reverse engineer a filehandle

smile4me has asked for the wisdom of the Perl Monks concerning the following question:

According to http://perldoc.perl.org/perlopentut.html: "Indirect filehandles also make it easy to pass filehandles to and return filehandles from subroutines."

Okay, so how do I find out which file this filehandle references? If the file name is dynamically generated, and all I have is a filehandle, how do I know which file is being written into?

For example:

#get a filehandle 
$fh = get_fh_from_file( generate_filename() ); 

if ( print_data( $fh, $data ) ) {
  say "data printed to file: $fh"; 
}
## got: data printed to file: *TRSC::FileUtils::FH;
## expected: '/path/to/file'
[download]

Thanks,
Steve

Comment on reverse engineer a filehandle Download Code

Replies are listed 'Best First'.
Re: reverse engineer a filehandle by tobyink (Canon) on Dec 09, 2011 at 23:17 UTC
As the first answer suggested, your best bet is to make sure you store the file name at the same time you open the filehandle. Failing that, this works on Linux, but is very non-portable: `open my $fh, '/etc/motd'; my $fd = fileno($fh); print readlink("/proc/$$/fd/$fd");` [download]	[reply] [d/l]
Re^2: reverse engineer a filehandle by smile4me (Beadle) on Dec 10, 2011 at 18:07 UTC
Thanks! the "readlink("/proc/$$/fd/$fd");" is what I was looking for.	[reply]
Re: reverse engineer a filehandle by TJPride (Pilgrim) on Dec 09, 2011 at 22:55 UTC
Best way is to store a hash of both the filehandle and its name: `use strict; use warnings; use Data::Dumper; sub openRandom { my $fh = { name => (int rand 1000) . '.txt' }; return $fh if open($fh->{'handle'}, ">$fh->{'name'}"); } print Dumper(openRandom());` [download]	[reply] [d/l]
Re: reverse engineer a filehandle by JavaFan (Canon) on Dec 10, 2011 at 18:47 UTC
"Indirect filehandles also make it easy to pass filehandles to and return filehandles from subroutines." Okay, so how do I find out which file this filehandle references? They way you pose the question makes me think the latter should follow from the former. I think how easy it is to pass around filehandles is complete unrelated to finding out file names. Do note you question may not have a unique answer: A "file"handle may actually read or write from a socket. Or a pipe. The file handle may be STDIN, STDOUT or STDERR. Or it may be some other redirection. Even if the file handle accesses a real file, the file may not have a name. Or the file may have more than one name.	[reply]