Possible issue with how aliases are resolved in JsonSchemaParser

[Thood42]

After looking deeper into the code while providing an alias mapping dictionary like so:
{ 'model_field_name': 'field_alias_name' }

I do not get the correct model outputs, It will only add the alias property to the model if the model_field_name and the field_alias_name values match.

I found this function that seems to handle the field_name and alias parameters while parsing the json objects:

Function Class: FieldNameResolver

Function file path: Line 271 -> #https://github.com/koxudaxi/datamodel-code-generator/blob/main/datamodel_code_generator/reference.py

    def get_valid_field_name_and_alias(
        self, field_name: str, excludes: Optional[Set[str]] = None
    ) -> Tuple[str, Optional[str]]:
        if field_name in self.aliases:
            return self.aliases[field_name], field_name
        valid_name = self.get_valid_name(field_name, excludes=excludes)
        return valid_name, None if field_name == valid_name else field_name

It first checks to see if the field_name is a key in the objects aliases variable, and if so returns a tuple like so:

return self.aliases[field_name], field_name

This implies that the tuple contains the alias from the mapping alias dictionary, and the field_name checked in the function.

So when the above function is called here:

File Path: Line 896 -> #https://github.com/koxudaxi/datamodel-code-generator/blob/main/datamodel_code_generator/parser/jsonschema.py

    def parse_object_fields(
        self, obj: JsonSchemaObject, path: List[str], module_name: Optional[str] = None
    ) -> List[DataModelFieldBase]:
        properties: Dict[str, Union[JsonSchemaObject, bool]] = (
            {} if obj.properties is None else obj.properties
        )
        requires: Set[str] = {*()} if obj.required is None else {*obj.required}
        fields: List[DataModelFieldBase] = []

        exclude_field_names: Set[str] = set()
        for original_field_name, field in properties.items():
            field_name, alias = self.model_resolver.get_valid_field_name_and_alias(
                original_field_name, exclude_field_names
            )

The values seem to be set in reverse, The alias is set to the field_name and the field_name is what the alias should be.

So with the context above, Am I using the library incorrectly when attempting to add the alias mapping dictionary?

I have tried reversing the key and value when providing the mapping data and in that case or the generation errors out and tries to make the alias the field name in the generated Pydantic model class.

[koxudaxi]

This is the intended behavior: the --aliases key is the schema field name and the value becomes the Python field name, with the original schema name automatically set as the Pydantic alias for correct serialization.