LinkFilter.php

Go to the documentation of this file.

<?php
use Wikimedia\Rdbms\LikeMatch;
 
class LinkFilter {
 
    static function matchEntry( Content $content, $filterEntry ) {
        if ( !( $content instanceof TextContent ) ) {
            // TODO: handle other types of content too.
            //      Maybe create ContentHandler::matchFilter( LinkFilter ).
            //      Think about a common base class for LinkFilter and MagicWord.
            return 0;
        }
 
        $text = $content->getNativeData();
 
        $regex = self::makeRegex( $filterEntry );
        return preg_match( $regex, $text );
    }
 
    private static function makeRegex( $filterEntry ) {
        $regex = '!http://';
        if ( substr( $filterEntry, 0, 2 ) == '*.' ) {
            $regex .= '(?:[A-Za-z0-9.-]+\.|)';
            $filterEntry = substr( $filterEntry, 2 );
        }
        $regex .= preg_quote( $filterEntry, '!' ) . '!Si';
        return $regex;
    }
 
    public static function makeLikeArray( $filterEntry, $protocol = 'http://' ) {
        $db = wfGetDB( DB_REPLICA );
 
        $target = $protocol . $filterEntry;
        $bits = wfParseUrl( $target );
 
        if ( $bits == false ) {
            // Unknown protocol?
            return false;
        }
 
        if ( substr( $bits['host'], 0, 2 ) == '*.' ) {
            $subdomains = true;
            $bits['host'] = substr( $bits['host'], 2 );
            if ( $bits['host'] == '' ) {
                // We don't want to make a clause that will match everything,
                // that could be dangerous
                return false;
            }
        } else {
            $subdomains = false;
        }
 
        // Reverse the labels in the hostname, convert to lower case
        // For emails reverse domainpart only
        if ( $bits['scheme'] === 'mailto' && strpos( $bits['host'], '@' ) ) {
            // complete email address
            $mailparts = explode( '@', $bits['host'] );
            $domainpart = strtolower( implode( '.', array_reverse( explode( '.', $mailparts[1] ) ) ) );
            $bits['host'] = $domainpart . '@' . $mailparts[0];
        } elseif ( $bits['scheme'] === 'mailto' ) {
            // domainpart of email address only, do not add '.'
            $bits['host'] = strtolower( implode( '.', array_reverse( explode( '.', $bits['host'] ) ) ) );
        } else {
            $bits['host'] = strtolower( implode( '.', array_reverse( explode( '.', $bits['host'] ) ) ) );
            if ( substr( $bits['host'], -1, 1 ) !== '.' ) {
                $bits['host'] .= '.';
            }
        }
 
        $like[] = $bits['scheme'] . $bits['delimiter'] . $bits['host'];
 
        if ( $subdomains ) {
            $like[] = $db->anyString();
        }
 
        if ( isset( $bits['port'] ) ) {
            $like[] = ':' . $bits['port'];
        }
        if ( isset( $bits['path'] ) ) {
            $like[] = $bits['path'];
        } elseif ( !$subdomains ) {
            $like[] = '/';
        }
        if ( isset( $bits['query'] ) ) {
            $like[] = '?' . $bits['query'];
        }
        if ( isset( $bits['fragment'] ) ) {
            $like[] = '#' . $bits['fragment'];
        }
 
        // Check for stray asterisks: asterisk only allowed at the start of the domain
        foreach ( $like as $likepart ) {
            if ( !( $likepart instanceof LikeMatch ) && strpos( $likepart, '*' ) !== false ) {
                return false;
            }
        }
 
        if ( !( $like[count( $like ) - 1] instanceof LikeMatch ) ) {
            // Add wildcard at the end if there isn't one already
            $like[] = $db->anyString();
        }
 
        return $like;
    }
 
    public static function keepOneWildcard( $arr ) {
        if ( !is_array( $arr ) ) {
            return $arr;
        }
 
        foreach ( $arr as $key => $value ) {
            if ( $value instanceof LikeMatch ) {
                return array_slice( $arr, 0, $key + 1 );
            }
        }
 
        return $arr;
    }
}