The actual file joining would be easier with the (unix) shell (or the windows equivalent of same, available from cygwin, ATT Research Labs, etc):

cat /some/full/path/file.abc.part*.txt > /some/full/path/file.abc.MERG
+ED.txt
[download]

Although that could go haywire if the sequencing is important and the file names don't sort "naturally" into the correct order -- e.g. if you have names like:

file.abc.part-1.txt
file.abc.part-10.txt
file.abc.part-11.txt
file.abc.part-2.txt
file.abc.part-3.txt
...
[download]

If that's how the file names go, you'll want to use Perl to sort things properly. How about:

#!/usr/bin/perl

use strict;

die "Usage: $0 /full/path/to/data/dir\n"
    unless ( @ARGV == 1 and -d $ARGV[0] );

my $basedir = shift;
chdir $basedir or die "chdir $basedir failed: $!";

my @parts = <*.part-*.txt>;

# or:
# opendir D, ".";
# my @parts = grep /\.part-\d+\./, readdir D;
# closedir D;

my %merged;

for my $part ( sort { my ($x) = ($a=~/part-(\d+)/);
                      my ($y) = ($b=~/part-(\d+)/);
                      $x <=> $y
                    } @parts )
{
    my ( $merge_key ) = ( $part =~ /^(.*)\.part-\d/ );
    push @{$merged{$merge_key}}, $part; 
}

for my $mrgfile ( sort keys %merged )
{
    open( OUT, ">", "$mrgfile.MERGED.txt" )
        or die "$mrgfile.MERGED.txt: $!";

    local $/;
    for my $partfile ( @{$merged{$mrgfile}} ) {
        $_ = do { open( I, $partfile ); <I> }
        close I;
        print OUT;
    }
    close OUT;
}
[download]

(not tested, but should be pretty close to what you want)