The basic rule is that you need to decode your strings before you apply text operations (like uc) to them, and then encode them when you do IO with them (ie print to STDOUT or a file).

It also seems you're coming up with an extra-complicated way of writing ucfirst.

So a piece of code that handles accents in UTF-8 input correctly could look like this:

# make sure that strings coming from STDIN are decoded:
binmode STDIN, ':encoding(UTF-8)';

# make sure that strings written to STDOUT are encoded:
binmode STDOUT, ':encoding(UTF-8)';

my $line = <STDIN>;
$line = ucfirst $line;

print $line;
[download]

Please read the longer introduction for more background and information.

Hi Moritz,

Thanks for this. I feel I'm pretty much doing as you suggest. However, I tried the link you suggested and copied some code (from about the middle of the page). Changed it a little and I get the following:

#!/usr/bin/perl
use warnings;
use strict;
use Encode qw(encode decode);

my $enc = 'utf-8'; # This script is stored as UTF-8
my $str = "úlcera\n";

# Byte strings:
print ucfirst $str; # prints 'úlcera', ucfirst didn't have any effect.

# text strings::
my $text_str = decode($enc, $str);
$text_str = ucfirst $text_str;
print encode($enc, $text_str); # prints '?lcera', ucfirst as specified
+.
[download]

Where the capitalised "Ú" is displayed as a white square symbol under Windows or a question mark symbol under Linux.

Do I need to configure anything on my machine to make this work?

Regards

Steve

Did you read the section Testing your Environment? Maybe your terminal isn't configured to use UTF-8, and your script isn't stored in UTF-8 either.

I don#t think that the windows console is ever configured by default to work with UTF-8.

Perl 6 - second systems done right

Somebody wrote:

#!/usr/bin/perl
use warnings;
use strict;
use Encode qw(encode decode);

my $enc = 'utf-8'; # This script is stored as UTF-8
my $str = "úlcera\n";

# Byte strings:
print ucfirst $str; # prints 'úlcera', ucfirst didn't have any effect.
[download]

Whoa, that’s never going to work! And you should (very very very almost) not ever need to be calling encode/decode yourself, either.

Honest, this is really very easy. Watch:

use utf8;
use strict;
use warnings;
use warnings FATAL => "utf8";
use feature "unicode_strings"; # or use v5.12 or superior
use open qw(:std :utf8);

print ucfirst("úlcera\n");
[download]

...very most assuredly does indeed print out Úlcera. Don’t go by appearances: trust only the numbers. Thus:

$ perl ultstertest
Úlcera

$ perl ulstertest | uniquote -x
\x{DA}lcera

$ perl ulstertest | uniquote -v
\N{LATIN CAPITAL LETTER U WITH ACUTE}lcera

$ perl ulstertest | uniquote -b
\xC3\x9Alcera
[download]

The outer pair of tests above risk only confusion; it is the inner pair that are wholly dispositive and convincing: trust the output of uniquote -v and uniquote -x to give you something you can actually read and depend on.

Like I said, just play it by the numbers.

--tom