#!/usr/local/bin/perl 

# tsh2ta.pl -f filename {-i interface} {-d displayflags} {-a alpha} {-t init_timeout} {-x exit time} {-b}
# 			{-s smallflow packets} {-m mediumflow packets} {-h alpha long}
# New timeout value = max interarrival time * L?h:k

### sections denoted by three comment characters (###) are modifications
### made by David Cheney for specific purposes related to timeout 
### research with Bo Ryu and Hans-Werner Braun.
###
$boff=2;		# default for the number of backoffs
@pkt_th = (10,6,3);  # default packet threshold for each timeout

###
$tdelta=1;              # synchronous array evaluation timeout in seconds
$tout_low = 4;                # default lowest flow timeout in seconds
$etime=9e99;            # forxed exit time in seconds

#-------------------------------------------------------------------------------
# command line parser

while(($id,@r)=split('=',$ARGV[$argc++])){
 if($id =~ "f"){                                        # input file name
  $I=$ARGV[$argc++];
  open(infile,$I) || die("Can't open input file\n");
#  open(infile,"gzip -dc $I |") || die("Can't open input file\n");
 }elsif($id =~ "t"){                                    # lowest (beginning) flow timeout
  $tout_low=$ARGV[$argc++];
 }elsif($id =~ "i"){                                    # input interface number
  $interface=$ARGV[$argc++];
 }elsif($id =~ "r"){                                    # number of exponential backoffs
   $boff=$ARGV[$argc++];
 }elsif($id =~ "m"){                                    # packet threshold for each timeout value
   @pkt_th = split(',' ,$ARGV[$argc++]);
 }elsif($id =~ "d"){                                    # display flags
  $DISPLAY=$ARGV[$argc++];
 }elsif($id =~ "x"){                                    # end time (for percentile cut)
  $etime=$ARGV[$argc++];
 }elsif($id =~ "b"){                                    # bidirectional flows (transactions)
  $bflag++;
 }else{
 printf stderr "Usage: $0 -f fname (-i #} {-d flags} {-x exit time} {-b} {-t lowest_timeout} {-r backoffs} {-m pkt_thresold (,,,)}\n";
 exit(1);
 }
}
if(!$I){
 printf stderr "Usage: $0 -f fname (-i #} {-d flags} {-x exit time} {-b} {-t lowest_timeout} {-r backoffs} {-m pkt_thresold (,,,)}\n";
 exit(1);
}

if($#pkt_th != $boff){
 printf stderr "Number of backoffs ($boff) does not match the number of packet thresholds\n";
 exit(1);
}

# print command line parameters
printf "P\t";
$argc=0;
while(($id,@r)=split('=',$ARGV[$argc++])){printf"%s ",$ARGV[$argc];}
printf"\n";
$tout = $tout_low * (2**$boff);

#@temp = join(',',@pkt_th);
#print "@temp\n";


#-------------------------------------------------------------------------------
while(read(infile,$record,44)){                 # read next record
#-----------------------------------------------------------------------
 $tpcount++;                                    # total packets
 $tstampsec  =vec($record,0,32);                # timestamp: seconds
 $tstampusec =vec($record,1,32);                # timestamp: microseconds
 $intf       =($tstampusec>>24)&0xff;           # get interface number
 if(($interface ne "") && ($intf != $interface)){next;} # for -i #
 $tstampusec =$tstampusec&0xffffff;             # mask out interface
 $tstamp     =($tstampusec/1e6)+$tstampsec;     # composite time stamp
 if(!$orgtstamp){$orgtstamp=$tstamp;$oldtstamp=$tstamp;}

 $sa =vec($record,5,32);                        # IP source address
 $da =vec($record,6,32);                        # IP destination address
 $pl =vec($record,5,16);                        # packet length
 $pr =vec($record,17,8);                        # IP protocol
 $hl =(vec($record,8,8)&0xf)*2;                 # header length for 16 bit offsets
 $sp =vec($record,$hl+4,16);                    # source port
 $dp =vec($record,$hl+5,16);                    # destination ports
 $fl =vec($record,($hl*2)+8+13,8)&0x3f;         # TCP flags

 if(($pr != 6) & ($pr != 17)){$sp=0;$dp=0;}     # unless TCP or UDP set ports=0
 if($pr != 6){$fl=0;}                           # only for TCP
#-----------------------------------------------------------------------
###if record exists, update it with new packet's info
 if(($ap,$ab,$ip,$ib,$at,$al)=split(':',$ta{$sa,$da,$pr,$sp,$dp})){
     $ta{$sa,$da,$pr,$sp,$dp}=join(':',++$ap,$ab+$pl,$ip,$ib,$at,$tstamp);
     ### dynamic timeout algorithm implementation:

     $life =  $tstamp - $at;

     # Flow classification
     # First, assign the highest class number. Then move to lower class number whenever it satisfies.
     for ($i=$boff; ($i >= 0) && ($life <= $tout); $i--) {
	 if ( ($ap == $pkt_th[$i]) && ($life <= $tout_low * (2**$i)) ) { 
	     $label{$sa,$da,$pr,$sp,$dp} = $i+1;
	     $dto{$sa,$da,$pr,$sp,$dp} = $tout_low * (2**$i);
	 } 
     }

#     print "$ta{$sa,$da,$pr,$sp,$dp}\ti=$i, timeout=$dto{$sa,$da,$pr,$sp,$dp}\n"; 

     if($DISPLAY =~ "U"){
	 $SA=sprintf("%d.%d.%d.%d",($sa>>24)&0xff,($sa>>16)&0xff,($sa>>8)&0xff,$sa&0xff);
	 $DA=sprintf("%d.%d.%d.%d",($da>>24)&0xff,($da>>16)&0xff,($da>>8)&0xff,$da&0xff);
	 printf"U\t%.6f\t%d\t%s\t%s\t%d\t%d\t%d\t%d\t%d\t%d\t%d\t%.6f\t%.6f\t%.6f\t%x\n",
	 $tstamp,$tf,$SA,$DA,$pr,$sp,$dp,$ap,$ab+$pl,$ip,$ib,$at,$tstamp,$tstamp-$at,$fl;
     }
 }
###else if record exists in the other direction, update it

 elsif((($ap,$ab,$ip,$ib,$at,$al)=split(':',$ta{$da,$sa,$pr,$dp,$sp})) && $bflag){
  $ta{$da,$sa,$pr,$dp,$sp}=join(':',$ap,$ab,++$ip,$ib+$pl,$at,$tstamp);
  ### dynamic timeout algorithm implementation:

     $life =  $tstamp - $at;

     # Flow classification
     # First, assign the highest class number. Then move to lower class number whenever the number of packets exceed higher threshold.
     for ($i=$boff; ($i >= 0) && ($life <= $tout); $i--) {
	 if ( (($ap+$ip) == $pkt_th[$i]) && ($life <= $tout_low * (2**$i)) ) { 
	     $label{$sa,$da,$pr,$sp,$dp} = $i+1;
	     $dto{$sa,$da,$pr,$sp,$dp} = $tout_low * (2**$i);
	 } 
     }

  if($DISPLAY =~ "U"){
   $SA=sprintf("%d.%d.%d.%d",($sa>>24)&0xff,($sa>>16)&0xff,($sa>>8)&0xff,$sa&0xff);
   $DA=sprintf("%d.%d.%d.%d",($da>>24)&0xff,($da>>16)&0xff,($da>>8)&0xff,$da&0xff);
   printf"U\t%.6f\t%d\t%s\t%s\t%d\t%d\t%d\t%d\t%d\t%d\t%d\t%.6f\t%.6f\t%.6f\t%x\n",
         $tstamp,$tf,$SA,$DA,$pr,$sp,$dp,$ap,$ab,$ip,$ib+$pl,$at,$tstamp,$tstamp-$at,$fl;
  }
###else create a new record
 }else{
     $ta{$sa,$da,$pr,$sp,$dp}=join(':',1,$pl,0,0,$tstamp,$tstamp);
     $label{$sa,$da,$pr,$sp,$dp} = 0;  ### initial flow label (meaning not labeled)
     ### new hash for dynamic timeout
     $dto{$sa,$da,$pr,$sp,$dp}=$tout;

     ###
     $tf++;
     if($DISPLAY =~ "C"){
         $SA=sprintf("%d.%d.%d.%d",($sa>>24)&0xff,($sa>>16)&0xff,($sa>>8)&0xff,$sa&0xff);
         $DA=sprintf("%d.%d.%d.%d",($da>>24)&0xff,($da>>16)&0xff,($da>>8)&0xff,$da&0xff);
         printf"C\t%.6f\t%d\t%s\t%s\t%d\t%d\t%d\t%d\t%d\t%d\t%d\t%.6f\t%.6f\t%.6f\t%x\n",
         $tstamp,$tf,$SA,$DA,$pr,$sp,$dp,1,$pl,0,0,$tstamp,$tstamp,0,$fl;
     }
 }
###once per second, timeout flows that have expired.
 if($tstamp >= ($oldtstamp+$tdelta)){
  $otf=$tf;
  foreach $fkey (keys %ta){
   ($ap,$ab,$ip,$ib,$at,$al)=split(':',$ta{$fkey});
   ### dynamic timeout replaces generic $tout
   if($al < ($tstamp-$dto{$fkey})){   # timing out transactions
    ($sa,$da,$pr,$sp,$dp)=split($;,$fkey);
    ###classification info
    $class = $label{$sa,$da,$pr,$sp,$dp};
    if($DISPLAY =~ "D"){
     $SA=sprintf("%d.%d.%d.%d",($sa>>24)&0xff,($sa>>16)&0xff,($sa>>8)&0xff,$sa&0xff);
     $DA=sprintf("%d.%d.%d.%d",($da>>24)&0xff,($da>>16)&0xff,($da>>8)&0xff,$da&0xff);
     printf"D\t%.6f\t%s\t%s\t%d\t%d\t%d\t%d\t%d\t%d\t%d\t%.6f\t%.6f\t%.6f\t%.6f\t%s\n",
          $tstamp,$SA,$DA,$pr,$sp,$dp,$ap,$ab,$ip,$ib,$at,$al,$al-$at,$tstamp-$al,$class;
    }
    delete $ta{$fkey};
    delete $dto{$fkey};
    delete $label{$fkey};
    $tf--;

   }
  }
  if($DISPLAY =~ "S"){
   printf "S\t%.6f\t%d\t%d\t%d\n",$tstamp,$otf,$tf,$otf-$tf;
  }
  $oldtstamp=$oldtstamp+$tdelta;
  if($tstamp >= ($orgtstamp+$etime)){last;}
 }
}

# print remaining records
if($DISPLAY =~ "X"){
 foreach $fkey (keys %ta){
  ($ap,$ab,$ip,$ib,$at,$al)=split(':',$ta{$fkey});
  ($sa,$da,$pr,$sp,$dp)=split($;,$fkey);
    ###classify as small medium or large
  $class = $label{$sa,$da,$pr,$sp,$dp};
  $SA=sprintf("%d.%d.%d.%d",($sa>>24)&0xff,($sa>>16)&0xff,($sa>>8)&0xff,$sa&0xff);
  $DA=sprintf("%d.%d.%d.%d",($da>>24)&0xff,($da>>16)&0xff,($da>>8)&0xff,$da&0xff);
  printf "X\t%.6f\t%s\t%s\t%d\t%d\t%d\t%d\t%d\t%d\t%d\t%.6f\t%.6f\t%.6f\t%.6f\t%s\n",
         $tstamp,$SA,$DA,$pr,$sp,$dp,$ap,$ab,$ip,$ib,$at,$al,$al-$at,$tstamp-$al,$class;
 }
}

sub numerically { $a <=> $b;}