to get rid of the OS-grep dependency, try utf8::is_utf8

use v5.12.0;
use warnings;
#use utf8;
binmode STDOUT, ":utf8";

print my $s = "\N{LATIN SMALL LETTER S WITH CARON}i\N{LATIN SMALL LETT
+ER C WITH CARON}";
say " is text" if utf8::is_utf8($s);

for my $i (0 .. 2) {
    print my $c = substr $s, $i, 1;
    say " is text" if utf8::is_utf8($c);
}
[download]

šič is text
š is text
i is text
č is text

(in pre-tags b/c of PM restrictions)

Please note that even after commenting use utf8 out, $s is still automatically flagged as text

The pragma is optional here (not recommended), because

• the functions are universally available in Perl
• there is no wide-character in the source-code