How to uppercase character within string 1

[pvollma]

I used a pattern to uppercase the first letter of each word, and then lowercase the rest, in a name and address file. This works fine, but, because these are names, there are special cases where I want a specific character within a word to be uppercase, such as the letter following a word that begins with "Mc". For example, if the name was "MCDONALD" in the original format, my first process changed this to "Mcdonald". What regex replacement can I use to search for words starting with "Mc" and then uppercase only the character after the "c"? Less importantly, a few of the names were input with a space between the "Mc" and the rest of the name, e.g., "MC DONALD". Is there a replacement pattern I can use to reliably merge them? Thanks!

 

[Rieekan]

Take a look at the function ucfirst();

- Rieekan

 

[TonyGroves]

How about:
[tt]
$name=ucfirst($name);
$name=~s/^Mc /Mc/;
$name=~s/^Mc(.)/Mc${\(uc $1)}/;
[/tt]

 

[rharsh]

This is probably a little easier to modify in case you need to add something other than 'Mc' to the list.

my @names = ('McDONALD', 'mac donald');
my @prefixes = ('Mc', 'Mac');

@names = map { "\L$_"} @names;    # LC the string
@names = map { "\u$_"} @names;    # UC first character

foreach my $prefix (@prefixes) {
    foreach (@names) {
        s/($prefix\s*)(\w)/$1\u$2/;
    }
}

 

[stevexff]

Whenever you are doing this kind of cleaning you have to make a trade off between the false positives and the false negatives. It's probably a good idea to put all the Mcs and Macs into a hash and print them out afterwards to get a list of unique names and counts. Then you can scan this list by eye to see if anything looks weird, and make a call on which is worse based on the counts. Had to do this once on 8 million customers. We got all the Mcs and Macs working perfectly, then discovered that we had created a 'MacHine Tools Ltd.' for a business customer...

 

[pvollma]

Since I was dealing with a line of text, with the Mc anywhere within the line, I used TonyGroves' idea with this modification:

$lineout =~ s/\bMc(.)/Mc${\(uc $1)}/g;

That worked just fine, and I checked through the text and found no anomolies. Thanks, Tony.

 

[mikevh]

$lineout =~ s/\bMc(.)/Mc${\(uc $1)}/g;

Interesting stuff on the RHS. You're passing $1 to the uc function, taking a reference to the string returned (\), dereferencing it as a scalar ($), and then concatting it to 'Mc'. Can't say I've ever seen that in an re before. I'm surprised it works without the /e modifier, but I guess the parens around (uc $1) force the function call to take place before the rest is eval'ed.

I think the following would do the same thing. (A little less complicated, maybe.)

$lineout =~ s/\b(Mc)(.)/$1.uc($2)/eg;