I am converting PyArrow types to their json-schema.org equivalent. The code works, all unit tests pass but MyPy gives the error
error: "object" has no attribute "get"
I have a dictionary as follows
from pyarrow import bool_, date32, float64, int64, string
DATATYPE_MAPPING = {
string(): {"json_type": "string", "example_value": "Words"},
bool_(): {"json_type": "bool", "example_value": True},
int64(): {"json_type": "integer", "example_value": 32},
float64(): {"json_type": "number", "example_value": 3.14},
date32(): {
"json_type": "string",
"example_value": "2020-02-01",
"format": "date",
},
}
This is used to translate data types from the results of PyArrow.csv.read_csv as follows
arrow_reader = csv.read_csv(
self.input_file,
read_options=csv_read_options,
parse_options=csv_parse_options,
)
self.column_headers = arrow_reader.column_names
# Build the list converting PyArrow types to json-schema.org types
data_type_list = [
DATATYPE_MAPPING[datatype].get("example_value", "string")
for datatype in arrow_reader.schema.types
]
MyPy does not understand that a DATATYPE_MAPPING[datatype] is a dictionary. How can I get instruct it that it is a dictionary? If not, can I get MyPy to ignore this problem?
Thanks to Joel for this.
from pyarrow import DataType, bool_, date32, float64, int64, string
DATATYPE_MAPPING: dict[DataType, dict[str, object]] = {
string(): {"json_type": "string", "example_value": "Words"},
bool_(): {"json_type": "bool", "example_value": True},
int64(): {"json_type": "integer", "example_value": 32},
float64(): {"json_type": "number", "example_value": 3.14},
date32(): {
"json_type": "string",
"example_value": "2020-02-01",
"format": "date",
},
}