2meet-data-optimizer/includes/diagnostic/class-tmdo-site-metrics-collector.php

<?php
/**
 * TMDO_Site_Metrics_Collector — daily EAV health snapshot writer.
 *
 * Writes structured rows to `wpdo_site_metrics` once per day (via
 * `wpdo_collect_site_metrics` cron action, scheduled at 05:00 UTC).
 *
 * Metric keys written per run:
 *   eav.postmeta_rows / eav.usermeta_rows / eav.termmeta_rows / eav.commentmeta_rows
 *   flat.hot_rows / flat.cold_rows / flat.warm_rows
 *   custom_tables.total_rows / custom_tables.table_count
 *   errors.last_24h / shadow_diffs.last_24h
 *
 * Monthly Summary (`TMDO_Monthly_Summary`) reads these rows for the
 * `zone_growth` section instead of querying live tables, so the monthly
 * rollup is fast even on large databases.
 *
 * @package WP_Data_Optimizer
 * @since   2.6.2
 */

declare(strict_types=1);

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

/**
 * Collects site-wide EAV health metrics and writes to wpdo_site_metrics.
 */
class TMDO_Site_Metrics_Collector {

	/**
	 * Cron hook name.
	 *
	 * @var string
	 */
	public const CRON_HOOK = 'wpdo_collect_site_metrics';

	/**
	 * How many days of daily rows to retain before pruning.
	 *
	 * @var int
	 */
	private const RETENTION_DAYS = 90;

	/**
	 * Register the cron handler and return the instance for chaining.
	 *
	 * @return void
	 */
	public static function register(): void {
		add_action( self::CRON_HOOK, array( __CLASS__, 'collect' ) );
	}

	/**
	 * Collect all site metrics and persist them to wpdo_site_metrics.
	 *
	 * Called by the daily cron. Safe to call manually (e.g., via WP-CLI).
	 *
	 * @param bool $dry_run When true, collect but do not write to the DB.
	 * @return array<string,int> Map of metric_key => metric_value collected.
	 */
	public static function collect( bool $dry_run = false ): array {
		global $wpdb;

		$now     = TMDO_DB::now();
		$metrics = array();

		// ── EAV row counts ────────────────────────────────────────────────
		$eav_tables = array(
			'eav.postmeta_rows'    => $wpdb->postmeta,
			'eav.usermeta_rows'    => $wpdb->usermeta,
			'eav.termmeta_rows'    => $wpdb->termmeta,
			'eav.commentmeta_rows' => $wpdb->commentmeta,
		);
		foreach ( $eav_tables as $key => $table ) {
			$count           = (int) $wpdb->get_var( "SELECT COUNT(*) FROM `{$table}`" ); // phpcs:ignore WordPress.DB
			$metrics[ $key ] = $count;
		}

		// ── Flat-table row counts (hot + cold dynamic tables, warm) ───────
		$hot_rows  = 0;
		$cold_rows = 0;

		if ( class_exists( 'TMDO_Schema_Registry' ) ) {
			$registry = TMDO_Schema_Registry::instance();
			foreach ( $registry->get_hot_post_types() as $pt ) {
				$ht     = $wpdb->prefix . 'wpdo_hot_' . sanitize_key( $pt );
				$exists = (int) $wpdb->get_var( $wpdb->prepare( 'SELECT COUNT(*) FROM information_schema.TABLES WHERE TABLE_SCHEMA = DATABASE() AND TABLE_NAME = %s', $ht ) ); // phpcs:ignore WordPress.DB
				if ( $exists ) {
					$hot_rows += (int) $wpdb->get_var( "SELECT COUNT(*) FROM `{$ht}`" ); // phpcs:ignore WordPress.DB
				}
			}
			foreach ( $registry->get_cold_post_types() as $pt ) {
				$ct     = $wpdb->prefix . 'wpdo_cold_' . sanitize_key( $pt );
				$exists = (int) $wpdb->get_var( $wpdb->prepare( 'SELECT COUNT(*) FROM information_schema.TABLES WHERE TABLE_SCHEMA = DATABASE() AND TABLE_NAME = %s', $ct ) ); // phpcs:ignore WordPress.DB
				if ( $exists ) {
					$cold_rows += (int) $wpdb->get_var( "SELECT COUNT(*) FROM `{$ct}`" ); // phpcs:ignore WordPress.DB
				}
			}
		}

		$warm_table                = $wpdb->prefix . 'wpdo_warm';
		$metrics['flat.hot_rows']  = $hot_rows;
		$metrics['flat.cold_rows'] = $cold_rows;
		$metrics['flat.warm_rows'] = (int) $wpdb->get_var( "SELECT COUNT(*) FROM `{$warm_table}`" ); // phpcs:ignore WordPress.DB

		// ── Custom table row counts ───────────────────────────────────────
		$custom_rows  = 0;
		$custom_count = 0;
		if ( class_exists( 'TMDO_Custom_Table_Registry' ) ) {
			foreach ( TMDO_Custom_Table_Registry::instance()->all() as $cfg ) {
				$tbl    = $wpdb->prefix . $cfg['table_name'];
				$exists = (int) $wpdb->get_var( $wpdb->prepare( 'SELECT COUNT(*) FROM information_schema.TABLES WHERE TABLE_SCHEMA = DATABASE() AND TABLE_NAME = %s', $tbl ) ); // phpcs:ignore WordPress.DB
				if ( $exists ) {
					$custom_rows += (int) $wpdb->get_var( "SELECT COUNT(*) FROM `{$tbl}`" ); // phpcs:ignore WordPress.DB
					++$custom_count;
				}
			}
		}
		$metrics['custom_tables.total_rows']  = $custom_rows;
		$metrics['custom_tables.table_count'] = $custom_count;

		// ── Error / shadow-diff activity (last 24 h) ─────────────────────
		$errors_table               = $wpdb->prefix . 'wpdo_errors';
		$metrics['errors.last_24h'] = (int) $wpdb->get_var(
			"SELECT COUNT(*) FROM `{$errors_table}` WHERE created_at >= DATE_SUB(UTC_TIMESTAMP(), INTERVAL 24 HOUR)" // phpcs:ignore WordPress.DB
		);

		$shadow_table                     = $wpdb->prefix . 'wpdo_shadow_diffs';
		$shadow_exists                    = (int) $wpdb->get_var( $wpdb->prepare( 'SELECT COUNT(*) FROM information_schema.TABLES WHERE TABLE_SCHEMA = DATABASE() AND TABLE_NAME = %s', $shadow_table ) ); // phpcs:ignore WordPress.DB
		$metrics['shadow_diffs.last_24h'] = $shadow_exists
			? (int) $wpdb->get_var( "SELECT COUNT(*) FROM `{$shadow_table}` WHERE ts >= DATE_SUB(UTC_TIMESTAMP(), INTERVAL 24 HOUR)" ) // phpcs:ignore WordPress.DB
			: 0;

		if ( $dry_run ) {
			return $metrics;
		}

		// ── Persist each metric to wpdo_site_metrics ─────────────────────
		$dest = TMDO_DB::table( 'wpdo_site_metrics' );
		foreach ( $metrics as $key => $value ) {
			$wpdb->insert(
				$dest,
				array(
					'collected_at' => $now,
					'metric_key'   => $key,
					'metric_value' => $value,
					'context'      => null,
				),
				array( '%s', '%s', '%d', '%s' )
			);
		}

		// ── Prune old rows beyond retention window ────────────────────────
		$cutoff = gmdate( 'Y-m-d H:i:s', strtotime( '-' . self::RETENTION_DAYS . ' days' ) );
		$wpdb->query( $wpdb->prepare( "DELETE FROM {$dest} WHERE collected_at < %s", $cutoff ) ); // phpcs:ignore WordPress.DB

		if ( class_exists( 'TMDO_Logger' ) ) {
			TMDO_Logger::info(
				'site_metrics_collected',
				array(
					'metric_count'  => count( $metrics ),
					'postmeta_rows' => $metrics['eav.postmeta_rows'] ?? 0,
					'hot_rows'      => $metrics['flat.hot_rows'] ?? 0,
					'custom_rows'   => $metrics['custom_tables.total_rows'] ?? 0,
				)
			);
		}

		return $metrics;
	}

	/**
	 * Get the most recent snapshot (latest collected_at timestamp).
	 *
	 * @return array<string,int>  metric_key => metric_value, or empty on miss.
	 */
	public static function get_latest_snapshot(): array {
		global $wpdb;
		$dest = TMDO_DB::table( 'wpdo_site_metrics' );

		$latest_ts = $wpdb->get_var( "SELECT MAX(collected_at) FROM `{$dest}`" ); // phpcs:ignore WordPress.DB
		if ( ! $latest_ts ) {
			return array();
		}

		$rows = (array) $wpdb->get_results(
			$wpdb->prepare(
				"SELECT metric_key, metric_value FROM `{$dest}` WHERE collected_at = %s", // phpcs:ignore WordPress.DB
				$latest_ts
			),
			ARRAY_A
		);

		$out = array();
		foreach ( $rows as $r ) {
			$out[ (string) $r['metric_key'] ] = (int) $r['metric_value'];
		}
		return $out;
	}

	/**
	 * Get daily metric history for a single key over the past N days.
	 *
	 * @param string $metric_key Metric key (e.g. 'eav.postmeta_rows').
	 * @param int    $days       Number of days of history to return (default 30).
	 * @return array<array{collected_at:string,value:int}> Oldest-first.
	 */
	public static function get_history( string $metric_key, int $days = 30 ): array {
		global $wpdb;
		$dest = TMDO_DB::table( 'wpdo_site_metrics' );

		// phpcs:disable WordPress.DB.PreparedSQL.InterpolatedNotPrepared -- $dest is a validated table name from TMDO_DB::table().
		$rows = (array) $wpdb->get_results(
			$wpdb->prepare(
				"SELECT collected_at, metric_value AS value
				   FROM `{$dest}`
				  WHERE metric_key = %s
				    AND collected_at >= DATE_SUB(UTC_TIMESTAMP(), INTERVAL %d DAY)
				  ORDER BY collected_at ASC",
				$metric_key,
				$days
			),
			ARRAY_A
		);
		// phpcs:enable WordPress.DB.PreparedSQL.InterpolatedNotPrepared

		return array_map(
			static fn( $r ) => array(
				'collected_at' => (string) $r['collected_at'],
				'value'        => (int) $r['value'],
			),
			$rows
		);
	}
}