#!/usr/bin/perl -w
#
# agent2mbox
# version 1.0 11/01/2003
#
# this is a converter for Forte Agent files to plain mbox files
# this will not convert filters or attachs, just the messages
# i dont use forte agent and i only have some .dat files to
# test but it seens that is working fine...
# if you find any bug, please email me to the email below
#
#   agent2mbox@caravela.homelinux.net  
#
# it will converte one .dat file at the time, so to convert all
# files do something like:
# 	for i in *.DAT ; do agent2mbox $i ; done
# if all in just one dir or
# 	find . -name \*.dat -exec agent2mbox {} \;
# to search and convert in several folders
#
# this converter was based in my kmx2mbox converter script that
# converts Mail Warrior .kmx mailboxes, so probably there are things
# in this script that arent really used to convert from agent
# if i have more free time i will hunt then down and remove then
# but they shouldnt do any harm too, just making the script more
# complex
#
# this was made to help Luis Ferreira from pt.comp.so.linux to
# export his old agent archive messages to any format usable by linux
#
# by: Daniel Mota Leite  higuita@Gmx.net  - Portugal
# tmp website: http://caravela.homelinux.net/~higuita/agent2mbox
# (dynamic DNS site, maybe offline sometimes)
#
#	Good luck and enjoy
#
#----------------------------------------------------------------------------
#This program is protected by the GNU General Public License.  Please see
#http://www.gnu.org for information regarding the GNU General Public License.
#----------------------------------------------------------------------------
#

$nome=shift;
unless ($nome) {$nome=""};
$mbox_nova=$nome;
$mbox_nova=~ s/.dat/.mbox/i;
$mbox_nova=~ s/.DAT/.mbox/i;
$ok=0;
$n=0;
$i=0;
@email="";
$from="";
$date="";
#$subject="";
#$to="";
$from_ok=0;
$date_ok=0;
$date="Sun, 1 Jan 1998 00:00:00 -0000";
$i_old=0;

# protection against screw up files with no .dat extension
# without this the script will overwrite all files without .dat
unless ( ( $nome=~ /.DAT/i) || ( $nome=~ /.dat/i) ) {
	print " $nome No .dat extension... Aborting\n";
	exit;
}

print "Reading $nome\n\n";

# read the mailbox $nome
open(DATA,   "<  $nome") || die("ERROR - can't open datafile: $!");
	@msg=<DATA>;
close(DATA);

# save the new mailbox with the correct translated name
if (($nome=~ /.DAT/i) || ( $nome=~ /.dat/i) ){
open(DATA , ">$mbox_nova ") || die("ERROR - cant open datafile: $!"); 
}

foreach $linha(@msg) {
	$i_old++;   #i_old is used to find the from and date from old, imported kmx 

#print "$n $i_old $i $linha";  #debug

	# escape first junk lines
	if ( $linha =~ /\t0\r$/ ){
			unless ( $n == 0 ) {
				print "$n -> $from $date\n";
				&save;
			}
			$i=0;
	                $from_ok=0;
	                $from="";
	                @email="";
			$ok=1;
	                $date_ok=0;
	                $i_old=0;
			$n++;
			$headers=0;
	}
	elsif ( $ok==1 )  {
		#verifica as headers
		unless ($headers == 1) {
			&check_headers;
			if ( $linha =~ /^X-Agent-Group: / ) {
				#fim das headers
				$headers=0;
			}
		}
		# ignore the multipart header, but not any possible message
		# also ignore any line with a NULL character 
		unless ( $linha =~ /\x00multipart\/single\x00/ || $linha =~ /\x00/){
			$linha =~ s/\r//g;
			$email[$i]=$linha;
			$i++;
		}
	}

}

&save;
&exit;

#######################
###### Modules ########
#######################

sub save{
# the mbox format need a "\nFrom name date" at start of each email 
# the "converted" header is just to fix some incomplete headers
# i had only one very old kmx mailbox that had this problem
# this helps this emails and doesnt do anything for the others 
# ignore if there is no email (ie: the last field in the new MW format) 
	unless  ( $email[0] eq "" ){
		$head="";
#
# one old hack for ancient kmx mailbox, should be needed by agent2mbox
#
#		if ( $from_ok==0 && $email[0] ne "" ) {
#	                $from= $from_old;
#			$from=~ s/\n//;
#			$from=~ s/\r//;
#			$subject=~ s/\n//;
#			$to=~ s/\n//;
#
# 			#fake missing header for the old imported kmx
#			$head= "From: $from\nDate: $date\nSubject: $subject\nX-Mailer: Mail Warrior 2\nTo: $to\n";
#			$head=~ s/\r/ /g;
#		}
		print DATA "\nFrom $from $date\nConverted: agent2mbox\n";
		print DATA $head;
		print DATA @email;
		print DATA "\n";  #just as precaution if its missing the newline in the last line of the email
	}
}



#closes files, display sumary, have a nice day... 8)
sub exit{
	close(DATA);
	print "\nConverted $n emails to mbox $mbox_nova\n\n";
	exit;
}



# check the From, Date and Content-Type: multipart (there is no attach, 
# this can screw some email clients)
sub check_headers{
	if ($linha =~ /^From: /i && $from_ok==0) {
		$from= $linha;
		unless ( $from =~ /\@/ ) { $from=" bogus\@email.none"; }
		$from=~ s/[<>]//ig;
		$from=~ s/^From:.*\s+([^\s]+@[^\s]+\.[\w\d]+)(.*)?/$1/i;
		$from=~ s/\n//;
		$from_ok=1;
	}
	elsif ($linha =~ /^Date: /i && $date_ok==0) {
		$date= $linha;
		$date=~ s/^Date: //i;
		$date=~ s/\n//;
		$date=~ s/,//;
		$date_ok=1;
	}
# hide multipart 
# shouldnt be needed in agent converter
	elsif ( $linha=~ /^Content-Type: multipart/i) {
		$linha=~ s/^Content-Type: multipart/encode-/i;
	}
}