Import Media from /Uploads, From old site to new site, Copy Paste Uploads Folder and Import

/**
 * Description: Adds an admin page that queues and imports original media files from existing uploads directories, skipping resized images.
 */

/**
 * Define an option key for the import queue.
 */
define( 'MY_MEDIA_IMPORT_QUEUE_OPTION', 'my_media_import_queue' );

/**
 * Add a custom "Import Media" submenu under the Media admin menu.
 */
add_action( 'admin_menu', function() {
	add_submenu_page(
		'upload.php',                      // Parent slug (Media menu).
		'Import Media',                    // Page title.
		'Import Media',                    // Menu title.
		'manage_options',                  // Capability.
		'custom-media-import',             // Menu slug.
		'my_custom_media_import_page_cb'   // Callback function.
	);
});

/**
 * Render the Import Media admin page.
 */
function my_custom_media_import_page_cb() {
	// Check for form submission.
	if ( isset( $_POST['import_media_nonce'] ) && 
	     wp_verify_nonce( $_POST['import_media_nonce'], 'import_media_action' ) ) {

		// Build the queue from existing upload folders.
		$queue = my_custom_setup_media_import_queue();

		if ( ! empty( $queue ) ) {
			// Schedule our cron event if not already scheduled.
			if ( ! wp_next_scheduled( 'my_media_import_cron_event' ) ) {
				wp_schedule_single_event( time() + 5, 'my_media_import_cron_event' );
			}
			echo '<div class="notice notice-success is-dismissible"><p>Import queued: ' . count( $queue ) . ' files to process.</p></div>';
		} else {
			echo '<div class="notice notice-info is-dismissible"><p>No new original files were found to import.</p></div>';
		}
	}

	// If there is a queue already saved, show the remaining count.
	$queue = get_option( MY_MEDIA_IMPORT_QUEUE_OPTION, array() );
	?>
	<div class="wrap">
		<h1>Import Media from Uploads</h1>
		<p>
			This tool scans the <code>wp-content/uploads/YYYY/MM</code> directories and queues <strong>only original files</strong> for import—skipping any resized versions (e.g. <code>image-150x150.jpg</code>).
			It then processes the queue in batches for optimized, background importing.
		</p>
		<form method="post">
			<?php wp_nonce_field( 'import_media_action', 'import_media_nonce' ); ?>
			<input type="submit" class="button button-primary" value="Start Import">
		</form>
		<?php if ( ! empty( $queue ) ) : ?>
			<p><strong>Import in progress:</strong> <?php echo count( $queue ); ?> file(s) remaining in the queue.</p>
		<?php endif; ?>
	</div>
	<?php
}

/**
 * Scan the uploads folders (YYYY/MM) and build a queue of original files.
 * Files matching the pattern -[width]x[height] (e.g., image-150x150.jpg) are skipped.
 *
 * @return array List of files to import.
 */
function my_custom_setup_media_import_queue() {
	$wp_uploads = wp_upload_dir();
	$base_dir   = $wp_uploads['basedir']; // e.g., /var/www/html/wp-content/uploads
	
	if ( ! is_dir( $base_dir ) ) {
		return array();
	}

	$queue = array();

	$years = scandir( $base_dir );
	if ( ! $years ) {
		return $queue;
	}

	foreach ( $years as $year ) {
		if ( $year === '.' || $year === '..' ) {
			continue;
		}

		$year_path = $base_dir . '/' . $year;
		if ( ! is_dir( $year_path ) || ! preg_match( '/^\d{4}$/', $year ) ) {
			continue;
		}

		$months = scandir( $year_path );
		if ( ! $months ) {
			continue;
		}

		foreach ( $months as $month ) {
			if ( $month === '.' || $month === '..' ) {
				continue;
			}

			$month_path = $year_path . '/' . $month;
			if ( ! is_dir( $month_path ) || ! preg_match( '/^\d{2}$/', $month ) ) {
				continue;
			}

			$files = scandir( $month_path );
			if ( ! $files ) {
				continue;
			}

			foreach ( $files as $file ) {
				if ( $file === '.' || $file === '..' ) {
					continue;
				}

				$full_path = $month_path . '/' . $file;
				if ( is_dir( $full_path ) ) {
					continue;
				}

				// Skip files that look like a resized image (e.g., image-150x150.jpg).
				if ( preg_match( '/-\d+x\d+(?=\.[^.]+$)/i', $file ) ) {
					continue;
				}

				// Check for valid media type.
				$filetype = wp_check_filetype( $file, null );
				if ( ! $filetype['type'] ) {
					continue;
				}

				$queue[] = array(
					'full_path' => $full_path,
					'filename'  => $file,
					'year'      => $year,
					'month'     => $month,
					'mime_type' => $filetype['type'],
				);
			}
		}
	}

	// Save the queue into an option.
	update_option( MY_MEDIA_IMPORT_QUEUE_OPTION, $queue );
	return $queue;
}

/**
 * WP Cron callback to process a batch of queued media imports.
 * Processes 50 files per run.
 */
function my_media_import_cron_event_handler() {
	$queue = get_option( MY_MEDIA_IMPORT_QUEUE_OPTION, array() );
	if ( empty( $queue ) ) {
		return;
	}

	$batch_size = 50;
	$processed  = 0;

	foreach ( $queue as $key => $item ) {
		if ( $processed >= $batch_size ) {
			break;
		}

		// Check if file exists; if not, remove it from the queue.
		if ( ! file_exists( $item['full_path'] ) ) {
			unset( $queue[ $key ] );
			continue;
		}

		// Check if the attachment already exists.
		if ( attachment_exists_by_filename( $item['filename'] ) ) {
			unset( $queue[ $key ] );
			continue;
		}

		// Import the file.
		$attach_id = import_file_as_attachment( $item['full_path'], $item['filename'], $item['mime_type'], $item['year'], $item['month'] );
		$processed++;

		// Remove processed item from the queue.
		unset( $queue[ $key ] );
	}

	// Re-index the queue.
	$queue = array_values( $queue );
	update_option( MY_MEDIA_IMPORT_QUEUE_OPTION, $queue );

	// If there are still queued files, schedule the next batch.
	if ( ! empty( $queue ) ) {
		wp_schedule_single_event( time() + 10, 'my_media_import_cron_event' );
	}
}
add_action( 'my_media_import_cron_event', 'my_media_import_cron_event_handler' );

/**
 * Helper function: Check whether an attachment with the same file name already exists.
 *
 * NOTE: This is a simple check that uses the attachment post title.
 *
 * @param string $filename File name.
 * @return bool True if already exists; false otherwise.
 */
function attachment_exists_by_filename( $filename ) {
	global $wpdb;
	$query = $wpdb->prepare(
		"SELECT COUNT(*) FROM $wpdb->posts 
		 WHERE post_type = 'attachment' 
		   AND post_title = %s",
		pathinfo( $filename, PATHINFO_FILENAME )
	);
	$count = $wpdb->get_var( $query );
	return ( $count > 0 );
}

/**
 * Helper function: Import a single file into the Media Library.
 *
 * The file's last modified time is used as the attachment's post_date.
 *
 * @param string $full_path Full file path.
 * @param string $filename  File name.
 * @param string $mime_type MIME type.
 * @param string $year      Year from folder structure.
 * @param string $month     Month from folder structure.
 * @return int|false Attachment ID on success, or false on failure.
 */
function import_file_as_attachment( $full_path, $filename, $mime_type, $year, $month ) {
	// Use the file's last modified time as post_date.
	$filetime   = filemtime( $full_path );
	$post_date  = ( $filetime ) ? date( 'Y-m-d H:i:s', $filetime ) : current_time( 'mysql' );

	$attachment = array(
		'guid'           => $full_path,
		'post_mime_type' => $mime_type,
		'post_title'     => pathinfo( $filename, PATHINFO_FILENAME ),
		'post_content'   => '',
		'post_status'    => 'inherit',
		'post_date'      => $post_date,
		'post_date_gmt'  => get_gmt_from_date( $post_date ),
	);

	// Insert the attachment into the database.
	$attach_id = wp_insert_attachment( $attachment, $full_path );

	if ( ! is_wp_error( $attach_id ) ) {
		require_once( ABSPATH . 'wp-admin/includes/image.php' );
		$attach_data = wp_generate_attachment_metadata( $attach_id, $full_path );
		if ( ! is_wp_error( $attach_data ) ) {
			wp_update_attachment_metadata( $attach_id, $attach_data );
		}
		return $attach_id;
	}

	return false;
}