Regex: How to optionally match something at beginning or end, but not both?

Question

I have situation where in the regular expression is something like this: ^b?A+b?$

So b may match at the start of the string 0 or 1 times, and A must match one or more times. Again b may match at the end of the string 0 or 1 times.

Now I want to modify this regular expression in such way that it may match b either at the start or at the end of the string, but not both.

How do I do this?

Answer 1

^(b+A+b?|b?A+b+)$

why doesn't that work?

Answer 2

Try this:

^(bA+|A+b?)$

This allows a b at the start and then at least one A, or some As at the start and optionally a b at the end. This covers all the possibilities and is slightly faster than the accepted answer in the case that it doesn't match as only one of the two options needs to be tested (assuming A cannot begin with b).

Just to be different from the other answers here, if the expression A is quite complex but b is simple, then you might want to do it using a negative lookahead to avoid repeating the entire expression for A in your regular expression:

^(b(?!.*b$))?A+b?$

The second might be more readable if your A is complex, but if performance is an issue I'd recommend the first method.

Answer 3

Theres a nice "or" operator in regexps you can use.

^(b?A+|A+b?)$

Answer 4

Although the other answers have covered the regex portion, for future reference, I find it can be easier to visualise these problems as the state machine the regex represents;

^b?A+b?$

= b can repeat at the start of the expression 0 or 1 times, and A can repeat more than once. Again b can repeat at the end of the expression 0 or 1 times.

= 1 b, infinite a, 0 b
OR 0 b, infinite a, 0 b
OR 0 b, infinite a, 1 b

---

"Now I want to modify this regular expression in such way that it will either allow repetition of "b" at the start or at the end, but not both."

This is "^(bA+|A+b?)$". You can see the top branch is "bA+", then the bottom branch "A+b?".