Question

I am trying to build a regex that should match for example.

b
abab
babab

but not

bb
babb
aaaba
abaaba

At the moment I have a(b)|b(a) and it is working for abab. I am missing the first and last letter, for example b or babab.

So I need to specify a alone or b alone or a letter at the end of a word (if the letter before it isn’t itself). But I can’t figure out how to do that.

I am using http://www.rexv.org/ (Perl PCRE) to try it.

Thanks guys but i forgot to mention:
An empty string can also be matched,
and i can only use the following

* ? +

|

()

.

Thanks guys!,

i suppose it isn’t possible without being able to specify beginning and end of the string to work correctly at http://www.rexv.org/

Answer 1

Try something like this:

^((?:(?:ab)*a?)|(?:(?:ba)*b?))$

Explained:

^(                   # beginning of the string
    (?:
        (?:ab)*      # matches any repeating `ab` group
        a?           # group can optionally end with an `a`
    )
    |
    (?:
        (?:ba)*      # matches any repeating `ba` group
        b?           # group can optionally end with a `b`
    )
)$                   # end of the string

I include the subgroups as non-capturing with the leading (?: using a full-capturing group around the entire regex. This will make sure to return you only the full-strings that match instead of the noise of each sub-group.

The caveat to this approach is that an “empty” string will also match.

UPDATE (limited set of characters)
Your limited-set of characters will still work with my pattern above, however, we’ll need to drop the non-matching group portion (?:). The regex will end up as:

(((ab)*a?)|((ba)*b?))

The caveat mentioned above is that it will also match an empty string, however, this appears to be what you need so we can add that to the bonus-list!

A slight issue with the set of characters you’re allowed to use is that you aren’t allowed to use the ^ and $ characters which indicate the start and end of a string, respectively. The problem with this is that, any sub-pattern that is matched (regardless of the regex you use) will flag the input as valid. I’m assuming that this is accounted for though.