WordPress: Auto-Import External Images to Media Library (Block Editor Compatible)

<?php
/**
 * Plugin Name: RR — Import External Images on Save
 * Description: On save (draft/publish), sideload external <img> to Media Library, rewrite src, and set alt/title/description from post title + increment.
 * Version: 2.3.0
 * 
 * Changelog:
 * 2.3.0 - Fixed block validation by removing srcset/sizes from saved content
 *       - Gutenberg adds srcset/sizes automatically on frontend render
 *       - Block attributes now perfectly match saved HTML
 *       - No more "unexpected or invalid content" errors
 * 2.2.0 - Complete block rebuild for validation
 * 2.1.0 - Fixed duplicate uploads
 * 2.0.0 - Major update: Full Gutenberg compatibility
 * 1.4.0 - Previous stable version
 */

if ( ! defined( 'ABSPATH' ) ) { exit; }

/**
 * ============================================
 * CONFIGURATION OPTIONS
 * ============================================
 */
define( 'RR_EXTIMG_SET_TITLE',       false );
define( 'RR_EXTIMG_SET_ALT',         false );
define( 'RR_EXTIMG_SET_CAPTION',     false );
define( 'RR_EXTIMG_SET_DESCRIPTION', false );
define( 'RR_EXTIMG_DEFAULT_SIZE',    'large' );

add_action( 'save_post', 'rr_import_external_images_on_save', 10, 3 );
add_action( 'rest_after_insert_post', 'rr_import_external_images_rest', 10, 3 );
add_action( 'rest_after_insert_page', 'rr_import_external_images_rest', 10, 3 );

function rr_import_external_images_rest( $post, $request, $creating ) {
    rr_import_external_images_on_save( $post->ID, $post, ! $creating );
}

function rr_import_external_images_on_save( $post_id, $post, $update ) {
    if ( defined( 'DOING_AUTOSAVE' ) && DOING_AUTOSAVE ) return;
    if ( wp_is_post_revision( $post_id ) || wp_is_post_autosave( $post_id ) ) return;
    if ( empty( $post ) || empty( $post->post_content ) ) return;

    rr_require_media_libs();

    $ptype = get_post_type( $post );
    if ( ! post_type_supports( $ptype, 'editor' ) ) return;

    $content = $post->post_content;
    
    if ( strlen( $content ) > 5000000 ) return;
    
    $content = apply_filters( 'rr_extimg_pre_process_content', $content, $post_id );
    
    // Get existing map
    $map = get_post_meta( $post_id, '_rr_extimg_map', true );
    if ( ! is_array( $map ) ) $map = [];

    $site_host = wp_parse_url( home_url(), PHP_URL_HOST );
    $changed = false;
    $image_counter = 1;

    // Process Gutenberg image blocks
    $content = preg_replace_callback(
        '#<!-- wp:image(\s+\{[^}]*\})?\s*-->.*?<!-- /wp:image -->#s',
        function( $matches ) use ( &$map, &$image_counter, $post_id, $post, $site_host, &$changed ) {
            $block = $matches[0];
            
            // Extract the current URL from JSON or img src
            $src = '';
            
            // Try JSON first
            if ( preg_match( '#"url"\s*:\s*"([^"]+)"#', $block, $url_match ) ) {
                $src = trim( html_entity_decode( stripslashes( $url_match[1] ) ) );
            }
            
            // Fallback to img src
            if ( empty( $src ) && preg_match( '#<img[^>]+src=(["\'])([^"\']+)\1#i', $block, $src_match ) ) {
                $src = trim( html_entity_decode( $src_match[2] ) );
            }
            
            if ( empty( $src ) || ! wp_http_validate_url( $src ) ) {
                return $block;
            }
            
            $host = wp_parse_url( $src, PHP_URL_HOST );
            if ( ! $host || strcasecmp( $host, $site_host ) === 0 ) return $block;

            // Check if already processed
            $attach_id = $map[ $src ] ?? 0;
            
            if ( ! $attach_id ) {
                // Check if image already exists in media library
                $existing_id = attachment_url_to_postid( $src );
                if ( $existing_id ) {
                    $attach_id = $existing_id;
                    rr_fill_attachment_texts_if_empty( $existing_id, $src, $post, $image_counter );
                } else {
                    // Download and sideload
                    $attach_id = rr_sideload_image_as_attachment( $src, $post_id, $post, $image_counter );
                    if ( is_wp_error( $attach_id ) || ! $attach_id ) {
                        return $block;
                    }
                }
                
                // Save to map
                $map[ $src ] = $attach_id;
            }

            if ( $attach_id ) {
                $new_block = rr_rebuild_image_block( $block, $attach_id );
                if ( $new_block !== $block ) {
                    $changed = true;
                    $image_counter++;
                    return $new_block;
                }
            }
            
            return $block;
        },
        $content
    );

    // Process any remaining standalone img tags (classic editor)
    $content = preg_replace_callback(
        '#<img[^>]+>#i',
        function( $matches ) use ( &$map, &$image_counter, $post_id, $post, $site_host, &$changed ) {
            $img_tag = $matches[0];
            
            if ( ! preg_match( '#\s+src=(["\'])([^"\']+)\1#i', $img_tag, $src_match ) ) {
                return $img_tag;
            }
            
            $src = trim( html_entity_decode( $src_match[2] ) );
            
            if ( ! wp_http_validate_url( $src ) ) return $img_tag;
            
            $host = wp_parse_url( $src, PHP_URL_HOST );
            if ( ! $host || strcasecmp( $host, $site_host ) === 0 ) return $img_tag;
            
            // Skip if already has wp-image class (already processed)
            if ( preg_match( '#wp-image-\d+#', $img_tag ) ) {
                return $img_tag;
            }

            $attach_id = $map[ $src ] ?? 0;
            
            if ( ! $attach_id ) {
                $existing_id = attachment_url_to_postid( $src );
                if ( $existing_id ) {
                    $attach_id = $existing_id;
                    rr_fill_attachment_texts_if_empty( $existing_id, $src, $post, $image_counter );
                } else {
                    $attach_id = rr_sideload_image_as_attachment( $src, $post_id, $post, $image_counter );
                    if ( is_wp_error( $attach_id ) || ! $attach_id ) {
                        return $img_tag;
                    }
                }
                $map[ $src ] = $attach_id;
            }

            if ( $attach_id ) {
                // Extract existing attributes
                $alt = '';
                $class = '';
                
                if ( preg_match( '#\s+alt=(["\'])([^"\']*)\1#i', $img_tag, $alt_match ) ) {
                    $alt = $alt_match[2];
                }
                if ( preg_match( '#\s+class=(["\'])([^"\']*)\1#i', $img_tag, $class_match ) ) {
                    $class = $class_match[2];
                }
                
                $new_img = rr_generate_wp_image_html( $attach_id, [
                    'alt' => $alt,
                    'class' => $class,
                ] );
                
                if ( $new_img !== $img_tag ) {
                    $changed = true;
                    $image_counter++;
                    return $new_img;
                }
            }
            
            return $img_tag;
        },
        $content
    );

    if ( $changed ) {
        remove_action( 'save_post', 'rr_import_external_images_on_save', 10 );
        remove_action( 'rest_after_insert_post', 'rr_import_external_images_rest', 10 );
        remove_action( 'rest_after_insert_page', 'rr_import_external_images_rest', 10 );
        
        wp_update_post( [ 'ID' => $post_id, 'post_content' => $content ] );
        update_post_meta( $post_id, '_rr_extimg_map', $map );
        
        add_action( 'save_post', 'rr_import_external_images_on_save', 10, 3 );
        add_action( 'rest_after_insert_post', 'rr_import_external_images_rest', 10, 3 );
        add_action( 'rest_after_insert_page', 'rr_import_external_images_rest', 10, 3 );
    }
}

/**
 * Completely rebuild the image block with proper WordPress structure
 */
function rr_rebuild_image_block( string $old_block, int $attach_id ) : string {
    $size = RR_EXTIMG_DEFAULT_SIZE;
    
    // Get image metadata
    $meta = wp_get_attachment_metadata( $attach_id );
    $full_url = wp_get_attachment_url( $attach_id );
    
    // Get the sized image
    $image = wp_get_attachment_image_src( $attach_id, $size );
    if ( ! $image ) {
        $image = wp_get_attachment_image_src( $attach_id, 'full' );
        $size = 'full';
    }
    
    if ( ! $image ) return $old_block;
    
    list( $src, $width, $height ) = $image;
    
    // Get alt text (from AI plugin or metadata)
    $alt = get_post_meta( $attach_id, '_wp_attachment_image_alt', true );
    if ( empty( $alt ) ) {
        // Try to extract from old block
        if ( preg_match( '#<img[^>]+alt=(["\'])([^"\']*)\1#i', $old_block, $alt_match ) ) {
            $alt = $alt_match[2];
        }
    }
    
    // Extract any existing caption
    $caption = '';
    if ( preg_match( '#<figcaption[^>]*>(.*?)</figcaption>#s', $old_block, $caption_match ) ) {
        $caption = trim( $caption_match[1] );
    }
    
    // Extract existing linkDestination if present
    $link_destination = 'none';
    if ( preg_match( '#"linkDestination"\s*:\s*"([^"]+)"#', $old_block, $link_match ) ) {
        $link_destination = $link_match[1];
    }
    
    // Extract href if image is linked
    $href = '';
    if ( preg_match( '#<a[^>]+href=(["\'])([^"\']+)\1#i', $old_block, $href_match ) ) {
        $href = $href_match[2];
    }
    
    // Build block attributes JSON
    $attrs = [
        'id' => $attach_id,
        'sizeSlug' => $size,
        'linkDestination' => $link_destination,
    ];
    
    // Add width/height if available from metadata
    if ( isset( $meta['sizes'][ $size ] ) ) {
        $attrs['width'] = (int) $meta['sizes'][ $size ]['width'];
        $attrs['height'] = (int) $meta['sizes'][ $size ]['height'];
    } elseif ( $size === 'full' && isset( $meta['width'] ) && isset( $meta['height'] ) ) {
        $attrs['width'] = (int) $meta['width'];
        $attrs['height'] = (int) $meta['height'];
    }
    
    $attrs_json = wp_json_encode( $attrs, JSON_UNESCAPED_SLASHES );
    
    // Build srcset and sizes
    $srcset = wp_get_attachment_image_srcset( $attach_id, $size );
    $sizes_attr = wp_get_attachment_image_sizes( $attach_id, $size );
    
    // Build the img tag - DO NOT include srcset/sizes
    // Gutenberg will add them automatically on the frontend
    $img_tag = sprintf(
        '<img src="%s" alt="%s" class="wp-image-%d"/>',
        esc_url( $src ),
        esc_attr( $alt ),
        $attach_id
    );
    
    // Wrap in link if needed
    if ( ! empty( $href ) ) {
        $img_tag = sprintf( '<a href="%s">%s</a>', esc_url( $href ), $img_tag );
    }
    
    // Build figure
    $figure_classes = 'wp-block-image size-' . $size;
    $figure = sprintf( '<figure class="%s">%s', esc_attr( $figure_classes ), $img_tag );
    
    if ( ! empty( $caption ) ) {
        $figure .= '<figcaption class="wp-element-caption">' . $caption . '</figcaption>';
    }
    
    $figure .= '</figure>';
    
    // Build complete block
    $new_block = sprintf(
        "<!-- wp:image %s -->\n%s\n<!-- /wp:image -->",
        $attrs_json,
        $figure
    );
    
    return $new_block;
}

/**
 * Generate proper WordPress image HTML with all required attributes
 * For classic editor content (not Gutenberg blocks)
 */
function rr_generate_wp_image_html( int $attach_id, array $original_data ) : string {
    $size = RR_EXTIMG_DEFAULT_SIZE;
    
    $image = wp_get_attachment_image_src( $attach_id, $size );
    if ( ! $image ) {
        $image = wp_get_attachment_image_src( $attach_id, 'full' );
    }
    
    if ( ! $image ) return '';
    
    list( $src, $width, $height ) = $image;
    
    $alt = get_post_meta( $attach_id, '_wp_attachment_image_alt', true );
    if ( empty( $alt ) && ! empty( $original_data['alt'] ) ) {
        $alt = $original_data['alt'];
    }
    
    $classes = [ "wp-image-{$attach_id}" ];
    
    if ( ! empty( $original_data['class'] ) ) {
        $existing_classes = explode( ' ', $original_data['class'] );
        foreach ( $existing_classes as $cls ) {
            $cls = trim( $cls );
            if ( $cls && ! preg_match( '/^wp-image-\d+$/', $cls ) ) {
                $classes[] = $cls;
            }
        }
    }
    
    $class_attr = implode( ' ', array_unique( $classes ) );
    
    // For classic editor, we CAN include srcset since it's not validated by Gutenberg
    $srcset = wp_get_attachment_image_srcset( $attach_id, $size );
    $sizes_attr = wp_get_attachment_image_sizes( $attach_id, $size );
    
    $img_html = sprintf(
        '<img src="%s" alt="%s" class="%s"',
        esc_url( $src ),
        esc_attr( $alt ),
        esc_attr( $class_attr )
    );
    
    if ( $srcset ) {
        $img_html .= sprintf( ' srcset="%s"', esc_attr( $srcset ) );
    }
    
    if ( $sizes_attr ) {
        $img_html .= sprintf( ' sizes="%s"', esc_attr( $sizes_attr ) );
    }
    
    $img_html .= '/>';
    
    return $img_html;
}

function rr_sideload_image_as_attachment( string $url, int $post_id, $post, int $counter ) {
    if ( strpos( $url, 'googleusercontent.com' ) !== false || 
         strpos( $url, 'docs.google.com' ) !== false ) {
        $url = rr_resolve_redirect( $url );
    }
    
    $path_parts = pathinfo( wp_parse_url( $url, PHP_URL_PATH ) );
    $ext = strtolower( $path_parts['extension'] ?? '' );
    $allowed = [ 'jpg', 'jpeg', 'png', 'gif', 'webp', 'svg', 'ico', 'bmp' ];
    
    if ( ! in_array( $ext, $allowed, true ) ) {
        return new WP_Error( 'invalid_image', 'Unsupported image format: ' . $ext );
    }
    
    $post_title = $post->post_title;
    if ( empty( $post_title ) ) {
        $post_title = 'image';
    }
    
    $custom_filename = rr_generate_filename( $post_title, $counter, $ext );
    
    $temp_file = download_url( $url );
    if ( is_wp_error( $temp_file ) ) {
        return $temp_file;
    }
    
    $file_array = [
        'name'     => $custom_filename,
        'tmp_name' => $temp_file,
    ];
    
    $id = media_handle_sideload( $file_array, $post_id );
    
    if ( file_exists( $temp_file ) ) {
        @unlink( $temp_file );
    }
    
    if ( is_wp_error( $id ) ) {
        return $id;
    }

    $path = get_attached_file( $id );
    if ( $path ) {
        $meta = wp_generate_attachment_metadata( $id, $path );
        if ( $meta ) wp_update_attachment_metadata( $id, $meta );
    }

    update_post_meta( $id, '_rr_source_url', esc_url_raw( $url ) );
    
    // Only fill texts if configured - allows AI plugins to run first
    rr_fill_attachment_texts_if_empty( $id, $url, $post, $counter );

    return (int) $id;
}

function rr_generate_filename( string $post_title, int $counter, string $ext ) : string {
    $slug = sanitize_title( $post_title );
    $slug = preg_replace( '/[^a-z0-9\-]/', '', $slug );
    $slug = substr( $slug, 0, 50 );
    
    if ( $counter > 1 ) {
        $filename = $slug . '-' . $counter . '.' . $ext;
    } else {
        $filename = $slug . '.' . $ext;
    }
    
    return $filename;
}

function rr_resolve_redirect( string $url ) : string {
    $response = wp_safe_remote_head( $url, [
        'timeout'     => 10,
        'redirection' => 5,
    ] );
    
    if ( is_wp_error( $response ) ) {
        return $url;
    }
    
    $final_url = wp_remote_retrieve_header( $response, 'location' );
    return $final_url ? $final_url : $url;
}

/**
 * Fill attachment metadata only if enabled and field is empty
 * This allows AI plugins and other metadata plugins to run first
 */
function rr_fill_attachment_texts_if_empty( int $attach_id, string $fallback_url = '', $post = null, int $counter = 1 ) : void {
    if ( ! RR_EXTIMG_SET_TITLE && ! RR_EXTIMG_SET_ALT && ! RR_EXTIMG_SET_CAPTION && ! RR_EXTIMG_SET_DESCRIPTION ) {
        return;
    }
    
    $label = '';
    
    if ( $post && ! empty( $post->post_title ) ) {
        $label = $post->post_title;
        if ( $counter > 1 ) {
            $label .= ' ' . $counter;
        }
    } else {
        $path = get_attached_file( $attach_id );
        $base = '';

        if ( $path && file_exists( $path ) ) {
            $pi   = pathinfo( $path );
            $base = $pi['filename'] ?? '';
        }
        if ( ! $base && $fallback_url ) {
            $pi   = pathinfo( wp_parse_url( $fallback_url, PHP_URL_PATH ) ?? '' );
            $base = $pi['filename'] ?? '';
        }
        
        $label = rr_humanize_filename( $base );
    }
    
    if ( $label === '' ) return;

    if ( RR_EXTIMG_SET_TITLE ) {
        $att = get_post( $attach_id );
        if ( $att && empty( trim( $att->post_title ) ) ) {
            wp_update_post( [
                'ID'         => $attach_id,
                'post_title' => $label,
            ] );
        }
    }

    if ( RR_EXTIMG_SET_ALT ) {
        $alt = get_post_meta( $attach_id, '_wp_attachment_image_alt', true );
        if ( empty( $alt ) ) {
            update_post_meta( $attach_id, '_wp_attachment_image_alt', $label );
        }
    }

    if ( RR_EXTIMG_SET_CAPTION ) {
        $cur_caption = get_post_field( 'post_excerpt', $attach_id );
        if ( empty( trim( $cur_caption ) ) ) {
            wp_update_post( [
                'ID'           => $attach_id,
                'post_excerpt' => $label,
            ] );
        }
    }
    
    if ( RR_EXTIMG_SET_DESCRIPTION ) {
        $cur_desc = get_post_field( 'post_content', $attach_id );
        if ( empty( trim( $cur_desc ) ) ) {
            wp_update_post( [
                'ID'           => $attach_id,
                'post_content' => $label,
            ] );
        }
    }
}

function rr_humanize_filename( string $name ) : string {
    $name = urldecode( $name );
    $name = preg_replace( '/[_\-]+/u', ' ', $name );
    $name = preg_replace( '/\s+/u', ' ', $name );
    $name = trim( $name );
    if ( $name === '' ) return '';
    
    if ( function_exists( 'mb_convert_case' ) ) {
        $name = mb_convert_case( $name, MB_CASE_TITLE, 'UTF-8' );
    } else {
        $name = ucwords( strtolower( $name ) );
    }
    
    return sanitize_text_field( $name );
}

function rr_require_media_libs() : void {
    if ( ! function_exists( 'download_url' ) ) {
        require_once ABSPATH . 'wp-admin/includes/file.php';
    }
    if ( ! function_exists( 'wp_generate_attachment_metadata' ) ) {
        require_once ABSPATH . 'wp-admin/includes/image.php';
    }
    if ( ! function_exists( 'media_sideload_image' ) ) {
        require_once ABSPATH . 'wp-admin/includes/media.php';
    }
}