argparse 是否支持多个独占参数? [英] Does argparse support multiple exclusive arguments?

问题描述

假设我有两组论点.您可以使用每个组中任意数量的参数，但不能在组之间混合参数.

Let's say I have two groups of arguments. You can use any number of arguments from each group but you cannot mix arguments between the groups.

有没有办法自定义 argparse 模块中的冲突参数?我试过使用 add_mutually_exclusive_group 方法，但这不是我想要的.

Is there a way to customize conflicting arguments in argparse module? I've tried using method add_mutually_exclusive_group but it's not what I'm looking for.

推荐答案

我提出了一个补丁(或者更确切地说是补丁)，可以让您测试参数的一般逻辑组合.http://bugs.python.org/issue11588.

I've proposed a patch (or rather patches) that would let you test for general logical combinations of arguments. http://bugs.python.org/issue11588.

我的想法的核心是在 parse_args 内部添加一个钩子，让用户测试所有参数的逻辑组合.此时它可以访问列表 seen 参数.这个列表在 parse_args 之外对你不可用(因此需要一个钩子).但是通过适当的defaults，您可以编写自己的使用args 命名空间的测试.

The core of my ideas there is to add a hook just inside parse_args that lets the user test for all logical combinations of arguments. At that it point it has access to a list seen arguments. This list is not available to you outside parse_args (hence the need for a hook). But with appropriate defaults, you can write your own tests that use the args Namespace.

实现通用 argparse 版本的困难包括:

The difficulties with implementing a general argparse version include:

a) 实现某种嵌套组(在您的情况下，多个 any 组嵌套在 xor 组中)

a) implementing some sort of nesting groups (in your case several any groups nesting within a xor group)

b) 在有意义的usage 行中显示这些组

b) displaying those groups in a meaningful usage line

现在，您最好的选择是使用子解析器(如果合适)实现您的问题，或者在解析后进行自己的测试.并编写自己的用法.

For now your best bet is either implement your problem with subparsers (if it fits), or do your own testing after parsing. And write your own usage.

这是一个可以在解析后应用于 args 命名空间的通用测试的草图

Here's a sketch of a generalizable test that could be applied to the args namespace after parsing

def present(a):
    # test whether an argument is 'present' or not
    # simple case, just check whether it is the default None or not
    if a is not None:
        return True
    else:
        return False

# sample namespace from parser
args = argparse.Namespace(x1='one',x2=None,y1=None,y2=3)

# a nested list defining the argument groups that need to be tested
groups=[[args.x1,args.x2],[args.y1,args.y2]]

# a test that applies 'any' test to the inner group
# and returns the number of groups that are 'present'
[any(present(a) for a in g) for g in groups].count(True)

如果count为0，则没有找到组，如果1找到一组，等等.我提到的hook在 bug issue 中进行相同类型的测试，只是使用不同的 present 测试.

If the count is 0, no groups are found, if 1 one group has been found, etc. The hook that I mentioned in the bug issue does the same sort of testing, just using a different present test.

如果计数>1，正常的互斥测试会反对.一个必需的组会反对 0 等.你也可以做类似

The normal mutually exclusive test would object if count >1. A required group would object to 0, etc. You could also do something like

if (present(args.x1) or present(args.x2)) and 
   (present(args.y1) or present(args.y2)): 
   parser.error('too many groups')

即.any,all,and,or 的某种组合.但是 count 是处理 xor 条件的好方法.

ie. some combination of any,all,and,or. But count is a nice way of handling the xor condition.

这篇关于argparse 是否支持多个独占参数?的文章就介绍到这了，希望我们推荐的答案对大家有所帮助，也希望大家多多支持IT屋！