Odd regex problem

[jwes]

I wanted to change chapter numbers to title case, e.g. TWENTY to Twenty or TWENTY-THREE to Twenty-Three. I came up with these patterns
Find:

([\p{Lu}])([\p{Lu}]+)((-)([\p{Lu}])([\p{Lu}]*))?

Replace:

\1\L\2\4\5\L\6

It works for TWENTY, but it turns TWENTY-THREE to Twenty-three. As an experiment, I put '|'s in the replace string like this:

\1|\L\2|\4|\5|\L\6

and I got

T|wenty|-|t|hree

, so I can see it is being split correctly, but the T in three is being lowercased. Any explanations why?

[BetterRed]

No idea from a technical perspective, but "Twenty-three" is the de-facto standard for Chapter numbers expressed in words - hyphenated sentence case.

BR

[DiapDealer]

You need to use an \E to terminate an \L or \U otherwise they'll just keep going. So I would suggest an \E immediately after the capture group that represents the '-' in the replace expression (or before it).

\1\L\2\E\4\5\L\6

Or:

\1\L\2\4\E\5\L\6

I'm not at a computer where I can test the replace expression exactly.

I also think you don't need the brackets [] around each individual \p{Lu} instance.

[jwes]

Originally Posted by DiapDealer

You need to use an \E to terminate an \L or \U otherwise they'll just keep going. So I would suggest an \E immediately after the capture group that represents the '-' in the replace expression (or before it).

\1\L\2\E\4\5\L\6

Or:

\1\L\2\4\E\5\L\6

I'm not at a computer where I can test the replace expression exactly.

I also think you don't need the brackets [] around each individual \p{Lu} instance.

Thank you, that fixed it. Is there documention about \L, \U, and \E? I didn't see it in the user guide or scanning through the links.

[DiapDealer]

Originally Posted by jwes

Thank you, that fixed it. Is there documention about \L, \U, and \E? I didn't see it in the user guide or scanning through the links.

In our user guide? I doubt it. We only give the briefest of intros to regex. But I'm sure it's in any thorough documentation for the PCRE flavor of regex.

[DNSB]

Originally Posted by jwes

Thank you, that fixed it. Is there documention about \L, \U, and \E? I didn't see it in the user guide or scanning through the links.

You might want to try Perl-compatible Regular Expressions (PCRE) for the documentation and Regex101 for as the page says, build, test and debug regex which is especially handy since it allows you to play with multiple flavours of RegEx.