#!/usr/bin/perl
use strict;
use Parse::AccessLogEntry;
use Data::Dumper;
my $P=Parse::AccessLogEntry::new();

my $devel;
$devel=1 if($ARGV[0] eq 'devel');

sub url_decode
{
  chomp $_[0];
  my %hash;
  my @parts = split(/&/, $_[0]);
  for (@parts)
  {
    my($a, $b) = split(/=/, $_);
    $a =~ s/%([a-fA-F0-9]{2})/chr(hex($1))/ge;
    $b =~ s/%([a-fA-F0-9]{2})/chr(hex($1))/ge;
    $b=~s/\+/ /g;
    $hash{$a} = $b;
  }
  return \%hash;
}

my %Fail;
my %Hits;
my %AgentHash;

while(<STDIN>)
 {
  my $H=$P->parse($_);
  next unless($H->{code} eq '200');
  my $Agent=$H->{agent};
  my $IP=$H->{host};

#I don't count!  Ignoring hits from work and home.
  next if($IP eq '160.33.43.75');
  next if($IP eq '69.227.0.244');

  my $File=$H->{file};
# Some things really mean other things.
  $File=~s/\/WebHome$/\//;
  $File=~s/\?.*//g;
  $File=~s/^\/bin\/view\//\//;
  $File=~s/^\/bin\/viewfile\//\//;
  $File=~s/^\/pub\//\//;
  $File=~s/^\/bin\/search\//\//;
  $File=~s/^\/bin\/natsearch\//\//;
  $File=~s/^\/bin\/attach\//\//;
  $File='/' if($File eq '/Gunwiki/');

#A lot of other things don't count either.
  next if($File eq '/viewauth/Sandbox/Comments');
  next if($File eq '/favicon.ico');
  next if($File eq '/robots.txt');
  next if($File=~m/\.css$/i);
  next if($File=~m/\.jpg$/i);
  next if($File=~m/\/WebRss/);
  next if($File=~m/\/pub\/TWiki\//);
  next if($File=~m/^\/TWiki\//);
  next if($File=~m/^\/Sandbox\//);
  next if($File=~m/^\/Main\//);
  next if($File=~m/^\/Personal\//);
  next if($File=~m/^\/bin\/login\//);

#We ignore administrative actions too.
  next if($File=~m/\/bin\/configure/);
  next if($File=~m/\/bin\/changes/);
  next if($File=~m/\/bin\/save/);
  next if($File=~m/\/bin\/edit\//);
  next if($File=~m/\/bin\/attach\//);
  next if($File=~m/\/bin\/rdiff\//);
  next if($File=~m/\/bin\/rename\//);
  next if($File=~m/\/admin\//);

#Disregarding 'oops', although perhaps this should be reported on
  next if($File=~m/\/bin\/oops\//);

#Bots shouldn't count!  Ignoring various user agents.
  next if($Agent=~m/^FAST MetaWeb Crawler/);
  next if($Agent=~m/^ichiro/);
  next if($Agent=~m/^psbot/);
  next if($Agent=~m/^SBIder/);
  next if($Agent=~m/^ilial\//);
  next if($Agent=~m/^NutchCVS/);
  next if($Agent=~m/^sproose/);
  next if($Agent=~m/^online link validator/);
  next if($Agent=~m/^Gigabot/);
  next if($Agent=~m/^TurnitinBot/);
  next if($Agent=~m/^Accoona-AI-Agent/);
  next if($Agent=~m/^Teemer/);
  next if($Agent=~m/^Speedy Spider/);
  next if($Agent=~m/^Yeti/);
  next if($Agent=~m/^ia_archiver-web.archive.org/);
  next if($Agent=~m/^SMBot/);
  next if($Agent=~m/\(Twiceler/);
  next if($Agent=~m/^SurveyBot/);
  next if($Agent=~m/^msnbot/);
  next if($Agent=~m/^MJ12bot/);
  next if($Agent=~m/^VadixBot/);
  next if($Agent=~m/Zyborg/);
  next if($Agent=~m/NextGenSearchBot/);
  next if($Agent=~m/Yahoo\! Slurp/);
  next if($Agent=~m/Googlebot/);
  next if($Agent=~m/SnapPreviewBot/);
  next if($Agent=~m/NimbleCrawler/);
  next if($Agent=~m/Ask Jeeves\/Teoma/);
  next if($Agent=~m/BecomeBot/);

  $AgentHash{$Agent}++;
  $Hits{$File}++;
#print "Valid: $_" if($devel);
 }

print "\nHits:\n";
foreach(sort { $Hits{$b}<=>$Hits{$a} } keys %Hits)
 {
  printf " %-s: %s\n", $_, $Hits{$_};
 }

exit unless($devel);
print "\nAgents:\n";
foreach(sort { $AgentHash{$b}<=>$AgentHash{$a} } keys %AgentHash)
 {
  printf " %-s: %s\n", $_, $AgentHash{$_};
 }