<?php
/**
 * Generates and serves /llms.txt and /llms-full.txt.
 *
 * @package AEL42A
 */

namespace AEL42A;

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

/**
 * llms.txt + llms-full.txt.
 *
 * llms.txt follows the llms.txt spec: H1 site name, blockquote description,
 * a short paragraph, then H2 link sections whose entries point at the
 * markdown variants of the pages. llms-full.txt is a concatenation of the
 * markdown documents of the most important pages (byte-capped).
 */
class Llms_Txt {

	const QUERY_VAR       = 'ael42a_llms';
	const TTL             = 12 * HOUR_IN_SECONDS;
	const FULL_BYTE_CAP   = 2000000;
	const FULL_ITEM_LIMIT = 50;

	/**
	 * Telemetry counters.
	 *
	 * @var Telemetry
	 */
	private $telemetry;

	/**
	 * Converter (shared with the controller behaviourally, cheap to own).
	 *
	 * @var Markdown_Converter
	 */
	private $converter;

	/**
	 * Constructor.
	 *
	 * @param Telemetry $telemetry Telemetry counters.
	 */
	public function __construct( $telemetry ) {
		$this->telemetry = $telemetry;
		$this->converter = new Markdown_Converter();
	}

	/**
	 * Hook everything up.
	 */
	public function register_hooks() {
		add_action( 'init', array( __CLASS__, 'add_rewrite_rules' ) );
		add_filter( 'query_vars', array( $this, 'register_query_vars' ) );
		add_action( 'parse_request', array( $this, 'maybe_serve' ), 0 );
		add_action( 'wp_update_nav_menu', array( $this, 'invalidate' ) );
		add_action( 'ael42a_daily', array( $this, 'regenerate' ) );
	}

	/**
	 * Rewrite rules for both files. Static for activation-time flushing.
	 * Skipped per file when the site already provides it from another
	 * source, so a competing handler's rule wins the pattern.
	 */
	public static function add_rewrite_rules() {
		if ( ! Existing_Content::is_external( 'llms_txt' ) ) {
			add_rewrite_rule( '^llms\.txt$', 'index.php?' . self::QUERY_VAR . '=txt', 'top' );
		}
		if ( ! Existing_Content::is_external( 'llms_full_txt' ) ) {
			add_rewrite_rule( '^llms-full\.txt$', 'index.php?' . self::QUERY_VAR . '=full', 'top' );
		}
	}

	/**
	 * Register the query var.
	 *
	 * @param string[] $vars Public query vars.
	 * @return string[]
	 */
	public function register_query_vars( $vars ) {
		$vars[] = self::QUERY_VAR;
		return $vars;
	}

	/**
	 * Serve at parse_request time - covers both the rewrite-matched case and
	 * a raw REQUEST_URI fallback for sites with broken rewrites.
	 *
	 * @param \WP $wp The WP environment instance.
	 */
	public function maybe_serve( $wp ) {
		// Detection probes must see what the site would serve WITHOUT us.
		if ( Existing_Content::is_probe_request() ) {
			return;
		}

		$which = isset( $wp->query_vars[ self::QUERY_VAR ] ) ? $wp->query_vars[ self::QUERY_VAR ] : '';

		if ( '' === $which ) {
			$path = $this->request_path();
			if ( '/llms.txt' === $path ) {
				$which = 'txt';
			} elseif ( '/llms-full.txt' === $path ) {
				$which = 'full';
			}
		}

		if ( 'txt' !== $which && 'full' !== $which ) {
			return;
		}

		// The site already provides this file from another source (physical
		// file or another plugin) - never override it.
		if ( Existing_Content::is_external( 'txt' === $which ? 'llms_txt' : 'llms_full_txt' ) ) {
			return;
		}

		$body = ( 'txt' === $which ) ? $this->get_llms_txt() : $this->get_llms_full_txt();

		$this->telemetry->increment( 'llms_txt_hits' );
		$this->telemetry->record_bot_hit_from_request();

		status_header( 200 );
		header( 'Content-Type: text/plain; charset=utf-8' );
		header( 'X-42A-AEL: ' . AEL42A_VERSION );

		echo $body; // phpcs:ignore WordPress.Security.EscapeOutput -- plain text body, not HTML.
		exit;
	}

	/**
	 * Current request path (no query string), relative to the site path.
	 *
	 * @return string
	 */
	private function request_path() {
		$uri = isset( $_SERVER['REQUEST_URI'] ) ? wp_unslash( $_SERVER['REQUEST_URI'] ) : ''; // phpcs:ignore WordPress.Security.ValidatedSanitizedInput -- compared against fixed strings only.
		$path = (string) wp_parse_url( $uri, PHP_URL_PATH );

		// Subdirectory installs: strip the home path prefix.
		$home_path = rtrim( (string) wp_parse_url( home_url( '/' ), PHP_URL_PATH ), '/' );
		if ( '' !== $home_path && 0 === strpos( $path, $home_path . '/' ) ) {
			$path = substr( $path, strlen( $home_path ) );
		}

		return '/' . ltrim( $path, '/' );
	}

	/**
	 * Get llms.txt content (cached).
	 *
	 * @return string
	 */
	public function get_llms_txt() {
		$cached = get_transient( Cache::LLMS_TXT );
		if ( is_string( $cached ) && '' !== $cached ) {
			return $cached;
		}

		$body = $this->build_llms_txt();
		set_transient( Cache::LLMS_TXT, $body, self::TTL );

		return $body;
	}

	/**
	 * Get llms-full.txt content (cached).
	 *
	 * @return string
	 */
	public function get_llms_full_txt() {
		$cached = get_transient( Cache::LLMS_FULL );
		if ( is_string( $cached ) && '' !== $cached ) {
			return $cached;
		}

		$body = $this->build_llms_full_txt();
		set_transient( Cache::LLMS_FULL, $body, self::TTL );

		return $body;
	}

	/**
	 * Build llms.txt.
	 *
	 * @return string
	 */
	public function build_llms_txt() {
		$site_name = get_bloginfo( 'name' );
		$tagline   = get_bloginfo( 'description' );

		$sections = array();

		// Main pages: front page + primary nav (fallback: published pages).
		$main_links = array();
		$front      = $this->front_page_link();
		if ( $front ) {
			$main_links[] = $front;
		}
		foreach ( $this->nav_pages( 20 ) as $post ) {
			if ( $front && (int) $front['id'] === (int) $post->ID ) {
				continue;
			}
			$main_links[] = $this->post_link( $post );
		}
		if ( ! empty( $main_links ) ) {
			$sections[] = array(
				'title' => __( 'Main Pages', 'ael-42a' ),
				'links' => $main_links,
			);
		}

		// Recent posts.
		$recent_links = array();
		foreach ( $this->recent_posts( 'post', 10 ) as $post ) {
			$recent_links[] = $this->post_link( $post );
		}
		if ( ! empty( $recent_links ) ) {
			$sections[] = array(
				'title' => __( 'Recent Posts', 'ael-42a' ),
				'links' => $recent_links,
			);
		}

		// WooCommerce product categories.
		if ( class_exists( 'WooCommerce' ) && taxonomy_exists( 'product_cat' ) ) {
			$product_links = array();
			$terms         = get_terms(
				array(
					'taxonomy'   => 'product_cat',
					'hide_empty' => true,
					'number'     => 15,
				)
			);
			if ( is_array( $terms ) ) {
				foreach ( $terms as $term ) {
					$term_url = get_term_link( $term );
					if ( is_wp_error( $term_url ) ) {
						continue;
					}
					$product_links[] = array(
						'id'          => 0,
						'title'       => $term->name,
						'url'         => $term_url,
						'description' => $this->trim_description( $term->description ? $term->description : sprintf( /* translators: %d: number of products. */ __( '%d products', 'ael-42a' ), (int) $term->count ) ),
					);
				}
			}
			if ( ! empty( $product_links ) ) {
				$sections[] = array(
					'title' => __( 'Products', 'ael-42a' ),
					'links' => $product_links,
				);
			}
		}

		// Optional resources.
		$sections[] = array(
			'title' => __( 'Optional', 'ael-42a' ),
			'links' => array(
				array(
					'id'          => 0,
					'title'       => 'llms-full.txt',
					'url'         => home_url( '/llms-full.txt' ),
					'description' => __( 'Full markdown content of the main pages of this site in one file', 'ael-42a' ),
				),
				array(
					'id'          => 0,
					'title'       => __( 'MCP server', 'ael-42a' ),
					'url'         => rest_url( '42a/v1/mcp' ),
					'description' => __( 'Model Context Protocol endpoint for structured access to this site', 'ael-42a' ),
				),
				array(
					'id'          => 0,
					'title'       => __( 'MCP discovery', 'ael-42a' ),
					'url'         => home_url( '/.well-known/mcp.json' ),
					'description' => __( 'MCP server discovery document', 'ael-42a' ),
				),
			),
		);

		/**
		 * Filter the llms.txt sections.
		 *
		 * @param array $sections Array of { title, links: [ { id, title, url, description } ] }.
		 */
		$sections = apply_filters( 'ael42a_llms_txt_sections', $sections );

		$lines   = array();
		$lines[] = '# ' . $site_name;
		$lines[] = '';
		if ( '' !== trim( (string) $tagline ) ) {
			$lines[] = '> ' . $this->one_line( $tagline );
			$lines[] = '';
		}
		$lines[] = sprintf(
			/* translators: 1: site name. */
			__( 'This file lists the key pages of %1$s for AI agents and LLMs. Every page on this site is also available as markdown: append ?format=markdown to any URL. A machine-readable MCP server is available for structured access.', 'ael-42a' ),
			$site_name
		);

		foreach ( $sections as $section ) {
			if ( empty( $section['links'] ) ) {
				continue;
			}
			$lines[] = '';
			$lines[] = '## ' . $this->one_line( $section['title'] );
			$lines[] = '';
			foreach ( $section['links'] as $link ) {
				$entry = '- [' . $this->one_line( $link['title'] ) . '](' . $link['url'] . ')';
				if ( ! empty( $link['description'] ) ) {
					$entry .= ': ' . $this->one_line( $link['description'] );
				}
				$lines[] = $entry;
			}
		}

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

	/**
	 * Build llms-full.txt: header + concatenated markdown documents.
	 *
	 * @return string
	 */
	public function build_llms_full_txt() {
		$site_name = get_bloginfo( 'name' );
		$tagline   = get_bloginfo( 'description' );
		$limit     = (int) get_option( 'ael42a_full_limit', self::FULL_ITEM_LIMIT );
		if ( $limit < 1 || $limit > self::FULL_ITEM_LIMIT ) {
			$limit = self::FULL_ITEM_LIMIT;
		}

		$header   = array();
		$header[] = '# ' . $site_name . ' - llms-full.txt';
		$header[] = '';
		if ( '' !== trim( (string) $tagline ) ) {
			$header[] = '> ' . $this->one_line( $tagline );
			$header[] = '';
		}
		$header[] = sprintf(
			/* translators: 1: generator id, 2: date. */
			__( 'Concatenated markdown of the main pages of this site. Generated by %1$s on %2$s. Individual pages: append ?format=markdown to any URL.', 'ael-42a' ),
			'ael-42a/' . AEL42A_VERSION,
			gmdate( 'Y-m-d\TH:i:s\Z' )
		);

		$out   = implode( "\n", $header );
		$bytes = strlen( $out );

		$seen  = array();
		$posts = array();

		$front_id = (int) get_option( 'page_on_front' );
		if ( $front_id > 0 ) {
			$front = get_post( $front_id );
			if ( $front instanceof \WP_Post && 'publish' === $front->post_status ) {
				$posts[] = $front;
			}
		}
		foreach ( $this->nav_pages( $limit ) as $page ) {
			$posts[] = $page;
		}
		foreach ( $this->recent_posts( 'post', $limit ) as $post ) {
			$posts[] = $post;
		}

		$count = 0;
		foreach ( $posts as $post ) {
			if ( $count >= $limit ) {
				break;
			}
			if ( isset( $seen[ $post->ID ] ) || post_password_required( $post ) ) {
				continue;
			}
			$seen[ $post->ID ] = true;

			$doc   = $this->converter->document( $post );
			$piece = "\n\n---\n\n" . rtrim( $doc );

			if ( $bytes + strlen( $piece ) > self::FULL_BYTE_CAP ) {
				break;
			}

			$out   .= $piece;
			$bytes += strlen( $piece );
			$count++;
		}

		return $out . "\n";
	}

	/**
	 * Invalidate both cached files (save_post is covered by Cache hooks).
	 */
	public function invalidate() {
		delete_transient( Cache::LLMS_TXT );
		delete_transient( Cache::LLMS_FULL );
	}

	/**
	 * Daily cron: rebuild both transients so cold hits stay fast.
	 */
	public function regenerate() {
		$this->invalidate();
		$this->get_llms_txt();
		$this->get_llms_full_txt();
	}

	/**
	 * Front page link row, if a static front page is configured.
	 *
	 * @return array|null
	 */
	private function front_page_link() {
		$front_id = (int) get_option( 'page_on_front' );
		if ( $front_id > 0 ) {
			$front = get_post( $front_id );
			if ( $front instanceof \WP_Post && 'publish' === $front->post_status ) {
				return $this->post_link( $front );
			}
			return null;
		}
		return array(
			'id'          => 0,
			'title'       => __( 'Home', 'ael-42a' ),
			'url'         => home_url( '/' ),
			'description' => $this->trim_description( get_bloginfo( 'description' ) ),
		);
	}

	/**
	 * Pages from the primary nav menu; fallback to all published pages.
	 *
	 * @param int $limit Max pages.
	 * @return \WP_Post[]
	 */
	private function nav_pages( $limit ) {
		$pages     = array();
		$locations = get_nav_menu_locations();
		$menu_id   = 0;

		foreach ( array( 'primary', 'main', 'header', 'menu-1', 'top' ) as $slot ) {
			if ( ! empty( $locations[ $slot ] ) ) {
				$menu_id = (int) $locations[ $slot ];
				break;
			}
		}
		if ( 0 === $menu_id && ! empty( $locations ) ) {
			$menu_id = (int) reset( $locations );
		}

		if ( $menu_id > 0 ) {
			$items = wp_get_nav_menu_items( $menu_id );
			if ( is_array( $items ) ) {
				foreach ( $items as $item ) {
					if ( count( $pages ) >= $limit ) {
						break;
					}
					if ( 'post_type' !== $item->type ) {
						continue;
					}
					$post = get_post( (int) $item->object_id );
					if ( $post instanceof \WP_Post && 'publish' === $post->post_status && is_post_publicly_viewable( $post ) ) {
						$pages[ $post->ID ] = $post;
					}
				}
			}
		}

		if ( empty( $pages ) ) {
			$all = get_pages(
				array(
					'sort_column' => 'menu_order,post_title',
					'number'      => $limit,
				)
			);
			if ( is_array( $all ) ) {
				foreach ( $all as $page ) {
					$pages[ $page->ID ] = $page;
				}
			}
		}

		return array_values( $pages );
	}

	/**
	 * Recent published posts of a type.
	 *
	 * @param string $post_type Post type.
	 * @param int    $limit     Max posts.
	 * @return \WP_Post[]
	 */
	private function recent_posts( $post_type, $limit ) {
		$query = new \WP_Query(
			array(
				'post_type'           => $post_type,
				'post_status'         => 'publish',
				'posts_per_page'      => $limit,
				'orderby'             => 'date',
				'order'               => 'DESC',
				'no_found_rows'       => true,
				'ignore_sticky_posts' => true,
			)
		);
		return $query->posts;
	}

	/**
	 * Link row for a post: markdown URL + short description.
	 *
	 * @param \WP_Post $post Post.
	 * @return array
	 */
	private function post_link( $post ) {
		return array(
			'id'          => $post->ID,
			'title'       => get_the_title( $post ),
			'url'         => Markdown_Controller::markdown_url( $post ),
			'description' => $this->post_description( $post ),
		);
	}

	/**
	 * Short plain-text description of a post.
	 *
	 * @param \WP_Post $post Post.
	 * @return string
	 */
	private function post_description( $post ) {
		$text = $post->post_excerpt;
		if ( '' === trim( (string) $text ) ) {
			$text = $post->post_content;
		}
		return $this->trim_description( $text );
	}

	/**
	 * Strip tags/shortcodes and trim to ~25 words.
	 *
	 * @param string $text Raw text.
	 * @return string
	 */
	private function trim_description( $text ) {
		$text = wp_strip_all_tags( strip_shortcodes( (string) $text ) );
		$text = preg_replace( '/\s+/u', ' ', $text );
		return wp_trim_words( trim( (string) $text ), 25, '...' );
	}

	/**
	 * Force a string onto one line.
	 *
	 * @param string $text Text.
	 * @return string
	 */
	private function one_line( $text ) {
		return trim( preg_replace( '/\s+/u', ' ', (string) $text ) );
	}
}