Under utf-8, the first 127 characters take up 1 byte, and higer numbered characters take a variable number of bytes (I'm not sure about the exact encoding, but IIRC it can take up to 4 bytes under the current unicode set). Under ascii and latin-1 all characters are encoded using 1 byte (8 bits). Under ucs-2 all characters take 2 bytes, and under ucs-4 all characters take 4 bytes.

More info: UTF8 ASCII as implemented in perl requires a second byte for codepoints 0x80 and higher, a third byte at 0x800, a fourth at 0x10000, a fifth at 0x200000, a sixth at 0x4000000 and a seventh at 0x80000000.

Note that this extends beyond the defined Unicode range, since we may store things other than Unicode characters in our strings - perl supports any integer that fits in a UV (32-bit or 64-bit unsigned integer, depending on your perl build) as a codepoint.

If I understand the code correctly (Perl_uvuni_to_utf8_flags() in utf8.c), higher codepoints (available only where perl is compiled with 64-bit integer support) use 7 bytes up to 0x1000000000, and a fixed 13 bytes for the rest.

Hugo

I guess I was not very clear with what I had written. In simple terms, what you are saying is correct and matter of fact thats exactly what I want.

Lets say I want to generate some random japanese characters which are of 2 bytes. Pl. note that I still don't know whether you can encode a japanese character in utf8 or utf16 or whatever the character set maybe.

Bottom line is, I dont really care about what language the characters get generated in. I shouldn't have used the term 'length'. What I meant was I want to generate a character string composed of characters of 2 bytes each, 4 bytes each etc.

Hope that clarifies things a bit.

BrowserUK, tall_man thanks for the response, but it doesn't quite solve my purpose. I hope this post adds a little more clarity to what I seek

Thanks everyone

Are you looking for something like this?

#! perl -slw
use strict;

## Adjust to suit your requirements
my %types = (
   lower => [ 'a'..'z' ],
   upper => [ 'A'..'Z' ],
   number=> [ '0'..'9' ],
   char  => [ 'a'..'z', 'A'..'Z', '0'..'9' ],
);

print join ' ', map{ 
    my( $type, $n ) = $_ =~ m[(\w+) length = (\d+)];
    join'', map{ 
        $types{ $type }[ rand @{ $types{ $type } } ] 
    } 1 .. $n;
} @ARGV;

__END__
P:\test>413131 "char length = 4" "number length = 3" "lower length = 6
+"
tAwY 828 xkppno

[12:35:08.56] P:\test>413131 "char length = 1" "number length = 10" "u
+pper length = 2"
j 3625181636 OB
[download]

print "char#\tlength1\tlength2\n";
for (32..500) {
    print;
    print "\t";
    print length chr;
    print "\t";
    {
        use bytes;
        print length chr;
        no bytes;
    };
    print "\n";
};
[download]

#!/usr/bin/perl -w
use strict;
 
my $len = 0;
if (@ARGV >= 1) {
  $len = $ARGV[0];
}
$len > 0 || die "Usage: genlen.pl number\n";
print "a" x $len;
[download]