Convert your dates to a common, comparable representation. I typically use "seconds since 0000 Jan 1, 1970" aka "unix time" or "GMT". See the docs for Time::Local or try your hand at one of the date manipulation modules from CPAN (DateTime, Date::Manip, Date::Parse, etc.)

For date parsing and comparisons have a look at Date::Manip which is the Rolls Royce of date-modules, what it lacks in raw speed, it delivers in convenience.

Generally, you could do as follows:

1. open your archive file for adding to it
2. open the first of your files for reading
3. read this file line by line
4. check the date in each line and if older than one month write it to the archive file, else push it onto a stack for temporary storage
5. At the end of the file; close it and reopen it for writing (this will clear the contents of this file!); else goto step 3
6. now write your temporary stack to this file (hint: use foreach)
7. close the file
8. open your next file; if no more files remain: close your archive file; else goto step 3

I think the original poster's code only gets files within the directory, not subdirectories. If the subdirectories are needed, then you could do a recursive solution, like -

use File::Find;
@ARGV = ('.') unless @ARGV;  #start at . unless given 
                             # as command line argument
sub ProcessFiles {
    &CheckAndFixFile($File::Find::name);
}
find(\&ProcessFiles, @ARGV);  # find is recursive

sub CheckAndFixFile {
    my $filename = shift @_;
    #Put code here to check if $filename meets criteria, and delete li
+ne if needed.
}
[download]

The CheckAndFixFile subroutine is where you would use Date::Manip, or whichever Date module you like, to compare dates. It's easier than writing your own regular expression, as you did in your post, and it's code that's been tested and works.

Seems like you could use stat on each file and compare the mtime with time() - (86400 * 30) (or whatever). There is even File::stat which gives you a nice readable interface to the stat() attributes.

Phil

Update:I see you need times as well as dates. Try Date::Calc which handles times too.

find / -type f -newer /etc/lastbackup > /etc/backuplist
[download]

#!/usr/bin/perl -w

use Date::Calc qw(Delta_Days);

my $dirname = ".";

my (@files, @log_lines);
my %months = ( 'Jan' => 1,
               'Feb' => 2,
               'Mar' => 3,
               'Apr' => 4,
               'May' => 5,
               'Jun' => 6,
               'Jul' => 7,
               'Aug' => 8,
               'Sep' => 9,
               'Oct' => 10,
               'Nov' => 11,
               'Dec' => 12,
             );

# get the list of text files
opendir DIR, "$dirname";                                      
@dircontent = grep { /\.txt$/ } readdir(DIR);
closedir DIR;

# Compute today's date
my @today = (localtime(time))[3..5];
$today[1]++;
$today[2] += 1900;

# foreach text file, compute lines condition
foreach $filename (@dircontent) {
  open FILE, "<$filename";
  while(<FILE>) {
    my $line = $_;
    # see if line matches date format
    if($line =~ /^\[(\w{3}?) (\d{2}?)[^\]]*(\d{4}?)(.*)$/) {
      # assumed 'one month older' as being over 30 days.
      # if the line is older than 30 days, save it into an array
      # and write it later. If not, write it to a temp file
      (Delta_Days($3, $months{$1}, $2, $today[2], $today[1], $today[0]
+) > 30 )
        ? push @log_lines, $line : keep_line($line);
    } else {
      keep_line($line);
    }
  }
  close FILE;
  # The following lines restore the initial file but without old lines
  unlink $filename;
  rename "temp.tmp", $filename;
}

# archive old lines
open ARCH, ">>archive.arc";
foreach $line (@log_lines) {
  print ARCH "$line";
}
close ARCH;

# Just append a line to temp file
sub keep_line {
  my $line = shift;
  open  TEMP, ">>temp.tmp";
  print TEMP "$line";
  close TEMP;
}
[download]