dobarkod / typedecorator Goto Github PK

As the title says, I get a TypeError: non-void function didn't return a value when I have a function with the decorator @returns(Nullable(T)) return None (where T is any type).

Python provides a standard module for emitting warnings. This is a reasonable action to take for a type mismatch issue. Please make this one of the built in options for responding to type mismatches.

I understand a list containing a single element can be checked as the document states as follows:

A list containing a single element, requiring that the value be a list of values, all matching the type signature of that element. For example, a type signature specifying a list of integers would be [int].

I was wondering if it is possible to check for a list containing multiple elements. For example, if a function expects a 2D coordinate, it would require the following type:

[int, int]

More precisely, I would like to check for the following type:

typedecorator.Union((int, int), [int, int])

but that does not seem possible as I'm getting TypeError: Invalid type signature. If I change my code a bit like:

typedecorator.Union((int, int), list)

it seems working okay but that's not quite what I want.

Using Python version 2.7

When you use type decorator with a function, then try to call that function using multiprocessing's apply_async function, it results in a pickling error. Working example below, this was tested with a clean virtualenv with only typedecorator and the standard libraries installed.

from multiprocessing import Pool
from typedecorator import *

@returns(int)
@params(a=int,b=int)
def f(a,b):
    print "Function called"
    return a + b

def main():
    pool = Pool(processes=4)
    pool.apply_async(f,[1,2])
    pool.close()
    pool.join()

if __name__ == '__main__':
    main()

This results in:

Exception in thread Thread-2:
Traceback (most recent call last):
  File "/usr/lib64/python2.7/threading.py", line 811, in __bootstrap_inner
    self.run()
  File "/usr/lib64/python2.7/threading.py", line 764, in run
    self.__target(*self.__args, **self.__kwargs)
  File "/usr/lib64/python2.7/multiprocessing/pool.py", line 342, in _handle_tasks
    put(task)
PicklingError: Can't pickle <type 'function'>: attribute lookup __builtin__.function failed

Removing the type decorations for the function 'f' fixes the problem. So far I really like your solution though, and will be implementing it where possible when moving forward. Keep up the great work!

Hi. Sorry to bother you again. I have the following situation:

class Thing(object):
    @params(self=other, thing=Thing)
    def do_something(self, thing):
        pass

The thing=Thing gives me an "Unresolved reference error". Is there syntax for this already and I'm just missing it? Thanks.

This would be an extremely useful extension.

Should we support contracts?

It would be straightforward to support contracts by supporting custom callables in type signatures. For example:

@params(a=int, b=lambda b: b != 0)
def div(a, b):
    return a / b

However, I find that syntax clunky (for contracts), compared to something like:

@contract
def div(a, b):
    return a / b

@div.requires
def div(a, b):
    assert isinstance(a, int)
    assert b != 0

But this being typedecorator, not a full-blown contracts library, I'm hesitant to grow it in that direction (unless there's popular demand).

Hi,

I'm just starting to play with this package, and it looks fantastic. However, I can't find any information about making it work with a class method.

For example:

class Foo(object):
    @returns(int)
    @params(a=int, b=int):
    def add(self, a, b):
        return a + b

returns a TypeError. And if I change it to @params(self=self, a=int, b=int) I get a NameError (name 'self' if not defined).

So what is the proper way to annotate the method signature of an object method?

Thanks!

class Foo:
    @typed
    def validate(self, string: str): # TypeError: Annotation doesn't match function signature
        ...

It'd be great if this would not give an error, but just limit the checking to the annotated variables.

A little feature request here: it would be great if it was possible to validate a class passed via function parameter using issubclass().