inc/htmlparse.inc

   1 <?php
   2
   3 class htmlparse {
   4
   5 function htmlparse($data)
   6 {
   7    /*
   8    if function finds anything unsafe,it will return
   9    FALSE and saves a reason info global variable $htmlparse
  10    */
  11
  12    global $htmlparse;
  13
  14    $data = StrToLower(" ".$data);
  15
  16    // tags, I don\14 need to close
  17    $unpaired = Array('br'=>1,
  18                      'br/'=>1, // fix later ;)
  19                      'li'=>1,
  20                      'hr'=>1,
  21                      'img'=>1,
  22                      'p'=>1
  23                      );
  24
  25    // allowed tags
  26    $allowed = Array('b'=>1,
  27                     'i'=>1,
  28                     'u'=>1,
  29                     'a'=>1,
  30                     'img'=>1,
  31                     'sup'=>1,
  32                     'sub'=>1,
  33                     'table'=>1,
  34                     'tr'=>1,
  35                     'td'=>1,
  36                     'font'=>1,
  37                     'ul'=>1,
  38                     'ol'=>1,
  39                     'li'=>1,
  40                     'tt'=>1,
  41                     'code'=>1,
  42                     'small'=>1,
  43                     'big'=>1,
  44                     'caption'=>1,
  45                     'thead'=>1,
  46                     'tfoot'=>1,
  47                     'col'=>1,
  48                     'colgroup'=>1,
  49                     'th'=>1,
  50                     'br'=>1,
  51                     'br/'=>1, // fix later
  52                     'hr'=>1,
  53                     'em'=>1,
  54                     'th'=>1,
  55                     'center'=>1,
  56                     'pre'=>1,
  57                     'xmp'=>1,
  58                     's'=>1,
  59                     'strong'=>1,
  60                     'legend'=>1,
  61                     'h1'=>1,
  62                     'h2'=>1,
  63                     'h3'=>1,
  64                     'h4'=>1,
  65                     'h5'=>1,
  66                     'h6'=>1,
  67                     'p'=>1,
  68                     'blockquote'=>1,
  69                     'div'=>1,
  70                     'span'=>1,
  71                     'fieldset'=>1
  72                     );
  73
  74    /*
  75    this part will go trought string and will ensure, if all tags are closed
  76    */
  77
  78    $tok = StrTok($data, '<');
  79    $tok = StrTok('<');
  80    while(!($tok === FALSE)){
  81       if(!StrStr($tok,'>')):
  82          $htmlparse = 'Chyba HTML syntaxe!';
  83          //$htmlparse = 'Wrong HTML syntax!';
  84          return 0;
  85       elseif(StrStr($tok,"<")):
  86          $htmlparse = 'Chyba HTML syntaxe!';
  87          //$htmlparse = 'Wrong HTML syntax!';
  88          return 0;
  89       endif;
  90       $tok = StrTok('<');
  91    }
  92
  93    /*
  94    main part of the function - it will check allowed tags, some parameters and so on...
  95    */
  96
  97    $tok = StrTok($data, '<');
  98    $i = 0;
  99    $j = 0;
 100    while(!($tok === FALSE)):
 101       if($i == 1):
 102          $tag = Split('>',$tok,2);
 103          $attrib = Split("[[:space:]>]",$tag[0],2);
 104          if($allowed[$attrib[0]] != 1 && $allowed[SubStr($attrib[0],1)] != 1): // if tag isn\14 in allowed array
 105             $htmlparse = 'Zakazany tag &lt;'.$attrib[0].'&gt;!';
 106             //$htmlparse = 'Forbidden tag &lt;'.$attrib[0].'&gt;!';
 107             return 0;
 108          endif;
 109          if('/'.$tags[$j] == $attrib[0]): // closing tag for last opening tag
 110             if($tags[$j] == 'table' && $opened_tables > 0):
 111                $opened_tables--;
 112             endif;
 113             $j--;
 114          elseif($tags[$j] == 'xmp'): // XMP tag...ignore eny other tags between them
 115          else:
 116             if(SubStr($attrib[0],0,1) == '/' && $unpaired[$tags[$j]]): // do I need to close the tag?
 117                $j--;
 118                continue;
 119             elseif(SubStr($attrib[0],0,1) == '/'): // am I closing something, I didn\14 open?
 120                $htmlparse = 'Chyba u tagu &lt;'.$tag[0].'&gt;! Zavirate tag, ktery jste neotevrel!';
 121                //$htmlparse = 'Error near tag &lt;'.$tag[0].'&gt;! Closing tag, that wasn\14 opened!';
 122                return 0;
 123             elseif(Ereg(' on',' '.$attrib[1])): // temporary solution for pernament problem...and it isn\14 suicide
 124                $htmlparse = 'JavaScript je na hovno!';
 125                //$htmlparse = 'JavaScript sux!';
 126                return 0;
 127             elseif(Ereg(' style',' '.$attrib[1])): // styles are forbidden - don\14 look at me THAT way ;)
 128                $htmlparse = 'Ten "style" se mi tam nezda!';
 129                //$htmlparse = '"styles" are forbidden!';
 130                return 0;
 131             elseif(Ereg('://',' '.$attrib[1]) && $attrib[0] != "img" && $attrib[0] != "a"): // adresses in attributes (except A and IMG tags) are forbidden
 132                $htmlparse = 'Neco se mi tam nelibi! To je hlaska HTML validace - nejedna se o nejakou cenzuru ;)';
 133                //$htmlparse = 'Forbidden usage of adresses in tags!';
 134                return 0;
 135             elseif((SubStr_Count($attrib[1],'"')%2) > 0): // are quotes closed? can do mess if they aren\14
 136                $htmlparse = 'Neuzavrel jste uvozovky uvnitr tagu &lt;'.$attrib[0].'&gt;!';
 137                //$htmlparse = 'Close quotes in tag &lt;'.$tag[0].'&gt;!';
 138                return 0;
 139             elseif(Ereg('\?',$attrib[1]) && $attrib[0] == 'img'): // don\14 allow parameters in IMG tags
 140                $htmlparse = 'Chyba u tagu &lt;img&gt; - nejsou povoleny parametry v adrese!';
 141                //$htmlparse = 'Error in tag &lt;img&gt; - parameters in image adresses are forbidden!';
 142                return 0;
 143             elseif(($attrib[0] == 'td' || $attrib[0] == 'tr') && $opened_tables == 0):
 144                $htmlparse = 'Strkej si ty tagy do vlastni tabulky, jo?';
 145                return 0;
 146             elseif($attrib[0] == 'table'):
 147                $opened_tables++;
 148             endif;
 149             $j++;
 150             $tags[$j] = $attrib[0];
 151          endif;
 152       endif;
 153       $tok = StrTok('<');
 154       $i = 1;
 155    endwhile;
 156
 157    /*
 158    just check, if all tags are properly closed
 159    */
 160
 161    while($j > 0):
 162       if($unpaired[$tags[$j]]):
 163          $j--;
 164          continue;
 165       else:
 166          $htmlparse = 'Neuzavrel jste tag &lt;'.$tags[$j].'&gt;!';
 167          //$htmlparse = 'Tag &lt;'.$tags[$j].'&gt; wasn\14 closed correctly!';
 168          return 0;
 169       endif;
 170    endwhile;
 171 return 1;
 172 }
 173
 174 }