includes/LinkFilter.php

   1 <?php
   2
   3 /**
   4  * Some functions to help implement an external link filter for spam control.
   5  *
   6  * TODO: implement the filter. Currently these are just some functions to help
   7  * maintenance/cleanupSpam.php remove links to a single specified domain. The
   8  * next thing is to implement functions for checking a given page against a big
   9  * list of domains.
  10  *
  11  * Another cool thing to do would be a web interface for fast spam removal.
  12  */
  13 class LinkFilter {
  14         /**
  15          * @static
  16          */
  17         static function matchEntry( $text, $filterEntry ) {
  18                 $regex = LinkFilter::makeRegex( $filterEntry );
  19                 return preg_match( $regex, $text );
  20         }
  21
  22         /**
  23          * @static
  24          */
  25         private static function makeRegex( $filterEntry ) {
  26                 $regex = '!http://';
  27                 if ( substr( $filterEntry, 0, 2 ) == '*.' ) {
  28                         $regex .= '(?:[A-Za-z0-9.-]+\.|)';
  29                         $filterEntry = substr( $filterEntry, 2 );
  30                 }
  31                 $regex .= preg_quote( $filterEntry, '!' ) . '!Si';
  32                 return $regex;
  33         }
  34
  35         /**
  36          * Make a string to go after an SQL LIKE, which will match the specified
  37          * string. There are several kinds of filter entry:
  38          *     *.domain.com    -  Produces http://com.domain.%, matches domain.com
  39          *                        and www.domain.com
  40          *     domain.com      -  Produces http://com.domain./%, matches domain.com
  41          *                        or domain.com/ but not www.domain.com
  42          *     *.domain.com/x  -  Produces http://com.domain.%/x%, matches
  43          *                        www.domain.com/xy
  44          *     domain.com/x    -  Produces http://com.domain./x%, matches
  45          *                        domain.com/xy but not www.domain.com/xy
  46          *
  47          * Asterisks in any other location are considered invalid.
  48          *
  49          * @static
  50          * @param $filterEntry String: domainparts
  51          * @param $prot        String: protocol
  52          */
  53          public static function makeLike( $filterEntry , $prot = 'http://' ) {
  54                 if ( substr( $filterEntry, 0, 2 ) == '*.' ) {
  55                         $subdomains = true;
  56                         $filterEntry = substr( $filterEntry, 2 );
  57                         if ( $filterEntry == '' ) {
  58                                 // We don't want to make a clause that will match everything,
  59                                 // that could be dangerous
  60                                 return false;
  61                         }
  62                 } else {
  63                         $subdomains = false;
  64                 }
  65                 // No stray asterisks, that could cause confusion
  66                 // It's not simple or efficient to handle it properly so we don't
  67                 // handle it at all.
  68                 if ( strpos( $filterEntry, '*' ) !== false ) {
  69                         return false;
  70                 }
  71                 $slash = strpos( $filterEntry, '/' );
  72                 if ( $slash !== false ) {
  73                         $path = substr( $filterEntry, $slash );
  74                         $host = substr( $filterEntry, 0, $slash );
  75                 } else {
  76                         $path = '/';
  77                         $host = $filterEntry;
  78                 }
  79                 // Reverse the labels in the hostname, convert to lower case
  80                 // For emails reverse domainpart only
  81                 if ( $prot == 'mailto:' && strpos($host, '@') ) {
  82                         // complete email adress
  83                         $mailparts = explode( '@', $host );
  84                         $domainpart = strtolower( implode( '.', array_reverse( explode( '.', $mailparts[1] ) ) ) );
  85                         $host = $domainpart . '@' . $mailparts[0];
  86                         $like = "$prot$host%";
  87                 } elseif ( $prot == 'mailto:' ) {
  88                         // domainpart of email adress only. do not add '.'
  89                         $host = strtolower( implode( '.', array_reverse( explode( '.', $host ) ) ) );
  90                         $like = "$prot$host%";
  91                 } else {
  92                         $host = strtolower( implode( '.', array_reverse( explode( '.', $host ) ) ) );
  93                         if ( substr( $host, -1, 1 ) !== '.' ) {
  94                                 $host .= '.';
  95                         }
  96                         $like = "$prot$host";
  97
  98                         if ( $subdomains ) {
  99                                 $like .= '%';
 100                         }
 101                         if ( !$subdomains || $path !== '/' ) {
 102                                 $like .= $path . '%';
 103                         }
 104                 }
 105                 return $like;
 106         }
 107 }
 108 ?>