pls_delete

harvest是一个下载html网页得机器人

: # *-*-perl-*-*
	eval 'exec perl -S $0 "$@"'
	if $running_under_some_shell;  
#
#  pls_delete - Deletes SOIF summary objects from PLWeb database
#
#  Darren Hardy, University of Colorado - Boulder, July 1995
#
#  pls_delete,v 1.1 1995/07/26 18:22:31 hardy Exp
#
$ENV{'HARVEST_HOME'} = "/usr/local/harvest" if (!defined($ENV{'HARVEST_HOME'}));
unshift(@INC, "$ENV{'HARVEST_HOME'}/lib");      # use local files 

$debug = 0;

&usage() if ($#ARGV < 3);

$bname = shift(@ARGV);
$dbgroup = shift(@ARGV);
$plsroot = shift(@ARGV);
$nreorg = shift(@ARGV);

$dbgroupdir = "$plsroot/DBGROUPS/$dbgroup";
$dbdir = "$plsroot/DBGROUPS/$dbgroup/databases/$bname";

die "plsroot: $plsroot: $!\n" if (! -d $plsroot);
die "dbgroup: $dbgroupdir: $!\n" if (! -d $dbgroupdir);

$n = 0;
while ($f = shift(@ARGV)) {
	warn "$f: $!\n", next if (! -r $f);
	&delete_file($f);
	$n++;
	if ($n > $nreorg) {
		&reorg();
		$n = 0;
	}
}

exit 0; 					## END OF PROGRAM ##

sub usage {
	print STDERR "Usage: pls_delete bname dbgroup plsroot nreorg file [file ...]\n";
	exit 1;
}

sub delete_file {
	local($f) = @_;
	system("plrmfile -v -g $dbgroup $dbdir/$bname $f");
}

sub reorg {
	print "Reorganization not implemented yet...\n";
}