perl perl_program.pl readline() on unopened filehandle at perl_program.pl line xx (#1) (W unopened) An I/O operation was attempted on a filehandle that was never initialized. You need to do an open(), a sysopen(), or a socket() call, or call a constructor from the FileHandle package. GLOB(0x5fc7210)ARRAY(0x603a130)

line xx: is $hash{$keyfile1} = \@arrayfile1;

You're confusing the filehandle with the record being read.

Using more code than is really necessary to demonstrate the point:

open my $filehandle, ...

while (my $record = <$filehandle>) {
    chomp $record;  # Remove the newline
    # Do stuff with the record text, e.g.
    my @fields = split /$field_separator/ => $record;
    ...
}
[download]

That would normally be coded more succinctly as:

open my $filehandle, ...

while (<$filehandle>) {
    chomp;  # Remove the newline
    # Do stuff with the record text, e.g.
    my @fields = split /$field_separator/;
    ...
}
[download]

In the second code fragment, all instances of $record (from the first fragment) are now $_, which (being the default for chomp, split, and many other functions) can be omitted.

And, just in case it wasn't obvious, that was for general enlightenment, not a recommendation to attempt your own version of Text::CSV. :-)

-- Ken

Why => in split /$field_separator/ => $record? To my eye that says paired item, but there it's just a plain ol' comma and would be clearer as:

    my @fields = split /$field_separator/, $record;
[download]

Personally I prefer making the record variable explicit, but I'd write the while expression as:

while (defined (my $record = <$filehandle>)) {
[download]

which is a closer match to the magic you get with while (<$filehandle>) {.

True laziness is hard work