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          * @deprecated Use makeLikeArray() and pass result to Database::buildLike() instead
  53          */
  54          public static function makeLike( $filterEntry , $prot = 'http://' ) {
  55                 wfDeprecated( __METHOD__ );
  56
  57                 $like = self::makeLikeArray( $filterEntry , $prot );
  58                 if ( !$like ) {
  59                         return false;
  60                 }
  61                 $dbw = wfGetDB( DB_MASTER );
  62                 $s = $dbw->buildLike( $like );
  63                 $m = false;
  64                 if ( preg_match( "/^ *LIKE '(.*)' *$/", $s, $m ) ) {
  65                         return $m[1];
  66                 } else {
  67                         throw new MWException( __METHOD__.': this DBMS is not supported by this function.' );
  68                 }
  69         }
  70
  71         /**
  72          * Make an array to be used for calls to Database::buildLike(), which will match the specified
  73          * string. There are several kinds of filter entry:
  74          *     *.domain.com    -  Produces http://com.domain.%, matches domain.com
  75          *                        and www.domain.com
  76          *     domain.com      -  Produces http://com.domain./%, matches domain.com
  77          *                        or domain.com/ but not www.domain.com
  78          *     *.domain.com/x  -  Produces http://com.domain.%/x%, matches
  79          *                        www.domain.com/xy
  80          *     domain.com/x    -  Produces http://com.domain./x%, matches
  81          *                        domain.com/xy but not www.domain.com/xy
  82          *
  83          * Asterisks in any other location are considered invalid.
  84          *
  85          * @static
  86          * @param $filterEntry String: domainparts
  87          * @param $prot        String: protocol
  88          */
  89          public static function makeLikeArray( $filterEntry , $prot = 'http://' ) {
  90                 $db = wfGetDB( DB_MASTER );
  91                 if ( substr( $filterEntry, 0, 2 ) == '*.' ) {
  92                         $subdomains = true;
  93                         $filterEntry = substr( $filterEntry, 2 );
  94                         if ( $filterEntry == '' ) {
  95                                 // We don't want to make a clause that will match everything,
  96                                 // that could be dangerous
  97                                 return false;
  98                         }
  99                 } else {
 100                         $subdomains = false;
 101                 }
 102                 // No stray asterisks, that could cause confusion
 103                 // It's not simple or efficient to handle it properly so we don't
 104                 // handle it at all.
 105                 if ( strpos( $filterEntry, '*' ) !== false ) {
 106                         return false;
 107                 }
 108                 $slash = strpos( $filterEntry, '/' );
 109                 if ( $slash !== false ) {
 110                         $path = substr( $filterEntry, $slash );
 111                         $host = substr( $filterEntry, 0, $slash );
 112                 } else {
 113                         $path = '/';
 114                         $host = $filterEntry;
 115                 }
 116                 // Reverse the labels in the hostname, convert to lower case
 117                 // For emails reverse domainpart only
 118                 if ( $prot == 'mailto:' && strpos($host, '@') ) {
 119                         // complete email adress
 120                         $mailparts = explode( '@', $host );
 121                         $domainpart = strtolower( implode( '.', array_reverse( explode( '.', $mailparts[1] ) ) ) );
 122                         $host = $domainpart . '@' . $mailparts[0];
 123                         $like = array( "$prot$host", $db->anyString() );
 124                 } elseif ( $prot == 'mailto:' ) {
 125                         // domainpart of email adress only. do not add '.'
 126                         $host = strtolower( implode( '.', array_reverse( explode( '.', $host ) ) ) );
 127                         $like = array( "$prot$host", $db->anyString() );
 128                 } else {
 129                         $host = strtolower( implode( '.', array_reverse( explode( '.', $host ) ) ) );
 130                         if ( substr( $host, -1, 1 ) !== '.' ) {
 131                                 $host .= '.';
 132                         }
 133                         $like = array( "$prot$host" );
 134
 135                         if ( $subdomains ) {
 136                                 $like[] = $db->anyString();
 137                         }
 138                         if ( !$subdomains || $path !== '/' ) {
 139                                 $like[] = $path;
 140                                 $like[] = $db->anyString();
 141                         }
 142                 }
 143                 return $like;
 144         }
 145
 146         /**
 147          * Filters an array returned by makeLikeArray(), removing everything past first pattern placeholder.
 148          * @static
 149          * @param $arr array: array to filter
 150          * @return filtered array
 151          */
 152         public static function keepOneWildcard( $arr ) {
 153                 if( !is_array( $arr ) ) {
 154                         return $arr;
 155                 }
 156
 157                 foreach( $arr as $key => $value ) {
 158                         if ( $value instanceof LikeMatch ) {
 159                                 return array_slice( $arr, 0, $key + 1 );
 160                         }
 161                 }
 162
 163                 return $arr;
 164         }
 165 }