<?php
if ( ! defined( 'ABSPATH' ) ) exit;

/**
 * Generates and serves /llms.txt
 */
class Semantic_LLMS_Txt {

    public static function init() {
        add_filter( 'query_vars',    [ __CLASS__, 'query_vars' ] );
        add_action( 'init',          [ __CLASS__, 'rewrite_rule' ] );
        add_action( 'template_redirect', [ __CLASS__, 'serve' ] );
    }

    public static function rewrite_rule() {
        add_rewrite_rule( '^llms\.txt$', 'index.php?semantic_llms_txt=1', 'top' );
    }

    public static function query_vars( $vars ) {
        $vars[] = 'semantic_llms_txt';
        return $vars;
    }

    public static function serve() {
        if ( ! get_query_var( 'semantic_llms_txt' ) ) return;
        if ( ! Semantic_Settings::llms_txt_enabled() ) {
            status_header( 404 );
            exit;
        }
        // Play nice: if another plugin/file owns /llms.txt and the user hasn't chosen
        // to override, step aside and let theirs serve.
        if ( Semantic_Compat::llms_txt_owned_elsewhere() && ! Semantic_Settings::llms_txt_override() ) {
            return;
        }

        header( 'Content-Type: text/plain; charset=utf-8' );
        header( 'X-Robots-Tag: noindex' );
        header( 'X-Content-Type-Options: nosniff' );
        // phpcs:ignore WordPress.Security.EscapeOutput.OutputNotEscaped -- Plain-text body (Content-Type: text/plain sent above); HTML escaping would corrupt the output and is not applicable.
        echo self::generate();
        exit;
    }

    public static function generate() {
        $home    = home_url();
        $name    = Semantic_Settings::org_name();
        $desc    = Semantic_Settings::org_description();

        $lines = [
            "# {$name}",
            '',
            "> {$desc}",
            '',
            '## AI Data Endpoints',
            '',
            '### Site-Wide',
            '',
            "- Organization Schema: {$home}/wp-json/ai/v1/organization",
            "- Entity Registry: {$home}/wp-json/ai/v1/entities/registry",
            "- Site Graph: {$home}/wp-json/ai/v1/site-graph",
            '',
            '### Per-Page (replace {id} with post ID or slug)',
            '',
            "- Context: {$home}/wp-json/ai/v1/context/{id}",
            "- Entities: {$home}/wp-json/ai/v1/entities/{id}",
            "- Schema: {$home}/wp-json/ai/v1/schema/{id}",
            "- Bundle: {$home}/wp-json/ai/v1/bundle/{id}",
            '',
        ];

        if ( Semantic_Settings::mcp_enabled() ) {
            $lines[] = '### MCP Server';
            $lines[] = '';
            $lines[] = "- Model Context Protocol endpoint: {$home}/wp-json/ai/v1/mcp";
            $lines[] = '  (Streamable HTTP, stateless — add as a custom connector in Claude / ChatGPT)';
            $lines[] = '';
        }

        $markdown_on = Semantic_Settings::markdown_enabled();
        if ( $markdown_on ) {
            $lines[] = '### Markdown';
            $lines[] = '';
            $lines[] = "- Full content (Markdown): {$home}/llms-full.txt";
            $lines[] = '- Per-page Markdown: append `.md` to any content URL below';
            $lines[] = '';
        }

        $lines[] = '## Content';
        $lines[] = '';

        // List published, public pages/posts (excluding password-protected and any
        // the user has marked out of the data layer) — same gate as every other
        // public surface.
        $types = Semantic_Settings::public_post_types();
        $posts = empty( $types ) ? [] : get_posts( [
            'post_type'      => $types,
            'post_status'    => 'publish',
            'has_password'   => false,
            'posts_per_page' => 500,
            'orderby'        => 'menu_order title',
            'order'          => 'ASC',
            'meta_query'     => [ [ 'key' => Semantic_Settings::EXCLUDE_META, 'compare' => 'NOT EXISTS' ] ],
        ] );

        foreach ( $posts as $post ) {
            $permalink = get_permalink( $post );
            if ( $markdown_on ) {
                $md = Semantic_Markdown::md_url( $post );
                $lines[] = $md
                    ? sprintf( '- [%s](%s) ([md](%s))', $post->post_title, $permalink, $md )
                    : sprintf( '- [%s](%s)', $post->post_title, $permalink );
            } else {
                $lines[] = sprintf( '- [%s](%s)', $post->post_title, $permalink );
            }
        }

        // Canonical entity reference pages — top entities by occurrence, restricted
        // to entities that appear in currently published, public content (matching
        // what the entity pages actually serve; non-public entities would 404).
        global $wpdb;
        $public_ids = Semantic_Entity_Registry::public_registry_ids();
        $entities   = [];
        if ( ! empty( $public_ids ) ) {
            $in       = implode( ',', array_map( 'intval', $public_ids ) );
            $entities = $wpdb->get_results(
                "SELECT entity_id, label FROM {$wpdb->prefix}semantic_entity_registry
                 WHERE id IN ($in) ORDER BY occurrence_count DESC LIMIT 100",
                ARRAY_A
            );
        }
        if ( ! empty( $entities ) ) {
            $lines[] = '';
            $lines[] = '## Entities';
            $lines[] = '';
            foreach ( $entities as $e ) {
                $lines[] = sprintf( '- [%s](%s)', $e['label'], $home . '/entities/' . $e['entity_id'] );
            }
        }

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