atoti.Session.read_parquet()

Session.read_parquet(path, /, *, client_side_encryption=None, columns={}, data_types={}, default_values={}, keys=frozenset({}), partitioning=None, table_name=None, **kwargs)

Read a Parquet file into a table.

Note

This is just a shortcut for:

parquet_load = tt.ParquetLoad(path, ...)
inferred_data_types = session.tables.infer_data_types(parquet_load)
table = session.create_table(
    table_name,
    data_types={**inferred_data_types, **data_types},
    default_values=...,
    keys=...,
    partitioning=...,
)
table.load(parquet_load)

The longer version unlocks better performance because it can be split to move the load() call inside a data_transaction().

Parameters:

• client_side_encryption (ClientSideEncryptionConfig | None) – See atoti.ParquetLoad.client_side_encryption.

• columns (Mapping[str, str]) – See atoti.ParquetLoad.columns.

• data_types (Mapping[str, DataType]) – See create_table()’s data_types.

• default_values (Mapping[str, bool | int | float | date | datetime | time | Sequence[bool] | Sequence[int] | Sequence[float] | Sequence[str] | str | None]) – See create_table()’s default_values.

• keys (Set[str] | Sequence[str]) – See create_table()’s keys.

• partitioning (str | None) – See create_table()’s partitioning.

• path (Path | str) – See atoti.ParquetLoad.path.

• table_name (str | None) – See create_table()’s name.

• kwargs (Unpack[_ReadParquetPrivateParameters])

Return type:

Table