def orc_type(field):
if pa.types.is_boolean(field):
return pyorc.Boolean()
elif pa.types.is_int8(field):
return pyorc.TinyInt()
elif pa.types.is_int16(field):
return pyorc.SmallInt()
elif pa.types.is_int32(field):
return pyorc.Int()
elif pa.types.is_int64(field):
return pyorc.BigInt()
elif pa.types.is_float32(field):
return pyorc.Float()
elif pa.types.is_float64(field):
return pyorc.Double()
elif pa.types.is_decimal(field):
return pyorc.Decimal(field.precision, field.scale)
elif pa.types.is_list(field):
return pyorc.Array(orc_type(field.value_type))
elif pa.types.is_timestamp(field):
return pyorc.Timestamp()
elif pa.types.is_date(field):
return pyorc.Date()
elif pa.types.is_binary(field):
return pyorc.Binary()
elif pa.types.is_string(field):
return pyorc.String()
else:
raise ValueError('Cannot Convert %s' % field)
pd.StringDtype(): ["string", "null"],
cudf.dtype("bool_"): "boolean",
cudf.dtype("int16"): "int",
cudf.dtype("int32"): "int",
cudf.dtype("int64"): "long",
cudf.dtype("O"): "string",
cudf.dtype("str"): "string",
cudf.dtype("float32"): "float",
cudf.dtype("float64"): "double",
cudf.dtype("<M8[ns]"): {"type": "long", "logicalType": "timestamp-millis"},
cudf.dtype("<M8[ms]"): {"type": "long", "logicalType": "timestamp-millis"},
cudf.dtype("<M8[us]"): {"type": "long", "logicalType": "timestamp-micros"},
}
PANDAS_TO_ORC_TYPES = {
cudf.dtype("int8"): pyorc.TinyInt(),
pd.Int8Dtype(): pyorc.TinyInt(),
pd.Int16Dtype(): pyorc.SmallInt(),
pd.Int32Dtype(): pyorc.Int(),
pd.Int64Dtype(): pyorc.BigInt(),
pd.BooleanDtype(): pyorc.Boolean(),
cudf.dtype("bool_"): pyorc.Boolean(),
cudf.dtype("int16"): pyorc.SmallInt(),
cudf.dtype("int32"): pyorc.Int(),
cudf.dtype("int64"): pyorc.BigInt(),
cudf.dtype("O"): pyorc.String(),
pd.StringDtype(): pyorc.String(),
cudf.dtype("float32"): pyorc.Float(),
cudf.dtype("float64"): pyorc.Double(),
cudf.dtype("<M8[ns]"): pyorc.Timestamp(),
cudf.dtype("<M8[ms]"): pyorc.Timestamp(),
np.dtype("<M8[ns]"): {
"type": "long",
"logicalType": "timestamp-millis"
},
np.dtype("<M8[ms]"): {
"type": "long",
"logicalType": "timestamp-millis"
},
np.dtype("<M8[us]"): {
"type": "long",
"logicalType": "timestamp-micros"
},
}
PANDAS_TO_ORC_TYPES = {
np.dtype("int8"): pyorc.TinyInt(),
pd.Int8Dtype(): pyorc.TinyInt(),
pd.Int16Dtype(): pyorc.SmallInt(),
pd.Int32Dtype(): pyorc.Int(),
pd.Int64Dtype(): pyorc.BigInt(),
pd.BooleanDtype(): pyorc.Boolean(),
np.dtype("bool_"): pyorc.Boolean(),
np.dtype("int16"): pyorc.SmallInt(),
np.dtype("int32"): pyorc.Int(),
np.dtype("int64"): pyorc.BigInt(),
np.dtype("O"): pyorc.String(),
pd.StringDtype(): pyorc.String(),
np.dtype("float32"): pyorc.Float(),
np.dtype("float64"): pyorc.Double(),
np.dtype("<M8[ns]"): pyorc.Timestamp(),
np.dtype("<M8[ms]"): pyorc.Timestamp(),