G'day Ineffectual,

I'm not actually familiar with IO::Uncompress::AnyUncompress so the following are just some troubleshooting hints as well as some guesswork based on the documentation (in particular, the OO Interface: Constructor Options section, referenced below).

The discrepancy between the line counts of 5,042,137 and 13,263 is obviously quite substantial. Perhaps some files are being read as a single string due to "Transparent => 1": see "... treat the whole file/buffer as a single data stream." in the docs.

Try moving the line counting code so it's counting individual files; maybe also check the size of the data being read. Something like this rough outline:

foreach ...
    ...
    my $linecount = 0;
    my $datasize = 0;
    while (...
        $datasize += length;
        ...
        ++$linecount;
    }
    close ...
    print $linecount, $datasize;
}
[download]

I see you have "AutoClose => 1"; however, the docs say "This option is only valid when the $input parameter is a filehandle.". Your $input is a filename, not a filehandle, so perhaps turn that off or remove it altogether (the default is 0).

Here's a few other things which probably (but not certainly) are unrelated to your current problem.

• Use the 3-argument form of open with a lexical filehandle (e.g. open my $in_fh, '-|', "gunzip ...).
• Heed the strong words of discouragement regarding use of Indirect Object Syntax in perlobj - Invoking Class Methods.
• Add a message to your die (2nd instance).
• Consider adding labels to your loops to clearly identify what next refers to.