2meet-data-optimizer/includes/migration/phases/class-tmdo-phase-verify-sample.php

<?php
/**
 * Migration phase: sample-verify flat table vs EAV for data integrity.
 *
 * @package TMDO
 * @since 3.0.1
 */

declare(strict_types=1);


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

/**
 * Samples a subset of entities and compares flat vs EAV values.
 * Aborts with RuntimeException if any divergences are found.
 */
class TMDO_Phase_Verify_Sample extends TMDO_Migration_Phase_Base {

	/**
	 * Minimum sample size regardless of total entity count.
	 *
	 * @var int
	 */
	const VERIFY_SAMPLE_MIN = 500;

	/**
	 * Fraction of total entities to sample in strict mode.
	 *
	 * @var float
	 */
	const VERIFY_SAMPLE_RATIO = 0.10;

	/**
	 * Returns the phase slug.
	 *
	 * @return string
	 */
	public function name(): string {
		return 'verify_sample';
	}

	/**
	 * Executes the phase.
	 *
	 * @param array $job Job array (by reference).
	 * @return array{status: string}
	 * @throws \RuntimeException If verification divergences are found.
	 */
	public function execute( array &$job ): array {
		global $wpdb;

		if ( ! empty( $job['options']['verify_24h'] ) ) {
			$elapsed = time() - ( $job['phase_started_at'] ?? time() );
			if ( ! isset( $job['phase_started_at'] ) ) {
				$job['phase_started_at'] = time();
				$this->log( $job, '⏸ 24h shadow_read window started — wizard will resume after window elapses.' );
				return array( 'status' => 'in_progress' );
			}
			if ( $elapsed < DAY_IN_SECONDS ) {
				return array( 'status' => 'in_progress' );
			}
			$this->log( $job, '⏳ 24h shadow_read window complete; running sample compare' );
		}

		$strict      = ! empty( $job['options']['verify_strict'] );
		$users_total = (int) $wpdb->get_var( "SELECT COUNT(*) FROM {$wpdb->users}" );
		$sample_size = $strict
			? (int) max( self::VERIFY_SAMPLE_MIN, ceil( $users_total * self::VERIFY_SAMPLE_RATIO ) )
			: 100;
		$sample_size = min( $sample_size, $users_total );
		$sample_ids  = $wpdb->get_col(
			$wpdb->prepare( "SELECT ID FROM {$wpdb->users} ORDER BY RAND() LIMIT %d", $sample_size )
		);

		$diffs        = 0;
		$compared     = 0;
		$managed_keys = $this->get_managed_keys();

		foreach ( $sample_ids as $uid ) {
			foreach ( $managed_keys as $key ) {
				$eav = $wpdb->get_var(
					$wpdb->prepare(
						"SELECT meta_value FROM {$wpdb->usermeta} WHERE user_id = %d AND meta_key = %s LIMIT 1", // phpcs:ignore WPDO.AntiEAV.no-direct-usermeta-select -- verify phase: must compare EAV source against flat table to confirm migration correctness
						$uid,
						$key
					)
				);

				// Skip keys with no EAV source-of-truth — only flag when EAV has data
				// but flat doesn't match (real backfill error).
				if ( null === $eav || '' === $eav ) {
					continue;
				}

				++$compared;
				$flat = TMDO_Hook_Bus::direct_read( $this->entity_type, (int) $uid, $key );

				if ( ! $this->values_loose_equal( $flat, $eav ) ) {
					++$diffs;
					if ( $diffs <= 3 ) {
						$this->log(
							$job,
							sprintf(
								'  ! diff uid=%d key=%s flat=%s eav=%s',
								$uid,
								$key,
								is_scalar( $flat ) ? (string) $flat : gettype( $flat ),
								is_scalar( $eav ) ? (string) $eav : gettype( $eav )
							)
						);
					}
				}
			}
		}

		$this->log(
			$job,
			sprintf(
				'Verify: sampled %d users × %d keys = %d compares, %d diff(s)',
				count( $sample_ids ),
				count( $managed_keys ),
				$compared,
				$diffs
			)
		);

		if ( $diffs > 0 ) {
			throw new \RuntimeException(
				sprintf(
					'Verification found %d divergence(s) across %d compares — aborting before destructive cleanup.',
					$diffs,
					$compared
				)
			);
		}

		return array( 'status' => 'ok' );
	}
}