<?php
/**
 * Search-term analytics: what visitors look for (and what they don't find).
 *
 * @package AccordionFaq
 */

namespace AccordionFaq\Analytics;

defined( 'ABSPATH' ) || exit;

/**
 * Class Searches
 */
class Searches {

	/**
	 * Record a search query (rate-limited, normalized, bounded).
	 *
	 * @param string $term    Raw query.
	 * @param int    $results Number of matches returned.
	 * @return void
	 */
	public static function record( $term, $results ) {
		$term = self::normalize( $term );
		if ( '' === $term ) {
			return;
		}

		// Rate limit: at most one recorded query per visitor per 3s to avoid
		// counting every keystroke of a debounced autocomplete.
		$ip  = isset( $_SERVER['REMOTE_ADDR'] ) ? sanitize_text_field( wp_unslash( $_SERVER['REMOTE_ADDR'] ) ) : '';
		$key = 'afaq_s_' . md5( $ip . '|' . $term );
		if ( get_transient( $key ) ) {
			return;
		}
		set_transient( $key, 1, 3 );

		global $wpdb;
		$table = Database::searches_table();
		$day   = gmdate( 'Y-m-d' );
		// phpcs:ignore WordPress.DB.DirectDatabaseQuery.DirectQuery -- writing to the plugin's own analytics table has no WordPress API alternative; the cache is invalidated below.
		$wpdb->query(
			$wpdb->prepare(
				'INSERT INTO %i (term, event_day, hits, results) VALUES (%s, %s, 1, %d)
				 ON DUPLICATE KEY UPDATE hits = hits + 1, results = %d',
				$table,
				$term,
				$day,
				(int) $results,
				(int) $results
			)
		);

		wp_cache_delete( 'last_changed', Database::CACHE_GROUP );
	}

	/**
	 * Normalize a query for storage: lowercase, collapse whitespace, trim length.
	 *
	 * @param string $term Raw query.
	 * @return string
	 */
	private static function normalize( $term ) {
		$term = trim( preg_replace( '/\s+/u', ' ', (string) $term ) );
		$term = function_exists( 'mb_strtolower' ) ? mb_strtolower( $term ) : strtolower( $term );
		$len  = function_exists( 'mb_strlen' ) ? mb_strlen( $term ) : strlen( $term );
		if ( $len < 2 ) {
			return '';
		}
		// Column is VARCHAR(120); keep it inside that bound.
		return function_exists( 'mb_substr' ) ? mb_substr( $term, 0, 120 ) : substr( $term, 0, 120 );
	}

	/**
	 * Top search terms over the last N days.
	 *
	 * @param int $limit Rows.
	 * @param int $days  Days back.
	 * @return array[] [ term, hits, results ].
	 */
	public static function top( $limit = 10, $days = 30 ) {
		return self::query( $limit, $days, false );
	}

	/**
	 * Top search terms that returned zero results (content gaps).
	 *
	 * @param int $limit Rows.
	 * @param int $days  Days back.
	 * @return array[] [ term, hits, results ].
	 */
	public static function no_results( $limit = 10, $days = 30 ) {
		return self::query( $limit, $days, true );
	}

	/**
	 * Shared aggregate query.
	 *
	 * @param int  $limit    Rows.
	 * @param int  $days     Days back.
	 * @param bool $zero_only Restrict to zero-result terms.
	 * @return array[]
	 */
	private static function query( $limit, $days, $zero_only ) {
		$last_changed = wp_cache_get_last_changed( Database::CACHE_GROUP );
		$cache_key    = 'searches:' . ( $zero_only ? 'zero:' : 'all:' ) . (int) $limit . ':' . (int) $days . ':' . $last_changed;
		$cached       = wp_cache_get( $cache_key, Database::CACHE_GROUP );
		if ( false !== $cached ) {
			return $cached;
		}

		global $wpdb;
		$table = Database::searches_table();
		$since = gmdate( 'Y-m-d', time() - ( (int) $days * DAY_IN_SECONDS ) );
		$limit = max( 1, min( 100, (int) $limit ) );

		if ( $zero_only ) {
			// phpcs:ignore WordPress.DB.DirectDatabaseQuery.DirectQuery -- reading the plugin's own analytics table has no WordPress API alternative; results are object-cached above.
			$rows = $wpdb->get_results(
				$wpdb->prepare(
					'SELECT term, SUM(hits) AS hits, MAX(results) AS results FROM %i WHERE event_day >= %s GROUP BY term HAVING MAX(results) = 0 ORDER BY hits DESC LIMIT %d',
					$table,
					$since,
					$limit
				),
				ARRAY_A
			);
		} else {
			// phpcs:ignore WordPress.DB.DirectDatabaseQuery.DirectQuery -- reading the plugin's own analytics table has no WordPress API alternative; results are object-cached above.
			$rows = $wpdb->get_results(
				$wpdb->prepare(
					'SELECT term, SUM(hits) AS hits, MAX(results) AS results FROM %i WHERE event_day >= %s GROUP BY term ORDER BY hits DESC LIMIT %d',
					$table,
					$since,
					$limit
				),
				ARRAY_A
			);
		}

		$out = array();
		foreach ( (array) $rows as $row ) {
			$out[] = array(
				'term'    => (string) $row['term'],
				'hits'    => (int) $row['hits'],
				'results' => (int) $row['results'],
			);
		}

		wp_cache_set( $cache_key, $out, Database::CACHE_GROUP, HOUR_IN_SECONDS );
		return $out;
	}
}