hustle-gdpr-cleanup-fix.php

<?php
/**
 * Plugin Name: Hustle GDPR Cleanup Fix (staging)
 * Description: Replaces hustle_general_data_protection_cleanup with a Forever-aware, batched callback. Does not edit plugin files.
 * Version: 1.0.0
 *
 * Install: copy this file to wp-content/mu-plugins/hustle-gdpr-cleanup-fix.php
 *
 * What it changes:
 * - Skips work when retention is Forever (string/int safe).
 * - Never loads every tracking ID into PHP.
 * - Anonymizes IPs with batched UPDATE ... LIMIT, not per-row SELECT tracking_id <.
 * - Deletes old rows in batches and stops before the PHP time limit.
 *
 * Keep the logger mu-plugin enabled so you can compare timings in
 * wp-content/hustle-gdpr-cleanup-debug.log
 */

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

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

if ( ! defined( 'HUSTLE_GDPR_FIX_BATCH_SIZE' ) ) {
	define( 'HUSTLE_GDPR_FIX_BATCH_SIZE', 500 );
}

if ( ! defined( 'HUSTLE_GDPR_FIX_TIME_BUDGET' ) ) {
	define( 'HUSTLE_GDPR_FIX_TIME_BUDGET', 20 );
}

/**
 * Log through the logger mu-plugin when present.
 *
 * @param string $message Message.
 * @param mixed  $data    Optional data.
 */
function hustle_gdpr_fix_log( $message, $data = null ) {
	if ( function_exists( 'hustle_gdpr_debug_log' ) ) {
		hustle_gdpr_debug_log( '[fix] ' . $message, $data );
		return;
	}

	$line = 'Hustle GDPR fix: ' . $message;
	if ( null !== $data ) {
		$line .= ' ' . wp_json_encode( $data );
	}
	// phpcs:ignore WordPress.PHP.DevelopmentFunctions.error_log_error_log
	error_log( $line );
}

/**
 * Whether a Forever flag should skip cleanup.
 *
 * @param mixed $value Stored setting.
 * @return bool
 */
function hustle_gdpr_fix_is_forever( $value ) {
	return '1' === (string) $value;
}

/**
 * Whether this job should no-op.
 *
 * @param mixed $forever Forever flag.
 * @param mixed $number  Retention number.
 * @return bool
 */
function hustle_gdpr_fix_should_skip( $forever, $number ) {
	return hustle_gdpr_fix_is_forever( $forever ) || 0 === absint( $number );
}

/**
 * Retain cutoff datetime using the same arithmetic as Hustle.
 *
 * @param mixed  $number Number.
 * @param string $unit   Unit.
 * @return string|false
 */
function hustle_gdpr_fix_retain_time( $number, $unit ) {
	$units = array( 'days', 'weeks', 'months', 'years' );
	if ( ! in_array( $unit, $units, true ) ) {
		return false;
	}

	return date_i18n( 'Y-m-d H:i:s', strtotime( '-' . absint( $number ) . ' ' . $unit ) );
}

/**
 * Table names.
 *
 * @return array<string,string>|false
 */
function hustle_gdpr_fix_tables() {
	if ( ! class_exists( 'Hustle_Db' ) ) {
		return false;
	}

	return array(
		'entries'      => Hustle_Db::entries_table(),
		'entries_meta' => Hustle_Db::entries_meta_table(),
		'tracking'     => Hustle_Db::tracking_table(),
	);
}

/**
 * Strip Hustle's original cleanup callback only.
 */
function hustle_gdpr_fix_replace_callback() {
	if ( ! HUSTLE_GDPR_FIX_ENABLED || ! class_exists( 'Hustle_General_Data_Protection' ) ) {
		return;
	}

	global $wp_filter;
	if ( empty( $wp_filter['hustle_general_data_protection_cleanup'] ) ) {
		add_action( 'hustle_general_data_protection_cleanup', 'hustle_gdpr_fix_cleanup', 10 );
		hustle_gdpr_fix_log( 'Original hook missing; registered batched cleanup only' );
		return;
	}

	$hook = $wp_filter['hustle_general_data_protection_cleanup'];
	foreach ( $hook->callbacks as $priority => $callbacks ) {
		foreach ( $callbacks as $callback ) {
			$function = $callback['function'];
			if ( is_array( $function ) && isset( $function[0], $function[1] )
				&& $function[0] instanceof Hustle_General_Data_Protection
				&& 'personal_data_cleanup' === $function[1]
			) {
				remove_action( 'hustle_general_data_protection_cleanup', $function, $priority );
				hustle_gdpr_fix_log( 'Removed Hustle_General_Data_Protection->personal_data_cleanup', array( 'priority' => $priority ) );
			}
		}
	}

	add_action( 'hustle_general_data_protection_cleanup', 'hustle_gdpr_fix_cleanup', 10 );
}

/**
 * Batched cleanup replacement.
 */
function hustle_gdpr_fix_cleanup() {
	$tables = hustle_gdpr_fix_tables();
	if ( ! $tables || ! class_exists( 'Hustle_Settings_Admin' ) ) {
		hustle_gdpr_fix_log( 'Hustle classes/tables unavailable; skipping' );
		return;
	}

	$settings = Hustle_Settings_Admin::get_privacy_settings();
	$started  = microtime( true );
	$deadline = $started + (float) HUSTLE_GDPR_FIX_TIME_BUDGET;
	$batch    = max( 1, (int) HUSTLE_GDPR_FIX_BATCH_SIZE );

	$plan = array(
		'submissions' => ! hustle_gdpr_fix_should_skip( $settings['retain_submission_forever'], $settings['submissions_retention_number'] ),
		'ip'          => ! hustle_gdpr_fix_should_skip( $settings['retain_ip_forever'], $settings['ip_retention_number'] ),
		'tracking'    => ! hustle_gdpr_fix_should_skip( $settings['retain_tracking_forever'], $settings['tracking_retention_number'] ),
	);

	hustle_gdpr_fix_log(
		'Batched cleanup start',
		array(
			'plan'        => $plan,
			'batch_size'  => $batch,
			'time_budget' => HUSTLE_GDPR_FIX_TIME_BUDGET,
		)
	);

	if ( ! $plan['submissions'] && ! $plan['ip'] && ! $plan['tracking'] ) {
		hustle_gdpr_fix_log( 'Nothing to do (all jobs Forever or 0). Returning.' );
		return;
	}

	$stats = array(
		'submissions_deleted' => 0,
		'entry_ips_cleared'   => 0,
		'tracking_ips_cleared'=> 0,
		'tracking_deleted'    => 0,
		'stopped_early'       => false,
	);

	if ( $plan['submissions'] ) {
		$retain = hustle_gdpr_fix_retain_time( $settings['submissions_retention_number'], $settings['submissions_retention_number_unit'] );
		if ( $retain ) {
			$stats['submissions_deleted'] = hustle_gdpr_fix_delete_old_entries( $tables, $retain, $batch, $deadline );
		}
	}

	if ( $plan['ip'] && microtime( true ) < $deadline ) {
		$retain = hustle_gdpr_fix_retain_time( $settings['ip_retention_number'], $settings['ip_retention_number_unit'] );
		if ( $retain ) {
			$ip_stats                     = hustle_gdpr_fix_anonymize_old_ips( $tables, $retain, $batch, $deadline );
			$stats['entry_ips_cleared']   = $ip_stats['entries'];
			$stats['tracking_ips_cleared']= $ip_stats['tracking'];
		}
	}

	if ( $plan['tracking'] && microtime( true ) < $deadline ) {
		$retain = hustle_gdpr_fix_retain_time( $settings['tracking_retention_number'], $settings['tracking_retention_number_unit'] );
		if ( $retain ) {
			$stats['tracking_deleted'] = hustle_gdpr_fix_delete_old_tracking( $tables, $retain, $batch, $deadline );
		}
	}

	$stats['stopped_early'] = microtime( true ) >= $deadline;
	$stats['seconds']       = round( microtime( true ) - $started, 3 );
	hustle_gdpr_fix_log( 'Batched cleanup end', $stats );
}

/**
 * Delete old submissions in ID batches.
 *
 * @param array  $tables   Tables.
 * @param string $retain   Cutoff datetime.
 * @param int    $batch    Batch size.
 * @param float  $deadline Unix microtime deadline.
 * @return int
 */
function hustle_gdpr_fix_delete_old_entries( $tables, $retain, $batch, $deadline ) {
	global $wpdb;

	$deleted = 0;
	$ids     = array();
	do {
		if ( microtime( true ) >= $deadline ) {
			break;
		}

		$ids = $wpdb->get_col( // phpcs:ignore WordPress.DB.DirectDatabaseQuery.DirectQuery,WordPress.DB.DirectDatabaseQuery.NoCaching
			$wpdb->prepare(
				"SELECT entry_id FROM `{$tables['entries']}` WHERE date_created < %s LIMIT %d", // phpcs:ignore WordPress.DB.PreparedSQL.InterpolatedNotPrepared
				$retain,
				$batch
			)
		);

		if ( empty( $ids ) ) {
			break;
		}

		$ids_sql = implode( ',', array_map( 'absint', $ids ) );
		$wpdb->query( "DELETE FROM `{$tables['entries_meta']}` WHERE entry_id IN ({$ids_sql})" ); // phpcs:ignore WordPress.DB.DirectDatabaseQuery.DirectQuery,WordPress.DB.DirectDatabaseQuery.NoCaching,WordPress.DB.PreparedSQL.InterpolatedNotPrepared
		$wpdb->query( "DELETE FROM `{$tables['entries']}` WHERE entry_id IN ({$ids_sql})" ); // phpcs:ignore WordPress.DB.DirectDatabaseQuery.DirectQuery,WordPress.DB.DirectDatabaseQuery.NoCaching,WordPress.DB.PreparedSQL.InterpolatedNotPrepared

		$deleted += count( $ids );
		hustle_gdpr_fix_log( 'Submissions batch deleted', array( 'batch' => count( $ids ), 'total' => $deleted ) );
	} while ( ! empty( $ids ) );

	return $deleted;
}

/**
 * Anonymize IPs in entries meta and tracking, in batches.
 *
 * @param array  $tables   Tables.
 * @param string $retain   Cutoff datetime.
 * @param int    $batch    Batch size.
 * @param float  $deadline Unix microtime deadline.
 * @return array
 */
function hustle_gdpr_fix_anonymize_old_ips( $tables, $retain, $batch, $deadline ) {
	global $wpdb;

	$ipv4 = function_exists( 'wp_privacy_anonymize_ip' ) ? wp_privacy_anonymize_ip( '1.2.3.4' ) : '0.0.0.0';
	$ipv6 = function_exists( 'wp_privacy_anonymize_ip' ) ? wp_privacy_anonymize_ip( '2001:db8::1' ) : '::';

	$entry_total    = 0;
	$tracking_total = 0;
	$ids            = array();
	$affected       = 0;

	do {
		if ( microtime( true ) >= $deadline ) {
			break;
		}

		$ids = $wpdb->get_col( // phpcs:ignore WordPress.DB.DirectDatabaseQuery.DirectQuery,WordPress.DB.DirectDatabaseQuery.NoCaching
			$wpdb->prepare(
				"SELECT em.meta_id
				FROM `{$tables['entries_meta']}` em
				INNER JOIN `{$tables['entries']}` e ON e.entry_id = em.entry_id
				WHERE em.meta_key = 'hustle_ip'
				AND e.date_created < %s
				AND em.meta_value <> ''
				AND em.meta_value <> %s
				AND em.meta_value <> %s
				LIMIT %d", // phpcs:ignore WordPress.DB.PreparedSQL.InterpolatedNotPrepared
				$retain,
				$ipv4,
				$ipv6,
				$batch
			)
		);

		if ( empty( $ids ) ) {
			break;
		}

		$ids_sql  = implode( ',', array_map( 'absint', $ids ) );
		$affected = (int) $wpdb->query( // phpcs:ignore WordPress.DB.DirectDatabaseQuery.DirectQuery,WordPress.DB.DirectDatabaseQuery.NoCaching
			$wpdb->prepare(
				"UPDATE `{$tables['entries_meta']}`
				SET meta_value = CASE
					WHEN meta_value LIKE %s THEN %s
					ELSE %s
				END
				WHERE meta_id IN ({$ids_sql})", // phpcs:ignore WordPress.DB.PreparedSQL.InterpolatedNotPrepared
				'%:%',
				$ipv6,
				$ipv4
			)
		);
		$entry_total += max( 0, $affected );
		hustle_gdpr_fix_log( 'Entry IP batch anonymized', array( 'batch' => $affected, 'total' => $entry_total ) );
	} while ( ! empty( $ids ) );

	do {
		if ( microtime( true ) >= $deadline ) {
			break;
		}

		$affected = (int) $wpdb->query( // phpcs:ignore WordPress.DB.DirectDatabaseQuery.DirectQuery,WordPress.DB.DirectDatabaseQuery.NoCaching
			$wpdb->prepare(
				"UPDATE `{$tables['tracking']}`
				SET ip = CASE
					WHEN ip LIKE %s THEN %s
					ELSE %s
				END
				WHERE date_created < %s
				AND ip IS NOT NULL
				AND ip <> ''
				AND ip <> %s
				AND ip <> %s
				LIMIT %d", // phpcs:ignore WordPress.DB.PreparedSQL.InterpolatedNotPrepared
				'%:%',
				$ipv6,
				$ipv4,
				$retain,
				$ipv4,
				$ipv6,
				$batch
			)
		);
		$tracking_total += max( 0, $affected );
		if ( $affected > 0 ) {
			hustle_gdpr_fix_log( 'Tracking IP batch anonymized', array( 'batch' => $affected, 'total' => $tracking_total ) );
		}
	} while ( $affected > 0 );

	return array(
		'entries'  => $entry_total,
		'tracking' => $tracking_total,
	);
}

/**
 * Delete old tracking rows in batches.
 *
 * @param array  $tables   Tables.
 * @param string $retain   Cutoff datetime.
 * @param int    $batch    Batch size.
 * @param float  $deadline Unix microtime deadline.
 * @return int
 */
function hustle_gdpr_fix_delete_old_tracking( $tables, $retain, $batch, $deadline ) {
	global $wpdb;

	$deleted  = 0;
	$affected = 0;
	do {
		if ( microtime( true ) >= $deadline ) {
			break;
		}

		$affected = (int) $wpdb->query( // phpcs:ignore WordPress.DB.DirectDatabaseQuery.DirectQuery,WordPress.DB.DirectDatabaseQuery.NoCaching
			$wpdb->prepare(
				"DELETE FROM `{$tables['tracking']}` WHERE date_created < %s LIMIT %d", // phpcs:ignore WordPress.DB.PreparedSQL.InterpolatedNotPrepared
				$retain,
				$batch
			)
		);
		$deleted += max( 0, $affected );
		if ( $affected > 0 ) {
			hustle_gdpr_fix_log( 'Tracking delete batch', array( 'batch' => $affected, 'total' => $deleted ) );
		}
	} while ( $affected > 0 );

	return $deleted;
}

add_action( 'plugins_loaded', 'hustle_gdpr_fix_replace_callback', 1000 );