<?php
defined( 'ABSPATH' ) || exit;

/**
 * Keyword Rank Tracker - Pro feature.
 * Tracks keyword positions over time with history and email alerts.
 *
 * Position data comes from two sources, in this order:
 *   1. Google Search Console (free, real click/impression-linked data) -
 *      used whenever GSC is connected AND has data for that query. GSC has
 *      nothing for brand-new pages or low-volume queries yet, which is the
 *      common case for a newly published page or a lower-volume dialect
 *      keyword, so this is not always enough on its own.
 *   2. A live SerpApi.com check (self-configured, optional) - used as a
 *      fallback whenever GSC is disconnected or has no data for that query.
 *      Requires a "Rank Tracker API Key" set under Settings; without one,
 *      positions simply aren't available outside GSC, same as before this
 *      fallback existed.
 *
 * Each tracked keyword can optionally be pinned to one of the plugin's 9
 * Arabic dialect regions (same list as the hreflang manager and dialect
 * adapter). When a region is set, both the GSC lookup (via its `country`
 * dimension) and the SerpApi fallback (via `gl`/`hl`) are scoped to that
 * market, so "how do I rank for this keyword in Kuwait" and "how do I rank
 * for it in Saudi" can be tracked as two separate rows for the same term.
 */
class AWBSEO_Rank_Tracker {

    const OPTION_KEYWORDS = 'awbseo_tracked_keywords';
    const TABLE_SUFFIX    = 'awb_rank_history';

    /**
     * Region -> search targeting. `gl` is the 2-letter country code SerpApi
     * (and Google's own Ads/Search APIs) use for geotargeting; `gsc_country`
     * is the 3-letter ISO code Search Console's `country` dimension expects
     * (a different code length, easy to mix up). 'msa' has no country - it's
     * the untargeted/formal-Arabic case, tracked without a location filter.
     *
     * Deliberately NOT using Google's old country-domain trick
     * (google.com.sa, google.ae, etc.) - Google has been phasing ccTLD
     * search domains out in favor of google.com with location handled via
     * `gl`/account settings, so a domain-based approach would be building on
     * a mechanism that's actively being retired.
     */
    const REGION_TARGETING = [
        'msa'       => [ 'gl' => '',   'gsc_country' => '',    'hl' => 'ar' ],
        'saudi'     => [ 'gl' => 'sa', 'gsc_country' => 'sau', 'hl' => 'ar' ],
        'egyptian'  => [ 'gl' => 'eg', 'gsc_country' => 'egy', 'hl' => 'ar' ],
        'kuwaiti'   => [ 'gl' => 'kw', 'gsc_country' => 'kwt', 'hl' => 'ar' ],
        'lebanese'  => [ 'gl' => 'lb', 'gsc_country' => 'lbn', 'hl' => 'ar' ],
        'uae'       => [ 'gl' => 'ae', 'gsc_country' => 'are', 'hl' => 'ar' ],
        'iraqi'     => [ 'gl' => 'iq', 'gsc_country' => 'irq', 'hl' => 'ar' ],
        'jordanian' => [ 'gl' => 'jo', 'gsc_country' => 'jor', 'hl' => 'ar' ],
        'moroccan'  => [ 'gl' => 'ma', 'gsc_country' => 'mar', 'hl' => 'ar' ],
    ];

    public static function init() {
        add_action( 'awbseo_daily_rank_check', [ __CLASS__, 'run_checks' ] );

        if ( ! wp_next_scheduled( 'awbseo_daily_rank_check' ) ) {
            wp_schedule_event( time(), 'daily', 'awbseo_daily_rank_check' );
        }
    }

    public static function create_table() {
        global $wpdb;
        $table   = $wpdb->prefix . self::TABLE_SUFFIX;
        $charset = $wpdb->get_charset_collate();
        $sql     = "CREATE TABLE IF NOT EXISTS {$table} (
            id         BIGINT UNSIGNED NOT NULL AUTO_INCREMENT,
            keyword    VARCHAR(255)    NOT NULL,
            position   SMALLINT        NOT NULL DEFAULT 0,
            url        VARCHAR(512)    NOT NULL,
            checked_at DATE            NOT NULL,
            PRIMARY KEY (id),
            KEY keyword (keyword),
            KEY checked_at (checked_at)
        ) {$charset};";
        require_once ABSPATH . 'wp-admin/includes/upgrade.php';
        dbDelta( $sql );
    }

    /**
     * The 9 dialect regions a keyword can be pinned to, for the admin UI's
     * dropdown. Same shape as AWBSEO_Hreflang_Manager::dialect_regions() -
     * reusing that method directly rather than duplicating its labels here,
     * since this is a plain code/label lookup, not one of the AI-prompt
     * hint blocks this codebase deliberately keeps as independent copies.
     */
    public static function regions() {
        return AWBSEO_Hreflang_Manager::dialect_regions();
    }

    public static function get_keywords() {
        return get_option( self::OPTION_KEYWORDS, [] );
    }

    public static function add_keyword( $keyword, $target_url = '', $region = '' ) {
        $keywords = self::get_keywords();
        $keyword  = sanitize_text_field( strtolower( trim( $keyword ) ) );
        if ( ! $keyword || isset( $keywords[ $keyword ] ) ) return false;
        $region = array_key_exists( $region, self::REGION_TARGETING ) ? $region : '';
        $keywords[ $keyword ] = [
            'url'        => esc_url_raw( $target_url ),
            'region'     => $region,
            'added'      => current_time( 'mysql' ),
            'alert_drop' => 5, // alert if drops more than 5 positions
        ];
        update_option( self::OPTION_KEYWORDS, $keywords );
        return true;
    }

    public static function remove_keyword( $keyword ) {
        $keywords = self::get_keywords();
        unset( $keywords[ sanitize_text_field( $keyword ) ] );
        update_option( self::OPTION_KEYWORDS, $keywords );
    }

    public static function get_history( $keyword, $days = 30 ) {
        global $wpdb;
        $table = $wpdb->prefix . self::TABLE_SUFFIX;
        return $wpdb->get_results(
            $wpdb->prepare(
                "SELECT position, url, checked_at FROM {$table}
                 WHERE keyword = %s AND checked_at >= DATE_SUB(CURDATE(), INTERVAL %d DAY)
                 ORDER BY checked_at ASC",
                $keyword, $days
            )
        );
    }

    /**
     * Fetches the current position for a single keyword and stores it.
     * Called immediately when a keyword is added so the first data point is
     * available without waiting for the next daily cron run.
     * Returns the position (int) or null if unavailable from any source.
     */
    public static function fetch_position( $keyword ) {
        $keywords = self::get_keywords();
        $config   = $keywords[ $keyword ] ?? [];

        $position = self::lookup_position( $keyword, $config['url'] ?? '', $config['region'] ?? '' );
        if ( $position === null ) return null;

        $position = (int) round( $position );
        self::store_position( $keyword, $position, $config['url'] ?? '' );
        return $position;
    }

    /**
     * Runs daily - checks positions for every tracked keyword.
     */
    public static function run_checks() {
        $keywords = self::get_keywords();
        if ( empty( $keywords ) ) return;

        foreach ( $keywords as $keyword => $config ) {
            $position = self::lookup_position( $keyword, $config['url'] ?? '', $config['region'] ?? '' );
            if ( $position === null ) continue;

            $position = (int) round( $position );
            self::store_position( $keyword, $position, $config['url'] ?? '' );
            self::maybe_send_alert( $keyword, $position, $config );
        }
    }

    /**
     * GSC first (real data, free), falling back to a live SerpApi check
     * whenever GSC is disconnected OR simply has no rows for this query yet
     * - which is the normal state for a brand-new page or a lower-volume
     * dialect keyword, not just for sites that skipped connecting GSC.
     * Returns a float|int position, or null if neither source has one.
     */
    private static function lookup_position( $keyword, $target_url, $region ) {
        $meta = self::REGION_TARGETING[ $region ] ?? [ 'gl' => '', 'gsc_country' => '', 'hl' => '' ];

        $gsc = AWBSEO_GSC::get_instance();
        if ( $gsc->is_connected() ) {
            $position = $gsc->get_keyword_position( $keyword, $meta['gsc_country'] );
            if ( $position !== null ) return $position;
        }

        return self::serp_check( $keyword, $target_url, $meta );
    }

    private static function store_position( $keyword, $position, $url ) {
        global $wpdb;
        $wpdb->replace(
            $wpdb->prefix . self::TABLE_SUFFIX,
            [
                'keyword'    => $keyword,
                'position'   => $position,
                'url'        => $url,
                'checked_at' => current_time( 'Y-m-d' ),
            ],
            [ '%s', '%d', '%s', '%s' ]
        );
    }

    private static function maybe_send_alert( $keyword, $current_pos, $config ) {
        global $wpdb;
        $table = $wpdb->prefix . self::TABLE_SUFFIX;

        // Get yesterday's position.
        $yesterday = $wpdb->get_var(
            $wpdb->prepare(
                "SELECT position FROM {$table}
                 WHERE keyword = %s AND checked_at = DATE_SUB(CURDATE(), INTERVAL 1 DAY)",
                $keyword
            )
        );

        if ( $yesterday === null ) return;
        $drop = (int) $current_pos - (int) $yesterday;
        if ( $drop < ( $config['alert_drop'] ?? 5 ) ) return;

        $admin_email = get_option( 'admin_email' );
        wp_mail(
            $admin_email,
            sprintf( '[AWBSEO] Ranking drop alert: "%s"', $keyword ),
            sprintf(
                "Your keyword \"%s\" dropped %d positions (from #%d to #%d).\n\nLog in to CitedSEO → Rank Tracker to view the full history.\n\n%s",
                $keyword, $drop, $yesterday, $current_pos,
                admin_url( 'admin.php?page=awb-seo-rank-tracker' )
            )
        );
    }

    /**
     * Live SERP position check via SerpApi.com (https://serpapi.com/) -
     * used as a fallback when GSC has no data for this query. Requires a
     * "Rank Tracker API Key" set under Settings; without one this always
     * returns null, exactly like before this integration existed, so a site
     * with no key configured behaves the same as it always did.
     *
     * Matches by host, not exact URL, since a ranking page can carry query
     * params or a trailing slash a strict string comparison would miss, and
     * falls back to the site's own domain when the keyword has no specific
     * target URL set (the "Target URL" field on the Rank Tracker screen is
     * optional).
     */
    private static function serp_check( $keyword, $target_url, array $region_meta ) {
        $settings = get_option( 'awbseo_settings', [] );
        $api_key  = trim( $settings['serp_api_key'] ?? '' );
        if ( ! $api_key || ! $keyword ) return null;

        $target_host = self::normalize_host( $target_url ?: home_url() );
        if ( ! $target_host ) return null;

        $query = [
            'engine'  => 'google',
            'q'       => $keyword,
            'num'     => 100,
            'device'  => 'desktop',
            'api_key' => $api_key,
        ];
        if ( ! empty( $region_meta['hl'] ) ) $query['hl'] = $region_meta['hl'];
        if ( ! empty( $region_meta['gl'] ) ) $query['gl'] = $region_meta['gl'];

        $response = wp_remote_get( 'https://serpapi.com/search.json?' . http_build_query( $query ), [
            'timeout' => 15,
        ] );

        if ( is_wp_error( $response ) || 200 !== (int) wp_remote_retrieve_response_code( $response ) ) {
            return null;
        }

        $body    = json_decode( wp_remote_retrieve_body( $response ), true );
        $results = $body['organic_results'] ?? [];
        if ( ! $results ) return null;

        foreach ( $results as $result ) {
            $result_host = self::normalize_host( $result['link'] ?? '' );
            if ( $result_host && $result_host === $target_host ) {
                $position = (int) ( $result['position'] ?? 0 );
                return $position > 0 ? $position : null;
            }
        }

        return null;
    }

    private static function normalize_host( $url ) {
        $host = wp_parse_url( $url, PHP_URL_HOST );
        if ( ! $host ) return '';
        return preg_replace( '/^www\./', '', strtolower( $host ) );
    }
}