_DATE`	`datetime.datetime` objects, timestamp string values
`NAIVE_DATETIME`	`datetime.datetime` objects, timestamp string values
`UTC_DATETIME`	`datetime.datetime` objects, timestamp string values
`BINARY`	`bytes`
`XML`	XML string
`STRING`	UTF-8 compatible string
`JSON`	JSON string

When a value does not match the expected shape, Connector SDK surfaces a conversion error.

Inferred column types

If a column is not defined in the schema, the Connector SDK infers the type from the data. This inference behavior is useful when you do not want to predeclare column type. The Connector SDK chooses the compatible type from the value it receives and falls back to string when it cannot infer something more specific.