feature request: custom serialization for specific column

[ivan-ngchakming]

Is there any way to specify a custom serialization function for one column by name? Like the serialize_types for custom type, something like

class CustomSerializerMixin(SerializerMixin):
    additional_serialize_rule = (('column1', lambda x: x),)

edit: my current get around is this, but it would be nice if this step could be done during serialization and reduce the redundant logic

class CustomSerializerMixin(SerializerMixin):

    def to_dict(self, *args, **kwargs):
        obj_dict = super().to_dict(*args, **kwargs)

        # ...custom logic to transform obj_dict

        return obj_dict

[gui-ito-makes]

I also had this idea when I had to format a column but wanted to keep its name:

def format_price(self):    
    return locale.currency(self.price, grouping=True)

serialize_rules = ('-price',  'format_price',)

My solution was to append the name after '/', changing the serialize_model method:

def format_price(self):    
    return locale.currency(self.price, grouping=True)

serialize_rules = ('-price',  'format_price/price',)

In serialize_model method:

for k in keys:
    var, key = k.split('/') if '/' in k else (k, k)
    if self.schema.is_included(key=var):  # TODO: Skip check if is NOT greedy
        v = getattr(value, var)
        logger.debug('Serialize key:%s type:%s model:%s', k, get_type(v), get_type(value))
        res[key] = self.fork(key=var, value=v)
    else:
        logger.debug('Skip key:%s of model:%s', k, get_type(value))

[n0nSmoker]

Available in the latest version on pypi

- Custom serialization for specific columns via `serialize_columns` parameter
  - Supports per-call customization: `to_dict(serialize_columns={'field': lambda v: ...})`
  - Supports model-level defaults: `serialize_columns = {'field': lambda v: ...}` class attribute
  - Custom serializers receive the field value and return the serialized result