#!/usr/local/bin/perl

# logcheck
$version ="0.56 ALPHA";

$logfile ="accesslog.txt";
require 'domain.pl';


@pairs = split(/&/, $ENV{'QUERY_STRING'});
foreach $pair (@pairs) {
   ($key, $value) = split(/=/, $pair);
   $key =~ tr/+/ /;
   $key =~ s/%([a-fA-F0-9][a-fA-F0-9])/pack("c",hex($1))/ge;
   $value =~ tr/+/ /;
   $value =~ s/%([a-fA-F0-9][a-fA-F0-9])/pack("c",hex($1))/ge;
   $FORM{$key} = $value;
   $value =~ s/([\\\'])/\\$1/g;
   eval("\$$key = \'$value\'");
}

print "Content-type: text/html\n\n";
&header;

unless ($type ne "" || $callday ne "" || $calluri ne "") {
   &main;
}
else{
   open (DATA,"$logfile");
   if ($type == 11){
      while(<DATA>){
         @data = split(/\t/,$_);
         &timing1;
      }
      &data_close;
      &timing_write1;
   }
   elsif ($type == 14){
      while(<DATA>){
         @data = split(/\t/,$_);
         &timing2;
      }
      &data_close;
      &timing_write4;
   }
   elsif ($type == 21){
      while(<DATA>){
         @data = split(/\t/,$_);
         &host1;
      }
      &data_close;
      &host_written;
   }
   elsif ($type == 22){
      while(<DATA>){
         @data = split(/\t/,$_);
         &host2;
      }
      &data_close;
      &host_written;
   }
   elsif ($type == 24){
      while(<DATA>){
         @data = split(/\t/,$_);
         &host3;
      }
      &data_close;
      &host_written3;
   }
   elsif ($type == 31){
      while(<DATA>){
         @data = split(/\t/,$_);
         &agent1;
      }
      &data_close;
      &agent_written;
   }
   elsif ($type == 32){
      while(<DATA>){
         @data = split(/\t/,$_);
         &agent2;
      }
      &data_close;
      &agent_written;
   }
   elsif ($type == 33){
      while(<DATA>){
         @data = split(/\t/,$_);
         &agent3;
      }
      &data_close;
      &agent_written;
   }
   elsif ($type == 41){
      while(<DATA>){
         @data = split(/\t/,$_);
         &referer1;
      }
      &data_close;
      &referer_written1;
   }
   elsif ($type == 42){
      while(<DATA>){
         @data = split(/\t/,$_);
         &referer2;
      }
      &data_close;
      &referer_written2;
   }
   elsif ($type == 43){
      while(<DATA>){
         @data = split(/\t/,$_);
         &referer3;
      }
      &data_close;
      &referer_written2;
   }
   elsif ($type == 51){
      while(<DATA>){
         @data = split(/\t/,$_);
         &uri1;
      }
      &data_close;
      &uri_written1;
   }
   elsif ($callday ne ""){
      if ($calluri ne "" && $calluri ne "total") {
         while(<DATA>){
            @data = split(/\t/,$_);
            &uri_referer2;
         }
         &data_close;
         &uri_referer_written1;
      }
      elsif ($calluri eq "total") {
         while(<DATA>){
            @data = split(/\t/,$_);
            &uri_referer3;
         }
         &data_close;
         &uri_referer_written2;
      }
      else {
         while(<DATA>){
            @data = split(/\t/,$_);
            &uri3;
         }
         close(DATA);
         &uri_written3;
      }
   }
   elsif ($calluri ne ""){
      while(<DATA>){
         @data = split(/\t/,$_);
         &uri_referer1;
      }
      &data_close;
      &uri_referer_written1;
   }
}
&footer;
exit;

## end

sub data_close {
   $number =$.;
   close(DATA);
}



sub main{
   print "<H2>logcheck</H2>\n";

   print "<UL>\n";
   print "<LI>TIME\n";
   print "<UL>\n";
   print "<LI><A HREF =\"datacheck.cgi?type=11\" TARGET =\"_self\">TYPE 1</A>\n";
   print "<LI><A HREF =\"datacheck.cgi?type=14\" TARGET =\"_self\">TYPE 4</A>\n";
   print "</UL>\n";
   print "<LI>REMOTE_HOST\n";
   print "<UL>\n";
   print "   <LI><A HREF =\"datacheck.cgi?type=21\" TARGET =\"_self\">TYPE 1</A>\n";
   print "   <LI><A HREF =\"datacheck.cgi?type=22\" TARGET =\"_self\">TYPE 2</A>\n";
   print "   <LI><A HREF =\"datacheck.cgi?type=24\" TARGET =\"_self\">TYPE 4</A>\n";
   print "</UL>\n";
   print "<LI>HTTP_USER_AGENT\n";
   print "<UL>\n";
   print "   <LI><A HREF =\"datacheck.cgi?type=31\" TARGET =\"_self\">TYPE 1</A>\n";
   print "   <LI><A HREF =\"datacheck.cgi?type=32\" TARGET =\"_self\">TYPE 2</A>\n";
   print "   <LI><A HREF =\"datacheck.cgi?type=33\" TARGET =\"_self\">TYPE 3</A>\n";
   print "</UL>\n";
   print "<LI>HTTP_REFERER\n";
   print "<UL>\n";
   print "<LI><A HREF =\"datacheck.cgi?type=41\" TARGET =\"_self\">TYPE 1</A>\n";
   print "<LI><A HREF =\"datacheck.cgi?type=42\" TARGET =\"_self\">TYPE 2</A>\n";
   print "<LI><A HREF =\"datacheck.cgi?type=43\" TARGET =\"_self\">TYPE 3</A>\n";
   print "</UL>\n";
   print "<LI><A HREF =\"datacheck.cgi?type=51\" TARGET =\"_self\">DOCUMENT_URI</A>\n";
   print "</UL>\n";

   print "\n";

}


sub header {
   print "<HTML>\n";
   print "<HEAD>\n";
   print "<TITLE>logcheck</TITLE>\n";
   print "</HEAD>\n";
   print "<BODY BGCOLOR =\"#ffffff\" TEXT =\"#000000\">\n";
   print "<NOBR>\n";
}


sub footer {
$time =times;
print "<BR>$time";
print "<DIV ALIGN =\"right\">version : $version</DIV>";

   print "</NOBR>\n";
   print "</BODY>\n";
   print "</HTML>\n";
   exit;
}


sub timing1{
   ($sec, $min, $hour, $mday, $mon, $year, $wday) = localtime($data[1]);
   $mon++;
   $hour =sprintf("%.2d",$hour);
   $mday =sprintf("%.2d",$mday);
   $mon =sprintf("%.2d",$mon);
   $year += 1900;
   $week{$wday} = ++$week{$wday};
   $hours{$hour} = ++$hours{$hour};
   $days{"$year\/$mon\/$mday"} = ++$days{"$year\/$mon\/$mday"};
   $week{"$year\/$mon\/$mday"} = $wday if $days{"$year\/$mon\/$mday"} == 1;
}


sub timing_write1{
   local($wday,$i,$data,$persent);
   local(@sortedweek, @sortedhours, @sorteddays, @week);

   print "<H2>week</H2>\n";
   @week =('Sun','Mon','Tue','Wed','Thu','Fri','Sat');
   @sortedweek = sort keys(%week);

   for($i = 0 ; $i < 7 ; $i++){
      $persent = &persent($week{$i}, $number);
      print "$week[$i]\n - $week{$i} ($persent \%)<BR>\n";
   }
   print "<HR>\n";

   print "<H2>hours</H2>\n";
   @sortedhours = sort keys(%hours);
   foreach(@sortedhours){
      $persent = &persent($hours{$_}, $number);
      print "$_ - $hours{$_} ($persent \%)<BR>\n";
   }
   print "<HR>\n";

   print "<H2>day</H2>\n";
   @sorteddays = sort keys(%days);
   foreach (@sorteddays){
      print "<A HREF =\"datacheck.cgi?callday=$_\">$_</A> $week[$week{$_}] $days{$_}<BR>\n";
   }
   print "<HR>\n";

   print "total:$number";
}



sub timing2 {
   ($sec, $min, $hour, $mday, $mon, $year, $wday) = localtime($data[1]);
   $mon++;
   $mday = sprintf("%.2d",$mday);
   $mon = sprintf("%.2d",$mon);
   $year += 1900;
   @week =('Sun','Mon','Tue','Wed','Thu','Fri','Sat');
   $wday = $week[$wday];
   $days{"$year\/$mon\/$mday"} = ++$days{"$year\/$mon\/$mday"};
   $week{"$year\/$mon\/$mday"} = $wday if $days{"$year\/$mon\/$mday"} == 1;
}

sub timing_write4 {
   local(@sorteddays, @week);

   print "<H2>day</H2>\n";
   @sorteddays = sort keys(%days);
   foreach(@sorteddays){
      print "<A HREF =\"datacheck.cgi?callday=$_\">$_</A> $week{$_} $days{$_}<BR>\n";
   }
   print "<BR>\ntotal:$number";
}

sub host1{
   $host{$data[2]} = ++$host{$data[2]};
}


sub host2{
   local($hostend, @hostdata);

   if($data[2] =~ /jp$/) {
      @hostdata = split(/\./, $data[2]);
      @hostdata = splice(@hostdata, $#hostdata - 2);
      $hostend = join(".", @hostdata);
   }
   elsif ($data[2] =~ /([0-9]+)\.([0-9]+)\.([0-9]+)\.([0-9]+)/) {
      $hostend = "ip_adress";
   }
   else {
      $hostend = &domain'host($data[2]);
   }

   $host{"\L$hostend\E"} = ++$host{"\L$hostend\E"};
}


sub host3{
   local($jp_hostend, @hostdata);

   if($data[2] !~/([0-9]+)\.([0-9]+)\.([0-9]+)\.([0-9]+)/){
      @hostdata = split(/\./, $data[2]);
      $host{"\L$hostdata[$#hostdata]\E"} = ++$host{"\L$hostdata[$#hostdata]\E"};
      if($data[2] =~ /jp$/) {
         @jp_hostend = splice(@hostdata, $#hostdata - 1);
         $jp_host{"\L$jp_hostend[0]\E"} = ++$jp_host{"\L$jp_hostend[0]\E"};
      }
   }
   else{
      $host{ip_adress} = ++$host{ip_adress};
   }
}


sub host_written{
   local($persent);
   local(@sortedhost);

   @sortedhost = sort by_host keys(%host);

   foreach  (@sortedhost){
      $persent = &persent($host{$_}, $number);
      print "$host{$_} ($persent \%) : $_<BR>\n";
   }
   print "<BR>\n";
   print "total:$number";
}


sub host_written3 {
   local($persent);
   local(@sortedhost, @sortedjp_host, @jp_domain, @domainjp, @domain);

   @sortedhost = sort by_host keys(%host);
   @domain = &domain'top_domain(@sortedhost);
   $i = 0;
   $j = 0;

   foreach (@sortedhost){
      $persent = &persent($host{$_}, $number);
      print "$host{$_} ($persent \%) : $_";
      print " - $domain[$i]" if $domain[$i] ne "0";
      print "<br>";

      $i++;

      if ($_ eq "jp") {
         @sortedjp_host = sort by_jp_host keys(%jp_host);
         @domainjp = &domain'jpsecond_domain(@sortedjp_host);
         print "<BLOCKQUOTE>\n";
         foreach(@sortedjp_host) {
            $persent = &persent($jp_host{$_}, $number);
            @jp_domain = split(/\./, $_);
            print "$jp_host{$_} ($persent \%) : $_\.jp";
            print " - $domainjp[$j]" if $domainjp[$j] ne "0";
            print "<BR>\n";
            $j++;
         }
         print "</BLOCKQUOTE>\n";
      }
   }
   print "<BR>total:$number\n\n";
}


sub by_host{
   if($host{$a} != $host{$b}){
      $host{$b} <=> $host{$a};
   }
   else{
      $a cmp $b;
   }
}

sub by_jp_host{
   if($jp_host{$a} != $jp_host{$b}){
      $jp_host{$b} <=> $jp_host{$a};
   }
   else{
      $a cmp $b;
   }
}


sub agent1{
   $agent{$data[3]} = ++$agent{$data[3]};
}


sub agent2{
   local(@agent);
   if($data[3] =~ /MSIE/){
      $data[3] =~ s/Mozilla\/[0-9]\.[0-9]+\s?\(compatible\;\s?//i;
      $data[3] =~ s/\..+//;
   }
   elsif($data[3] =~ /compatible/){
      @agent = split(/ /,$_);
      $data[3] = $agent[2];
      $data[3] =~ s/\..+//;
   }
   elsif($data[3] =~/DreamPassport/){
      @agent = split(/[\(\)]/,$_);
      $data[3] = $agent[1];
   }
   elsif($data[3] =~ /WebAuto/ || $data[3] =~ /JustView/){
      $data[3] =~ s/\..+//;
   }
   elsif($data[3] =~ /WWWC/){
      $data[3] =~ s/\/.+//;
   }
   elsif($data[3] =~ /sharp wd browser/){
      $data[3] =~ s/\(.+//;
   }
   elsif($data[3] =~ /[Mm]ozilla/){
      $data[3] =~ s/mozilla/Mozilla/g if $data[3] =~ /mozilla/;
      $data[3] =~ s/\..+//;
   }
   $agent{$data[3]} = ++$agent{$data[3]};
}

sub agent3{
   if($data[3] =~ /compatible/ || $data[3] =~ /Microsoft Internet Explorer/){
      $data[3] ="MSIE";
   }
   elsif($data[3] =~/DreamPassport/){
      @agent = split(/[\(\)]/,$_);
      $data[3] = $agent[1];
      $data[3] =~ s/\/.+//;
   }
   elsif($data[3] =~ /GetYou!/){
      $data[3] =~ s/\s.+//;
   }
   elsif($data[3] =~ /mozilla/i){
      $data[3] ="Mozilla";
   }
   else{
      $data[3] =~ s/\/.+//;
}
   $agent{$data[3]} = ++$agent{$data[3]};
}


sub agent_written{
   local($persent);
   local(@sortedagent);

   @sortedagent = sort by_agent keys(%agent);

   foreach(@sortedagent){
      $persent = &persent($agent{$_}, $number);
      print "$agent{$_} ($persent \%) : ";
      if ($_ ne ""){
         print "$_";
      }
      print "<BR>\n";
   }
   print "<BR>\ntotal:$number";
}

sub by_agent{
   if($agent{$a} != $agent{$b}){
      $agent{$b} <=> $agent{$a};
   }
   else{
      $a cmp $b;
   }
}



sub referer1{
   $referer{$data[4]} = ++$referer{$data[4]};
}

sub referer2{
   $data[4] =~ s/\#.+//;
   $data[4] =~ s/\?.+/\*/;
   $data[4] =~ s/%7E/\~/i;
   $data[4] =~ s/:[0-9]+//;
   $referer{$data[4]} = ++$referer{$data[4]};
}

sub referer3{
   $data[4] =~ s/\#.+//;
   $data[4] =~ s/\?.+/\*/;
   $data[4] =~ s/%7E/\~/i;
   $data[4] =~ s/:[0-9]+//;

   if($data[4] =~ /yahoo.co.jp/i){
      $data[4] = 'http://www.yahoo.co.jp/*';
   }
   elsif($data[4] =~ /goo.ne.jp/i){
      $data[4] = 'http://www.goo.ne.jp/*';
   }
   elsif($data[4] =~ /navi.ntt.co.jp/i){
      $data[4] = 'http://navi.ntt.co.jp/*';
   }
   elsif($data[4] =~ /www.acara.com/i || $data[4] =~ /www.achara.com/i || $data[4] =~ /webdew.rnet.or.jp/){
      $data[4] = 'http://www.acara.com/*';
   }
   elsif($data[4] =~ /csj.co.jp\/csjindex/i || $data[4] =~ /csj.co.jp\/search/i){
      $data[4] = 'http://csj.co.jp/csjindex/*';
   }
   elsif($data[4] =~ /www.infoseek.co.jp/i){
      $data[4] = 'http://www.infoseek.co.jp/*';
   }
   elsif($data[4] =~ /www.lycos.co.jp/i){
      $data[4] = 'http://www.lycos.co.jp/*';
   }
   elsif($data[4] =~ /www.dragon.co.jp/i){
      $data[4] = 'http://www.dragon.co.jp/*';
   }
   elsif($data[4] =~ /file:/i){
      $data[4] = 'file';
   }
   $referer{$data[4]} = ++$referer{$data[4]};
}



sub referer_written1{
   local($persent);
   local(@sortedreferer);

   @sortedreferer = sort by_referer keys(%referer);

   foreach(@sortedreferer){
      $persent = &persent($referer{$_}, $number);
      print "$referer{$_} ($persent \%) : ";
      if ($_ ne ""){
         print "<A HREF =\"$_\" TARGET =\"_top\">$_</A>";
      }
      print "<BR>\n";
   }
   print "<BR>total:$number";
}


sub referer_written2{
   local($persent, $url);
   local(@sortedreferer);

   @sortedreferer = sort by_referer keys(%referer);

   foreach(@sortedreferer){
      $persent = &persent($referer{$_}, $number);
      print "$referer{$_} ($persent \%) : ";
      if ($_ ne ""){
         $url = $_;
         $url =~ s/\*//g if $_ =~ /\*/;
         print "<A HREF =\"$url\" TARGET =\"_top\">$_</A>";
      }
      print "<BR>\n";
   }
   print "<BR>total:$number";
}


sub by_referer{
   if($referer{$a} != $referer{$b}){
      $referer{$b} <=> $referer{$a};
   }
   else{
      $a cmp $b;
   }
}


sub uri1{
   $uri{$data[5]} = ++$uri{$data[5]};
   ++$numbers;
}


sub uri3{
   ($sec,$min,$hour,$mday,$mon,$year,$wday) = localtime($data[1]);
   $mon++;
   $mday =sprintf("%.2d",$mday);
   $mon =sprintf("%.2d",$mon);
   $year += 1900;

   if ($callday eq "$year\/$mon\/$mday") {
      $uri{$data[5]} = ++$uri{$data[5]};
   $hour =sprintf("%.2d",$hour);
   $hours{$hour} = ++$hours{$hour};
   $days{"$year\/$mon\/$mday"} = ++$days{"$year\/$mon\/$mday"};

      ++$number;
   }
}


sub uri_written1{
   local($persent);
   local(@sorteduri);

   @sorteduri = sort by_uri keys(%uri);

   foreach(@sorteduri){
      $persent = &persent($uri{$_}, $number);
      print "$uri{$_} ($persent \%) : <A HREF =\"datacheck.cgi?calluri=$_\" TARGET =\"_top\">$_</A><BR>\n";
   }
   print "total:$number\n";
}


sub uri_written3 {
   local($persent);
   local(@sortedreferer, @sortedhours);

   print "$callday<BR>\n";
   print "<H2>hours</H2>\n";
   @sortedhours = sort keys(%hours);
   foreach(@sortedhours){
      $persent = &persent($hours{$_}, $number);
      print "$_ - $hours{$_} ($persent \%)<BR>\n";
   }
   print "<HR>\n";

   print "<H2>access source</H2>\n";
   @sorteduri = sort by_uri keys(%uri);
   foreach(@sorteduri){
      $persent = &persent($uri{$_}, $number);
      print "$uri{$_} ($persent \%) - <A HREF =\"datacheck.cgi?calluri=$_&callday=$callday\" TARGET =\"_top\">$_</A><BR>\n";
   }
   print "<A HREF =\"datacheck.cgi?calluri=total&callday=$callday\" TARGET =\"_top\">total</A>:$number\n";
}


sub by_uri{
   if($uri{$a} != $uri{$b}){
      $uri{$b} <=> $uri{$a};
   }
   else{
      $a cmp $b;
   }
}



sub uri_referer1{
   if ($data[5] eq "$calluri\n") {
      $data[4] =~ s/\#.+//;
      $data[4] =~ s/%7E/\~/i;
      $uri_referer{$data[4]} = ++$uri_referer{$data[4]};
      ++$numbers;
   }
}


sub uri_referer2{
   ($sec,$min,$hour,$mday,$mon,$year,$wday) = localtime($data[1]);
   $mon++;
   $mday =sprintf("%.2d",$mday);
   $mon =sprintf("%.2d",$mon);
   $year += 1900;

   if ($data[5] eq "$calluri\n" && $callday eq "$year\/$mon\/$mday") {
      $data[4] =~ s/\#.+//;
      $data[4] =~ s/%7E/\~/i;
      $uri_referer{$data[4]} = ++$uri_referer{$data[4]};
      ++$numbers;
   }
}


sub uri_referer3{
   ($sec,$min,$hour,$mday,$mon,$year,$wday) = localtime($data[1]);
   $mon++;
   $mday =sprintf("%.2d",$mday);
   $mon =sprintf("%.2d",$mon);
   $year += 1900;

   if ($callday eq "$year\/$mon\/$mday") {
      $data[4] =~ s/\#.+//;
      $data[4] =~ s/%7E/\~/i;
      $uri_referer{$data[4]} = ++$uri_referer{$data[4]};
      ++$numbers;
   }
}


sub uri_referer_written1{
   local($persent);
   local(@sorteduri_referer);

   print "<A HREF =\"$calluri\">$calluri</A><BR>\n";

   @sorteduri_referer = sort by_uri_referer keys(%uri_referer);

   foreach(@sorteduri_referer){
      $persent = &persent($uri_referer{$_}, $numbers);
      print "$uri_referer{$_} ($persent \%) : <A HREF =\"$_\" TARGET =\"_top\">$_</A><BR>\n";
   }
   print "total:$numbers\n";
}


sub uri_referer_written2{
   local($persent);
   local(@sorteduri_referer);

   @sorteduri_referer = sort by_uri_referer keys(%uri_referer);

   print "$callday<BR>\n";

   foreach(@sorteduri_referer){
      $persent = &persent($uri_referer{$_}, $numbers);
      print "$uri_referer{$_} ($persent \%) : <A HREF =\"$_\" TARGET =\"_top\">$_</A><BR>\n";
   }
   print "total:$numbers\n";
}


sub by_uri_referer{
   if($uri_referer{$a} != $uri_referer{$b}){
      $uri_referer{$b} <=> $uri_referer{$a};
   }
   else{
      $a cmp $b;
   }
}

sub persent {
   local($group, $number) = @_;
   if ($number == 0) {
      0;
   }
   else {
      int(($group / $number) * 10000) / 100;
   }
}