匹配Java正则表达式中被可选组包围的内容

Question

I'm having trouble wrapping my head around how a particular Java regex should be written. The regex will be used in a sequence, and will match sections ending with / .

The problem is that using a simple split won't work because the text before the / can optionally be surrounded by ~ . If it is, then the text inside can match anything - including / and ~ . The key here is the ending ~/ , which is the only way to escape this 'anything goes' sequence if it begins with ~ .

Because the regex pattern will be used in a sequence (ie (xxx)+ ), I can't use ^ or $ for non-greedy matching.

Example matches:

• foo/
• ~foo~/
• ~foo/~/
• ~foo~~/
• ~foo/bar~/

and some that wouldn't match:

• foo~//
• ~foo~/bar~/
• ~foo/
• foo~/ (see edit 2)

Is there any way to do this without being redundant with my regexes? What would be the best way to think about matching this? Java doesn't have a conditional modifier ( ? ) so that complicated things in my head a bit more.

EDIT : After working on this in the meantime, the regex ((?:\\~())?)(((?!((?!\\2)/|\\~/)).)+)\\1/ gets close but #6 doesn't match.

EDIT 2 : After Steve pointed out that there is ambiguity, it became clear #6 shouldn't match.

Answer 1

I don't think that this is a solvable problem. From your givens, these are all acceptable:

~foo/~/
~foo/
foo~/

So, now, let's consider this combination:

~foo/foo~/

What happens here? We have combined the second example and the third example to create an instance of the first example. How do you suggest a correct splitting? As far as I can tell, there's no way to tell if we should be taking the entire expression as one or two valid expressions. Hence, I don't think it's possible to break it up accurately based on your listed restrictions.