pls_newdb

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_newdb - Creates a new PLWeb database, and deletes the old one
#
#  Darren Hardy, University of Colorado - Boulder, July 1995
#
#  pls_newdb,v 1.1 1995/07/26 18:22:33 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 != 2);

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

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

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

&delete_prev_dbase();
&show_dbase();
&create_dbase();

exit 0; 					## END OF PROGRAM ##

sub usage {
	print STDERR "Usage: pls_newdb bname dbgroup plsroot\n";
	exit 1;
}

# delete the old PLWeb database $bname in $dbgroup
sub delete_prev_dbase {
	if (-e "${dbdir}/${bname}.def") {
		print "Deleting previous database for $bname.\n" if ($debug);
		&do_system("rm -f $dbdir/${bname}*");
	} else {
		print "No previous database for $bname.\n" if ($debug);
	}
	return if (! -d $dbgroupdir);
	$tmpfile = "/tmp/pls_newdb.$$";

	# remove references from database.tab and dblist.cpl
	foreach $f ("database.tab", "dblist.cpl") {
		next if (! -e "$dbgroupdir/$f");
		&do_system("cp $dbgroupdir/$f $tmpfile");
		&do_system("rm -f $dbgroupdir/$f");
		&do_system("grep -v $bname $tmpfile > $dbgroupdir/$f");
		unlink($tmpfile);
	}
}

sub show_dbase {
	# create configuration files for dbgroup
	open(OUT, ">>$dbgroupdir/database.tab") || 
		die "$dbgroupdir/database.tab: $!\n";
	print OUT "Show $bname\n";
	close(OUT);
}


sub create_dbase {
	print "Creating database for $bname\n" if ($debug);
	mkdir("$dbdir", 0755) if (! -d "$dbdir");
	&do_system("touch $dbdir/$bname.def");
	&do_system("cd $dbdir; plcreate -v -g $dbgroup $bname");
}

sub do_system {
	local($cmd) = @_;
	print "DOING: $cmd\n" if ($debug);
	return system($cmd);
}