It isn't really csv if it's "vertical" .. I think you should definitely go "horizontal" with it so that it's just a standard "table".

date,count,type,state,client,quality,program,duration
d1,c1,t1,s1,k1,q1,p1,d1
d2,c2,t2,s2,k2,q2,p2,d2
...
[download]

Now, this is easily parsable using Text::CSV or similar. But actually, i'd recommend DBD::CSV and use the power of SQL for your needs.

# get dates w/dups, and the dup count
SELECT `date`, count(*) as num_dups FROM foo GROUP BY `date` HAVING co
+unt(*) > 1

# get everything for one date
SELECT * FROM foo WHERE `date` = ?

# gather the unique combinations for each date.
my %dates;
my $rows = $dbh->selectall_arrayref("select * from foo", {Slice=>{}}, 
+); # (might not want to suck it all in at once, though)
foreach my $row (@$rows){
  push @{ $dates{ $row->{date} } }, $row;
}
print Dumper \%dates;
[download]

(i also recommend changing your column names so that they're not sql keywords like 'count' or 'date')