Fix HTML entities in XML feeds

functions.php

/**
 * Fix HTML entities in XML feeds
 */
add_filter( 'adt_product_feed_xml_attribute_handling', function( $handled, $product, $key, $value, $feed_config, $channel_attributes, $feed ) {
    if ( $handled || ! is_object( $product ) || ! is_object( $feed ) ) {
        return $handled;
    }

    $target_feed_id = 37281;
    $is_target_feed = false;
    
    if ( method_exists( $feed, 'get_id' ) ) {
        try {
            $feed_id = (int) $feed->get_id();
            $is_target_feed = ( $feed_id === $target_feed_id );
        } catch ( \Exception $e ) {
            // If get_id() fails, check by channel name instead
            if ( isset( $feed_config['name'] ) && ( $feed_config['name'] === 'Skroutz' || $feed_config['name'] === 'Custom Feed' ) ) {
                $is_target_feed = true;
            }
        }
    } else {
        // Fallback: check by channel name if get_id() method doesn't exist
        if ( isset( $feed_config['name'] ) && ( $feed_config['name'] === 'Skroutz' || $feed_config['name'] === 'Custom Feed' ) ) {
            $is_target_feed = true;
        }
    }
    
    if ( ! $is_target_feed ) {
        return $handled;
    }

    // Skip variations field (handled separately)
    if ( 'skroutz_variations' === $key ) {
        return $handled;
    }

    // Skip if field already exists in XML
    if ( isset( $product->$key ) ) {
        return $handled;
    }

    // Only process string values
    if ( ! is_string( $value ) && ! is_numeric( $value ) ) {
        return $handled;
    }

    // Convert to string
    $decoded = (string) $value;

    // Skip empty values
    if ( '' === trim( $decoded ) ) {
        return $handled;
    }

    try {
        // Decode HTML entities (handles double/triple encoding)
        $previous   = '';
        $iterations = 0;
        $max_iterations = 10;
        
        while ( $decoded !== $previous && $iterations < $max_iterations ) {
            $previous = $decoded;
            $decoded  = html_entity_decode( $decoded, ENT_QUOTES | ENT_XML1 | ENT_HTML5, 'UTF-8' );
            $iterations++;
        }

        // Validate UTF-8 and remove invalid XML characters
        $decoded = wp_check_invalid_utf8( $decoded, true );
        
        if ( class_exists( '\AdTribes\PFP\Helpers\Sanitization' ) ) {
            $decoded = \AdTribes\PFP\Helpers\Sanitization::sanitize_utf8_for_xml( $decoded );
        }

        // Sanitize element name to prevent XML injection
        $sanitized_key = preg_replace( '/[^a-zA-Z0-9_\-:]/', '', $key );
        if ( '' === $sanitized_key ) {
            $sanitized_key = 'value';
        }

        // Use DOMDocument to add the node with proper UTF-8 handling
        // This ensures plain UTF-8 text without HTML entities
        $dom = dom_import_simplexml( $product );
        
        if ( $dom && $dom->ownerDocument ) {
            try {
                $child = $dom->ownerDocument->createElement( $sanitized_key );
                if ( $child ) {
                    $text_node = $dom->ownerDocument->createTextNode( $decoded );
                    if ( $text_node ) {
                        $child->appendChild( $text_node );
                        $dom->appendChild( $child );
                        return true; // Successfully handled, skip default processing
                    }
                }
            } catch ( \DOMException $e ) {
                // Log error but don't break feed generation
                if ( function_exists( 'wc_get_logger' ) ) {
                    $logger = wc_get_logger();
                    $logger->warning( 
                        'DOM operation failed in HTML entities fix: ' . $e->getMessage(),
                        array( 'source' => 'woo-product-feed-pro', 'key' => $key )
                    );
                }
                // Fall through to SimpleXML fallback
            }
        }

        // Fallback to SimpleXML if DOM operations fail
        // Note: This will still encode entities, but at least the feed won't break
        try {
            $fallback_value = htmlspecialchars( $decoded, ENT_XML1, 'UTF-8' );
            $product->addChild( $sanitized_key, $fallback_value );
            return true; // Handled, skip default processing
        } catch ( \Exception $e ) {
            // If even SimpleXML fails, log and let default handling proceed
            if ( function_exists( 'wc_get_logger' ) ) {
                $logger = wc_get_logger();
                $logger->error( 
                    'Failed to add XML element in HTML entities fix: ' . $e->getMessage(),
                    array( 'source' => 'woo-product-feed-pro', 'key' => $key )
                );
            }
            // Return false to let default handling proceed
            return $handled;
        }

    } catch ( \Exception $e ) {
        // Catch any unexpected errors and log them
        if ( function_exists( 'wc_get_logger' ) ) {
            $logger = wc_get_logger();
            $logger->error( 
                'Unexpected error in HTML entities fix: ' . $e->getMessage(),
                array( 'source' => 'woo-product-feed-pro', 'key' => $key )
            );
        }
        // Return false to let default handling proceed
        return $handled;
    }

}, 9, 7 );