#!/usr/bin/perl
#
# This script parses the default logfile format produced by rsync when running
# as a daemon with transfer logging enabled.  It also parses a slightly tweaked
# version of the default format where %o has been replaced with %i.
#
# This script is derived from the xferstats script that comes with wuftpd.  See
# the usage message at the bottom for the options it takes.
#
# Andrew Tridgell, October 1998

use Getopt::Long;

# You may wish to edit the next line to customize for your default log file.
$usage_file = "/var/log/rsyncd";

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

$hourly_report = 0; 
$domain_report = 0;
$total_report = 0;
$depth_limit = 9999;
$only_section = '';

&Getopt::Long::Configure('bundling');
&usage if !&GetOptions(
    'hourly-report|h' => \$hourly_report,
    'domain-report|d' => \$domain_report,
    'domain|D:s' => \$only_domain,
    'total-report|t' => \$total_report,
    'depth-limit|l:i' => \$depth_limit,
    'real|r' => \$real,
    'anon|a' => \$anon,
    'section|s:s' => \$only_section,
    'file|f:s' => \$usage_file,
);

$anon = 1 if !$real && !$anon;

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

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

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

line: while (<LOG>) {

my $syslog_prefix = '\w\w\w +\d+ \d\d:\d\d:\d\d \S+ rsyncd';
my $rsyncd_prefix = '\d\d\d\d/\d\d/\d\d \d\d:\d\d:\d\d ';

   next unless ($day,$time,$op,$host,$module,$file,$bytes)
      = m{^
	  ( \w\w\w\s+\d+ | \d+/\d\d/\d\d ) \s+ # day
	  (\d\d:\d\d:\d\d) \s+                 # time
	  [^[]* \[\d+\]:? \s+                  # pid (ignored)
	  (send|recv|[<>]f\S+) \s+             # op (%o or %i)
	  (\S+) \s+                            # host
	  \[\d+\.\d+\.\d+\.\d+\] \s+           # IP (ignored)
	  (\S+) \s+                            # module
	  \(\S*\) \s+                          # user (ignored)
	  (.*) \s+                             # file name
	  (\d+)                                # file length in bytes
	  $ }x;

   # TODO actually divide the data by into send/recv categories
   if ($op =~ /^>/) {
      $op = 'send';
   } elsif ($op =~ /^</) {
      $op = 'recv';
   }

   $daytime = $day;
   $hour = substr($time,0,2); 

   $file = $module . "/" . $file;

   $file =~ s|//|/|mg;

   @path = split(/\//, $file);

   $pathkey = "";
   for ($i=0; $i <= $#path && $i <= $depth_limit; $i++) {
	$pathkey = $pathkey . "/" . $path[$i];
   }

   if ($only_section ne '') {
       next unless (substr($pathkey,0,length($only_section)) eq $only_section);
   }

   $host =~ tr/A-Z/a-z/;

   @address = split(/\./, $host);

   $domain = $address[$#address];
   if ( int($address[0]) > 0 || $#address < 2 )
      { $domain = "unresolved"; }

   if ($only_domain ne '') {
       next unless (substr($domain,0,length($only_domain)) eq $only_domain);
   }


#   printf("c=%d day=%s bytes=%d file=%s path=%s\n", 
#	   $#line, $daytime, $bytes, $file, $pathkey);

   $xferfiles++;                                # total files sent
   $xfertfiles++;                               # total files sent
   $xferfiles{$daytime}++;                      # files per day
   $groupfiles{$pathkey}++;                     # per-group accesses
   $domainfiles{$domain}++;

   $xferbytes{$daytime}   += $bytes;          # bytes per day
   $domainbytes{$domain}  += $bytes;		# xmit bytes to domain
   $xferbytes             += $bytes;          # total bytes sent
   $groupbytes{$pathkey}  += $bytes;          # per-group bytes sent

   $xfertfiles{$hour}++;                        # files per hour
   $xfertbytes{$hour}     += $bytes;          # bytes per hour
   $xfertbytes            += $bytes;          # 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   Percent Of  Percent Of
     Date        Files Sent   MB  Sent    Files Sent  Bytes Sent
---------------  ----------  -----------  ----------  ----------
.

format line1 =
@<<<<<<<<<<<<<<  @>>>>>>>>>  @>>>>>>>>>>  @>>>>>>>    @>>>>>>>  
$date,           $nfiles,    $nbytes/(1024*1024), $pctfiles,  $pctbytes
.

$^ = top1;
$~ = line1;

foreach $date (sort datecompare keys %xferbytes) {

   $nfiles   = $xferfiles{$date};
   $nbytes   = $xferbytes{$date};
   $pctfiles = sprintf("%8.2f", 100*$xferfiles{$date} / $xferfiles);
   $pctbytes = sprintf("%8.2f", 100*$xferbytes{$date} / $xferbytes);
   write;
}

if ($total_report) {
format top2 =

Total Transfers from each Archive Section (By bytes)

                                                           - Percent -
     Archive Section                   NFiles     MB      Files   Bytes
------------------------------------- ------- ----------- ----- -------
.

format line2 =
@<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<< @>>>>>> @>>>>>>>>>> @>>>>   @>>>>
$section,                 $files,    $bytes/(1024*1024),     $pctfiles, $pctbytes
.

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

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 ( $xferfiles < 1 ) { $xferfiles = 1; }
if ( $xferbytes < 1 ) { $xferbytes = 1; }
}

if ($domain_report) {
format top3 =

Total Transfer Amount By Domain

             Number Of    Number of    Percent Of  Percent Of
Domain Name  Files Sent    MB Sent     Files Sent  Bytes Sent
-----------  ----------  ------------  ----------  ----------
.

format line3 =
@<<<<<<<<<<  @>>>>>>>>>  @>>>>>>>>>>>  @>>>>>>>    @>>>>>>>  
$domain,     $files,     $bytes/(1024*1024), $pctfiles,  $pctbytes
.

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

foreach $domain (sort domnamcompare keys %domainfiles) {

   if ( $domainsecs{$domain} < 1 ) { $domainsecs{$domain} = 1; }

   $files = $domainfiles{$domain};
   $bytes = $domainbytes{$domain};
   $pctfiles = sprintf("%8.2f", 100 * $domainfiles{$domain} / $xferfiles);
   $pctbytes = sprintf("%8.2f", 100 * $domainbytes{$domain} / $xferbytes);
   write;

}

}

if ($hourly_report) {

format top8 =

Hourly Transmission Statistics

                 Number Of    Number of   Percent Of  Percent Of
     Time        Files Sent    MB  Sent   Files Sent  Bytes Sent
---------------  ----------  -----------  ----------  ----------
.

format line8 =
@<<<<<<<<<<<<<<  @>>>>>>>>>  @>>>>>>>>>>  @>>>>>>>    @>>>>>>>  
$hour,           $nfiles,    $nbytes/(1024*1024), $pctfiles,  $pctbytes
.


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

foreach $hour (sort keys %xfertbytes) {

   $nfiles   = $xfertfiles{$hour};
   $nbytes   = $xfertbytes{$hour};
   $pctfiles = sprintf("%8.2f", 100*$xfertfiles{$hour} / $xferfiles);
   $pctbytes = sprintf("%8.2f", 100*$xfertbytes{$hour} / $xferbytes);
   write;
}
}
exit(0);

sub datecompare {
    $a cmp $b;
}

sub domnamcompare {

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

}

sub bytecompare {

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

}

sub faccompare {

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

}

sub usage
{
    die <<EOT;
USAGE: rsyncstats [options]

OPTIONS:
  -f FILENAME   Use FILENAME for the log file.
  -h            Include report on hourly traffic.
  -d            Include report on domain traffic.
  -t            Report on total traffic by section.
  -D DOMAIN     Report only on traffic from DOMAIN.
  -l DEPTH      Set DEPTH of path detail for sections.
  -s SECTION    Set SECTION to report on. For example, "-s /pub"
                will report only on paths under "/pub".
EOT
}