#!/usr/bin/perl -w
# $Id: legalrpt.pl,v 1.1 2001/01/31 22:21:13 cpatti Exp $
# This script scans a specified directory tree for illegal filenames from
# the Perforce SCM's perspective.
# 12-23-00 -CAP

use File::Find;
use POSIX;
use Getopt::Long;

# Options
my $verbose='';
my $nuke='';

my $tdcnt=0;
my $bogcnt=0;
my $upcnt=0;
my $twincnt=0;
my $pristinecnt=0;

sub maybenuke
{
	my $victim = shift;
	if ($nuke)
	{
		if (-d $victim)
		{
			print LOGFILE "Can't unlink $victim as it's a directory\n";
			return;
		}
		
		if (unlink($victim) ne 1)
		{
			print LOGFILE "Can't unlink $victim\n";
			return;
		}
	}
}

##### MAIN PROGRAM ######
if (($#ARGV lt 1) || (! -d $ARGV[$#ARGV-1]))
{
	print STDERR "Usage: $0 [--verbose|-v] [--nuke|-n] <dir to search> <file to log to>\n";	
	exit;
}

GetOptions('verbose'=>\$verbose,'nuke'=>\$nuke);

$dir = $ARGV[$#ARGV-1];
$logfile = $ARGV[$#ARGV];
	
die "Could not open log: $logfile"
	unless open(LOGFILE,">$logfile");

if ($verbose)
{
	print LOGFILE "FILES CONTAINING UNPRINTABLES\n"; 
	print LOGFILE "-----------------------------\n\n";
}


sub perflegal
{
	$unprintable = (!POSIX::isprint($_));
	$gotthreedots = /\.\.\./;
	$boguschar = m{[@#*%/"']};
	$attictwin = (-f "$File::Find::dir/Attic/$_");

	
	if ($verbose)
	{
		if ($unprintable)
		{
			print LOGFILE "$File::Find::name - unprintable\n";
			$upcnt++;
			maybenuke($File::Find::name);			
		}
		elsif ($gotthreedots)
		{
			print LOGFILE "$File::Find::name - has three dots\n";
			$tdcnt++;
			maybenuke($File::Find::name);						
		}
		elsif ($boguschar)
		{
			print LOGFILE "$File::Find::name - has illegal printable chars\n";
			$bogcnt++;
			maybenuke($File::Find::name);						
		}
		elsif ($attictwin)
		{
			print LOGFILE "$File::Find::name has a twin in the attic!\n";
			$twincnt++;
			maybenuke("$File::Find::dir/Attic/$_");
		}
		else
		{
			# These files are OK :)
			$pristinecnt++;
		}
	}
	else 			# We're not in verbose mode.
	{
		if ($unprintable || $gotthreedots || $boguschar)
		{
			maybenuke($File::Find::name);						
			print LOGFILE "$File::Find::name\n";
		}
	}
}

find(\&perflegal,$dir);

if ($verbose)
{
	print LOGFILE "\n\nThere were:\n";
	print LOGFILE "$upcnt filenames containing unprintable characters\n";
	print LOGFILE "$tdcnt filenames containing the illegal sequence ...\n";
	print LOGFILE "$bogcnt filenames containing other illegal chars like #\n";
	print LOGFILE "$twincnt filenames have twins in the attic!\n";
	print LOGFILE "$pristinecnt filenames that are A-OK for import to Perforce\n";
}