canonical_args

canonical_args is a package designed to provide some certainty around abstract method calls. Consider, for instance, that we need to call one of many possible methods for a package we do not control. Each of these methods has the same arguments, but the potential values change depending on the function. We can write canonical_args arg specs for each of these methods, allowing us some clarity as to what each argument needs to be (types, values, etc.) when we execute dynamically:

{
        "args": [
                {
                        "name": "argument1",
                        "type": int,
                        "values": "range(0, 15)"
                },
                {
                        "name": "argument2",
                        "type": "one([int, float, str])",
                        "values": {
                                "int": ">0",
                                "float": ">0",
                                "str": ["A", "B", "C"]
                        }
                }
        ],
        "kwargs": {
                "loss_function": {
                        "type": str,
                        "values": ["quadratic", "0-1"]
                }
        }
}

We can associate this spec with a method, either by registering it (if we do not control the method source):

from canonical_args import register_spec

# associates the spec to the method
register_spec(somemethod, spec)

# method instance method returns the registered spec
print somemethod.get_spec()

or by decorating a method, if we do control it (let’s say for a dynamically imported method handler sub-method).

from canonical_args import argspec

@arg_spec(spec, register=True)
def ourmethod(argument1, argument2, loss_function="quadratic"):
        pass

print ourmethod.get_spec()

This could potentially be of great use to dynamically generate frontend code with type and value-checking code. The specs themselves could be stored in a file or database, allowing for fully dynamic method calls:

from canonical_args import check_args
import pymongo

conn = pymongo.MongoClient("localhost", 27017)

def handle(message_type, *args, **kwargs):
        spec = conn.somedatabase.arg_specs.find_one(
                {"message_type": message_type})
        subhandler = conn.somedatabase.handlers.find_one(
                {"message_type": message_type})

        # use canonical_args to check the unknown arguments
        # against the retrieved spec. will raise AssertionError
        # if fails.
        check_args(spec, *args, **kwargs)

        # if no errors raised, fire the retrieved handler method
        return subhandler(*args, **kwargs)

def get_handler_spec(message_type):
        """
        get the arg spec without executing the function. can
        be used at front end (eg. HTML) for generating an
        appropriate form for method calls.
        """
        return conn.somedatabase.handlers.find_one(
                {"message_type": message_type})

The code above does not register the spec directly to the subhandler method, as it may not always be desirable to do so. The choice is yours.

pip install canonical_args

git clone https://github.com/shonteag/canonical_args
nosetests