Storing open filehandles in a hash

jpfarmer has asked for the wisdom of the Perl Monks concerning the following question:

Recently, I was working on a program that takes one input file and splits it into n other files depending on the content of each line. I figured the most efficient way to do this would be to keep all of my filehandles in a hash, so I wrote the following:

while (<>){
  # do some stuff
  if (defined $fhs{$conn}){
       local *FH = $fhs{$conn};
       print FH $_;
  } else {
       local *FH;
       open(FH, '>', "conn-$conn") or die($!);
       print FH $_;
       $fhs{$conn} = \*FH{IO};
  }
}
[download]

I seem to be able to build up the file handles the way I expected, but by the time I get to the if block, the filehandle is closed. Is there a way I can keep it open? or is there a better way to do this?

Thanks!

Comment on Storing open filehandles in a hash Select or Download Code

Replies are listed 'Best First'.
Re: Storing open filehandles in a hash by holli (Abbot) on Aug 16, 2006 at 14:21 UTC
Avoid the globs and use IO::File instead. Then everything will work fine. `while (<>){ # do some stuff unless defined ( $fhs{$conn} ) { $fhs{$conn} = IO::File->new(); die $! unless $fhs{$conn}->open(">conn-$conn"); } $fhs{$conn}->print ($_); }` [download] Update: added code holli, /regexed monk/	[reply] [d/l]
Re: Storing open filehandles in a hash by ikegami (Patriarch) on Aug 16, 2006 at 14:37 UTC
`while (<>) { defined $fhs{$conn} or open($fhs{$conn}, '>', "conn-$conn") or die("Unable to create file conn-$conn: $!\n"); print { $fhs{$conn} } $_; }` [download]	[reply] [d/l]
Re^2: Storing open filehandles in a hash by kovacsbv (Novice) on Nov 20, 2008 at 14:59 UTC
Can you shed some light on what the braces around $fhs{$conn} do in this case? It must have something to do with the file handle and print string being space separated, but I've been wondering for a while why we need those if we're using anything more complex than $variable_name but don't otherwise.	[reply]
Re^3: Storing open filehandles in a hash by ikegami (Patriarch) on Nov 20, 2008 at 22:14 UTC
See the last paragraph of `print`'s documentation. To keep the parser sane, `print` is very limited as to what it can accept for the file handle argument. It accepts a scalar variable (`$fh`) or a bare word (`FH`), and that's probably it unless you wrap the expression in curlies. Inside the curlies can be any Perl expression, including the a hash lookup as desired by the OP. `print { $fh } ...; print $fh ...; # Shorthand print { FH } ...; print FH ...; # Same thing, but also works under strict 'vars'. print { FH } ...; print { $a[$i] } ...; print { $h{$k} } ...; print { $o->get_fh() } ...; print { $e ? STDERR : *STDOUT } ...; etc.` [download]	[reply] [d/l] [select]