Using Java remove unbalanced/unpartnered parenthesis

问题

I want to remove all the "unpartnered" or unpaired parenthesises from a string.

exampleStr = back-pay) zyzhk1219(17) zyzhk1329 zyzhk1595(15) zyzhk1988 zyzhk2004 zyzhk2131) jswioj((testsjkldf

The expected "parenthesis balanced" string should be

back-pay zyzhk1219(17) zyzhk1329 zyzhk1595(15) zyzhk1988 zyzhk2004 zyzhk2131 jswiojtestsjkldf

I saw some ruby based solution on stackoverflow. But, couldn't find one I can use in java.

回答1:

How it might be done in pseudo-code:

initialize parenLevel = 0
for each character in string
    if char is ( increment parenLevel
    if char is )
        if parenLevel = 0, remove bad paren
        else, decrement parenLevel
next

initialize parenLevel = 0
for each character in string in reverse
    if char is ) increment parenLevel
    if char is (
        if parenLevel = 0, remove bad paren
        else, decrement parenLevel
next

How it might be implemented in practice: http://ideone.com/K3s0X

Sample result:

back-pay zyzhk1219(17) zyzhk1329 zyzhk1595(15) zyzhk1988 zyzhk2004 zyzhk2131 jswiojtestsjkldf

回答2:

This works correctly on your example string:

s = s.replaceAll("([^()]*(?:\\([^()]*\\)[^()]*)*)[()]", "$1");

Or, in more readable form:

(
  [^()]*          # some non-parentheses
  (?:
    \([^()]*\)    # a matched pair of parens
    [^()]*        # some more non-parens
  )*
)
[()]              # one unpaired paren

I'm assuming you don't want to allow nested pairs of parentheses. For example, in this string:

"abc(def(ghi)jkl)mno"

...the first ( and the last ) should be removed, leaving this:

"abcdef(ghi)jklmno"

If you do want to allow nesting, there's no pure regex solution--at least, not in Java.

来源：https://stackoverflow.com/questions/9898455/using-java-remove-unbalanced-unpartnered-parenthesis