<?php
/**
 * Table creation, upgrades and scheduled cleanup.
 *
 * @package Citolume
 */

defined( 'ABSPATH' ) || exit;

/**
 * The crawler hit table name.
 *
 * @return string
 */
function citolume_table() {
	global $wpdb;

	return $wpdb->prefix . 'citolume_crawler_hits';
}

/**
 * Runs on activation.
 *
 * @return void
 */
function citolume_activate() {
	citolume_install();

	if ( ! wp_next_scheduled( 'citolume_prune_hits' ) ) {
		wp_schedule_event( time() + HOUR_IN_SECONDS, 'daily', 'citolume_prune_hits' );
	}

	// The virtual /llms.txt route is a rewrite rule, so the rules have to be
	// rebuilt before it will resolve.
	citolume_register_rewrites();
	flush_rewrite_rules();
}

/**
 * Runs on deactivation.
 *
 * @return void
 */
function citolume_deactivate() {
	wp_clear_scheduled_hook( 'citolume_prune_hits' );
	wp_clear_scheduled_hook( 'citolume_ping' );
	flush_rewrite_rules();
}

add_action( 'plugins_loaded', 'citolume_maybe_install' );
/**
 * Creates or upgrades the table when the schema version moves.
 *
 * Activation alone is not enough: a plugin updated in place never re-runs its
 * activation hook.
 *
 * @return void
 */
function citolume_maybe_install() {
	if ( (string) get_option( 'citolume_schema_version' ) === CITOLUME_SCHEMA_VERSION ) {
		return;
	}

	citolume_install();
}

/**
 * Creates the crawler hit table.
 *
 * Hits are aggregated per bot, per URL, per day rather than stored one row per
 * request. A busy site can be crawled thousands of times a week, and nothing in
 * the product needs request level detail.
 *
 * @return void
 */
function citolume_install() {
	global $wpdb;

	require_once ABSPATH . 'wp-admin/includes/upgrade.php';

	$table   = citolume_table();
	$collate = $wpdb->get_charset_collate();

	// url_hash rather than the URL itself carries the unique index: a URL long
	// enough to matter would overflow the index size limit on older MySQL.
	$sql = "CREATE TABLE {$table} (
		id bigint(20) unsigned NOT NULL AUTO_INCREMENT,
		bot varchar(32) NOT NULL,
		url varchar(255) NOT NULL,
		url_hash char(32) NOT NULL,
		hit_date date NOT NULL,
		hits int(10) unsigned NOT NULL DEFAULT 1,
		PRIMARY KEY  (id),
		UNIQUE KEY bot_url_date (bot,url_hash,hit_date),
		KEY hit_date (hit_date),
		KEY bot_date (bot,hit_date)
	) {$collate};";

	dbDelta( $sql );

	update_option( 'citolume_schema_version', CITOLUME_SCHEMA_VERSION, true );
}

add_action( 'citolume_prune_hits', 'citolume_prune_hits' );
/**
 * Drops hits older than the retention window.
 *
 * @return void
 */
function citolume_prune_hits() {
	global $wpdb;

	$days = (int) apply_filters( 'citolume_retention_days', (int) get_option( 'citolume_retention_days', 90 ) );
	$days = max( 7, $days );

	$table = citolume_table();

	$wpdb->query( // phpcs:ignore WordPress.DB.DirectDatabaseQuery -- scheduled cleanup of the plugin's own table.
		$wpdb->prepare(
			"DELETE FROM {$wpdb->prefix}citolume_crawler_hits WHERE hit_date < %s", // phpcs:ignore WordPress.DB.PreparedSQL.InterpolatedNotPrepared -- table name is built from $wpdb->prefix.
			gmdate( 'Y-m-d', time() - ( $days * DAY_IN_SECONDS ) )
		)
	);
}