Simplified test case:
I have a script that takes two arguments. The first is a list of integers. The second is a single integer that must be contained in the first argument's set of integers.
For example:
$ python argtest.py --valid_nums 1 2 3 --num 2
Should work, but:
$ python argtest.py --valid_nums 1 2 3 --num 4
Should not work, since num is not in valid_nums. However, I have had some difficulty (read: I suspect it is more trouble than it's worth, but I really want it to work) implementing this functionality.
Implementation attempt:
import argparse
parser = argparse.ArgumentParser()
parser.add_argument('--valid_nums',
type=int,
nargs='+')
args = parser.parse_args()
print "Numbers: ", args.valid_nums
parser.add_argument('--num',
type=int,
choices=args.valid_nums)
args = parser.parse_args()
print args
Actual output:
$ python argtesttest.py --valid_nums 1 2 3 --num 2
usage: argtesttest.py [-h] [--valid_nums VALID_NUMS [VALID_NUMS ...]]
argtesttest.py: error: unrecognized arguments: --num 2
Desired output:
$ python argtesttest.py --valid_nums 1 2 3 --num 2
Namespace(num=2, valid_nums=[1, 2, 3])
Now, (I think) the issue here is that I cannot add a new argument to the parser after I have called parse_args(), which yields the error about an unrecognized argument, but I can't think of a way around it. Is there any way to break up the processing of arguments such that it isn't all at once?
Obviously this would be pretty trivial to do if I only called parse_args() once and handled the container membership checking myself, but I'd like to get it working 'natively' using argparse's built-in error-checking.
Any ideas?
解决方案
you need to use parser.parse_known_args() as opposed to parser.parse_args(), or add all the arguments to parser before calling parse_args().
parse_args() needs to understand all the current arguments in sys.argv, which already contains --num at the time of first parsing, hence the exception.