LinkFilter.php

Go to the documentation of this file.

<?php
use Wikimedia\IPUtils;
use Wikimedia\Rdbms\LikeMatch;
 
class LinkFilter {
    public const VERSION = 1;
 
    public static function matchEntry( Content $content, $filterEntry, $protocol = 'http://' ) {
        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->getText();
 
        $regex = self::makeRegex( $filterEntry, $protocol );
        return preg_match( $regex, $text );
    }
 
    private static function makeRegex( $filterEntry, $protocol ) {
        $regex = '!' . preg_quote( $protocol, '!' );
        if ( substr( $filterEntry, 0, 2 ) == '*.' ) {
            $regex .= '(?:[A-Za-z0-9.-]+\.|)';
            $filterEntry = substr( $filterEntry, 2 );
        }
        $regex .= preg_quote( $filterEntry, '!' ) . '!Si';
        return $regex;
    }
 
    public static function supportsIDN() {
        return is_callable( 'idn_to_utf8' ) && defined( 'INTL_IDNA_VARIANT_UTS46' );
    }
 
    private static function indexifyHost( $host ) {
        // NOTE: If you change the output of this method, you'll probably have to increment self::VERSION!
 
        // Canonicalize.
        $host = rawurldecode( $host );
        if ( $host !== '' && self::supportsIDN() ) {
            // @todo Add a PHP fallback
            $tmp = idn_to_utf8( $host, IDNA_DEFAULT, INTL_IDNA_VARIANT_UTS46 );
            if ( $tmp !== false ) {
                $host = $tmp;
            }
        }
        $okChars = 'a-zA-Z0-9\\-._~!$&\'()*+,;=';
        if ( StringUtils::isUtf8( $host ) ) {
            // Save a little space by not percent-encoding valid UTF-8 bytes
            $okChars .= '\x80-\xf4';
        }
        $host = preg_replace_callback(
            '<[^' . $okChars . ']>',
            static function ( $m ) {
                return rawurlencode( $m[0] );
            },
            strtolower( $host )
        );
 
        // IPv6? RFC 3986 syntax.
        if ( preg_match( '/^\[([0-9a-f:*]+)\]$/', rawurldecode( $host ), $m ) ) {
            $ip = $m[1];
            if ( IPUtils::isValid( $ip ) ) {
                return 'V6.' . implode( '.', explode( ':', IPUtils::sanitizeIP( $ip ) ) ) . '.';
            }
            if ( substr( $ip, -2 ) === ':*' ) {
                $cutIp = substr( $ip, 0, -2 );
                if ( IPUtils::isValid( "{$cutIp}::" ) ) {
                    // Wildcard IP doesn't contain "::", so multiple parts can be wild
                    $ct = count( explode( ':', $ip ) ) - 1;
                    return 'V6.' .
                        implode( '.', array_slice( explode( ':', IPUtils::sanitizeIP( "{$cutIp}::" ) ), 0, $ct ) ) .
                        '.*.';
                }
                if ( IPUtils::isValid( "{$cutIp}:1" ) ) {
                    // Wildcard IP does contain "::", so only the last part is wild
                    return 'V6.' .
                        substr( implode( '.', explode( ':', IPUtils::sanitizeIP( "{$cutIp}:1" ) ) ), 0, -1 ) .
                        '*.';
                }
            }
        }
 
        // Regularize explicit specification of the DNS root.
        // Browsers seem to do this for IPv4 literals too.
        if ( substr( $host, -1 ) === '.' ) {
            $host = substr( $host, 0, -1 );
        }
 
        // IPv4?
        $b = '(?:0*25[0-5]|0*2[0-4][0-9]|0*1[0-9][0-9]|0*[0-9]?[0-9])';
        if ( preg_match( "/^(?:{$b}\.){3}{$b}$|^(?:{$b}\.){1,3}\*$/", $host ) ) {
            return 'V4.' . implode( '.', array_map( static function ( $v ) {
                return $v === '*' ? $v : (int)$v;
            }, explode( '.', $host ) ) ) . '.';
        }
 
        // Must be a host name.
        return implode( '.', array_reverse( explode( '.', $host ) ) ) . '.';
    }
 
    public static function makeIndexes( $url ) {
        // NOTE: If you change the output of this method, you'll probably have to increment self::VERSION!
 
        // NOTE: refreshExternallinksIndex.php assumes that only protocol-relative URLs return more
        // than one index, and that the indexes for protocol-relative URLs only vary in the "http://"
        // versus "https://" prefix. If you change that, you'll likely need to update
        // refreshExternallinksIndex.php accordingly.
 
        $bits = wfParseUrl( $url );
        if ( !$bits ) {
            return [];
        }
 
        // Reverse the labels in the hostname, convert to lower case, unless it's an IP.
        // For emails turn it into "domain.reversed@localpart"
        if ( $bits['scheme'] == 'mailto' ) {
            $mailparts = explode( '@', $bits['host'], 2 );
            if ( count( $mailparts ) === 2 ) {
                $domainpart = self::indexifyHost( $mailparts[1] );
            } else {
                // No @, assume it's a local part with no domain
                $domainpart = '';
            }
            $bits['host'] = $domainpart . '@' . $mailparts[0];
        } else {
            $bits['host'] = self::indexifyHost( $bits['host'] );
        }
 
        // Reconstruct the pseudo-URL
        $index = $bits['scheme'] . $bits['delimiter'] . $bits['host'];
        // Leave out user and password. Add the port, path, query and fragment
        if ( isset( $bits['port'] ) ) {
            $index .= ':' . $bits['port'];
        }
        $index .= $bits['path'] ?? '/';
        if ( isset( $bits['query'] ) ) {
            $index .= '?' . $bits['query'];
        }
        if ( isset( $bits['fragment'] ) ) {
            $index .= '#' . $bits['fragment'];
        }
 
        if ( $bits['scheme'] == '' ) {
            return [ "http:$index", "https:$index" ];
        } else {
            return [ $index ];
        }
    }
 
    public static function getQueryConditions( $filterEntry, array $options = [] ) {
        $options += [
            'protocol' => 'http://',
            'oneWildcard' => false,
            'prefix' => 'el',
            'db' => null,
        ];
 
        // First, get the like array
        $like = self::makeLikeArray( $filterEntry, $options['protocol'] );
        if ( $like === false ) {
            return $like;
        }
 
        // Get the constant prefix (i.e. everything up to the first wildcard)
        $trimmedLike = self::keepOneWildcard( $like );
        if ( $options['oneWildcard'] ) {
            $like = $trimmedLike;
        }
        if ( $trimmedLike[count( $trimmedLike ) - 1] instanceof LikeMatch ) {
            array_pop( $trimmedLike );
        }
        $index = implode( '', $trimmedLike );
 
        $p = $options['prefix'];
        $db = $options['db'] ?: wfGetDB( DB_REPLICA );
 
        // Build the query
        $l = strlen( $index );
        if ( $l >= 60 ) {
            // The constant prefix is larger than el_index_60, so we can use a
            // constant comparison.
            return [
                "{$p}_index_60" => substr( $index, 0, 60 ),
                "{$p}_index" . $db->buildLike( $like ),
            ];
        }
 
        // The constant prefix is smaller than el_index_60, so we use a LIKE
        // for a prefix search.
        return [
            "{$p}_index_60" . $db->buildLike( $index, $db->anyString() ),
            "{$p}_index" . $db->buildLike( $like ),
        ];
    }
 
    public static function makeLikeArray( $filterEntry, $protocol = 'http://' ) {
        $db = wfGetDB( DB_REPLICA );
        $like = [];
 
        $target = $protocol . $filterEntry;
        $bits = wfParseUrl( $target );
        if ( !$bits ) {
            return false;
        }
 
        $subdomains = false;
        if ( $bits['scheme'] === 'mailto' && strpos( $bits['host'], '@' ) ) {
            // Email address with domain and non-empty local part
            $mailparts = explode( '@', $bits['host'], 2 );
            $domainpart = self::indexifyHost( $mailparts[1] );
            if ( $mailparts[0] === '*' ) {
                $subdomains = true;
                $bits['host'] = $domainpart . '@';
            } else {
                $bits['host'] = $domainpart . '@' . $mailparts[0];
            }
        } else {
            // Non-email, or email with only a domain part.
            $bits['host'] = self::indexifyHost( $bits['host'] );
            if ( substr( $bits['host'], -3 ) === '.*.' ) {
                $subdomains = true;
                $bits['host'] = substr( $bits['host'], 0, -2 );
            }
        }
 
        $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;
    }
}