Add possible solution to the "Reverse Parenthesis" problem

[zxul767]

I added this solution because I noticed the observations mentioned that you needed a stack and a queue to solve it. You don't really need a queue, as the proposed solution shows.

I also added a few extra test cases that test extreme cases, which is where most candidates typically trip up. I think they can give us a good chance to test their debugging skills after they've implemented a partially correct solution.

Please have a look and let me know your feedback (@oryws I'm adding you as a reviewer since you've recently interviewed someone using this problem)

• @oryws
• @bobmazanec
• @legtzp
• @sergiolagunaswize

Thanks

[zxul767]

@oryws @bobmazanec Adding some additional test cases is important, since there are some solutions that will pass simple tests but will definitely fail for more elaborate ones. For example, I just interviewed a candidate who wrote a rather concise, "partially correct" solution:

def revert_words_in_parenthesis(word):
    opening = []
    closing = []
    for idc, char in enumerate(word):
        if char == '(':
            opening.append(idc)
        if char == ')':
            closing.append(idc)
    for s, e in zip(reversed(opening), closing):
        word[s + 1: e] = reversed(word[s + 1: e])
    return ''.join(char for char in word if char != '(' and char != ')')

def test():
    examples = ['', '()' ,'(())(((())))', '(bar)', '((bar))', 'wi(ez)(((il)))(en)']
    expected = ['', '', '', 'rab', 'bar', 'wizeline']
    results = [revert_words_in_parenthesis(list(word)) for word in examples]
    for i in range(len(results)):
        print(results[i] == expected[i], results[i], expected[i])

    examples = ['foobarbaz', 'foo(bar)baz', 'foo(bar(baz))blim', '(tr(e)e)']
    results = [revert_words_in_parenthesis(list(word)) for word in examples]
    expected = ['foobarbaz', 'foorabbaz', 'foobazrabblim', 'eert'] 
    for i in range(len(results)):
        print(results[i] == expected[i], results[i], expected[i])

with the following output:

(True, '', '')
(True, '', '')
(True, '', '')
(True, 'rab', 'rab')
(True, 'bar', 'bar')
(False, 'wineilze', 'wizeline')
(True, 'foobarbaz', 'foobarbaz')
(True, 'foorabbaz', 'foorabbaz')
(True, 'foobazrabblim', 'foobazrabblim')
(True, 'eert', 'eert')

[legtzp]

@zxul767 detecting a malformed expression using a stack is clear for me, but could you please elaborate on how would you detect one using recursion? I like this problem :p

[zxul767]

One way to detect this in a recursive implementation would be to return from the recursive call when you find a closing parenthesis (or when you reach the end of the string) but without consuming the token, and then check --in the code the made the recursive call-- that you have a closing parenthesis. If you don't have a closing parenthesis, then you ran into a malformed expression.

Let me know if this makes sense

[sergiolagunaswize]

I'd like this to say something like: Two possible solutions for this problem are using recursion or using a stack, so we say that we are not married with only two approaches to this problem (I thought in one solution using sort, and I am not doubting there are even more)

[zxul767]

I agree, I'll update the statement to that.

[jcperez]

I added that problem to the original set of questions, but Bob added to this Github repo.

The solution says a queue because you need some kind of buffer to keep the current string. This problem is from the OMI contest back in 2004. Contestants can only use C or Pascal, that's why you cannot do something like reversed(word[s + 1: e]) in a "easy" way.

But... I'm ok with removing it from the text, just want to give more context :)

[bobmazanec]

Nice test cases! I added them to the problem in our HackerRank library.

[jcperez]

Should we merge this?

[zxul767]

@jcperez I think we can... @bobmazanec has set as a guideline that if there are no further comments after a while, it's probably time to go on and merge.

Who's supposed to merge though? The original submitter or the reviewers?

[bobmazanec]

Who's supposed to merge though? The original submitter or the reviewers?

I suggest the submitter — presumably they are "eager" for it.

Have arguments for/against and other another idea?
Please submit a PR against https://github.com/wizeline/wize-docs/blob/master/CONTRIBUTING.md 🤓

[jcperez]

@zxul767 Please merge then :) Thanks!