<?php
/**
 * Generates /llms.txt and /llms-full.txt for LLM content discovery.
 *
 * @see https://llmstxt.org/
 */

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

class AIREWP_LLMS_Txt {

    /**
     * Both documents, built once and kept for the dynamic answer.
     *
     * Building them costs two WP_Query runs over up to a hundred posts each,
     * which is why the URL used to answer 404 without a file rather than
     * build on every hit. The cache removes that cost: one build per rebuild
     * of the files, or per twelve hours, whichever comes first.
     */
    private const CACHE     = 'airewp_llms_cache';
    private const CACHE_TTL = 12 * HOUR_IN_SECONDS;

    public function __construct() {
        // Generation triggered by main plugin
    }

    /**
     * Build both documents from the current content.
     *
     * @return array{llms.txt: string, llms-full.txt: string}
     */
    public static function build( array $settings ): array {
        $pages = self::get_public_pages();
        $posts = self::get_public_posts();

        return [
            'llms.txt'      => self::build_compact( $settings, $pages, $posts ),
            'llms-full.txt' => self::build_full( $settings, $pages, $posts ),
        ];
    }

    /**
     * Generate both llms.txt and llms-full.txt.
     *
     * The cache is filled before the files are written, so the dynamic answer
     * is never older than the file next to it.
     *
     * @return bool True when both files were written. The caller needs to know:
     *              a rebuild that silently failed leaves the previous version's
     *              file, including any password-protected content it carried,
     *              in place.
     */
    public static function generate( array $settings ): bool {
        $built = self::build( $settings );
        self::remember( $built );

        $compact_ok = AI_Ready_WP::write_file( AIREWP_ROOT_PATH . 'llms.txt', $built['llms.txt'] );
        $full_ok    = AI_Ready_WP::write_file( AIREWP_ROOT_PATH . 'llms-full.txt', $built['llms-full.txt'] );

        // Both writes always run; no short-circuit.
        return $compact_ok && $full_ok;
    }

    /**
     * The document as served without a file: from the cache, or built now.
     *
     * A cache written by another version of the plugin is not trusted; the
     * format may have changed with it.
     *
     * @param string $which 'llms.txt' or 'llms-full.txt'.
     */
    public static function cached( string $which, array $settings ): string {
        $cache = get_transient( self::CACHE );
        if ( is_array( $cache ) && isset( $cache[ $which ] ) && AIREWP_VERSION === ( $cache['version'] ?? '' ) ) {
            return (string) $cache[ $which ];
        }

        $built = self::build( $settings );
        self::remember( $built );

        return $built[ $which ] ?? '';
    }

    /**
     * Drop the cached documents, for when the feature is switched off.
     */
    public static function forget(): void {
        delete_transient( self::CACHE );
    }

    private static function remember( array $built ): void {
        set_transient(
            self::CACHE,
            $built + [ 'version' => AIREWP_VERSION, 'built_at' => time() ],
            self::CACHE_TTL
        );
    }

    /**
     * Build compact llms.txt.
     */
    private static function build_compact( array $settings, array $pages, array $posts ): string {
        $lines = [];

        // Header — fall back to site name/tagline, skip the blockquote when
        // there is no description at all (a bare "> " line is invalid llms.txt).
        $name = $settings['company_name'] ?: get_bloginfo( 'name' );
        $desc = $settings['company_desc'] ?: get_bloginfo( 'description' );

        $lines[] = '# ' . $name;
        $lines[] = '';
        if ( '' !== trim( (string) $desc ) ) {
            $lines[] = '> ' . $desc;
            $lines[] = '';
        }

        // Services
        if ( ! empty( $settings['services'] ) ) {
            $lines[] = '## Services';
            $lines[] = '';
            $services = array_map( 'trim', explode( ',', $settings['services'] ) );
            foreach ( $services as $service ) {
                $lines[] = '- ' . $service;
            }
            $lines[] = '';
        }

        // Pages
        if ( ! empty( $pages ) ) {
            $lines[] = '## Pages';
            $lines[] = '';
            foreach ( $pages as $page ) {
                $lines[] = sprintf( '- [%s](%s)', $page['title'], $page['url'] );
            }
            $lines[] = '';
        }

        // Posts (latest 20)
        if ( ! empty( $posts ) ) {
            $lines[] = '## Articles';
            $lines[] = '';
            $recent = array_slice( $posts, 0, 20 );
            foreach ( $recent as $post ) {
                $lines[] = sprintf( '- [%s](%s)', $post['title'], $post['url'] );
            }
            if ( count( $posts ) > 20 ) {
                $lines[] = sprintf( '- ... and %d more articles', count( $posts ) - 20 );
            }
            $lines[] = '';
        }

        // Contact
        $lines[] = '## Contact';
        $lines[] = '';
        if ( ! empty( $settings['company_email'] ) ) {
            $lines[] = '- Email: ' . $settings['company_email'];
        }
        if ( ! empty( $settings['company_phone'] ) ) {
            $lines[] = '- Phone: ' . $settings['company_phone'];
        }
        $lines[] = '- Website: ' . home_url();
        $lines[] = '';

        // AI-discovery pointers — only advertise endpoints that are actually
        // enabled, otherwise llms.txt links to 404s on default installs.
        $lines[] = '## AI Discovery';
        $lines[] = '';
        if ( ! empty( $settings['enable_agents_json'] ) ) {
            $lines[] = sprintf( '- [agents.json](%s) — AI Agent Tool Discovery', home_url( '/.well-known/agents.json' ) );
        }
        if ( ! empty( $settings['enable_agent_card'] ) ) {
            $lines[] = sprintf( '- [agent-card.json](%s) — A2A Protocol v0.3', home_url( '/.well-known/agent-card.json' ) );
            $lines[] = sprintf( '- [agent.json](%s) — A2A v0.3 (canonical)', home_url( '/.well-known/agent.json' ) );
        }
        if ( ! empty( $settings['enable_mcp_json'] ) ) {
            $lines[] = sprintf( '- [mcp.json](%s) — MCP Server Discovery', home_url( '/.well-known/mcp.json' ) );
        }
        if ( ! empty( $settings['enable_rest_api'] ) ) {
            $lines[] = sprintf( '- [REST API](%s) — AI agent endpoints', home_url( '/wp-json/airewp/v1/' ) );
        }
        $lines[] = sprintf( '- [llms-full.txt](%s) — Full content version', home_url( '/llms-full.txt' ) );
        $lines[] = '';

        // Footer
        $lines[] = '---';
        $lines[] = sprintf( 'Generated by AI-Ready WP %s | %s', AIREWP_VERSION, gmdate( 'Y-m-d' ) );

        return implode( "\n", $lines );
    }

    /**
     * Build full llms-full.txt with content.
     */
    private static function build_full( array $settings, array $pages, array $posts ): string {
        $lines = [];

        // Header — same fallback rules as build_compact().
        $name = $settings['company_name'] ?: get_bloginfo( 'name' );
        $desc = $settings['company_desc'] ?: get_bloginfo( 'description' );

        $lines[] = '# ' . $name;
        $lines[] = '';
        if ( '' !== trim( (string) $desc ) ) {
            $lines[] = '> ' . $desc;
            $lines[] = '';
        }

        // Services with descriptions
        if ( ! empty( $settings['services'] ) ) {
            $lines[] = '## Services';
            $lines[] = '';
            $services = array_map( 'trim', explode( ',', $settings['services'] ) );
            foreach ( $services as $service ) {
                $lines[] = '### ' . $service;
                $lines[] = '';
            }
        }

        // Pages with full content
        if ( ! empty( $pages ) ) {
            $lines[] = '## Pages';
            $lines[] = '';
            foreach ( $pages as $page ) {
                $lines[] = '### ' . $page['title'];
                $lines[] = '';
                $lines[] = 'URL: ' . $page['url'];
                $lines[] = '';
                if ( ! empty( $page['excerpt'] ) ) {
                    $lines[] = $page['excerpt'];
                    $lines[] = '';
                }
                if ( ! empty( $page['content'] ) ) {
                    // to_plain_text(), not wp_strip_all_tags(): the latter
                    // leaves shortcodes standing, so every page built with a
                    // page builder used to land here as its raw shortcode tree.
                    $content = AI_Ready_WP::to_plain_text( $page['content'] );
                    // Limit content to prevent massive files
                    if ( strlen( $content ) > 2000 ) {
                        $content = substr( $content, 0, 2000 ) . '...';
                    }
                    $lines[] = $content;
                    $lines[] = '';
                }
            }
        }

        // Posts with excerpts
        if ( ! empty( $posts ) ) {
            $lines[] = '## Articles';
            $lines[] = '';
            foreach ( $posts as $post ) {
                $lines[] = '### ' . $post['title'];
                $lines[] = '';
                $lines[] = 'URL: ' . $post['url'];
                if ( ! empty( $post['date'] ) ) {
                    $lines[] = 'Date: ' . $post['date'];
                }
                $lines[] = '';
                if ( ! empty( $post['excerpt'] ) ) {
                    $lines[] = $post['excerpt'];
                    $lines[] = '';
                }
            }
        }

        // Contact
        $lines[] = '## Contact';
        $lines[] = '';
        if ( ! empty( $settings['company_name'] ) ) {
            $lines[] = '- Company: ' . $settings['company_name'];
        }
        if ( ! empty( $settings['company_email'] ) ) {
            $lines[] = '- Email: ' . $settings['company_email'];
        }
        if ( ! empty( $settings['company_phone'] ) ) {
            $lines[] = '- Phone: ' . $settings['company_phone'];
        }
        if ( ! empty( $settings['company_address'] ) ) {
            $addr = $settings['company_address'];
            if ( ! empty( $settings['company_zip'] ) ) {
                $addr .= ', ' . $settings['company_zip'];
            }
            if ( ! empty( $settings['company_city'] ) ) {
                $addr .= ' ' . $settings['company_city'];
            }
            if ( ! empty( $settings['company_country'] ) ) {
                $addr .= ', ' . $settings['company_country'];
            }
            $lines[] = '- Address: ' . $addr;
        }
        $lines[] = '- Website: ' . home_url();
        $lines[] = '';

        // Footer
        $lines[] = '---';
        $lines[] = sprintf( 'Generated by AI-Ready WP %s | %s', AIREWP_VERSION, gmdate( 'Y-m-d' ) );

        return implode( "\n", $lines );
    }

    /**
     * Get all public pages.
     *
     * 'has_password' => false is what keeps locked pages out of llms-full.txt.
     * That file used to carry $post->post_content of password-protected pages
     * verbatim (2000 characters of it), and llms-full.txt is a static file that
     * anyone can fetch. The check cannot be post_password_required() here:
     * generation runs for whoever saved a post or for WP-Cron, so it would ask
     * "does this session know the password" — the wrong question for a file
     * that is served to the whole world.
     */
    private static function get_public_pages(): array {
        $query = new WP_Query( [
            'post_type'      => 'page',
            'post_status'    => 'publish',
            'has_password'   => false,
            'posts_per_page' => 100,
            'orderby'        => 'menu_order',
            'order'          => 'ASC',
        ] );

        $pages = [];
        foreach ( $query->posts as $post ) {
            if ( ! self::is_publishable( $post ) ) {
                continue;
            }
            $pages[] = [
                'title'   => $post->post_title,
                'url'     => get_permalink( $post ),
                // The excerpt needs the same cleaning as the content, and for a
                // post it is the ONLY text that reaches llms-full.txt. Core's
                // wp_trim_excerpt() calls strip_shortcodes(), which does nothing
                // when the page builder is not loaded, so the raw tree came
                // through here even after the content path was fixed.
                'excerpt' => AI_Ready_WP::to_plain_text( get_the_excerpt( $post ) ),
                'content' => $post->post_content,
            ];
        }
        wp_reset_postdata();

        return $pages;
    }

    /**
     * Get all public posts.
     */
    private static function get_public_posts(): array {
        $query = new WP_Query( [
            'post_type'      => 'post',
            'post_status'    => 'publish',
            'has_password'   => false,
            'posts_per_page' => 100,
            'orderby'        => 'date',
            'order'          => 'DESC',
        ] );

        $posts = [];
        foreach ( $query->posts as $post ) {
            if ( ! self::is_publishable( $post ) ) {
                continue;
            }
            $posts[] = [
                'title'   => $post->post_title,
                'url'     => get_permalink( $post ),
                'date'    => get_the_date( 'Y-m-d', $post ),
                'excerpt' => AI_Ready_WP::to_plain_text( get_the_excerpt( $post ) ),
            ];
        }
        wp_reset_postdata();

        return $posts;
    }

    /**
     * Second gate, right where the content is read.
     *
     * The query argument above is the real filter; this one guards the case
     * where another plugin rewrites the query through pre_get_posts or
     * posts_where and the condition silently disappears. Nothing that carries
     * a password reaches the generated files.
     */
    private static function is_publishable( $post ): bool {
        return $post instanceof WP_Post
            && 'publish' === $post->post_status
            && '' === (string) $post->post_password;
    }
}