How to handle calling functions from cli

[Rodrick Brown]

I have a bunch of sub routines that run independently to perform various system checks on my servers. I wanted to get an opinion on the following code I have about 25 independent checks and I'm adding the ability to disable certain checks that don't apply to certain hosts.


m = { 'a': 'checkDisks()',
'b': 'checkMemSize()',
'c': 'checkBondInterfaces()'
}

parser = argparse.ArgumentParser(description='Parse command line args.')
parser.add_argument('-x', action="store", dest="d")
r = parser.parse_args(sys.argv[1:])

runlist = [ c for c in m.keys() if c not in r.d ]
for runable in runlist:
eval(m[runable])

I'm using temp variable names for now until I find an approach I like.

Is this a good approach ? It doesn't look too pretty and to be honest feels awkward?

Sent from my iPhone

[Chris Rebert]

You should make use of the fact that functions are first-class objects
in Python:

m = { 'a': checkDisks,
'b': checkMemSize,
'c': checkBondInterfaces }
# …
for runable in runlist:
m[runable]()


Cheers,
Chris

[Chris Angelico]

On Sat, Feb 25, 2012 at 9:16 AM, Rodrick Brown <rodric...@gmail.com> wrote:
> m = { 'a': 'checkDisks()',
>          'b': 'checkMemSize()',
>          'c': 'checkBondInterfaces()'
>    }
>

>    runlist = [ c for c in m.keys() if c not in r.d ]
>    for runable in runlist:
>        eval(m[runable])

It's a reasonable technique. Does have the downside that your
functions will be called in an unpredictable order, though. If that's
a problem, replace the dictionary with a tuple of tuples (and then
just take off the .items() in the list comp).

I would be inclined to avoid eval, especially if none of your
functions need parameters. Just hold references to the functions
themselves:

checks = {
'a': checkDisks,
'b': checkMemSize,
'c': checkBondInterfaces, # note that this comma is perfectly
legal - all these lines can be structured identically
}

[func[option]() for option,func in checks.items() if option not in r.d]

ChrisA