Mehmet Aktas created FLINK-34203:
------------------------------------
Summary: PyFlink `get_field_data_types()` supports only `precision
= 3` for `TimestampType`
Key: FLINK-34203
URL: https://issues.apache.org/jira/browse/FLINK-34203
Project: Flink
Issue Type: Technical Debt
Components: API / Python
Reporter: Mehmet Aktas
PyFlink schema's `get_field_data_types()` set `precision` to 3 for `TIMESTAMP`
regardless of the input precision – copied the relevant pieces of code below.
There does not seem to be any reason for this implicit assumption. It led to an
error in our project, forcing us to implement a wrapper around
`get_field_data_types()`.
pyflink/table/table_schema.py: get_field_data_types():
{code:java}
def get_field_data_types(self) -> List[DataType]:
"""
Returns all field data types as a list. :return: A list of all
field data types.
"""
return [_from_java_data_type(item) for item in
self._j_table_schema.getFieldDataTypes()] {code}
pyflink/table/types.py: _from_java_data_type():
{code:java}
def _from_java_data_type(j_data_type):
"""
Converts Java DataType to Python DataType.
"""
...
if is_instance_of(j_data_type, gateway.jvm.AtomicDataType):
logical_type = j_data_type.getLogicalType()
...
elif is_instance_of(logical_type, gateway.jvm.TimestampType):
data_type = DataTypes.TIMESTAMP(precision=3,
nullable=logical_type.isNullable())
...{code}
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
