why doesn't argparse custom type function work with nargs?

Question

These lines work:

parser = argparse.ArgumentParser(prog='PROG')
parser.add_argument('foo', nargs='+', type=lambda x:x.split('/'))
parser.parse_args(['3/5', '4/6']) # output Namespace(foo=[['3', '5'], ['4', '6']])

But the following don't. Why?

The only difference is that this time the type caster uses a list comprehension to convert strings to integers.

parser = argparse.ArgumentParser(prog='PROG')
parser.add_argument('foo', nargs='+', type=lambda x:[[int(a), int(b)] for a,b in x.split('/')])
parser.parse_args(['3/5', '4/6']) # Raise error: argument foo: invalid <lambda> value: '3/5'

Answer 1

Basically, what you're trying to do here is

for a, b in '3/5'.split('/')...

which is a ValueError ("not enough values to unpack"). argparse hides actual errors and responds with a generic ArgumentError if there's anything wrong with your type function. You might want to use an actual def instead to debug it:

def test(x):
    try:
        return [[int(a), int(b)] for a,b in x.split('/')]
    except Exception as e:
        print(e)

parser = argparse.ArgumentParser(prog='PROG')
parser.add_argument('foo', nargs='+', type=test)
parser.parse_args(['3/5', '4/6'])