#!/usr/local/bin/perl
#
#	Uploader
#	V1.1 pl 5 - 19 Feb 1997
#
#   Written by
#	Jouko Valta   (jopi@x2ftp.oulu.fi)
#
#
#   Get uploader's e-mail address from xferlog files.
#

$Usage="\
	Options:\
	  -h	Print Usage.\
	  -d	Debug\
	  -f	Force\
	 -L,-w	Specify output Logfile\
	-r,-nw	Read only. Do not update log file.\
	-u,-in	Find uploads only.\
	  -q	Quiet mode.\
	  -v	Verbose mode.\
	  -#	0 to 9 xferlog files to read.\
	  xxx	search pattern\n\n";

#
#	Return Values
#	    0	Completed OK.
#	    1	Error occurred while processing.
#
#    BUGS: Modifying the upload log may result in previous week's data
#	   left unread.
#
#    Performance: Creating uploadlog  ~ 6:20
#		  Updating uploadlog  ~ 0:__
#
#
#	V1.0 - 28 Jun 1996
#	    Print filename, transfer mode (a/b), and uploader's address.
#
#	V1.1 pl 2 - 15 Oct 1996
#	    Create separate incoming log file to reduce workload.
#	    Print real hostname if different from e-mail host.
#	    Print upload date as well.
#	    Added '-u' option to skip download attempts.
#
#	V1.1 pl 3 - 11 Jan 1997
#	    Logfile moved to /pc/log, except upon failure onto /var/tmp.
#
#	V1.1 pl 4 -  6 Feb 1997
#	    -d Debug option
#	    -f option to silence errors.
#	    -r option to disable log file update.
#	    Parse mail domains
#
#	V1.1 pl 5 -  6 Feb 1997
#	    Strip out '-' before login name
#	    Print file size
#


$EGREP =  "/bin/egrep";
$ZCAT  =  "/usr/local/bin/gunzip --stdout";

## Filenames

$LOGHOME = "/pc/log";

if (! -d $LOGHOME || ! -w $LOGHOME) {
    printf STDERR "Cannot access $LOGHOME. Please, check the setting or run\n create_dosadm_logs script.\n";
    $LOGHOME = "/var/tmp";
}

$XFERLOG	= "/ftp-service/adm/xferlog";
$UPLOADLOG	= $LOGHOME . "/uploadlog";
$TMPF		= "/var/tmp/upldtmp";


$InMode  = 0;
$Verbose = 1;
$Debug   = 0;
$Force   = 0;
$Readonly= 0;
$Weeks   = 9;


### Command line options
while ($_ = $ARGV[0], /^-/) {
	shift;
	last if /^--$/;

	if (/^-L|^-w/)	{ $UPLOADLOG = $ARGV[0]; shift; next; }
	if (/^-d/)	{ ++$Debug;	next; }
	if (/^-f/)	{ ++$Force;	next; }
	if (/^-r|^-nw/)	{ ++$Readonly;	next; }
	if (/^-q/)	{ $Verbose = 0;	next; }
	if (/^-v/)	{ ++$Verbose;	next; }
	if (/^-u|^-in/)	{ $InMode = 1;	next; }
	if (/^-(\d)/)	{ $Weeks = $1;	next; }

	if (/^-h/)	{  die $Usage; }
	if (/^-/)	{ print STDERR "Unsupported argument near $_\n";
			  die $Usage; }
}

if ($Weeks < 0 || $Weeks > 9) {
    printf "Number of weeks to update is invalid.\n";
}

# -----------------------------------------------------------------------------

# Update uploads log file

if (!$ARGV[0] || grep(/incoming/, $ARGV[0]) ) {  # Do we use the log at all ?

    @ustat = stat ($UPLOADLOG);
    @xstat = stat ($XFERLOG);
    @xstat1 = stat ($XFERLOG.1);


# Create

    if ( ! -e $UPLOADLOG || -z $UPLOADLOG ) {   # First time, create the log

	system ("rm -f $TMPF");
	printf STDERR "\nCreating logfile '$UPLOADLOG'.\n";

	for ($n = $Weeks; $n >= 2; --$n) {
	    $Verbose && printf STDERR "Reading $XFERLOG.$n.gz\n";
	    system ("$ZCAT $XFERLOG.$n.gz | $EGREP \"/etc/|incoming\" >> $TMPF");
	}  # for

	$Verbose && printf STDERR "Reading $XFERLOG.1\n";
	system ("$EGREP \"/etc/|incoming\" $XFERLOG.1 >> $TMPF");

	$Verbose && printf STDERR "Reading $XFERLOG\n";
	system ("$EGREP \"/etc/|incoming\" $XFERLOG   >> $TMPF");

	$Verbose && printf STDERR "Removing index fecthes...\n\n";
	system ("$EGREP -v \"incoming/README.uploader|/README.UPLOAD|/.message|/.cache|please_supply_the_desc\" $TMPF > $UPLOADLOG");

	system ("rm $TMPF");
	system ("chgrp -f ftpadm $UPLOADLOG");
	chmod 0660, $UPLOADLOG;
    }  # create

    else {

# Update uploadlog
	# Check if  xferlog.1 and the older files need to be updated.
	# All log after the last updater are read to a temp file.
	#
	# BUGS: Modifying the upload log may result in previous week's data
	#   left unread.

	system ("rm -f $TMPF");

	if ($xstat1[9] > $ustat[9]) {		# This file is never written to
	    printf STDERR "Updating $UPLOADLOG.\n";

	    for ($n = $Weeks; $n >= 2; --$n) {
		@xstat1 = stat ("$XFERLOG.$n.gz");
		if ($xstat1[9] > $ustat[9]) {
		    $Verbose && printf STDERR "Reading $XFERLOG.$n.gz\n";
		    system ("$ZCAT $XFERLOG.$n.gz | $EGREP \"/etc/|incoming\" >> $TMPF");
		} # if
	    }  # for

	    # we tested this one already.

	    $Verbose && printf STDERR "Reading $XFERLOG.1\n";
	    system ("$EGREP \"/etc/|incoming\" $XFERLOG.1 >> $TMPF");
	} # update xferlog.1 info

# Update current

	# Update new info from the current xferlog.
	# Always true when the previous section is true as well.

	if (($xstat[9] - $ustat[9]) > 14000) {		#  ~ 4 hours
	    $Verbose && printf STDERR "Reading $XFERLOG\n";
	    system ("$EGREP \"/etc/|incoming\" $XFERLOG   >> $TMPF");

print "DEBUG: reading completed.\n";

	    # Find the first new entry, and copy the rest to uploadlog.

#	    $lastold ="`/bin/tail -1 $UPLOADLOG`";
	    system ("/bin/tail -1 $UPLOADLOG > /tmp/last_upload");
	    open (inf, "/tmp/last_upload") || printf STDERR "Cannot stat $UPLOADLOG";
	    $lastold = <inf>;
	    close inf;
	    system ("/bin/rm -f /tmp/last_upload");
print "DEBUG: Last line '$lastold'";

	    open (inf, $TMPF);
	    open (outf, ">>$UPLOADLOG");
	    $count = 0;
	    while (<inf>) {
		(/$lastold/) && break;   # skip to the last old line
	    }
print "DEBUG: $_";
	    if ($_ ne $lastold) {
		printf STDERR "\n***ERROR: uploadlog is not continuous.\n";
		seek(inf, 0, 0);
	    }

	    printf outf "### Updated  %s", `/bin/date`;
	    while (<inf>) {
#print "DEBUG: $_";

		if (/incoming\/README.uploader|README.UPLOAD|.message|.cache|please_supply_the_desc/) {
		    next;
		}

		print outf $_;
		++$count;
	    }

	    system ("rm -f $TMPF");
	    $Verbose && printf STDERR "Read $count lines.\n";
	}  # update current
    }  # update


    # Syntax check. (Note: this block is enabled if !ARGV[0] )

    if (!$ARGV[0]) {
	printf STDERR "uploader: No file to check. Aborting.\n";
	exit 1;			# Make sure we don't accept empty field
    }
}  #  log created or updated


# ---------------------------------------------------------------------------

# Scan selected transfer log

if (!open (M, "$EGREP -i $ARGV[0] $UPLOADLOG |")) {
    print STDERR "$0: Cannot read uploader data log.\n";
    exit 1;
}

# Parse line

$found = 0;

while (<M>) {  #  date       year          host    size   file    ascii  inout        user    ftp    0   *
    if (/^(\S+ \S+ \S+ \S+\s+\S+)\s+\S+\s+(\S+)\s+(\S+)\s+(\S+)\s+(\S) _ (\S) \S\s+\-*(\S+)\s+\S+\s+\S+ \*$/) {
	$date = $1;
	$host = $2;
	$size = $3;
	$file = $4;
	$mode = $5;		# ascii / binary
	$inout= $6;
	$mail = $7;
	if ($InMode && $inout ne 'i') { next; }

	do parse_fields($mail);
    }
}

exit 0;


sub parse_fields {
    local $address = $mail;

    ($user, $edomain) = split /\@/, $address;

    $Debug && print "DEBUG: $user\t$edomain\t$host\n\n";

    if (!$edomain) {
	$edomain = $host;
	$Force || printf STDERR "NO HOST SPECIFIED. Assuming '$user\@$host'\n";
    }
    # $edomain expected as domain or parallel machine for $host
    #
    elsif (! ($host =~ /$edomain/) ) {
#    elsif (! ($host =~ /$edomain/) && $edomain ne $host) {

	($Verbose && !$Force) &&
	    printf STDERR "DIFFERENT HOST SPECIFIED. Real '$host'; claimed '$edomain'\n";
	printf "$file\t$mode\t$size\t$user\@$edomain\t$host\t\[$date\]\n";
	return;
    }

    printf "$file\t$mode\t$size\t$user\@$edomain\t*\t[$date]\n";
}