Built-in function for chaining types to provide inside Bear.run() as annotation

[Makman2]

Originated from #1855, where multiple-types shall be supported. coala already has a way to define your own parsing function:

def bool_or_int(value):
    if value == "False":
        return False
    val = int(value)
    if val == 0:
        raise ValueError
    return val

def run(value: bool_or_int):
    pass

The task of this issue to implement a generic function that works like this:

def run(value: typechain(bool, int))
    pass

So it takes a sequence of types and tries to parse them in order. If nothing works, raise a ValueError.

[adtac]

Would it be enough to have a helper function that does this job inside the function instead of verifying at the function prototype definition?

Example:

def run(val_1, val_2): # unrestricted
    types.validate(val_1, [bool, int])
    types.validate(val_2, [string])

And types.validate raises ValueError if it doesn't match. I have no knowledge on how to validate types directly at the argument definition line.

[sils]

the thing is if we have it as the annotation then coala will auto convert
it and the user doesn't have to care any further, this way he'd have to
manually error out if something's inconveertible and stuff

Sincerely,

Lasse Schuirmann

contact@viperdev.io
http://viperdev.io/

2016-03-13 13:28 GMT+01:00 Adhityaa Chandrasekar notifications@github.com:

Would it be enough to have a helper function that does this job inside
the function instead of verifying at the function prototype definition?

Example:

def run(val_1, val_2): # unrestricted
types.validate(val_1, [bool, int])
types.validate(val_2, [string])

And types.validate raises ValueError if it doesn't match. I have no
knowledge on how to validate types directly at the argument definition line.

—
Reply to this email directly or view it on GitHub
#1859 (comment)
.

[adtac]

No, we can let the function raise an error if the variable doesn't match the type. The user doesn't need to do anything IMO.

[Makman2]

Having this inside the signature is way better since coala is then able to provide more help for that bear. Inside the body coala can't influence/see anything. Also your validation function would do pretty much the same, so having the extra signature advantage is good keep^^

[adtac]

Is a function prototype like

@enforce(int, float)
def run(arg1, arg2):
    ...

make the enforced variable types available to bear help?

[sils]

right now it's more:

def run(arg1: int, arg2: float):
    ...

and coala will always give the bear an int and a float, if conversion is not possible the bear won't be executed

[Uran198]

Where does this function have to live?

[sils]

I think StringConverter would be the right place

[Makman2]

@sils1297 what about directly inside Bear? since it's general purpose type-chaining^^

[sils]

it's general purpose, it's not at all tied to bears

[Uran198]

I want to get assigned :)