my $days = 30;
for (glob '*') {
unlink if -f and -M > $days;
}
That uses the file test operators to check if each name rperesents an ordinary file, and to see if the file is more than 30 days old. Adjust $days (can be fractional) for age, and the argument of glob for path and any specializations like file extension.
| [reply]
[d/l] |
I don't know why, but I find -M and friends confusing, because they work the 'wrong' way for my brane, by making older files have higher values, whereas stat and friends use epoch time and newer files have higher values. So I'd use stat like below. There's a little more code, but it's easier for me to read:
unlink grep { (stat($_))[9] < time - ($days * 86400) } @files;
where 86400 is the number of seconds in a day, so we unlink all files from the @files array where their modification time earlier than $days days old. | [reply]
[d/l] |
...I find -M and friends confusing, because they work the 'wrong' way for my brane...
I agree, but for a different reason: the fact that calculations are done from script startup time can lead to unexpected results for long-running scripts, or when running under mod_perl.
That's why I personally prefer to use stat() as well: changing $^T is too much "action at a distance" to me.
On the other hand, being able to recognize 86400 as a synonym for a day, is a typicial sysadmin deformity to me ;-)
Liz
| [reply] |
There is no option in DOS to delete by date.
It's possible to do this in DOS (or at the Windows
command prompt, which I suspect is what you really
mean) by combining some things, but it's actually
easier to do it in pure Perl.
for (<C:\\newfiles\\*.htm>) {
unlink if (-M > $somenumberofdays);
}
This doesn't recursively traverse the directory
structure, but you can do that also using readdir
and/or the -d and/or -f file test operators.
With readdir, it's even possible to do this in
a cross-platform way without worrying about what
character is the directory separator on your
platform (backslash on DOS and Windows, slash on
POSIX systems, colon I think on older MacOSes,
and then there's VMS...)
Also, I think $somenumberofdays doesn't have to
be an integer, so you could delete files more
than (say) twelve hours old by setting it to 0.5,
unless I am mistaken. If you need to traverse
the directory tree and need more help with it,
post a reply here saying so.
Like I said, it's possible to
use backticks to shell out to COMMAND.COM and get
this done with standard MS command-line tools,
but it's the more difficult way to go. Since you
already have Perl on the system, use it.
$;=sub{$/};@;=map{my($a,$b)=($_,$;);$;=sub{$a.$b->()}}
split//,".rekcah lreP rehtona tsuJ";$\=$ ;->();print$/
| [reply]
[d/l]
[select] |
The easiest way is to use find2perl. If there is something available to you, what is the point to reinvent it ;-?
If you are familiar with unix, this makes you feel back to home. It is similar to unix find command, and you will love it.
It is a tool coming with Perl, you can find it under your perl bin directory. On Windows, it is a .bat file. Even on unix, you can use it, as find2perl is faster than the find command.
| [reply] |
Hi
you could even look into something as simple as:
get absolute path of filename input
use the stat command and slice the array subscript containg the file creation date.
ctime File creation time (C lang. time_t value)
then you could just remove all leading and trailing spaces from the date input and that received from stat.
and do a
if( $date =~ m/$date1/)
{ unlink file; }
HTH | [reply]
[d/l] |
try the lstat(file)[10] function to get file creation date?
you mean there's any easier way?
Edited by castaway - added code tags. | [reply]
[d/l] |