<?php
/**
 * Optional 404 log (off by default).
 *
 * @package Unbreak
 */

namespace Unbreak\Redirects;

use Unbreak\Install\Schema;
use Unbreak\Support\Clock;
use Unbreak\Support\Settings;

defined( 'ABSPATH' ) || exit;

/**
 * Logs 404 paths with one INSERT ... ON DUPLICATE KEY UPDATE per request. Never stores IPs.
 */
final class NotFoundLogger {

	public const MAX_ROWS = 5000;

	/**
	 * Max logged 404s per minute when a persistent object cache is available.
	 */
	private const PER_MINUTE = 120;

	/**
	 * Register hooks.
	 */
	public function register(): void {
		add_action( 'template_redirect', array( $this, 'maybe_log' ), 99 );
	}

	/**
	 * Log the current request if it is a 404 and logging is enabled.
	 */
	public function maybe_log(): void {
		if ( ! is_404() || ! Settings::get( 'log_404' ) ) {
			return;
		}
		$uri      = isset( $_SERVER['REQUEST_URI'] ) ? (string) wp_unslash( $_SERVER['REQUEST_URI'] ) : ''; // phpcs:ignore WordPress.Security.ValidatedSanitizedInput.InputNotSanitized -- Normalized below.
		$referrer = wp_get_raw_referer();
		$this->log( $uri, is_string( $referrer ) ? $referrer : '' );
	}

	/**
	 * Record a 404 hit.
	 *
	 * @param string $uri      Request URI.
	 * @param string $referrer Referrer URL.
	 */
	public function log( string $uri, string $referrer = '' ): bool {
		$path = Path::normalize( $uri, false );
		if ( null === $path || strlen( $path ) > 2048 || $this->rate_limited() ) {
			return false;
		}
		$referrer = '' !== $referrer ? substr( esc_url_raw( $referrer ), 0, 2048 ) : null;

		global $wpdb;
		$table = Schema::table( 'log_404' );
		$now   = Clock::mysql();
		// One statement. New paths are only inserted while the table holds fewer than MAX_ROWS rows, so the
		// log cannot grow without bound even without an object cache; known paths always count the hit.
		$hash = sha1( $path );
		// phpcs:ignore WordPress.DB.DirectDatabaseQuery.DirectQuery,WordPress.DB.DirectDatabaseQuery.NoCaching -- Single upsert per 404 request.
		$wpdb->query(
			$wpdb->prepare(
				'INSERT INTO %i (path, path_hash, referrer, hits, first_seen, last_seen)
				SELECT %s, %s, %s, 1, %s, %s FROM DUAL
				WHERE ( SELECT COUNT(*) FROM %i AS c ) < %d OR EXISTS ( SELECT 1 FROM %i AS e WHERE e.path_hash = %s )
				ON DUPLICATE KEY UPDATE hits = hits + 1, last_seen = VALUES(last_seen), referrer = COALESCE(VALUES(referrer), referrer)',
				$table,
				$path,
				$hash,
				$referrer,
				$now,
				$now,
				$table,
				self::MAX_ROWS,
				$table,
				$hash
			)
		);
		return true;
	}

	/**
	 * Per-minute cap, only with a persistent object cache (without one, a counter would cost queries).
	 */
	private function rate_limited(): bool {
		if ( ! wp_using_ext_object_cache() ) {
			return false;
		}
		$key = 'unbreak_404_' . gmdate( 'YmdHi', Clock::now() );
		wp_cache_add( $key, 0, 'unbreak', 120 );
		return (int) wp_cache_incr( $key, 1, 'unbreak' ) > self::PER_MINUTE;
	}

	/**
	 * Keep at most MAX_ROWS rows (least-hit and oldest go first) and drop rows older than the retention period.
	 */
	public static function trim(): void {
		global $wpdb;
		$table = Schema::table( 'log_404' );
		$days  = (int) Settings::get( 'retention_days' );
		// phpcs:disable WordPress.DB.PreparedSQL.InterpolatedNotPrepared,WordPress.DB.DirectDatabaseQuery.DirectQuery,WordPress.DB.DirectDatabaseQuery.NoCaching -- Daily maintenance on a plugin table.
		$wpdb->query( $wpdb->prepare( 'DELETE FROM %i WHERE last_seen < %s', $table, Clock::to_mysql( Clock::now() - $days * DAY_IN_SECONDS ) ) );
		$total = (int) $wpdb->get_var( $wpdb->prepare( 'SELECT COUNT(*) FROM %i', $table ) );
		if ( $total > self::MAX_ROWS ) {
			$wpdb->query( $wpdb->prepare( 'DELETE FROM %i ORDER BY hits ASC, last_seen ASC LIMIT %d', $table, $total - self::MAX_ROWS ) );
		}
		// phpcs:enable
	}

	/**
	 * Paginated list, most hits first.
	 *
	 * @param int    $page     Page.
	 * @param int    $per_page Rows per page.
	 * @param string $search   Search.
	 * @return array{items:array<int,array<string,mixed>>,total:int}
	 */
	public static function paginate( int $page = 1, int $per_page = 50, string $search = '' ): array {
		global $wpdb;
		$table  = Schema::table( 'log_404' );
		$like   = '%' . $wpdb->esc_like( $search ) . '%';
		$offset = max( 0, ( $page - 1 ) * $per_page );
		// phpcs:disable WordPress.DB.DirectDatabaseQuery.DirectQuery,WordPress.DB.DirectDatabaseQuery.NoCaching -- Plugin table.
		$total = (int) $wpdb->get_var( $wpdb->prepare( "SELECT COUNT(*) FROM %i WHERE ( %s = '' OR path LIKE %s )", $table, $search, $like ) );
		$rows  = (array) $wpdb->get_results( $wpdb->prepare( "SELECT id, path, referrer, hits, first_seen, last_seen FROM %i WHERE ( %s = '' OR path LIKE %s ) ORDER BY hits DESC, last_seen DESC LIMIT %d OFFSET %d", $table, $search, $like, $per_page, $offset ), ARRAY_A );
		// phpcs:enable
		$items = array();
		foreach ( $rows as $row ) {
			$items[] = array(
				'id'         => (int) $row['id'],
				'path'       => (string) $row['path'],
				'referrer'   => $row['referrer'],
				'hits'       => (int) $row['hits'],
				'first_seen' => $row['first_seen'],
				'last_seen'  => $row['last_seen'],
			);
		}
		return array(
			'items' => $items,
			'total' => $total,
		);
	}

	/**
	 * Delete one row, or all rows when $id is 0.
	 *
	 * @param int $id Row ID.
	 */
	public static function delete( int $id = 0 ): void {
		global $wpdb;
		$table = Schema::table( 'log_404' );
		if ( 0 === $id ) {
			$wpdb->query( $wpdb->prepare( 'TRUNCATE TABLE %i', $table ) ); // phpcs:ignore WordPress.DB.DirectDatabaseQuery.DirectQuery,WordPress.DB.DirectDatabaseQuery.NoCaching,WordPress.DB.DirectDatabaseQuery.SchemaChange -- Clear the plugin's own log.
			return;
		}
		$wpdb->delete( $table, array( 'id' => $id ), array( '%d' ) ); // phpcs:ignore WordPress.DB.DirectDatabaseQuery.DirectQuery,WordPress.DB.DirectDatabaseQuery.NoCaching -- Plugin table.
	}
}