Lets say you have a string where you want to ucfirst each word. You could do that with the following:

$string = s/\b(\w+)\b/ucfirst($1)/eg;
[download]

You could modify the above regexp to handle apostrophes and hyphens by using negative lookbehind:

$string = s/(?<![-'])\b(\w+)\b/ucfirst($1)/eg;
[download]

The above substitution regex will apply ucfirst to any cluster of "word" characters as long as they're not immediately preceeded by an apostrophe or a hyphen. That way you don't end up with Frank'S Place when you really want Frank's Place.

Negative lookbehind with a character class is the right tool, as opposed to positive lookbehind for a negated character class, because you only care that an apostrophe or hyphen doesn't appear before the 'word' cluster. A negated character class inside of a positive lookbehind would require that a "non-apostrophe, non-hyphen" exist in front of the word cluster, which is going to foul everything up. You don't want that. ;)

You can skip the /e modifier by using the \u escape. And you don't need to capture the whole word; the first letter will do.

perl -ple 's/\b(?<![-'])(\w)/\u$1/g'
[download]

-sauoq
"My two cents aren't worth a dime.";

#!/usr/local/bin/perl -w
use strict;

while (<DATA>) {
    s/(\w+)/\u\L$1/g;
    print "$_";
}
# or just a one liner -
# s/(\w+)/\u\L$1/g, print for <DATA>;

__DATA__
foo bar
ned KelLy
james bond
[download]

Foo Bar
Ned Kelly
James Bond
[download]

What you want is called title case.

Unfortunately, the function associated with it, ucfirst, only does its magic on the first letter of the string, not the first letter of each word.

Try

join ' ' , map ucfirst, split ' '
[download]

Like with upper case, the character escape for title case is \u. It has the same problem as ucfirst above.

Edit:

Never EVER use tr/a-z/A-Z/ for any uppercasing stuff. It doesn't play well with unicode. Not that you were going to, but it's always good to check.

I like your approach too though.

So we have a s/// (substitution) method, and a join / split / map method. As you mentioned, tr/// isn't a good candidate, but I couldn't resist at least tossing out there the thought of tr/// applied to the first letter of words by using pos along with m/\b[a-z]/g (to locate position of first letters of words) along with substr (as an lvalue to constrain the effects of tr///). ...it could work, but sounds like madness. ;)

TI(always)MTOWTDI



Dave

"If I had my life to live over again, I'd be a plumber." -- Albert Einstein

perldoc -q capitalize
[download]

Abigail