#! /usr/bin/perl
#
# This script looks at all files in the directories that contain
# references to attachments and compares this to the file in the 
# Attach directory.  
#
$debug = 0;
$verbose = 0;
$outname = "";
$list_orphans = 0;
$move_orphans = 0;
$LISTFD = STDOUT;
$attachdir = "Attach";
$newdir    = "Attach-old2";
sub ListAttachments {
    my $filename = $_[0];
    print "Checking $filename...\n" if $verbose;
    open (FD, "<$filename" ) || die "Could not open $filename\n";
    # There are a number of formats for the lising of an attachment:
    # ^Content-Disposition: attachment;[\s\n]filename="?.*"?
    #    Note the location of the allowed newline, and the optional quotes
    #    May also have a : followed by other data, e.g.,
    #    : CHEP summary.doc: 00000001,1fe.....
    #    (but this may be an invalid name that does not have an
    #    attachment?)
    # This should match an entry of the form
    # ^Attachment Converted: "?name"?
    #   Note that the quotes are optional
    while (<FD>) {
	if (/^Attachment [Cc]onverted:\s+(\S.*)/) {
	    $candidate = $1;
	    if ($candidate =~ /\"([^\"]*)\"/) {
		$attachment = $1;
	    }
	    else {
		$attachment = $candidate;
	    }
	    print "$attachment\n" if $verbose;
	    # Remove any directories
	    $attachment =~ /.*\\([^\\]*)/;
	    $shortname = $1;
	    print $LISTFD "$shortname\n" if $verbose;
	    $allfiles{$shortname} = 1;
	}
# 	elsif (/Content-[Dd]isposition:\s+attachment;\s+filename=\"?(.*)\"?/) {
# 	    $attachment = $1;
# 	    print "$attachment\n" if $verbose;
# 	    # Remove any directories
# 	    $attachment =~ /.*\\([^\\]*)/;
# 	    $shortname = $1;
# 	    print $LISTFD "$shortname\n" if $verbose;
# 	    $allfiles{$shortname} = 1;
# 	}
    }
}

sub ProcessDir {
    my $dirname = $_[0];
    print "Processing directory $dirname\n" if $verbose;
    opendir( DIR, $dirname ) || die "Could not open directory $dirname\n";
    my @files = readdir( DIR );
    closedir DIR;
    foreach my $file (@files) {
	if ($file eq "." || $file eq "..") { next; }
	if (-d "$dirname/$file") {
	    &ProcessDir( "$dirname/$file" );
	}
	elsif ($file =~ /\.mbx/) {
            print "# file $dirname/$file\n" if $debug;
	    &ListAttachments( "$dirname/$file" );
	}
    }
}

#
# -outfile=file-for-list -attach=dir -move -movedir=dir directory-to-check
# where -move moves orphans
foreach $_ (@ARGV) {
    if (/^-outfile=(.*)/) { 
	$outname = $1; 
	open( LISTFD, ">$outname" ) || die "could not open $outname\n";
	$LISTFD = LISTFD;
	next;
    }
    if (/^-attach=(.*)/) {
	$attachdir = $1;
	opendir (DIRIN,$attachdir);
	foreach $file (readdir(DIRIN)) {
	    $allfiles{$file} = 0;
	}
	closedir (DIRIN);
	$list_orphans = 1;
	next;
    }
    if (/^-movedir=(.*)/) {
	$newdir = $1;
	$move_orphans = 1;
	next;
    }
    # 
    # Need to replace this with a read of the $attachdir directory
#      if (/^-inlist=(.*)/) {
#  	$infile = $1;
#  	open (FDIN, "<$infile" ) || die "Cannot open $infile\n";
#  	$list_orphans = 1;
#  	while (<FDIN>) {
#  	    s/\r\n//;
#  	    $allfiles{$_} = 0;
#  	}
#  	next;
#      }
    if (/^-move/) {
	$move_orphans = 1;
	next;
    }
    if (/^-quiet/) {
	$verbose = 0;
	$list_orphans = 0;
	next;
    }
    $filename = $_;
    print "Looking at $filename\n" if $debug;
    if (-d $filename) {
	print "Entering directory $filename...\n" if $debug;
	&ProcessDir( $filename );
    }
    else {
        print "# file $filename\n" if $debug;
	&ListAttachments( $filename );
    }
}

if ($list_orphans) {
    my $totsize = 0;
    foreach $file (keys(%allfiles)) {
	if ($allfiles{$file} == 0) {
	    ($d,$d,$d,$d,$d,$d,$d,$size,$d,$d,$d,$d,$d) = stat "$attachdir/$file";
	    print "Orphan\t\"$file\"\t$size\n";
	    $totsize += $size;
	}
    }
    print "Total size of orphans = $totsize bytes\n";
} 
if ($move_orphans) {
    if (! -d $newdir) {
	mkdir $newdir || die "Cannot make $newdir";
    }
    foreach $file (keys(%allfiles)) {
	if ($allfiles{$file} == 0) {
	    if (-f "$attachdir/$file") {
		rename "$attachdir/$file","$newdir/$file";
	    }
	}
    }
}