I have an expression that should either be wrapped with quotation marks from both sides or not at all, and I want to create a regex to capture this expression.
The point is, I don't want to duplicate the expression.
For example, if the expression is any single letter in english then the duplicated regex would be
^[a-z]|'[a-z]'$
Now both a and 'a' would be accepted.
Is there a way to define this regex without duplicating the expression?
How about:
^('?)[a-z]\1$
In group 1, match ' or nothing. And after the [a-z] part, match what we matched in group 1 (\1). So either:
', in which case \1 matches ' --> ^'[a-z]'$, or;\1 matches nothing --> ^[a-z]$Therefore, this matches the same things as ^([a-z]|'[a-z]')$..
What about this?
^(?:'|(?=.*[^']$))[a-z]'?$
Actually this is broken, as it also matches 'a. This is better, but a bit longer too:
^(?:'(?=.*')|(?=[^']+$))[a-z]'?$
Explanation of this second, working version:
[a-z]'?$ matches the letter optionally followed by a ' at the end of the line. Whether this single quote is there or not is governed by the other part of the regex^(?:'(?=.*')|(?=[^']+$)) matches a ' if the line also ends with ', or just a bunch of non-single quotes (well, in reality just one)
(?:…) is a non-capturing group'(?=.*') is the alternative that matches and consumes the leading ' if the line end with a ' too; the reason why (?=.*') and not (?=.*'$) is just that the $ is already imposed by the other piece of the regex(?=[^']+$) doesn't consume anything, but only asserts that the line does not have single quotesAnother option could asserting that the string does not consists of a single ' using a single negative lookahead.
^(?![^'\r\n]*'[^'\r\n]*$)'?[a-z]'?$
Explanation
^ Start of string(?! Negative lookahead, assert what from this position at the right is not
[^'\r\n]*'[^'\r\n]*$ Match a single occurrence of ' using a negated character class) Close lookahead'?[a-z]'? Match [a-z] between optional '$ End of string