If you run the code below (bare in mind it drops a table called testutf) does it work? Are the files uni.out and utf.out the same afterwards?

You can always change it to put more Asian characters in the initial string.

#!/usr/bin/perl -w
use strict;
use warnings;
use DBI qw(:utils);
use charnames ':full';
use Encode;

binmode(STDOUT, ":utf8");
my $str = "\x{263a}xxx" . chr(0x05d0) . "\N{ARABIC LETTER ALEF}";
print $str, "\n";
print join(" ", unpack("H*", $str)), "\n";
print "length(str) = ", length($str), "\n";
print "bytes::length(str) = ", bytes::length($str), "\n";

print "utf8::is_utf8 = ", utf8::is_utf8($str) ? 1 : 0, "\n";
print "data_string_desc: ", data_string_desc($str),"\n";

open OUT, ">uni.out";
binmode(OUT, ":utf8");
print OUT "$str\n";

my $dbh = DBI->connect("dbi:Oracle:XX", "XX", "XX",{oracharset => 'AL3
+2UTF8', ChopBlanks => 1});
$dbh->do("drop table testutf");
$dbh->do("create table testutf (a char(100))");
my $sth = $dbh->prepare("insert into testutf values (?)");
$sth->execute($str);
$sth = $dbh->prepare("select * from testutf");
$sth->execute;

my @row = $sth->fetchrow_array;
print "data_string_desc (after fetch): ", data_string_desc($row[0]),"\
+n";
print join(" ", unpack("H*", $row[0])), "\n";

open OUT, ">utf.out";
binmode (OUT, ":utf8");
print OUT $row[0];
close OUT;
[download]

We use unicode in Oracle all the time with japanese, chinese, arabic and many other languages and it works fine so long as the database and character set are set to AL32UTF8. Also bare in mind there are oracle database downloads (I think for Oracle XE) which are not labeled "international" and don't do unicode.

Lastly, what you see when you print unicode to your terminal is no indication of whether the data you retrieved is correct or not - that largely depends on how your terminal is set up and whether your system can display the characters you have output.