Custom WP-CLI command to import a CSV file into a post type

class-cli-import-csv.php

<?php
/**
 * ExamplePackage\CLI_Import_CSV Class.
 *
 * @package ExamplePackage
 */

namespace ExamplePackage;

if ( ! defined( '\WP_CLI' ) ) {
	return;
}

use WP_CLI;
use WPCOM_VIP_CLI_Command;

/**
 * Adds import-csv-example CLI command.
 */
class CLI_Import_CSV {

	/**
	 * Default associative args.
	 *
	 * @var array
	 */
	public array $defaults = array();

	/**
	 * List of items used for the named args.
	 *
	 * This will be added to the assoc_args using the provided keys.
	 *
	 * @var array
	 */
	public array $args = array();

	/**
	 * The combined associative array of args.
	 *
	 * @var array
	 */
	public array $assoc_args = array();

	/**
	 * Rows from the CSV file.
	 *
	 * @var array
	 */
	public array $rows = array();

	/**
	 * The current row.
	 *
	 * @var array
	 */
	public array $row = array();

	/**
	 * Posts to be added.
	 *
	 * @var array
	 */
	public array $posts = array();

	/**
	 * Post IDs that have been inserted.
	 *
	 * @var array
	 */
	public array $post_ids = array();

	/**
	 * The number of posts that have been inserted.
	 *
	 * @var int
	 */
	public int $count = 0;

	/**
	 * CSV constructor.
	 */
	public function __construct() {
		$this->defaults = array(
			'site-id'   => 0,
			'post-type' => null,
			'dry-run'   => ''
		);

		$this->args = array(
			'data-source'
		);
	}

	/**
	 * Sets the $assoc_args property.
	 *
	 * Uses the $args parameter to set the named args in the $assoc_args property
	 *
	 * @param array $args       The arguments.
	 * @param array $assoc_args Associative array of args.
	 */
	public function set_args( array $args, array $assoc_args ): void {
		$this->assoc_args = array_merge( $this->defaults, $assoc_args );

		$count = 0;

		if ( ! empty( $this->args ) ) {
			foreach ( $this->args as $name ) {
				$this->assoc_args[ $name ] = empty( $args[ $count ] ) ? '' : $args[ $count ++ ];
			}
		}
	}

	/**
	 * Checks to see if a flag is set and returns true if so otherwise false.
	 *
	 * @param string $flag The item to check.
	 *
	 * @return bool
	 */
	public function get_flag_value( string $flag ): bool {
		return ! empty( $this->assoc_args[ $flag ] );
	}

	/**
	 * Imports a CSV file into a post type.
	 *
	 * Only supports attachment post type. Useful when orphan media files don't have
	 * associated attachment post type data in WordPress.
	 *
	 * Important – File URLs must use the "/uploads/YYYY/MM/" directory structure.
	 *
	 * ## OPTIONS
	 *
	 * <data-source>
	 * : Local server absolute path to the csv file to import.
	 *
	 * [--site-id=<id>]
	 * : Network site ID.
	 *
	 * [--post-type=<slug>]
	 * : Post type slug (e.g., attachment).
	 *
	 * [--dry-run]
	 * : Indicates to test the import instead of doing the import.
	 *
	 * ## EXAMPLES
	 *
	 * // VIP Local Development Environment
	 * wp import-csv-example /wp/wp-content/themes/theme/attachments.csv --site-id=2
	 * --post-type=attachment
	 *
	 * // VIP Environment
	 * wp import-csv-example /var/www/wp-content/themes/theme/attachments.csv --site-id=2
	 * --post-type=attachment
	 *
	 * @param array $args       The arguments.
	 * @param array $assoc_args Associative array of args.
	 */
	public function __invoke( array $args, array $assoc_args ): void {
		if ( empty( $args[0] ) ) {
			WP_CLI::error( __( 'The file is required.' ) );
		}

		$this->set_args( $args, $assoc_args );

		if ( ! $this->get_flag_value( 'site-id' ) ) {
			WP_CLI::error( __( 'The site ID is required.' ) );
		}

		if ( ! $this->get_flag_value( 'post-type' ) ) {
			WP_CLI::error( __( 'The post type is required.' ) );
		}

		if ( ! $this->assoc_args['dry-run'] ) {
			WP_CLI::confirm( 'Are you sure you want to import into ' . get_site_url( $this->assoc_args['site-id'] ) . '?' );
		}

		if ( ! defined( 'WP_IMPORTING' ) ) {
			define( 'WP_IMPORTING', true );
		}

		if ( class_exists( 'WPCOM_VIP_CLI_Command' ) ) {
			WPCOM_VIP_CLI_Command::start_bulk_operation();
		}

		switch_to_blog( $this->assoc_args['site-id'] );

		WP_CLI::log( WP_CLI::colorize( '%bParsing CSV file:%n ' ) . $this->assoc_args['data-source'] );

		$this->open_file();

		if ( 'attachment' === $this->assoc_args['post-type'] ) {
			$this->prepare_attachment_posts();
		}
		
		WP_CLI::log( WP_CLI::colorize( '%bInserting ' . count( $this->posts ) . ' post(s)%n' ) . '...' );

		$this->insert_posts();
		$this->finish();

		if ( class_exists( 'WPCOM_VIP_CLI_Command' ) ) {
			WPCOM_VIP_CLI_Command::end_bulk_operation();
		}

		restore_current_blog();
	}

	/**
	 * Adds attachment posts from the import to the posts' property.
	 */
	public function prepare_attachment_posts(): void {
		foreach ( $this->rows as $this->row ) {
			$guid          = $this->row[0];
			$filename      = basename( $guid );
			$wp_filetype   = wp_check_filetype( $filename, null );
			$wp_upload_dir = wp_upload_dir();

			preg_match( '(\d{4}\/\d\d)', $guid, $retro_post_date_matches );

			$filepath = trailingslashit( $retro_post_date_matches[0] ) . $filename;

			/**
			 * Post date should match the initial /YYYY/MM/ directory structure.
			 */
			$retro_post_date = explode( '/', $retro_post_date_matches[0] );
			$retro_post_date = strtotime( $retro_post_date[0] . '-' . $retro_post_date[1] . '-01' );
			$retro_post_date = gmdate( 'Y-m-d H:i:s', $retro_post_date );

			$this->posts[] = array(
				'data' => array(
					'guid'           => $guid,
					'post_mime_type' => $wp_filetype['type'],
					'post_title'     => preg_replace( '/\.[^.]+$/', '', basename( $filename ) ),
					'post_content'   => '',
					'post_status'    => 'inherit',
					'post_date'      => $retro_post_date,
					'post_author'    => 1,
					'post_type'      => 'attachment'
				),
				'meta' => array(
					'file_path'     => trailingslashit( $wp_upload_dir['basedir'] ) . $filepath,
					'attached_file' => $filepath
				)
			);
		}
	}

	/**
	 * Opens the CSV file and sets the rows' property.
	 */
	public function open_file(): void {
		$content = file( $this->assoc_args['data-source'] );

		if ( $content ) {
			$this->rows = array_map( 'str_getcsv', $content );
		} else {
			WP_CLI::error( __( 'The file was not found on the server.' ) );
		}
	}

	/**
	 * Inserts the posts.
	 *
	 * If dry run is enabled the post data is output instead.
	 */
	public function insert_posts(): void {
		foreach ( $this->posts as $post ) {
			if ( $this->assoc_args['dry-run'] ) {
				sprintf(
					'%1$s%3$s%2$s%3$s',
					esc_html__( 'Prepared Post Data:' ),
					print_r( $post, false ), // phpcs:ignore WordPress.PHP.DevelopmentFunctions.error_log_print_r
					PHP_EOL
				);
			} else {
				$post_id = wp_insert_attachment( $post['data'] );

				$this->post_ids[] = $post_id;

				$attach_data = wp_generate_attachment_metadata( $post_id, $post['meta']['file_path'] );
				wp_update_attachment_metadata( $post_id, $attach_data );

				add_post_meta( $post_id, '_wp_attached_file', $post['meta']['attached_file'], true );

				if ( is_wp_error( $post_id ) ) {
					WP_CLI::warning(
						sprintf(
							esc_html__( 'Failed to import %s' ),
							print_r( $post, false ) // phpcs:ignore WordPress.PHP.DevelopmentFunctions.error_log_print_r
						)
					);
				} else {
					WP_CLI::success(
						sprintf(
							esc_html__( 'Imported post %d' ),
							$post_id
						)
					);
				}
			}

			$this->count ++;
		}
	}

	/**
	 * Outputs the success message.
	 */
	public function finish(): void {
		if ( $this->assoc_args['dry-run'] ) {
			WP_CLI::success(
				sprintf(
					esc_html__( '%d post(s) would have been imported' ),
					$this->count
				)
			);
		} else {
			WP_CLI::success(
				sprintf(
					esc_html__( '%d post(s) were imported' ),
					$this->count
				)
			);

			WP_CLI::log( 
				WP_CLI::colorize( '%bOptionally validate counts with WP-CLI command: %n' ) . 
				'wp post list --post__in=' . implode( ',', $this->post_ids ) . ' --post_type=' . $this->assoc_args['post-type'] . ' --format=count'
			);
		}
	}
}

$command = new CLI_Import_CSV();

WP_CLI::add_command( 'import-csv-example', $command );