X-Git-Url: https://git.heureux-cyclage.org/?a=blobdiff_plain;f=includes%2FLinkFilter.php;h=53841df12e10211f860c572120f15568b7b830da;hb=0a45ed6a91a3acc15cb6cc58a2b343a4bed0de16;hp=0008c6a1e28480d7372d75ddae74e62e4cc9bc77;hpb=34912bff40a595a9c34fa68683be53d6669e146a;p=lhc%2Fweb%2Fwiklou.git

diff --git a/includes/LinkFilter.php b/includes/LinkFilter.php
index 0008c6a1e2..53841df12e 100644
--- a/includes/LinkFilter.php
+++ b/includes/LinkFilter.php
@@ -3,29 +3,38 @@
 /**
  * Some functions to help implement an external link filter for spam control.
  * 
- * TODO: implement the filter. Currently these are just some functions to help
- * maintenance/cleanupSpam.php remove links to a single specified domain. The 
- * next thing is to implement functions for checking a given page against a big 
+ * @todo implement the filter. Currently these are just some functions to help
+ * maintenance/cleanupSpam.php remove links to a single specified domain. The
+ * next thing is to implement functions for checking a given page against a big
  * list of domains.
  *
  * Another cool thing to do would be a web interface for fast spam removal.
  */
 class LinkFilter {
+
 	/**
-	 * @static
+	 * Check whether $text contains a link to $filterEntry
+	 *
+	 * @param $text String: text to check
+	 * @param $filterEntry String: domainparts, see makeRegex() for more details
+	 * @return Integer: 0 if no match or 1 if there's at least one match
 	 */
-	function matchEntry( $text, $filterEntry ) {
+	static function matchEntry( $text, $filterEntry ) {
 		$regex = LinkFilter::makeRegex( $filterEntry );
 		return preg_match( $regex, $text );
 	}
 
 	/**
-	 * @static
+	 * Builds a regex pattern for $filterEntry.
+	 *
+	 * @param $filterEntry String: URL, if it begins with "*.", it'll be
+	 *        replaced to match any subdomain
+	 * @return String: regex pattern, for preg_match()
 	 */
-	function makeRegex( $filterEntry ) {
+	private static function makeRegex( $filterEntry ) {
 		$regex = '!http://';
 		if ( substr( $filterEntry, 0, 2 ) == '*.' ) {
-			$regex .= '([A-Za-z0-9.-]+\.|)';
+			$regex .= '(?:[A-Za-z0-9.-]+\.|)';
 			$filterEntry = substr( $filterEntry, 2 );
 		}
 		$regex .= preg_quote( $filterEntry, '!' ) . '!Si';
@@ -33,22 +42,61 @@ class LinkFilter {
 	}
 
 	/**
-	 * Make a string to go after an SQL LIKE, which will match the specified 
+	 * Make a string to go after an SQL LIKE, which will match the specified
 	 * string. There are several kinds of filter entry:
-	 *     *.domain.com    -  Produces http://com.domain.%, matches domain.com 
+	 *     *.domain.com    -  Produces http://com.domain.%, matches domain.com
 	 *                        and www.domain.com
 	 *     domain.com      -  Produces http://com.domain./%, matches domain.com
 	 *                        or domain.com/ but not www.domain.com
-	 *     *.domain.com/x  -  Produces http://com.domain.%/x%, matches 
+	 *     *.domain.com/x  -  Produces http://com.domain.%/x%, matches
 	 *                        www.domain.com/xy
-	 *     domain.com/x    -  Produces http://com.domain./x%, matches 
+	 *     domain.com/x    -  Produces http://com.domain./x%, matches
 	 *                        domain.com/xy but not www.domain.com/xy
 	 *
 	 * Asterisks in any other location are considered invalid.
 	 * 
-	 * @static
+	 * @param $filterEntry String: domainparts
+	 * @param $prot        String: protocol
+	 * @return String
+	 * @deprecated Use makeLikeArray() and pass result to Database::buildLike() instead
 	 */
-	function makeLike( $filterEntry ) {
+	 public static function makeLike( $filterEntry , $prot = 'http://' ) {
+		wfDeprecated( __METHOD__ );
+
+		$like = self::makeLikeArray( $filterEntry , $prot );
+		if ( !$like ) {
+			return false;
+		}
+		$dbw = wfGetDB( DB_MASTER );
+		$s = $dbw->buildLike( $like );
+		$m = false;
+		if ( preg_match( "/^ *LIKE '(.*)' *$/", $s, $m ) ) {
+			return $m[1];
+		} else {
+			throw new MWException( __METHOD__.': this DBMS is not supported by this function.' );
+		}
+	}
+
+	/**
+	 * Make an array to be used for calls to DatabaseBase::buildLike(), which
+	 * will match the specified string. There are several kinds of filter entry:
+	 *     *.domain.com    -  Produces http://com.domain.%, matches domain.com
+	 *                        and www.domain.com
+	 *     domain.com      -  Produces http://com.domain./%, matches domain.com
+	 *                        or domain.com/ but not www.domain.com
+	 *     *.domain.com/x  -  Produces http://com.domain.%/x%, matches
+	 *                        www.domain.com/xy
+	 *     domain.com/x    -  Produces http://com.domain./x%, matches
+	 *                        domain.com/xy but not www.domain.com/xy
+	 *
+	 * Asterisks in any other location are considered invalid.
+	 *
+	 * @param $filterEntry String: domainparts
+	 * @param $prot        String: protocol
+	 * @return Array to be passed to DatabaseBase::buildLike() or false on error
+	 */
+	 public static function makeLikeArray( $filterEntry , $prot = 'http://' ) {
+		$db = wfGetDB( DB_MASTER );
 		if ( substr( $filterEntry, 0, 2 ) == '*.' ) {
 			$subdomains = true;
 			$filterEntry = substr( $filterEntry, 2 );
@@ -61,7 +109,7 @@ class LinkFilter {
 			$subdomains = false;
 		}
 		// No stray asterisks, that could cause confusion
-		// It's not simple or efficient to handle it properly so we don't 
+		// It's not simple or efficient to handle it properly so we don't
 		// handle it at all.
 		if ( strpos( $filterEntry, '*' ) !== false ) {
 			return false;
@@ -74,19 +122,53 @@ class LinkFilter {
 			$path = '/';
 			$host = $filterEntry;
 		}
-		$host = strtolower( implode( '.', array_reverse( explode( '.', $host ) ) ) );
-		if ( substr( $host, -1, 1 ) !== '.' ) {
-			$host .= '.';
+		// Reverse the labels in the hostname, convert to lower case
+		// For emails reverse domainpart only
+		if ( $prot == 'mailto:' && strpos($host, '@') ) {
+			// complete email adress 
+			$mailparts = explode( '@', $host );
+			$domainpart = strtolower( implode( '.', array_reverse( explode( '.', $mailparts[1] ) ) ) );
+			$host = $domainpart . '@' . $mailparts[0];
+			$like = array( "$prot$host", $db->anyString() );
+		} elseif ( $prot == 'mailto:' ) {
+			// domainpart of email adress only. do not add '.'
+			$host = strtolower( implode( '.', array_reverse( explode( '.', $host ) ) ) );	
+			$like = array( "$prot$host", $db->anyString() );			
+		} else {
+			$host = strtolower( implode( '.', array_reverse( explode( '.', $host ) ) ) );	
+			if ( substr( $host, -1, 1 ) !== '.' ) {
+				$host .= '.';
+			}
+			$like = array( "$prot$host" );
+
+			if ( $subdomains ) {
+				$like[] = $db->anyString();
+			}
+			if ( !$subdomains || $path !== '/' ) {
+				$like[] = $path;
+				$like[] = $db->anyString();
+			}
 		}
-		$like = "http://$host";
-		
-		if ( $subdomains ) {
-			$like .= '%';
+		return $like;
+	}
+
+	/**
+	 * Filters an array returned by makeLikeArray(), removing everything past first pattern placeholder.
+	 *
+	 * @param $arr array: array to filter
+	 * @return filtered array
+	 */
+	public static function keepOneWildcard( $arr ) {
+		if( !is_array( $arr ) ) {
+			return $arr;
 		}
-		if ( !$subdomains || $path !== '/' ) {
-			$like .= $path . '%';
+
+		foreach( $arr as $key => $value ) {
+			if ( $value instanceof LikeMatch ) {
+				return array_slice( $arr, 0, $key + 1 );
+			}
 		}
-		return $like;
+
+		return $arr;
 	}
 }
-?>