#!/usr/bin/perl
# a40402.pl 2008-12-4 by akebia
# Nippongo search string decoder for Webalizer
# http://nyan.co.uk/akebia/item/81

use Jcode;

@search_engine = (
  ['[^/]*/search\?',      'q=', '=', 'q=cache:.+\+.*', 'h', 'www.google.com', 's', '/([&\?])q=cache:.+\+/$1q=/', ';'],
  ['[^/]*/translate_c\?', 'q=', '=', 'prev=/.*',       'h', 'www.google.com', ';'],
  ['google\.', 'q=', '=', 'prev=/.*', 's', '/\/[^\/]*(\?|\?.*&)prev=\//\//', 'u', ';'],
  ['google\.', 'q=', '=', 'ie=[^uU].*'],
  ['google\.', 'q=', '=', 'ie=[uU].*', '.'],
  ['google\.', 'q=', 'e', 'btnG=.*\x8c\x9f\x8d\xf5.*'],
  ['google\.', 'q=', 'e', 'btnG=.*\xe6\xa4\x9c\xe7\xb4\xa2.*', '.'],
  ['goo\.ne\.jp/',      'MT=', '!', 'IE=UTF-8'],
  ['yahoo\.co\.jp/',    'p=',  '!', 'ei=UTF-8'],
  ['yahoofs\.jp/',      'p=',  '!', 'ei=UTF-8'],
  ['livedoor\.com/',    'q=',  '!', 'ie=utf8'],
  ['fresheye\.com/',    'kw='],
  ['excite\.co\.jp/',   'search='],
  ['excite-cache\.jp/', 'search=', '=', 'preurl=.+', 's', '/.*[&\?]preurl=//', 'u'],
  ['nifty\.com/',       'Text='],
  ['biglobe\.ne\.jp/',  'q='],
  ['msn\.co\.jp/',      'q=',  '=', 'cp=(932|50220|51932)'],
  ['ceek\.jp/',         'q='],
  ['[^/]*/search/cache\?', 'q=',  'h', 'search.yahoo.com', 's', '/\/search\/cache\?/\/search\?/', '.'],
  ['[^/]*/webcp.asp\?',    'q=',  'h', 'ask.jp', 's', '/\/webcp\.asp\?/\/web\.asp\?/', '.'],
  ['.*', '', '.'],
);

$normalize = 0;
$zenkaku_moji = "　！＃＄’（）＊＋，－．／０１２３４５６７８９：；＜＞？＠ＡＢＣＤＥＦＧＨＩＪＫＬＭＮＯＰＱＲＳＴＵＶＷＸＹＺ［］＾＿‘ａｂｃｄｅｆｇｈｉｊｋｌｍｎｏｐｑｒｓｔｕｖｗｘｙｚ｛｜｝";
$hankaku_moji = " !#\$'()*+,-./0123456789:;<>?\@ABCDEFGHIJKLMNOPQRSTUVWXYZ[]^_`abcdefghijklmnopqrstuvwxyz{|}";

if($ARGV[0] eq '-n'){$normalize = 1;}

sub url_decode_x #($s)
{
  $_[0] =~ s/\+|\x22|\x0d\x0a|\x0d|\x0a|%22|%25|%26|%3d|%0d%0a|%0d|%0a/ /gi;
  $_[0] =~ s/%([a-f\d]{2})/pack('H2',$1)/egi;
}

sub check_encode #($a, $s, \@e, $i)
{
  my $a = \$_[0]; my $s = \$_[1];
  my $e =  $_[2]; my $i =  $_[3];

  while($$e[$i] ne ''){
       if($$e[$i] eq '!'){if($$s =~ /[\?&]$$e[$i + 1](&|\s*)/){return 1;}}
    elsif($$e[$i] eq '='){if($$s !~ /[\?&]$$e[$i + 1](&|\s*)/){return 1;}}
    elsif($$e[$i] eq 'n'){if($$a =~ /[\?&]$$e[$i + 1](&|\s*)/){return 1;}}
    elsif($$e[$i] eq 'e'){if($$a !~ /[\?&]$$e[$i + 1](&|\s*)/){return 1;}}
    elsif($$e[$i] eq '.'){return 2;}
    elsif($$e[$i] eq ';'){return 1;}
    elsif($$e[$i] eq 'h'){$$s =~ s/^http:\/\/[^\/]*\//http:\/\/$$e[$i + 1]\//;}
    elsif($$e[$i] eq 's'){eval("\$\$s =~ s$$e[$i + 1]");}
    elsif($$e[$i] eq 'u'){$$s =~ s/%([a-f\d]{2})/pack('H2',$1)/egi;
                          $$a = $$s; url_decode_x($$a);      $i--;}
    $i += 2;
  }
    return 0;
}

my $a, @a, $e, @e, @f, @g, $r;

while(<STDIN>){
    @a = split(/\x22/);
    if($a[6] eq "\n"){$a[6] = " \n";}
  for($e = 1; $e <= 5; $e += 2){
    if($a[$e] eq '' || $a[$e] eq '-'){next;}
    $a[$e] =~ s/([\$\@])/\\$1/g; $a[$e] = eval('"'.$a[$e].'"');
    $a[$e] =~ s/\x22/%22/gi;
  } if($a[3] eq '' || $a[3] eq '-'){print join("\x22", @a); next;}
    $a = $a[3]; url_decode_x($a);
  foreach $e (@search_engine){
    if($a[3] !~ /^http:\/\/([^\/]+\.|)$$e[0]/){next;}
    $r = check_encode($a, $a[3], $e, 2);
    if($r == 1){next;} url_decode_x($a[3]); if($r == 2){last;}
    @f = split(/[&\?]$$e[1]/, $a[3]); if($f[1] eq ''){next;}
    @g = split(/&/, $f[1]); $g[0] = Jcode->new(\$g[0])->utf8;
    $f[1] = join("&", @g);
    $a[3] = join(substr($a[3], length($f[0]), 1).$$e[1], @f);
    last;
  }
    if($normalize){$a[3] = Jcode->new(\$a[3], 'utf8')->h2z->tr($zenkaku_moji, $hankaku_moji)->utf8;}
    print join("\x22", @a);
}