<?php
/**
 * Controller: Traffic Aggregator.
 *
 * Reads buffered traffic data and writes aggregated stats to the database.
 *
 * @package advan
 *
 * @since 5.1.0
 */

declare(strict_types=1);

namespace ADVAN\Controllers;

use ADVAN\Entities\Traffic_Site_Stats_Entity;
use ADVAN\Entities\Traffic_Page_Stats_Entity;
use ADVAN\Entities\Traffic_Referrer_Stats_Entity;
use ADVAN\Entities\Traffic_Paths_Entity;
use ADVAN\Entities\Traffic_Referrer_Urls_Entity;
use ADVAN\Entities\Traffic_Tech_Entity;
use ADVAN\Entities\Traffic_Tech_Stats_Entity;

// Exit if accessed directly.
if ( ! defined( 'ABSPATH' ) ) {
	exit;
}

if ( ! class_exists( '\ADVAN\Controllers\Traffic_Aggregator' ) ) {
	/**
	 * Aggregates buffered traffic data into the database tables.
	 *
	 * @since 5.1.0
	 */
	class Traffic_Aggregator {

		/**
		 * Option key for realtime pageview counts (last 30 minutes).
		 *
		 * @var string
		 */
		public const REALTIME_OPTION = ADVAN_PREFIX . 'traffic_realtime';

		/**
		 * Run the aggregation cycle.
		 *
		 * Claims the buffer file, reads all records, accumulates in memory,
		 * then writes to DB in batch.
		 *
		 * @return int Number of records processed.
		 *
		 * @since 5.1.0
		 */
		public static function run(): int {
			// Clean up any stale busy files first.
			Traffic_Buffer::cleanup_stale();

			// Claim the buffer.
			$busy_file = Traffic_Buffer::claim_buffer();
			if ( false === $busy_file ) {
				return 0;
			}

			// Process the buffer in chunks to cap memory usage (R-05).
			$handle      = null;
			$total_count = 0;
			$realtime    = array();
			$chunk_size  = 5000;

			global $wpdb;
			$use_transaction = ( $wpdb->dbh instanceof \mysqli );

			while ( true ) {
				$records = Traffic_Buffer::read_buffer_chunk( $busy_file, $handle, $chunk_size );

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

				// Accumulate in memory for this chunk.
				$site_stats     = array();
				$page_stats     = array();
				$referrer_stats = array();
				$tech_stats     = array();

				foreach ( $records as $record ) {
					$date     = isset( $record['date'] ) ? $record['date'] : gmdate( 'Y-m-d' );
					$path     = isset( $record['path'] ) ? $record['path'] : '/';
					$referrer = isset( $record['referrer'] ) ? $record['referrer'] : '';
					$is_new   = ! empty( $record['is_new'] ) ? 1 : 0;

					if ( ! isset( $site_stats[ $date ] ) ) {
						$site_stats[ $date ] = array(
							'visitors'  => 0,
							'pageviews' => 0,
						);
					}
					$site_stats[ $date ]['pageviews'] += 1;
					$site_stats[ $date ]['visitors']  += $is_new;

					if ( ! isset( $page_stats[ $date ][ $path ] ) ) {
						$page_stats[ $date ][ $path ] = array(
							'visitors'  => 0,
							'pageviews' => 0,
						);
					}
					$page_stats[ $date ][ $path ]['pageviews'] += 1;
					$page_stats[ $date ][ $path ]['visitors']  += $is_new;

					if ( '' !== $referrer ) {
						if ( ! isset( $referrer_stats[ $date ][ $referrer ] ) ) {
							$referrer_stats[ $date ][ $referrer ] = array(
								'visitors'  => 0,
								'pageviews' => 0,
							);
						}
						$referrer_stats[ $date ][ $referrer ]['pageviews'] += 1;
						$referrer_stats[ $date ][ $referrer ]['visitors']  += $is_new;
					}

					$device_type = isset( $record['device_type'] ) ? $record['device_type'] : 'desktop';
					$browser     = isset( $record['browser'] ) ? $record['browser'] : 'Other';
					$os          = isset( $record['os'] ) ? $record['os'] : 'Other';
					$tech_key    = $device_type . '|' . $browser . '|' . $os;

					if ( ! isset( $tech_stats[ $date ][ $tech_key ] ) ) {
						$tech_stats[ $date ][ $tech_key ] = array(
							'device_type' => $device_type,
							'browser'     => $browser,
							'os'          => $os,
							'visitors'    => 0,
							'pageviews'   => 0,
						);
					}
					$tech_stats[ $date ][ $tech_key ]['pageviews'] += 1;
					$tech_stats[ $date ][ $tech_key ]['visitors']  += $is_new;

					// Track unique visitors for realtime by IP hash.
					$ip_hash   = isset( $record['ip_hash'] ) ? $record['ip_hash'] : md5( 'unknown' . gmdate( 'Y-m-d' ) );
					$timestamp = isset( $record['timestamp'] ) ? (int) $record['timestamp'] : time();
					if ( ! isset( $realtime[ $ip_hash ] ) || $realtime[ $ip_hash ] < $timestamp ) {
						$realtime[ $ip_hash ] = $timestamp;
					}
				}

				// Write this chunk to DB in a transaction.
				if ( $use_transaction ) {
					$wpdb->query( 'START TRANSACTION' );
				}

				try {
					self::write_aggregated_stats( $site_stats, $page_stats, $referrer_stats, $tech_stats );

					if ( $use_transaction ) {
						$wpdb->query( 'COMMIT' );
					}
				} catch ( \Throwable $e ) {
					if ( $use_transaction ) {
						$wpdb->query( 'ROLLBACK' );
					}

					// Close handle if still open.
					if ( null !== $handle && is_resource( $handle ) ) {
						// phpcs:ignore WordPress.WP.AlternativeFunctions.file_system_operations_fclose
						fclose( $handle );
					}

					// Don't remove buffer on failure — data can be retried.
					return 0;
				}

				$total_count += count( $records );

				// Free chunk memory before reading the next.
				unset( $site_stats, $page_stats, $referrer_stats, $tech_stats, $records );

				// If handle was closed by read_buffer_chunk (EOF reached), stop.
				if ( null === $handle ) {
					break;
				}
			}

			if ( 0 === $total_count ) {
				Traffic_Buffer::remove_buffer( $busy_file );
				return 0;
			}

			// Only remove buffer after successful DB commit.
			Traffic_Buffer::remove_buffer( $busy_file );

			// Update realtime counter via transient (P1).
			self::update_realtime( $realtime );

			return $total_count;
		}

		/**
		 * Write aggregated stats to the database entities.
		 *
		 * Extracted to share between run() and write_batch().
		 *
		 * @param array $site_stats     Site-level stats.
		 * @param array $page_stats     Per-page stats.
		 * @param array $referrer_stats Per-referrer stats.
		 * @param array $tech_stats     Per-tech-combo stats.
		 *
		 * @return void
		 *
		 * @since 5.1.0
		 */
		private static function write_aggregated_stats( array $site_stats, array $page_stats, array $referrer_stats, array $tech_stats ): void {
			foreach ( $site_stats as $date => $stats ) {
				Traffic_Site_Stats_Entity::upsert( $date, $stats['visitors'], $stats['pageviews'] );
			}

			foreach ( $page_stats as $date => $paths ) {
				foreach ( $paths as $path => $stats ) {
					$path_id = Traffic_Paths_Entity::get_or_create( $path );
					if ( $path_id > 0 ) {
						Traffic_Page_Stats_Entity::upsert( $date, $path_id, $stats['visitors'], $stats['pageviews'] );
					}
				}
			}

			foreach ( $referrer_stats as $date => $referrers ) {
				foreach ( $referrers as $referrer_url => $stats ) {
					$referrer_id = Traffic_Referrer_Urls_Entity::get_or_create( $referrer_url );
					if ( $referrer_id > 0 ) {
						Traffic_Referrer_Stats_Entity::upsert( $date, $referrer_id, $stats['visitors'], $stats['pageviews'] );
					}
				}
			}

			foreach ( $tech_stats as $date => $combos ) {
				foreach ( $combos as $stats ) {
					$tech_id = Traffic_Tech_Entity::get_or_create( $stats['device_type'], $stats['browser'], $stats['os'] );
					if ( $tech_id > 0 ) {
						Traffic_Tech_Stats_Entity::upsert( $date, $tech_id, $stats['visitors'], $stats['pageviews'] );
					}
				}
			}
		}

		/**
		 * Write a batch of records directly to the database (P2 — in-memory buffer flush).
		 *
		 * Accumulates records first, then writes in a single pass to reduce query count.
		 *
		 * @param array $records Array of pageview data arrays.
		 *
		 * @return void
		 *
		 * @since 5.1.0
		 */
		public static function write_batch( array $records ): void {
			if ( empty( $records ) ) {
				return;
			}

			$site_stats     = array();
			$page_stats     = array();
			$referrer_stats = array();
			$tech_stats     = array();
			$realtime       = array();

			foreach ( $records as $record ) {
				$date     = isset( $record['date'] ) ? $record['date'] : gmdate( 'Y-m-d' );
				$path     = isset( $record['path'] ) ? $record['path'] : '/';
				$referrer = isset( $record['referrer'] ) ? $record['referrer'] : '';
				$is_new   = ! empty( $record['is_new'] ) ? 1 : 0;

				if ( ! isset( $site_stats[ $date ] ) ) {
					$site_stats[ $date ] = array(
						'visitors'  => 0,
						'pageviews' => 0,
					);
				}
				$site_stats[ $date ]['pageviews'] += 1;
				$site_stats[ $date ]['visitors']  += $is_new;

				if ( ! isset( $page_stats[ $date ][ $path ] ) ) {
					$page_stats[ $date ][ $path ] = array(
						'visitors'  => 0,
						'pageviews' => 0,
					);
				}
				$page_stats[ $date ][ $path ]['pageviews'] += 1;
				$page_stats[ $date ][ $path ]['visitors']  += $is_new;

				if ( '' !== $referrer ) {
					if ( ! isset( $referrer_stats[ $date ][ $referrer ] ) ) {
						$referrer_stats[ $date ][ $referrer ] = array(
							'visitors'  => 0,
							'pageviews' => 0,
						);
					}
					$referrer_stats[ $date ][ $referrer ]['pageviews'] += 1;
					$referrer_stats[ $date ][ $referrer ]['visitors']  += $is_new;
				}

				$device_type = isset( $record['device_type'] ) ? $record['device_type'] : 'desktop';
				$browser     = isset( $record['browser'] ) ? $record['browser'] : 'Other';
				$os          = isset( $record['os'] ) ? $record['os'] : 'Other';
				$tech_key    = $device_type . '|' . $browser . '|' . $os;

				if ( ! isset( $tech_stats[ $date ][ $tech_key ] ) ) {
					$tech_stats[ $date ][ $tech_key ] = array(
						'device_type' => $device_type,
						'browser'     => $browser,
						'os'          => $os,
						'visitors'    => 0,
						'pageviews'   => 0,
					);
				}
				$tech_stats[ $date ][ $tech_key ]['pageviews'] += 1;
				$tech_stats[ $date ][ $tech_key ]['visitors']  += $is_new;

				// Track unique visitors for realtime by IP hash.
				$ip_hash   = isset( $record['ip_hash'] ) ? $record['ip_hash'] : md5( 'unknown' . gmdate( 'Y-m-d' ) );
				$timestamp = isset( $record['timestamp'] ) ? (int) $record['timestamp'] : time();
				if ( ! isset( $realtime[ $ip_hash ] ) || $realtime[ $ip_hash ] < $timestamp ) {
					$realtime[ $ip_hash ] = $timestamp;
				}
			}

			self::write_aggregated_stats( $site_stats, $page_stats, $referrer_stats, $tech_stats );
			self::update_realtime( $realtime );
		}

		/**
		 * Update the realtime pageview counter option.
		 *
		 * @param array $new_visitors IP hash => last-seen timestamp map.
		 *
		 * @return void
		 *
		 * @since 5.1.0
		 */
		private static function update_realtime( array $new_visitors ): void {
			// Use options (DB-backed) instead of transients to avoid data loss
			// on sites with volatile object caches (Redis/Memcached) that may
			// evict the key between cron writes and dashboard reads.
			$existing = \get_option( self::REALTIME_OPTION, array() );
			if ( ! is_array( $existing ) ) {
				$existing = array();
			}

			// Merge new visitors — keep the latest timestamp per IP hash.
			foreach ( $new_visitors as $ip_hash => $timestamp ) {
				if ( ! isset( $existing[ $ip_hash ] ) || $existing[ $ip_hash ] < $timestamp ) {
					$existing[ $ip_hash ] = $timestamp;
				}
			}

			// Prune visitors not seen in the last 30 minutes.
			$cutoff = time() - 1800;
			foreach ( $existing as $ip_hash => $timestamp ) {
				if ( $timestamp < $cutoff ) {
					unset( $existing[ $ip_hash ] );
				}
			}

			\update_option( self::REALTIME_OPTION, $existing, false );
		}

		/**
		 * Get the realtime unique visitor count (last 30 minutes).
		 *
		 * Bypasses the persistent object cache (Redis / Memcached) so the
		 * admin dashboard always reads the freshest value written by the
		 * aggregation cron — even when that cron runs in a separate process.
		 * Stale entries are pruned on every read to keep the option compact.
		 *
		 * @return int
		 *
		 * @since 5.1.0
		 */
		public static function get_realtime_count(): int {
			// Flush any stale persistent-cache copy so get_option() hits the DB.
			\wp_cache_delete( self::REALTIME_OPTION, 'options' );

			$data = \get_option( self::REALTIME_OPTION, array() );
			if ( ! is_array( $data ) ) {
				return 0;
			}

			$cutoff = time() - 1800;
			$total  = 0;
			$pruned = false;

			foreach ( $data as $ip_hash => $timestamp ) {
				if ( $timestamp >= $cutoff ) {
					++$total;
				} else {
					unset( $data[ $ip_hash ] );
					$pruned = true;
				}
			}

			// Write back pruned data so the option stays compact and the
			// cache is refreshed with the authoritative value.
			if ( $pruned ) {
				\update_option( self::REALTIME_OPTION, $data, false );
			}

			return $total;
		}

		/**
		 * Prune old traffic data beyond retention period.
		 *
		 * @param int $days Number of days to retain.
		 *
		 * @return void
		 *
		 * @since 5.1.0
		 */
		public static function prune( int $days = 365 ): void {

			$cutoff_date = gmdate( 'Y-m-d', strtotime( "-{$days} days" ) );

			// Use delete_query from entities — it has the 1146 error recovery built in.
			Traffic_Site_Stats_Entity::delete_query(
				'DELETE FROM `' . Traffic_Site_Stats_Entity::get_table_name() . '` WHERE date < %s',
				array( $cutoff_date )
			);
			Traffic_Page_Stats_Entity::delete_query(
				'DELETE FROM `' . Traffic_Page_Stats_Entity::get_table_name() . '` WHERE date < %s',
				array( $cutoff_date )
			);
			Traffic_Referrer_Stats_Entity::delete_query(
				'DELETE FROM `' . Traffic_Referrer_Stats_Entity::get_table_name() . '` WHERE date < %s',
				array( $cutoff_date )
			);

			// Clean orphaned paths.
			$paths_table = Traffic_Paths_Entity::get_table_name();
			$stats_table = Traffic_Page_Stats_Entity::get_table_name();

			// phpcs:ignore WordPress.DB.PreparedSQL.InterpolatedNotPrepared
			Traffic_Paths_Entity::delete_query(
				"DELETE p FROM `{$paths_table}` p LEFT JOIN `{$stats_table}` s ON p.id = s.path_id WHERE s.path_id IS NULL"
			);

			// Clean orphaned referrer URLs.
			$urls_table      = Traffic_Referrer_Urls_Entity::get_table_name();
			$ref_stats_table = Traffic_Referrer_Stats_Entity::get_table_name();

			// phpcs:ignore WordPress.DB.PreparedSQL.InterpolatedNotPrepared
			Traffic_Referrer_Urls_Entity::delete_query(
				"DELETE u FROM `{$urls_table}` u LEFT JOIN `{$ref_stats_table}` s ON u.id = s.referrer_id WHERE s.referrer_id IS NULL"
			);

			// Prune tech stats.
			Traffic_Tech_Stats_Entity::delete_query(
				'DELETE FROM `' . Traffic_Tech_Stats_Entity::get_table_name() . '` WHERE date < %s',
				array( $cutoff_date )
			);

			// Clean orphaned tech combos.
			$tech_table       = Traffic_Tech_Entity::get_table_name();
			$tech_stats_table = Traffic_Tech_Stats_Entity::get_table_name();

			// phpcs:ignore WordPress.DB.PreparedSQL.InterpolatedNotPrepared
			Traffic_Tech_Entity::delete_query(
				"DELETE t FROM `{$tech_table}` t LEFT JOIN `{$tech_stats_table}` s ON t.id = s.tech_id WHERE s.tech_id IS NULL"
			);
		}
	}
}