#! /usr/bin/perl
# ---------------------------------------------------------------------------
#
# USAGE: xferstats <options>
#
# OPTIONS:
#       -f <filename>   Use <filename> for the log file
#       -r              include real users 
#       -a              include anonymous users 
#       -h		include report on hourly traffic
#       -d		include report on domain traffic
#       -t		report on total traffic by section
#	-i 		report incomming traffic only (uploads)
#	-o		report outgoing traffic only (download)
#       -D <domain>     report only on traffic from <domain>
#				this option leads to problems with the local
#				domain: e.g. test.com is counted under test
#				and not recognized under com, -D com will give
#				you only statistics about com excluding
#				test.com! use -A com for correct results.
#       -A <address>    report only on traffic from addresses ends matching
#				<address> e.g. -A test.domain.com will report
#				only on addresses ending with test.domain.com
#       -l <depth>      Depth of path detail for sections
#       -s <section>    Section to report on, For example: -s /pub will report
#				only on paths under /pub
#       -u <user>       Report traffic for specified user.
#
# ---------------------------------------------------------------------------
# 05.09.01: minor change by Jamie Fifield (fifield@chebucto.ns.ca):
# - added option u which singles out the data for a single user
#
# ---------------------------------------------------------------------------
# 30.09.98: changes by Jan Menzel (jan.menzel@gmx.net):
#  -documented problems with option -D
#  -fixed same problems with spaces in the filenames (lines which are not
#     exactly 16 elements long are skipped in earlier versions)
#  -added option A which compares the addresses end with a given pattern
#  -added option i and o which reports either incoming or outgoing traffic only

@mydom = split(/\./, `dnsdomainname`);
$mydom2 = pop(@mydom); chop($mydom2);
$mydom1 = pop(@mydom);

# If you want to specify $mydom1 and $mydom2 manually you should edit the
# next two lines to customize for your domain. This will allow your domain
# to be separated in the domain listing.
# $mydom1 = "debian";
# $mydom2 = "org";

# edit the next line to customize for your default log file
$usage_file = "/var/log/xferlog";

# Edit the following lines for default report settings.
# Entries defined here will be over-ridden by the command line.

$opt_h = 0; 
$opt_d = 0;
$opt_t = 1;
$opt_l = 3;

require 'getopts.pl';
&Getopts('f:rahdD:l:s:A:iou:');

if ($opt_r) { $real = 1;}
if ($opt_a) { $anon = 1;}
if ($real == 0 && $anon == 0) { $anon = 1; }
if ($opt_f) {$usage_file = $opt_f;}
if ($opt_i) {$opt_i = 1;}
if ($opt_o) {$opt_o = 1;}

open (LOG,$usage_file) || die "Error opening usage log file: $usage_file\n";

if ($opt_o) {
   print "Transfer Totals include outgoing traffic only.\n\n";
}

if ($opt_i) {
   print "Transfer Totals include incoming traffic only.\n\n";
}

if ($opt_D) {
   $opt_D =~ tr/A-Z/a-z/;
   print "Transfer Totals include the '$opt_D' domain only.\n";
   print "All other domains are filtered out for this report.\n\n";
}

if ($opt_A) {
   $opt_A =~ tr/A-Z/a-z/;
   print "Transfer Totals include the addresses ending in '$opt_A' only.\n";
   print "All other addresses are filtered out for this report.\n\n";
}

if ($opt_s) {
   print "Transfer Totals include the '$opt_s' section only.\n";
   print "All other sections are filtered out for this report.\n\n";
}

if ($opt_u) {
   print "Transfer Totals are for user '$opt_u' only.\n";
   print "All other users are filtered out for this report.\n\n";
}

line: while (<LOG>) {

   @line = split;
   # is the first entry week day abbreviation?
   next if (length("$line[0]") != 3);

   # check whether there is a valid 'username'
   if ($line[$#line-7] eq "a" or $line[$#line-7] eq "b") {
      # yes, there i
      # offset points to the first element just behind the filename
      $offset = $#line - 7;
   } elsif ($line[$#line-6] eq "a" or $line[$#line-6] eq "b") {
      $offset = $#line - 6
   } elsif ($line[$#line-5] =~ /^(a|g|r)$/) {
      $offset = $#line - 5;
   } else {
      next;
   }
   next if (!$opt_u && !$anon && $line[$offset+3] eq "a");
   next if (!$opt_u && !$real && $line[$offset+3] eq "r");
   next if ($opt_i && $line[$offset+2] ne "i");
   next if ($opt_o && $line[$offset+2] ne "o");
   next if ($opt_u && $line[$offset+4] ne $opt_u);

   $daytime = substr($_, 0, 10) . substr($_, 19, 5);
   $time = substr($_,11,2); 

   if ($line[8] eq "\.") { $line[8] = "/unreadable/filename";}
   next if (substr($line[8],0,length("$opt_s")) ne "$opt_s");
   $line[8] = substr($line[8],length("$opt_s"));
   @path = split(/\//, $line[8]);

#
# Why was the original xferstats dropping leading 1 character path
# segments???
#
#  while (length($path[1]) <= 1) {
#     shift @path;
#     next line if ($#path == -1);
#  }

# Things in the top-level directory are assumed to be informational files

   if ($#path == 1)
      { $pathkey = "Index/Informational Files"; }
      else {
	$pathkey = "";
	for ($i=1; $i <= $#path-1 && $i <= $opt_l;$i++) {
		$pathkey = $pathkey . "/" . $path[$i];
		}
	}

   $line[6] =~ tr/A-Z/a-z/;

   @address = split(/\./, $line[6]);

   $domain = $address[$#address];
   if ($domain eq "$mydom2" && $address[$#address-1] eq "$mydom1")
      { $domain = $mydom1 . "." . $mydom2; }
   if ( int($address[0]) > 0 || $#address < 2 )
      { $domain = "unresolved"; }

   $count = 1;
   if ($opt_D and substr($domain,0,length("$opt_D")) ne "$opt_D" ) {
      $count = 0;
   }
   if ($opt_A and substr($line[6],length("$line[6]")-length("$opt_A")) ne "$opt_A" ) {
      $count = 0;
   }


   if ($count) {

   $systemfiles{$line[6]}++;                    # Systems Accessing the Arc
   $xferfiles++;                                # total files sent
   $xfertfiles++;                               # total files sent
   $xferfiles{$daytime}++;                      # files per day
   $groupfiles{$pathkey}++;                     # per-group accesses
   $domainfiles{$domain}++;

   $xfersecs{$daytime}    += $line[5];          # xmit seconds per day
   $domainsecs{$domain}   += $line[5];		# xmit seconds for domain
   $xferbytes{$daytime}   += $line[7];          # bytes per day
   $domainbytes{$domain}  += $line[7];		# xmit bytes to domain
   $xferbytes             += $line[7];          # total bytes sent
   $groupbytes{$pathkey}  += $line[7];          # per-group bytes sent

   $xfertfiles{$time}++;                        # files per hour
   $xfertsecs{$time}      += $line[5];          # xmit seconds per hour
   $xfertbytes{$time}     += $line[7];          # bytes per hour
   $xfertbytes            += $line[7];          # total bytes sent
   }
}
close LOG;

@syslist = keys(systemfiles);
@dates = sort datecompare keys(xferbytes);

if ($xferfiles == 0) {die "There was no data to process.\n";}


print "TOTALS FOR SUMMARY PERIOD ", $dates[0], " TO ", $dates[$#dates], "\n\n";
printf ("Files Transmitted During Summary Period  %12.0f\n", $xferfiles);
printf ("Bytes Transmitted During Summary Period  %12.0f\n", $xferbytes); 
printf ("Systems Using Archives                   %12.0f\n\n", $#syslist+1);

printf ("Average Files Transmitted Daily          %12.0f\n",
   $xferfiles / ($#dates + 1));
printf ("Average Bytes Transmitted Daily          %12.0f\n",
   $xferbytes / ($#dates + 1));

format top1 =

Daily Transmission Statistics

                 Number Of    Number of    Average    Percent Of  Percent Of
     Date        Files Sent  Bytes  Sent  Xmit  Rate  Files Sent  Bytes Sent
---------------  ----------  -----------  ----------  ----------  ----------
.

format line1 =
@<<<<<<<<<<<<<<  @>>>>>>>>>  @>>>>>>>>>>  @>>>>>>>>>  @>>>>>>>    @>>>>>>>  
$date,           $nfiles,    $nbytes,     $avgrate,   $pctfiles,  $pctbytes
.

$^ = top1;
$~ = line1;

# sort daily traffic by bytes sendt
#foreach $date ( sort datecompare keys(nbytes) ) {
foreach $date ( sort datecompare keys(xferbytes) ) {

   $nfiles = $xferfiles{$date};
   $nbytes = $xferbytes{$date};
   if ($xfersecs{$date}) {
      $avgrate = sprintf("%5.1f KB/s", $xferbytes{$date}/$xfersecs{$date}/1024);
   } else {
      $avgrate = sprintf("undefined");
   }
   $pctfiles = sprintf("%8.2f", 100*$xferfiles{$date}/$xferfiles);
   $pctbytes = sprintf("%8.2f", 100*$xferbytes{$date}/$xferbytes);
   write;
}

if ($opt_t) {
format top2 =

Total Transfers from each Archive Section (By bytes)

                                                 ---- Percent  Of ----
     Archive Section      Files Sent Bytes Sent  Files Sent Bytes Sent
------------------------- ---------- ----------- ---------- ----------
.

format line2 =
@<<<<<<<<<<<<<<<<<<<<<<<< @>>>>>>>>> @>>>>>>>>>> @>>>>>>>   @>>>>>>>
$section,                 $files,    $bytes,     $pctfiles, $pctbytes
.

$| = 1;
$- = 0;
$^ = top2;
$~ = line2;

# sort total transfer for each archive by # files transfered
foreach $section ( sort bytecompare keys(groupfiles) ) {

   $files = $groupfiles{$section};
   $bytes = $groupbytes{$section};
   $pctbytes = sprintf("%8.2f", 100 * $groupbytes{$section} / $xferbytes);
   $pctfiles = sprintf("%8.2f", 100 * $groupfiles{$section} / $xferfiles);
   write;

}

}

if ($opt_d) {
format top3 =

Total Transfer Amount By Domain

             Number Of    Number of     Average    Percent Of  Percent Of
Domain Name  Files Sent   Bytes Sent   Xmit  Rate  Files Sent  Bytes Sent
-----------  ----------  ------------  ----------  ----------  ----------
.

format line3 =
@<<<<<<<<<<  @>>>>>>>>>  @>>>>>>>>>>>  @>>>>>>>>>  @>>>>>>>    @>>>>>>>  
$domain,     $files,     $bytes,       $avgrate,   $pctfiles,  $pctbytes
.

$- = 0;
$^ = top3;
$~ = line3;

# sort amount per domain by files
foreach $domain ( sort domnamcompare keys(domainfiles) ) {

   $files = $domainfiles{$domain};
   $bytes = $domainbytes{$domain};
   if ($domainsecs{$domain}) {
      $avgrate = sprintf("%5.1f KB/s", $domainbytes{$domain}/$domainsecs{$domain}/1024);
   } else {
      $avgrate = sprintf("undefined");
   }
   $pctfiles = sprintf("%8.2f", 100 * $domainfiles{$domain} / $xferfiles);
   $pctbytes = sprintf("%8.2f", 100 * $domainbytes{$domain} / $xferbytes);
   write;

}

print "\n";

}

if ($opt_h) {

format top8 =

Hourly Transmission Statistics

                 Number Of    Number of    Average    Percent Of  Percent Of
     Time        Files Sent  Bytes  Sent  Xmit  Rate  Files Sent  Bytes Sent
---------------  ----------  -----------  ----------  ----------  ----------
.

format line8 =
@<<<<<<<<<<<<<<  @>>>>>>>>>  @>>>>>>>>>>  @>>>>>>>>>  @>>>>>>>    @>>>>>>>  
$time,           $nfiles,    $nbytes,     $avgrate,   $pctfiles,  $pctbytes
.


$| = 1;
$- = 0;
$^ = top8;
$~ = line8;

# sort hourly transmission by sent bytes
foreach $time ( sort keys(xfertbytes) ) {

   $nfiles   = $xfertfiles{$time};
   $nbytes   = $xfertbytes{$time};
   if ($xfertsecs{$time}) {
      $avgrate  = sprintf("%5.1f KB/s", $xfertbytes{$time}/$xfertsecs{$time}/1024);
   } else {
      $avgrate  = sprintf("undefined");
   }
   $pctfiles = sprintf("%8.2f", 100*$xfertfiles{$time} / $xferfiles);
   $pctbytes = sprintf("%8.2f", 100*$xfertbytes{$time} / $xferbytes);
   write;
}
}
exit(0);

sub datecompare {

   $date1  = substr($a, 11, 4) * 4800;
   $date2  = substr($b, 11, 4) * 4800;
   $date1 += index("JanFebMarAprMayJunJulAugSepOctNovDec",substr($a, 4, 3)) / 3 * 400;
   $date2 += index("JanFebMarAprMayJunJulAugSepOctNovDec",substr($b, 4, 3)) / 3 * 400;
   $date1 += substr($a, 8, 2);
   $date2 += substr($b, 8, 2);
   $date1 - $date2;

}

sub domnamcompare {

   $sdiff = length($a) - length($b);
   ($sdiff < 0) ? -1 : ($sdiff > 0) ? 1 : ($a lt $b) ? -1 : ($a gt $b) ? 1 : 0;

}

sub bytecompare {

   $bdiff = $groupbytes{$b} - $groupbytes{$a};
   ($bdiff < 0) ? -1 : ($bdiff > 0) ? 1 : ($a lt $b) ? -1 : ($a gt $b) ? 1 : 0;

}

sub faccompare {

   $fdiff = $fac{$b} - $fac{$a};
   ($fdiff < 0) ? -1 : ($fdiff > 0) ? 1 : ($a lt $b) ? -1 : ($a gt $b) ? 1 : 0;

}