robotstxt-mediaaudit/includes/Internal/AttachmentIndexer.php

<?php
/**
 * Indexes image attachments into mra_media_index.
 *
 * @package MediaRightsAudit\Internal
 */

namespace MediaRightsAudit\Internal;

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

use MediaRightsAudit\Core\Queue\Scheduler;

/**
 * Populates mra_media_index with one row per image attachment.
 *
 * Processing order: oldest-first (ASC by attachment ID).
 * Idempotent: safely re-runnable; existing rows are never overwritten.
 */
class AttachmentIndexer {

	/**
	 * Action Scheduler hook name for background indexing.
	 */
	const AS_HOOK = 'mra_internal_index_batch';

	/**
	 * Indexes one batch of not-yet-indexed image attachments.
	 *
	 * @param int $batch_size Maximum number of attachments to process.
	 *
	 * @return int Number of attachments newly added to the index.
	 */
	public static function index_batch( int $batch_size = 50 ): int {
		global $wpdb;

		// phpcs:disable WordPress.DB.DirectDatabaseQuery
		$rows = $wpdb->get_results(
			$wpdb->prepare(
				"SELECT p.ID, p.guid, p.post_mime_type
				 FROM {$wpdb->posts} p
				 LEFT JOIN {$wpdb->prefix}mra_media_index i ON p.ID = i.attachment_id
				 WHERE p.post_type = 'attachment'
				   AND p.post_mime_type LIKE %s
				   AND p.post_status != 'trash'
				   AND i.attachment_id IS NULL
				 ORDER BY p.ID ASC
				 LIMIT %d",
				'image/%',
				$batch_size
			),
			ARRAY_A
		);
		// phpcs:enable WordPress.DB.DirectDatabaseQuery

		if ( ! $rows ) {
			return 0;
		}

		// Prime postmeta cache so get_attached_file() and wp_get_attachment_url() are fast.
		$ids = array_map( 'intval', array_column( $rows, 'ID' ) );
		update_meta_cache( 'post', $ids );

		$now   = current_time( 'mysql', true );
		$count = 0;

		foreach ( $rows as $row ) {
			$attachment_id = intval( $row['ID'] );
			$guid          = strval( $row['guid'] );
			$mime_type     = strval( $row['post_mime_type'] );

			$file_url  = wp_get_attachment_url( $attachment_id );
			$file_url  = $file_url ? $file_url : $guid;
			$file_path = get_attached_file( $attachment_id );
			$file_size = ( $file_path && file_exists( $file_path ) )
				? intval( filesize( $file_path ) )
				: 0;

			$inserted = $wpdb->insert( // phpcs:ignore WordPress.DB.DirectDatabaseQuery
				$wpdb->prefix . 'mra_media_index',
				array(
					'attachment_id' => $attachment_id,
					'file_url'      => $file_url,
					'file_name'     => self::extract_filename( $file_url ),
					'mime_type'     => $mime_type,
					'file_size'     => $file_size,
					'created_at'    => $now,
				),
				array( '%d', '%s', '%s', '%s', '%d', '%s' )
			);

			if ( false !== $inserted ) {
				++$count;
			}
		}

		return $count;
	}

	/**
	 * Extracts the bare filename from a URL.
	 *
	 * Uses PHP's parse_url() so this method is testable without WordPress.
	 *
	 * @param string $url Absolute or relative URL.
	 *
	 * @return string Filename, or the original string if parsing fails.
	 */
	public static function extract_filename( string $url ): string {
		$path = parse_url( $url, PHP_URL_PATH ); // phpcs:ignore WordPress.WP.AlternativeFunctions.parse_url_parse_url
		return is_string( $path ) ? basename( $path ) : basename( $url );
	}

	/**
	 * Returns the number of image attachments not yet in the index.
	 *
	 * @return int
	 */
	public static function get_pending_count(): int {
		global $wpdb;

		$result = $wpdb->get_var( // phpcs:ignore WordPress.DB.DirectDatabaseQuery
			"SELECT COUNT(p.ID)
			 FROM {$wpdb->posts} p
			 LEFT JOIN {$wpdb->prefix}mra_media_index i ON p.ID = i.attachment_id
			 WHERE p.post_type = 'attachment'
			   AND p.post_mime_type LIKE 'image/%'
			   AND p.post_status != 'trash'
			   AND i.attachment_id IS NULL"
		);

		return intval( $result );
	}

	/**
	 * Returns the number of attachments currently in the index.
	 *
	 * @return int
	 */
	public static function get_indexed_count(): int {
		global $wpdb;

		$result = $wpdb->get_var( // phpcs:ignore WordPress.DB.DirectDatabaseQuery
			"SELECT COUNT(*) FROM {$wpdb->prefix}mra_media_index"
		);

		return intval( $result );
	}

	/**
	 * Processes one scheduled batch and re-queues or hands off to UsageScanner.
	 *
	 * Called by Action Scheduler via the mra_internal_index_batch hook.
	 *
	 * @return void
	 */
	public static function process_scheduled_batch(): void {
		self::index_batch();

		if ( self::get_pending_count() > 0 ) {
			Scheduler::schedule_single( self::AS_HOOK );
		} else {
			Scheduler::schedule_single( UsageScanner::AS_HOOK );
		}
	}

	/**
	 * Enqueues a background indexing job if one is not already pending.
	 *
	 * @return void
	 */
	public static function schedule(): void {
		if ( ! Scheduler::has_pending( self::AS_HOOK ) ) {
			Scheduler::schedule_single( self::AS_HOOK );
		}
	}

	/**
	 * Deletes all rows from mra_media_index and mra_media_usage, forcing a full re-index on the next run.
	 *
	 * @return int Number of index rows removed.
	 */
	public static function reset_all(): int {
		global $wpdb;

		// phpcs:disable WordPress.DB.DirectDatabaseQuery
		$count = (int) $wpdb->get_var( "SELECT COUNT(*) FROM {$wpdb->prefix}mra_media_index" );
		$wpdb->query( "DELETE FROM {$wpdb->prefix}mra_media_usage" );
		$wpdb->query( "DELETE FROM {$wpdb->prefix}mra_media_index" );
		// phpcs:enable WordPress.DB.DirectDatabaseQuery

		return $count;
	}
}