<?php
/**
 * Pro: tracked citation monitoring. The free "Live citation check" (see
 * Evoka_Byok_Client) answers one question once. This saves a list of
 * questions the user cares about and checks all of them automatically every
 * week (piggybacking the existing re-audit cron), building a history per
 * question - the "is AI mentioning me, over time, for the things I actually
 * care about" feature every paid AEO/GEO tool is built around.
 *
 * Still $0 backend: every check uses the user's own OpenAI/Perplexity key,
 * same as the free check. This only adds scheduling + storage + a crude,
 * clearly-labelled "other domains mentioned" extraction - not a second
 * provider integration.
 *
 * @package Evoka
 */

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

/**
 * Stores tracked questions and their citation history, and runs them.
 */
class Evoka_Citation_Tracker {

	const QUERIES_OPTION = 'evoka_tracked_queries';
	const HISTORY_OPTION = 'evoka_tracked_history';
	const MAX_QUERIES    = 20;
	const MAX_HISTORY    = 26; // ~6 months of weekly checks.

	/**
	 * All tracked queries.
	 *
	 * @return array List of {id, query, created}.
	 */
	public function queries() {
		return (array) get_option( self::QUERIES_OPTION, array() );
	}

	/**
	 * Add a tracked query.
	 *
	 * @param string $query Question text.
	 * @return array|WP_Error The new query row, or an error.
	 */
	public function add( $query ) {
		$query = trim( wp_strip_all_tags( (string) $query ) );
		if ( '' === $query ) {
			return new WP_Error( 'evoka_empty_query', __( 'Please enter a question to track.', 'evoka' ) );
		}
		$queries = $this->queries();
		if ( count( $queries ) >= self::MAX_QUERIES ) {
			/* translators: %d: maximum number of tracked questions */
			return new WP_Error( 'evoka_too_many', sprintf( __( 'You can track up to %d questions. Remove one before adding another.', 'evoka' ), self::MAX_QUERIES ) );
		}
		foreach ( $queries as $q ) {
			if ( 0 === strcasecmp( $q['query'], $query ) ) {
				return new WP_Error( 'evoka_duplicate', __( 'That question is already tracked.', 'evoka' ) );
			}
		}
		$row       = array(
			'id'      => wp_generate_password( 8, false ),
			'query'   => $query,
			'created' => time(),
		);
		$queries[] = $row;
		update_option( self::QUERIES_OPTION, $queries, false );
		return $row;
	}

	/**
	 * Remove a tracked query and its history.
	 *
	 * @param string $id Query id.
	 * @return bool True if a query was removed.
	 */
	public function remove( $id ) {
		$queries  = $this->queries();
		$filtered = array_values(
			array_filter(
				$queries,
				function ( $q ) use ( $id ) {
					return $q['id'] !== $id;
				}
			)
		);
		$removed  = count( $filtered ) !== count( $queries );
		if ( $removed ) {
			update_option( self::QUERIES_OPTION, $filtered, false );
			$history = get_option( self::HISTORY_OPTION, array() );
			unset( $history[ $id ] );
			update_option( self::HISTORY_OPTION, $history, false );
		}
		return $removed;
	}

	/**
	 * History for one tracked query, oldest first.
	 *
	 * @param string $id Query id.
	 * @return array
	 */
	public function history( $id ) {
		$history = get_option( self::HISTORY_OPTION, array() );
		return isset( $history[ $id ] ) ? (array) $history[ $id ] : array();
	}

	/**
	 * Run every tracked query now and append to history. No-ops quietly if
	 * no BYOK key/provider is configured - tracking is opt-in on top of an
	 * already-opt-in feature, so a missing key is not an error state.
	 *
	 * @return int Number of queries actually run.
	 */
	public function run_all() {
		$byok = new Evoka_Byok_Client();
		if ( ! $byok->is_configured() ) {
			return 0;
		}

		$ran = 0;
		foreach ( $this->queries() as $q ) {
			if ( ! is_wp_error( $this->run_one( $q['id'], $byok ) ) ) {
				++$ran;
			}
		}
		return $ran;
	}

	/**
	 * Run one tracked query now and append to its history.
	 *
	 * @param string                   $id   Query id.
	 * @param Evoka_Byok_Client|null $byok Reuse a client, or null to create one.
	 * @return array|WP_Error The recorded history point, or an error.
	 */
	public function run_one( $id, $byok = null ) {
		$queries = $this->queries();
		$query   = null;
		foreach ( $queries as $q ) {
			if ( $q['id'] === $id ) {
				$query = $q;
				break;
			}
		}
		if ( null === $query ) {
			return new WP_Error( 'evoka_not_found', __( 'That tracked question no longer exists.', 'evoka' ) );
		}

		$byok = $byok ? $byok : new Evoka_Byok_Client();
		if ( ! $byok->is_configured() ) {
			return new WP_Error( 'evoka_no_provider', __( 'Add your own API key and choose a provider in Evoka -> Settings first.', 'evoka' ) );
		}

		$result = $byok->check_citation( $query['query'] );
		if ( is_wp_error( $result ) ) {
			return $result;
		}

		$point = array(
			'date'          => gmdate( 'Y-m-d' ),
			'cited'         => (bool) $result['cited'],
			'other_domains' => $this->other_domains( $result['answer'], $result['sources'], $result['domain'] ),
		);

		$history          = get_option( self::HISTORY_OPTION, array() );
		$history[ $id ]   = isset( $history[ $id ] ) ? $history[ $id ] : array();
		$history[ $id ][] = $point;
		$history[ $id ]   = array_slice( $history[ $id ], -self::MAX_HISTORY );
		update_option( self::HISTORY_OPTION, $history, false );

		return $point;
	}

	/**
	 * Extract other domains mentioned in an answer/its sources, excluding the
	 * site's own domain. Crude pattern match on the answer text plus any
	 * source URLs - a directional "who else got mentioned" signal, not a
	 * verified competitor list, so the UI must label it as such.
	 *
	 * @param string $answer  Raw answer text.
	 * @param array  $sources Source URLs returned with the answer.
	 * @param string $domain  The site's own domain (excluded from results).
	 * @return array Up to 5 domains.
	 */
	public function other_domains( $answer, $sources, $domain ) {
		$found = array();

		foreach ( (array) $sources as $url ) {
			$host = (string) wp_parse_url( $url, PHP_URL_HOST );
			if ( $host ) {
				$found[] = strtolower( preg_replace( '/^www\./i', '', $host ) );
			}
		}

		if ( preg_match_all( '/\b([a-z0-9][a-z0-9-]{0,61}(?:\.[a-z0-9][a-z0-9-]{0,61}){0,3}\.(?:com|org|net|io|ai|co|dev|app|edu|gov))\b/i', (string) $answer, $m ) ) {
			foreach ( $m[1] as $host ) {
				$found[] = strtolower( preg_replace( '/^www\./i', '', $host ) );
			}
		}

		$own   = strtolower( preg_replace( '/^www\./i', '', (string) $domain ) );
		$found = array_values(
			array_unique(
				array_filter(
					$found,
					function ( $h ) use ( $own ) {
						return '' !== $own ? ( $h !== $own && ! Evoka_Byok_Client::mentions_domain( $h, $own ) ) : true;
					}
				)
			)
		);

		return array_slice( $found, 0, 5 );
	}
}