Abstract away from pandera about dac HOT 3 CLOSED

From the example above, it is probably a good idea to clearly isolate the columns, so that Schema can contain other objects, such as list of valid values used in validate.

class Columns:
    col1 = "column_1_name"
    col2 = ...

class Schema:
    @classmethod
    def validate(cls) -> None:
        pass

    @classmethod
    def example(cls) -> object:
        pass

Things that should work (need to check):

• class inheritance of both Columns and Schema
• definition starting from pandera schema

from dac.

I checked and everything works as expected. If one has a pandera SchemaModel/DataFrameModel, can simply do

class PASchema(pa.DataFrameModel):
    ...

Columns = PASchema
Schema = PASchema

Obviously in this way there are some "side effects" (Columns will have methods like evaluate and example, and Schema will contain the column names), but I think it is ok. Potentially one could do something more sophisticated to separate the two.

The idea is to set Columns and Schema as Protocols in the dac library, to allow as much freedom as wanted to the users.

from dac.

Eventually I realized that I could not include Columns (actually, should have named it Fields anyway, to account for non-tabular data), because there is no simple way to check that it is provided correctly.

compatibility with pandera schemas is guaranteed.

from dac.