2meet-data-optimizer/includes/class-tmdo-postmeta-cleaner.php

<?php
/**
 * TMDO_Postmeta_Cleaner — wp_postmeta garbage cleanup (v2.9.0 Phase 0).
 *
 * Identifies and removes three classes of low-value rows from wp_postmeta
 * that bloat the table without serving any business purpose:
 *
 *   - transients  — stale `_transient_*` and `_transient_timeout_*` rows
 *                   (often left by HivePress model version cache)
 *   - wp_old_date — WP core internal record of post date changes (no app value)
 *   - edit_locks  — `_edit_lock` rows whose lock timestamp is > 24h old
 *                   (orphaned from interrupted edit sessions)
 *
 * Runs before any Entity Bridge migration so subsequent ratio measurements
 * reflect real data, not garbage. Pure DB layer — no Hook Bus / Entity Bridge
 * coupling so cleanup is safe even when post entity bridge is disabled.
 *
 * @package WP_Data_Optimizer
 * @since   2.9.0
 */

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

/**
 * Wp_postmeta garbage cleanup (v2.9.0 Phase 0).
 */
class TMDO_Postmeta_Cleaner {

	public const TARGET_TRANSIENTS  = 'transients';
	public const TARGET_WP_OLD_DATE = 'wp_old_date';
	public const TARGET_EDIT_LOCKS  = 'edit_locks';
	public const TARGET_ALL         = 'all';

	public const VALID_TARGETS = array(
		self::TARGET_TRANSIENTS,
		self::TARGET_WP_OLD_DATE,
		self::TARGET_EDIT_LOCKS,
		self::TARGET_ALL,
	);

	/**
	 * Seconds after which an _edit_lock is considered stale.
	 * WP refreshes locks on a 15-second heartbeat; 24h is intentionally
	 * conservative to avoid disturbing any active edit session.
	 */
	private const EDIT_LOCK_STALE_THRESHOLD = 86400;

	/**
	 * Count rows that would be cleaned for the given target.
	 *
	 * @param string $target One of TARGET_* constants.
	 * @return array{transients:int, wp_old_date:int, edit_locks:int, total:int}
	 * @throws InvalidArgumentException When $target is not a valid target.
	 */
	public static function count_garbage( string $target = self::TARGET_ALL ): array {
		self::assert_valid_target( $target );

		$counts = array(
			'transients'  => 0,
			'wp_old_date' => 0,
			'edit_locks'  => 0,
			'total'       => 0,
		);

		if ( self::target_includes( $target, self::TARGET_TRANSIENTS ) ) {
			$counts['transients'] = self::count_transients();
		}
		if ( self::target_includes( $target, self::TARGET_WP_OLD_DATE ) ) {
			$counts['wp_old_date'] = self::count_wp_old_date();
		}
		if ( self::target_includes( $target, self::TARGET_EDIT_LOCKS ) ) {
			$counts['edit_locks'] = self::count_stale_edit_locks();
		}

		$counts['total'] = $counts['transients'] + $counts['wp_old_date'] + $counts['edit_locks'];
		return $counts;
	}

	/**
	 * Delete garbage rows for the given target.
	 *
	 * @param string $target One of TARGET_* constants.
	 * @return array{transients:int, wp_old_date:int, edit_locks:int, total:int}
	 * @throws InvalidArgumentException When $target is not a valid target.
	 */
	public static function delete_garbage( string $target = self::TARGET_ALL ): array {
		self::assert_valid_target( $target );

		$deleted = array(
			'transients'  => 0,
			'wp_old_date' => 0,
			'edit_locks'  => 0,
			'total'       => 0,
		);

		if ( self::target_includes( $target, self::TARGET_TRANSIENTS ) ) {
			$deleted['transients'] = self::delete_transients();
		}
		if ( self::target_includes( $target, self::TARGET_WP_OLD_DATE ) ) {
			$deleted['wp_old_date'] = self::delete_wp_old_date();
		}
		if ( self::target_includes( $target, self::TARGET_EDIT_LOCKS ) ) {
			$deleted['edit_locks'] = self::delete_stale_edit_locks();
		}

		$deleted['total'] = $deleted['transients'] + $deleted['wp_old_date'] + $deleted['edit_locks'];
		return $deleted;
	}

	/**
	 * Whether $target selects $bucket (i.e. target=all or target=bucket).
	 *
	 * @param string $target Selected target.
	 * @param string $bucket Bucket constant (TARGET_TRANSIENTS / WP_OLD_DATE / EDIT_LOCKS).
	 * @return bool
	 */
	private static function target_includes( string $target, string $bucket ): bool {
		return self::TARGET_ALL === $target || $bucket === $target;
	}

	/**
	 * Validate target parameter.
	 *
	 * @param string $target Target to validate.
	 * @return void
	 * @throws InvalidArgumentException When $target is not in VALID_TARGETS.
	 */
	private static function assert_valid_target( string $target ): void {
		if ( in_array( $target, self::VALID_TARGETS, true ) ) {
			return;
		}
		// Exception messages bubble up to PHP's error handler / WP_CLI; not user output.
		$msg = sprintf( 'Invalid target "%s". Valid: %s', $target, implode( ', ', self::VALID_TARGETS ) );
		throw new InvalidArgumentException( $msg ); // phpcs:ignore WordPress.Security.EscapeOutput.ExceptionNotEscaped
	}

	// phpcs:disable WordPress.DB.DirectDatabaseQuery.DirectQuery,WordPress.DB.DirectDatabaseQuery.NoCaching,WordPress.DB.PreparedSQL.InterpolatedNotPrepared -- Internal cleanup class: $wpdb->postmeta is WP-managed and never user input. PreparedSQL.InterpolatedNotPrepared fires on multi-line $wpdb->prepare() literals where the only interpolation is the trusted table name; user-controlled values use placeholders.

	/**
	 * Count rows matching transient meta_key patterns in wp_postmeta.
	 *
	 * @return int
	 */
	private static function count_transients(): int {
		global $wpdb;
		$table = $wpdb->postmeta;
		return (int) $wpdb->get_var(
			"SELECT COUNT(*) FROM `{$table}` WHERE meta_key LIKE '\\_transient\\_%' OR meta_key LIKE '\\_transient\\_timeout\\_%'"
		);
	}

	/**
	 * Delete rows matching transient meta_key patterns from wp_postmeta.
	 *
	 * @return int Affected row count.
	 */
	private static function delete_transients(): int {
		global $wpdb;
		$table = $wpdb->postmeta;
		return (int) $wpdb->query(
			"DELETE FROM `{$table}` WHERE meta_key LIKE '\\_transient\\_%' OR meta_key LIKE '\\_transient\\_timeout\\_%'"
		);
	}

	/**
	 * Count _wp_old_date rows in wp_postmeta.
	 *
	 * @return int
	 */
	private static function count_wp_old_date(): int {
		global $wpdb;
		$table = $wpdb->postmeta;
		return (int) $wpdb->get_var(
			"SELECT COUNT(*) FROM `{$table}` WHERE meta_key = '_wp_old_date'"
		);
	}

	/**
	 * Delete _wp_old_date rows from wp_postmeta.
	 *
	 * @return int Affected row count.
	 */
	private static function delete_wp_old_date(): int {
		global $wpdb;
		$table = $wpdb->postmeta;
		return (int) $wpdb->query(
			"DELETE FROM `{$table}` WHERE meta_key = '_wp_old_date'"
		);
	}

	/**
	 * Count stale _edit_lock rows (lock_ts older than 24h) in wp_postmeta.
	 *
	 * @return int
	 */
	private static function count_stale_edit_locks(): int {
		global $wpdb;
		$table  = $wpdb->postmeta;
		$cutoff = time() - self::EDIT_LOCK_STALE_THRESHOLD;
		// _edit_lock format is "<unix_ts>:<user_id>"; SUBSTRING_INDEX extracts the timestamp.
		return (int) $wpdb->get_var(
			$wpdb->prepare(
				"SELECT COUNT(*) FROM `{$table}` WHERE meta_key = '_edit_lock' AND CAST(SUBSTRING_INDEX(meta_value, ':', 1) AS UNSIGNED) < %d",
				$cutoff
			)
		);
	}

	/**
	 * Delete stale _edit_lock rows (lock_ts older than 24h) from wp_postmeta.
	 *
	 * @return int Affected row count.
	 */
	private static function delete_stale_edit_locks(): int {
		global $wpdb;
		$table  = $wpdb->postmeta;
		$cutoff = time() - self::EDIT_LOCK_STALE_THRESHOLD;
		return (int) $wpdb->query(
			$wpdb->prepare(
				"DELETE FROM `{$table}` WHERE meta_key = '_edit_lock' AND CAST(SUBSTRING_INDEX(meta_value, ':', 1) AS UNSIGNED) < %d",
				$cutoff
			)
		);
	}

	// phpcs:enable WordPress.DB.DirectDatabaseQuery.DirectQuery,WordPress.DB.DirectDatabaseQuery.NoCaching,WordPress.DB.PreparedSQL.InterpolatedNotPrepared
}