A) eval isn't supposed to capture those sort of errors.
B) if you read the friendly man pages it mentions that backticks sets the $! error variable, so you should be checking that.
C) Why are you bothering to fork a new process and execute du? What's wrong with the built in perl operators for determing file size? Like -s? | [reply]
[d/l] |
i understand what you're saying.
my $str=-s "upload.log";
print $str;
How can i calculate the size for multiple files without looping like,
i have upload1.log, upload2.log, upload3.log.
how can i get the sizes without looping ?.
How can i use upload* pattern in the -s option ?. I tried but it's not return anything.
Can you please ?
- kulls
| [reply]
[d/l] |
@size = map { -s $_ } glob("upload*.log");
| [reply]
[d/l] |
As I've said before, using perl as a shell script language is not really the best use of perl, nor your CPU. Especially when the pure-perl solutions can be just as fast (or faster) and easier to get right.
Check out cog's module, Filesys::DiskUsage. It can do this much more simply, and much more perlishly.
use Filesys::DiskUsage qw/du/;
my $fullpath = '/tmp/dir*';
my $sum = du({'sector-size' => 4096, 'symlink-size' => 4096}, glob $fu
+llpath);
I'm using a size of 4096 above as that seems to be a normal size nowadays for some reason. If you want the individual entries' sizes, you can also pass in 'make-hash' => 1 as an option in the hashref of options. Then the answers should come back in a hash where the key is the file/directory, and the value is the size. | [reply]
[d/l]
[select] |
If your original "du" command is covering a lot of sibling directories that match "/tmp/dir*", and if each of those has a any amount of subdirectory structure, then I would probably stick with your backtick approach, but it can be very simple:
my @size = `du -sk /tmp/dir* | cut -f1`;
Note that in a list context (e.g. assigning to an array), the backticks will split on line breaks for you, and retain the line-feed at the end of each line (each element of the array).
If the "du" fails there, @size will be empty, but the error report from du will go to your STDERR. If you want to catch the error report, and you are using a bourne-style shell (as opposed to a csh-style shell), you can redirect the shell's stderr, as explained at length in the description of "qx" (backtick opertator) in "perldoc perlop":
my @size = `du -sk /tmp/dir* 2>&1 | cut -f1`;
This will put the error message as the sole element of @size. Then again, if you somehow get warnings on some files/paths, like "permission denied", these will be mixed in with size numbers for paths that had no problems -- which makes @size kind of a mess... So you could redirect the shell's stderr to a file (if you want to read it later from the perl script) or just redirect to /dev/null (if you don't need to read it). | [reply]
[d/l]
[select] |
| [reply] |