<?php
/**
 * Automatic llms.txt generator.
 *
 * Updates automatically every time content is published.
 *
 * @package Aidentity_GEO
 * @since 1.0.0
 */

// Block direct access.
if ( ! defined( 'ABSPATH' ) ) {
    exit;
}

/**
 * llms.txt generation class.
 *
 * @since 1.0.0
 */
class Aidentity_GEO_LLMS_TXT {

    /**
     * Single class instance.
     *
     * @var Aidentity_GEO_LLMS_TXT|null
     */
    private static $instance = null;

    /**
     * Returns the single class instance.
     *
     * @since 1.0.0
     * @return Aidentity_GEO_LLMS_TXT
     */
    public static function get_instance() {
        if ( null === self::$instance ) {
            self::$instance = new self();
        }
        return self::$instance;
    }

    /**
     * Constructor.
     *
     * @since 1.0.0
     */
    private function __construct() {
        // Regenerate when a post is published, updated or deleted.
        add_action( 'publish_post', array( $this, 'schedule_regeneration' ) );
        add_action( 'publish_page', array( $this, 'schedule_regeneration' ) );
        add_action( 'wp_trash_post', array( $this, 'schedule_regeneration' ) );
        add_action( 'delete_post', array( $this, 'schedule_regeneration' ) );

        // Hook for scheduled regeneration.
        add_action( 'aidentity_geo_regenerate_llms_txt', array( $this, 'generate_file' ) );

        // Intercept requests to /llms.txt and /llms-full.txt.
        // NOTE: this class is instantiated on 'init' (priority 10) by the main
        // plugin class, so hooking 'init' at an earlier priority would never
        // fire. 'parse_request' runs after 'init' but BEFORE the main query,
        // the 404 handling and canonical/guess redirects, so the file is
        // served with a proper HTTP 200 status (required by the
        // PageSpeed/Lighthouse Agentic Browsing audit).
        add_action( 'parse_request', array( $this, 'serve_file' ), 0 );
    }

    /**
     * Schedules file regeneration.
     *
     * @since 1.0.0
     */
    public function schedule_regeneration() {
        // Use a small delay to avoid multiple regenerations.
        if ( ! wp_next_scheduled( 'aidentity_geo_regenerate_llms_txt' ) ) {
            wp_schedule_single_event( time() + 30, 'aidentity_geo_regenerate_llms_txt' );
        }
    }

    /**
     * Serves the llms.txt file.
     *
     * @since 1.0.0
     */
    public function serve_file() {
        $settings = aidentity_geo_get_settings();

        if ( empty( $settings['llms_txt_enabled'] ) ) {
            return;
        }

        // Sanitize and clean the request URI.
        $request = isset( $_SERVER['REQUEST_URI'] ) ? sanitize_text_field( wp_unslash( $_SERVER['REQUEST_URI'] ) ) : '';
        $request = trim( $request, '/' );

        // Strip query string.
        $request = strtok( $request, '?' );

        if ( 'llms.txt' !== $request && 'llms-full.txt' !== $request ) {
            return;
        }

        // Keep page-cache plugins (WP Rocket, W3TC, LiteSpeed...) away from this route.
        if ( ! defined( 'DONOTCACHEPAGE' ) ) {
            define( 'DONOTCACHEPAGE', true );
        }

        if ( 'llms.txt' === $request ) {
            $this->output_llms_txt();
            exit;
        }

        $this->output_llms_full_txt();
        exit;
    }

    /**
     * Outputs the llms.txt file.
     *
     * @since 1.0.0
     */
    private function output_llms_txt() {
        status_header( 200 );
        header( 'Content-Type: text/plain; charset=utf-8' );
        header( 'Cache-Control: public, max-age=3600' );
        header( 'X-Generated-By: Aidentity GEO ' . AIDENTITY_GEO_VERSION );

        $content = get_transient( 'aidentity_geo_llms_txt' );

        if ( false === $content ) {
            $content = $this->generate_llms_txt();
            set_transient( 'aidentity_geo_llms_txt', $content, HOUR_IN_SECONDS );
        }

        echo $content; // phpcs:ignore WordPress.Security.EscapeOutput.OutputNotEscaped -- Plain text output.
    }

    /**
     * Outputs the llms-full.txt file.
     *
     * @since 1.0.0
     */
    private function output_llms_full_txt() {
        status_header( 200 );
        header( 'Content-Type: text/plain; charset=utf-8' );
        header( 'Cache-Control: public, max-age=3600' );
        header( 'X-Generated-By: Aidentity GEO ' . AIDENTITY_GEO_VERSION );

        $content = get_transient( 'aidentity_geo_llms_full_txt' );

        if ( false === $content ) {
            $content = $this->generate_llms_full_txt();
            set_transient( 'aidentity_geo_llms_full_txt', $content, HOUR_IN_SECONDS );
        }

        echo $content; // phpcs:ignore WordPress.Security.EscapeOutput.OutputNotEscaped -- Plain text output.
    }

    /**
     * Generates the files and stores them in cache.
     *
     * @since 1.0.0
     */
    public function generate_file() {
        // Clear the cache.
        delete_transient( 'aidentity_geo_llms_txt' );
        delete_transient( 'aidentity_geo_llms_full_txt' );

        // Pre-generate the files.
        $llms_txt  = $this->generate_llms_txt();
        $llms_full = $this->generate_llms_full_txt();

        set_transient( 'aidentity_geo_llms_txt', $llms_txt, HOUR_IN_SECONDS );
        set_transient( 'aidentity_geo_llms_full_txt', $llms_full, HOUR_IN_SECONDS );
    }

    /**
     * Generates the llms.txt content.
     *
     * @since 1.0.0
     * @return string File content.
     */
    private function generate_llms_txt() {
        $site_name        = get_bloginfo( 'name' );
        $site_description = get_bloginfo( 'description' );

        // Header with entity information.
        $output  = "# {$site_name}\n\n";
        $output .= "> {$site_description}\n\n";
        $output .= '**' . esc_html__( 'Site:', 'aidentity-generative-engine-optimizer' ) . '** ' . home_url() . "\n\n";

        // Links to AI resources.
        $output .= '## ' . esc_html__( 'Resources for AI', 'aidentity-generative-engine-optimizer' ) . "\n\n";
        $output .= '- [' . esc_html__( 'Full content', 'aidentity-generative-engine-optimizer' ) . '](' . home_url( '/llms-full.txt' ) . ")\n";
        $output .= '- [Sitemap XML](' . home_url( '/wp-sitemap.xml' ) . ")\n\n";

        // Main pages.
        $pages = get_posts(
            array(
                'post_type'      => 'page',
                'posts_per_page' => 20,
                'post_status'    => 'publish',
                'orderby'        => 'menu_order',
                'order'          => 'ASC',
                'meta_query'     => array( // phpcs:ignore WordPress.DB.SlowDBQuery.slow_db_query_meta_query
                    'relation' => 'OR',
                    array(
                        'key'     => '_geo_exclude',
                        'compare' => 'NOT EXISTS',
                    ),
                    array(
                        'key'     => '_geo_exclude',
                        'value'   => '1',
                        'compare' => '!=',
                    ),
                ),
            )
        );

        if ( ! empty( $pages ) ) {
            $output .= '## ' . esc_html__( 'Main Pages', 'aidentity-generative-engine-optimizer' ) . "\n\n";
            foreach ( $pages as $page ) {
                $description = $this->get_post_description( $page );
                $url         = get_permalink( $page );
                $output     .= "- [{$page->post_title}]({$url})";
                if ( $description ) {
                    $output .= ": {$description}";
                }
                $output .= "\n";
            }
            $output .= "\n";
        }

        // Posts grouped by category.
        $categories = get_categories(
            array(
                'orderby'    => 'count',
                'order'      => 'DESC',
                'number'     => 10,
                'hide_empty' => true,
            )
        );

        foreach ( $categories as $category ) {
            $posts = get_posts(
                array(
                    'post_type'      => 'post',
                    'posts_per_page' => 10,
                    'post_status'    => 'publish',
                    'category'       => $category->term_id,
                    'meta_query'     => array( // phpcs:ignore WordPress.DB.SlowDBQuery.slow_db_query_meta_query
                        'relation' => 'OR',
                        array(
                            'key'     => '_geo_exclude',
                            'compare' => 'NOT EXISTS',
                        ),
                        array(
                            'key'     => '_geo_exclude',
                            'value'   => '1',
                            'compare' => '!=',
                        ),
                    ),
                )
            );

            if ( ! empty( $posts ) ) {
                $output .= "## {$category->name}\n\n";
                foreach ( $posts as $post ) {
                    $description = $this->get_post_description( $post );
                    $url         = get_permalink( $post );
                    $output     .= "- [{$post->post_title}]({$url})";
                    if ( $description ) {
                        $output .= ": {$description}";
                    }
                    $output .= "\n";
                }
                $output .= "\n";
            }
        }

        // Footer.
        $output .= "---\n";
        $output .= sprintf(
            /* translators: %1$s: plugin version, %2$s: current date and time */
            esc_html__( 'Generated by Aidentity GEO v%1$s | %2$s', 'aidentity-generative-engine-optimizer' ),
            AIDENTITY_GEO_VERSION,
            current_time( 'Y-m-d H:i' )
        ) . "\n";

        return $output;
    }

    /**
     * Generates the llms-full.txt content.
     *
     * @since 1.0.0
     * @return string File content.
     */
    private function generate_llms_full_txt() {
        $site_name        = get_bloginfo( 'name' );
        $site_description = get_bloginfo( 'description' );

        $output  = "# {$site_name}\n\n";
        $output .= "> {$site_description}\n\n";
        $output .= "---\n\n";

        // All published content.
        $posts = get_posts(
            array(
                'post_type'      => array( 'post', 'page' ),
                'posts_per_page' => 100,
                'post_status'    => 'publish',
                'orderby'        => 'date',
                'order'          => 'DESC',
                'meta_query'     => array( // phpcs:ignore WordPress.DB.SlowDBQuery.slow_db_query_meta_query
                    'relation' => 'OR',
                    array(
                        'key'     => '_geo_exclude',
                        'compare' => 'NOT EXISTS',
                    ),
                    array(
                        'key'     => '_geo_exclude',
                        'value'   => '1',
                        'compare' => '!=',
                    ),
                ),
            )
        );

        foreach ( $posts as $post ) {
            $output .= "## {$post->post_title}\n\n";
            $output .= '**URL:** ' . get_permalink( $post ) . "\n";
            $output .= '**' . esc_html__( 'Date:', 'aidentity-generative-engine-optimizer' ) . '** ' . get_the_date( 'Y-m-d', $post ) . "\n\n";

            // Convert content to simplified markdown.
            $content = $this->html_to_markdown( $post->post_content );
            $output .= $content . "\n\n";
            $output .= "---\n\n";
        }

        return $output;
    }

    /**
     * Gets the description of a post.
     *
     * @since 1.0.0
     * @param WP_Post $post Post object.
     * @return string Post description.
     */
    private function get_post_description( $post ) {
        // Try the excerpt first.
        if ( ! empty( $post->post_excerpt ) ) {
            return wp_trim_words( $post->post_excerpt, 20, '...' );
        }

        // Try Yoast meta description.
        $yoast_desc = get_post_meta( $post->ID, '_yoast_wpseo_metadesc', true );
        if ( ! empty( $yoast_desc ) ) {
            return wp_trim_words( $yoast_desc, 20, '...' );
        }

        // Try RankMath meta description.
        $rank_desc = get_post_meta( $post->ID, 'rank_math_description', true );
        if ( ! empty( $rank_desc ) ) {
            return wp_trim_words( $rank_desc, 20, '...' );
        }

        // Fallback: first words of the content.
        $content = wp_strip_all_tags( $post->post_content );
        if ( ! empty( $content ) ) {
            return wp_trim_words( $content, 20, '...' );
        }

        return '';
    }

    /**
     * Converts HTML to Markdown.
     *
     * @since 1.0.0
     * @param string $html HTML content.
     * @return string Markdown content.
     */
    private function html_to_markdown( $html ) {
        // Remove shortcodes.
        $content = strip_shortcodes( $html );

        // Process Gutenberg blocks.
        $content = do_blocks( $content );

        // Convert headings.
        $content = preg_replace( '/<h1[^>]*>(.*?)<\/h1>/is', '# $1' . "\n\n", $content );
        $content = preg_replace( '/<h2[^>]*>(.*?)<\/h2>/is', '## $1' . "\n\n", $content );
        $content = preg_replace( '/<h3[^>]*>(.*?)<\/h3>/is', '### $1' . "\n\n", $content );
        $content = preg_replace( '/<h4[^>]*>(.*?)<\/h4>/is', '#### $1' . "\n\n", $content );

        // Convert paragraphs.
        $content = preg_replace( '/<p[^>]*>(.*?)<\/p>/is', '$1' . "\n\n", $content );

        // Convert lists.
        $content = preg_replace( '/<li[^>]*>(.*?)<\/li>/is', '- $1' . "\n", $content );
        $content = preg_replace( '/<\/?[ou]l[^>]*>/is', "\n", $content );

        // Convert links.
        $content = preg_replace( '/<a[^>]*href=["\']([^"\']*)["\'][^>]*>(.*?)<\/a>/is', '[$2]($1)', $content );

        // Convert bold/italic.
        $content = preg_replace( '/<strong[^>]*>(.*?)<\/strong>/is', '**$1**', $content );
        $content = preg_replace( '/<b[^>]*>(.*?)<\/b>/is', '**$1**', $content );
        $content = preg_replace( '/<em[^>]*>(.*?)<\/em>/is', '*$1*', $content );
        $content = preg_replace( '/<i[^>]*>(.*?)<\/i>/is', '*$1*', $content );

        // Strip any remaining HTML tags.
        $content = wp_strip_all_tags( $content );

        // Clean multiple whitespace.
        $content = preg_replace( '/\n{3,}/', "\n\n", $content );
        $content = trim( $content );

        return $content;
    }
}