#!/bin/perl

use File::Path;

# Constants
my $from="Distribute message from:";
my $to="Distributed:";

if ( $#ARGV != 0 ) {
  usage();
  exit 1;
}

#
# Parse the command line argument
#
$dir = shift @ARGV;

if ( ! -e $dir ) {
  print "'$dir' does not exists. Exiting the script.\n";
  exit 1;
} elsif (! -d $dir ) {
  print "'$dir' is not a valid directory. Exiting the script.\n";
  exit 1;
}

my $outdir = "$dir/filtered";
my $logdir = "$dir/logdir";

if ( -e "$outdir" ) {
  if ( -f "$outdir" ) {
    print "'$outdir' is a file. Rename the file.\n";
    exit 0;
  } else {
    rmtree("$outdir", 0) || die "Could not delete '$outdir' $!\n";
  }
}

mkpath("$outdir") || die "Could not create '$outdir' $!\n";


if ( -e "$logdir" ) {
  if ( -f "$logdir" ) {
    print "'$logdir' is a file. Rename the file.\n";
    exit 0;
  } else {
    rmtree("$logdir", 0) || die "Could not delete '$logdir' $!\n";
  }
}

mkpath("$logdir") || die "Could not create '$logdir' $!\n";

opendir(DIR, "$dir") || die "Can't open directory '$dir' $!\n";
@files = readdir(DIR);
closedir(DIR);

foreach $file (@files) {
  print "Processing $dir/$file\n";
  if ( -f "$dir/$file") {
    open(FIN, "<$dir/$file");
    open(FOUT, ">$outdir/$file");
    open(LOG, ">$logdir/$file");

    my %map = ();
    my %lmap = ();

    while ($line = <FIN>) {
      chomp($line);
      doLog("Processing", $line);
      if ( $line =~ /^[0-9]/m) {
        my ($time, $data) = removetimestamp($line);
        my ($key, $value) = keyValue($data);
        if (defined $key) {
          if ( $value =~ /^$from/ ) {
            my $sender = getSenderName($value);
            if (defined $map{$key}) {
              $val = $map{$key};
              doLog("End of sender", $val);
              my $length = $val;
              $val = substr($val, 0, ($length - 1));
              print FOUT "$val\n";
              $map{$key} = "$sender,";
              $lmap{$key} = $line;
              doLog("Replacing", $sender);
            } else {
              doLog("New Entry", "$data");
              $map{$key} = "$sender,";
              $lmap{$key} = $line;
            }
          } elsif ( $value =~ /^$to/ ) {
            my $recipient = getRecipientName($value);
            if (defined $map{$key} ) {
              $val = $map{$key};
              doLog("Adding recipeint:", "$recipient to $val");
              $val .= "$recipient,";
              $map{$key} = $val;
            } else {
              doLog("Ignoring", $line);
            }
          } else {
            doLog("Ignoring", $line);
          }
        } 
      } else {
        doLog("Ignoring", $line);
      }
    }

    for $mkey ( keys %lmap ) {
       doLog("Incomplete", $lmap{$mkey});
    }

    close FIN;
    close FOUT;
    close LOG;
  }

  print "Processed $dir/$file\n";
}

sub removetimestamp() {
  my ($line) = @_;
  my $ind = index($line, " ");
  if ( $ind != "-1" ) {
    $time = substr($line, 0, $ind);
    $line = substr($line, $ind + 1);
    return ($time, $line);
  }
}


sub keyValue() {
  my ($line) = @_;
  my $ind = index($line, " ");
  if ($ind != -1) {
   my $key = substr($line, 0, $ind);
   my $value = substr($line, $ind + 1);
   return ($key, $value);
  }
}

sub getSenderName() {
  my ($from_line) = @_;
  my $ind = index($from_line, ":");
  if ( $ind != "-1" ) {
    $sender = substr($from_line, $ind + 1);
    $sender =~ s/^\s+//; #remove leading spaces
    $sender =~ s/\s+$//; #remove trailing spaces
    return $sender;
  }
}

sub getRecipientName() {
  my ($to_line) = @_;
  my $ind = index($to_line, ":");
  if ( $ind != "-1" ) {
    $recipient = substr($to_line, $ind + 1);
    $recipient =~ s/^\s+//; #remove leading spaces
    $recipient =~ s/\s+$//; #remove trailing spaces
    return $recipient;
  }
}

sub doLog() {
  my ($msg, $line) = @_;
  print LOG "$msg: $line\n";
}

sub usage() {
  print "Usage:\n";
  print " cmd: perl <path to>dmining.pl <directory-name>\n";
  print "  where:\n";
  print "    directory-name: the absolute or relative path to raw data\n";
}