<?php
/**
 * llms.txt generation and endpoint handling.
 *
 * @package AgizoAI_Agentic
 */

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

/**
 * llms.txt generator.
 */
class AgizoAI_Agentic_LLMS {

	/**
	 * Settings.
	 *
	 * @var AgizoAI_Agentic_Settings
	 */
	private $settings;

	/**
	 * SEO integrations helper.
	 *
	 * @var AgizoAI_Agentic_Integrations|null
	 */
	private $integrations;

	/**
	 * Constructor.
	 *
	 * @param AgizoAI_Agentic_Settings          $settings Settings helper.
	 * @param AgizoAI_Agentic_Integrations|null $integrations SEO integrations helper.
	 */
	public function __construct( AgizoAI_Agentic_Settings $settings, AgizoAI_Agentic_Integrations $integrations = null ) {
		$this->settings     = $settings;
		$this->integrations = $integrations;
	}

	/**
	 * Register pretty endpoints.
	 *
	 * @return void
	 */
	public function register_rewrite_rules() {
		add_rewrite_rule( '^llms\.txt$', 'index.php?agizoai_agentic_llms=1', 'top' );
		add_rewrite_rule( '^llms-full\.txt$', 'index.php?agizoai_agentic_llms_full=1', 'top' );
	}

	/**
	 * Register query vars.
	 *
	 * @param array<int,string> $vars Query vars.
	 * @return array<int,string>
	 */
	public function register_query_vars( $vars ) {
		$vars[] = 'agizoai_agentic_llms';
		$vars[] = 'agizoai_agentic_llms_full';

		return $vars;
	}

	/**
	 * Append llms.txt reference to robots.txt when enabled.
	 *
	 * @param string $output robots.txt output.
	 * @param bool   $public Whether site discourages search engines.
	 * @return string
	 */
	public function append_robots_txt( $output, $public ) {
		unset( $public );

		if ( ! $this->settings->get( 'enabled' ) || ! $this->settings->get( 'add_robots_reference' ) ) {
			return $output;
		}

		$output .= "\n# AgizoAI Agentic Browsing Optimizer\n";
		$output .= 'LLMs: ' . home_url( '/llms.txt' ) . "\n";

		if ( $this->settings->get( 'full_enabled' ) ) {
			$output .= 'LLMs-Full: ' . home_url( '/llms-full.txt' ) . "\n";
		}

		return $output;
	}

	/**
	 * Render endpoint if requested.
	 *
	 * @return void
	 */
	public function maybe_render_endpoint() {
		$is_llms      = (bool) get_query_var( 'agizoai_agentic_llms' );
		$is_llms_full = (bool) get_query_var( 'agizoai_agentic_llms_full' );

		if ( ! $is_llms && ! $is_llms_full ) {
			return;
		}

		if ( ! $this->settings->get( 'enabled' ) || ( $is_llms_full && ! $this->settings->get( 'full_enabled' ) ) ) {
			status_header( 404 );
			nocache_headers();
			header( 'Content-Type: text/plain; charset=utf-8' );
			echo esc_html__( 'Not enabled.', 'agizoai-agentic-browsing-optimizer' );
			exit;
		}

		$content = $is_llms_full ? $this->generate_full() : $this->generate();

		status_header( 200 );
		nocache_headers();
		header( 'Content-Type: text/plain; charset=utf-8' );
		header( 'X-AgizoAI-Agentic: ' . AGIZOAI_AGENTIC_VERSION );
		echo $content; // phpcs:ignore WordPress.Security.EscapeOutput.OutputNotEscaped -- Plain text endpoint generated and escaped by generator.
		exit;
	}

	/**
	 * Clear cached generated content.
	 *
	 * @return void
	 */
	public function clear_cache() {
		delete_transient( 'agizoai_agentic_llms_txt' );
		delete_transient( 'agizoai_agentic_llms_full_txt' );
		$this->settings->update( 'last_generated', current_time( 'mysql' ) );
	}

	/**
	 * Generate standard llms.txt.
	 *
	 * @param bool $force Force no cache.
	 * @return string
	 */
	public function generate( $force = false ) {
		if ( ! $force ) {
			$cached = get_transient( 'agizoai_agentic_llms_txt' );
			if ( is_string( $cached ) && '' !== $cached ) {
				return $cached;
			}
		}

		$options = $this->settings->get_options();
		$mode    = $options['generation_mode'];

		if ( 'manual' === $mode ) {
			$content = $this->normalise_text( $options['manual_llms'] );
			if ( '' === trim( $content ) ) {
				$content = $this->generate_auto( false );
			}
		} elseif ( 'hybrid' === $mode ) {
			$content = $this->normalise_text( $options['manual_llms'] );
			if ( '' !== trim( $content ) ) {
				$content .= "\n\n---\n\n";
			}
			$content .= $this->generate_auto( false );
		} else {
			$content = $this->generate_auto( false );
		}

		$content = $this->normalise_text( $content );
		set_transient( 'agizoai_agentic_llms_txt', $content, DAY_IN_SECONDS );

		return $content;
	}

	/**
	 * Generate llms-full.txt.
	 *
	 * @param bool $force Force no cache.
	 * @return string
	 */
	public function generate_full( $force = false ) {
		if ( ! $force ) {
			$cached = get_transient( 'agizoai_agentic_llms_full_txt' );
			if ( is_string( $cached ) && '' !== $cached ) {
				return $cached;
			}
		}

		$content = $this->generate_auto( true );
		$content = $this->normalise_text( $content );
		set_transient( 'agizoai_agentic_llms_full_txt', $content, DAY_IN_SECONDS );

		return $content;
	}

	/**
	 * Build automatic llms text.
	 *
	 * @param bool $full Whether full mode.
	 * @return string
	 */
	private function generate_auto( $full = false ) {
		$options      = $this->settings->get_options();
		$site_name    = get_bloginfo( 'name' );
		$description  = get_bloginfo( 'description' );
		$language     = get_bloginfo( 'language' );
		$home_url     = home_url( '/' );
		$max_items    = $full ? absint( $options['full_max_items'] ) : absint( $options['max_items'] );
		$site_purpose = trim( (string) $options['site_purpose'] );
		$site_type    = isset( $options['site_type'] ) ? sanitize_key( $options['site_type'] ) : 'custom';
		$custom_intro = trim( (string) $options['custom_intro'] );
		$date         = current_time( 'Y-m-d' );

		$lines   = array();
		$lines[] = '# ' . $this->escape_markdown_line( $site_name );

		if ( '' !== $description ) {
			$lines[] = '> ' . $this->escape_markdown_line( $description );
		}

		$lines[] = '';
		$lines[] = '## Site information';
		$lines[] = '- Site URL: ' . esc_url_raw( $home_url );
		$lines[] = '- Language: ' . $this->escape_markdown_line( $language );
		$lines[] = '- Generated: ' . $this->escape_markdown_line( $date );
		$lines[] = '- Generated by: AgizoAI Agentic Browsing Optimizer v' . AGIZOAI_AGENTIC_VERSION;

		$lines[] = '- Site type: ' . $this->escape_markdown_line( $site_type );

		if ( '' !== $site_purpose ) {
			$lines[] = '- Site purpose: ' . $this->escape_markdown_line( $site_purpose );
		}

		if ( '' !== $custom_intro ) {
			$lines[] = '';
			$lines[] = '## Guidance for AI agents';
			$lines[] = $this->paragraph( $custom_intro );
		}

		$lines[] = '';
		$lines[] = '## Key URLs';
		$lines[] = '- [Home](' . esc_url_raw( $home_url ) . ')';

		$page_for_posts = (int) get_option( 'page_for_posts' );
		if ( $page_for_posts > 0 ) {
			$blog_url = get_permalink( $page_for_posts );
			if ( $blog_url ) {
				$lines[] = '- [Blog](' . esc_url_raw( $blog_url ) . ')';
			}
		}

		$privacy_url = get_privacy_policy_url();
		if ( $privacy_url ) {
			$lines[] = '- [Privacy Policy](' . esc_url_raw( $privacy_url ) . ')';
		}

		$sitemap_url = $this->get_sitemap_url();
		if ( $sitemap_url ) {
			$lines[] = '- [XML Sitemap](' . esc_url_raw( $sitemap_url ) . ')';
		}

		$terms = $this->get_terms_for_llms( $options['include_taxonomies'], $full ? 200 : 30 );
		if ( ! empty( $terms ) ) {
			$lines[] = '';
			$lines[] = '## Public sections';
			foreach ( $terms as $term ) {
				$term_link = get_term_link( $term );
				if ( is_wp_error( $term_link ) ) {
					continue;
				}
				$lines[] = '- [' . $this->escape_markdown_link_text( $term->name ) . '](' . esc_url_raw( $term_link ) . ') - ' . $this->escape_markdown_line( $term->taxonomy );
			}
		}

		$posts = $this->get_posts_for_llms( $options['include_post_types'], $max_items, (bool) $options['include_private'] );
		if ( ! empty( $posts ) ) {
			$lines[] = '';
			$lines[] = $full ? '## Public content index' : '## Important and recent public content';
			foreach ( $posts as $post ) {
				if ( $this->should_skip_post( $post ) ) {
					continue;
				}

				$title   = get_the_title( $post );
				$link    = get_permalink( $post );
				$type    = get_post_type( $post );
				$updated = get_the_modified_date( 'Y-m-d', $post );

				if ( ! $link || '' === $title ) {
					continue;
				}

				$line = '- [' . $this->escape_markdown_link_text( $title ) . '](' . esc_url_raw( $link ) . ')';
				$line .= ' - ' . $this->escape_markdown_line( $type );
				if ( $updated ) {
					$line .= ', updated ' . $this->escape_markdown_line( $updated );
				}

				$summary = $this->get_post_summary( $post );
				if ( '' !== $summary ) {
					$line .= '. ' . $this->escape_markdown_line( $summary );
				}

				$lines[] = $line;
			}
		}

		if ( ! empty( $options['webmcp_enabled'] ) && ! empty( $options['webmcp_form_metadata_enabled'] ) ) {
			$lines[] = '';
			$lines[] = '## Public form and WebMCP notes';
			$lines[] = '- Declarative WebMCP-style form metadata is enabled where supported by the plugin.';
			$lines[] = '- Form submissions should be treated as state-changing actions and require explicit user confirmation.';
			if ( ! empty( $options['webmcp_search_form_enabled'] ) ) {
				$lines[] = '- The public search form may be exposed as a read-only search_public_content-style tool.';
			}
		}

		if ( ! empty( $options['accessibility_scanner_enabled'] ) ) {
			$lines[] = '';
			$lines[] = '## Accessibility for agents';
			$lines[] = '- The site is configured to scan public content for labels, names, alt text, link clarity and heading structure.';
			$lines[] = '- Agents should prefer elements with clear accessible names and labelled fields when interpreting public pages.';
		}

		if ( ! empty( $options['layout_scanner_enabled'] ) ) {
			$lines[] = '';
			$lines[] = '## Layout stability for agents';
			$lines[] = '- The site is configured to scan public content for media, embeds and ad slots that may cause layout shifts.';
			$lines[] = '- Agents should account for ads, sticky bars, popups and late-loading media when interpreting visible page positions.';
			if ( ! empty( $options['layout_helper_css_enabled'] ) && ! empty( $options['layout_reserve_ad_slots'] ) ) {
				$lines[] = '- Common ad-slot reservation helper CSS is enabled to reduce unexpected content movement.';
			}
		}

		if ( ! empty( $options['agent_tools_enabled'] ) && ! empty( $options['agent_tools_rest_enabled'] ) ) {
			$lines[] = '';
			$lines[] = '## Public read-only agent tools';
			$lines[] = '- Tool registry: ' . esc_url_raw( rest_url( 'agizoai-agentic/v1/tools' ) );
			$lines[] = '- Mode: public read-only. These endpoints are for discovery and public content retrieval only.';
			$lines[] = '- Available tool families may include site information, public search, public content summaries, recent content, navigation menus, public sections and public WooCommerce product discovery when enabled.';
		}

		if ( ! empty( $options['ecommerce_product_tools_enabled'] ) && post_type_exists( 'product' ) ) {
			$lines[] = '';
			$lines[] = '## WooCommerce product-discovery notes';
			$lines[] = '- Public read-only product discovery may be available at: ' . esc_url_raw( rest_url( 'agizoai-agentic/v1/products' ) );
			$lines[] = '- Product tools are limited to published public product data such as titles, URLs, public prices, stock status, summaries, categories and images.';
			$lines[] = '- Agents must not assume access to carts, checkout, accounts, orders, customers, addresses or payment actions.';
		}

		if ( $this->integrations && ! empty( $options['integrations_enabled'] ) && ! empty( $options['integration_add_seo_notes_to_llms'] ) ) {
			$summary = $this->integrations->summarize_integrations();
			$lines[] = '';
			$lines[] = '## SEO metadata for agents';
			$lines[] = '- SEO metadata integration is enabled for agent summaries and public discovery files.';
			$lines[] = '- Enabled metadata sources: ' . $this->escape_markdown_line( implode( ', ', (array) $summary['enabled_sources'] ) );
			$lines[] = '- Agents should prefer canonical URLs, public meta descriptions and index/follow hints when available.';
			if ( ! empty( $options['integration_expose_seo_metadata_in_tools'] ) ) {
				$lines[] = '- Public read-only content tools may include SEO metadata fields for published content.';
			}
		}

		$lines[] = '';
		$lines[] = '## Agentic browsing notes';
		$lines[] = '- Public content only is intended for LLM and AI-agent discovery.';
		$lines[] = '- Private, draft, password-protected and noindex content should not be treated as public agent-facing content.';
		$lines[] = '- For any action that changes site state, agents should request explicit user confirmation.';

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

	/**
	 * Get posts for llms output.
	 *
	 * @param array<int,string> $post_types Post types.
	 * @param int               $limit Limit.
	 * @param bool              $include_private Include private content.
	 * @return array<int,WP_Post>
	 */
	private function get_posts_for_llms( array $post_types, $limit, $include_private = false ) {
		$post_status = $include_private ? array( 'publish', 'private' ) : array( 'publish' );

		if ( empty( $post_types ) ) {
			$post_types = array( 'post', 'page' );
		}

		$args = array(
			'post_type'              => $post_types,
			'post_status'            => $post_status,
			'posts_per_page'         => absint( $limit ),
			'orderby'                => 'modified',
			'order'                  => 'DESC',
			'no_found_rows'          => true,
			'ignore_sticky_posts'    => true,
			'update_post_meta_cache' => true,
			'update_post_term_cache' => false,
			'suppress_filters'       => false,
		);

		$posts = get_posts( $args );

		return is_array( $posts ) ? $posts : array();
	}

	/**
	 * Get public terms.
	 *
	 * @param array<int,string> $taxonomies Taxonomies.
	 * @param int               $limit Limit.
	 * @return array<int,WP_Term>
	 */
	private function get_terms_for_llms( array $taxonomies, $limit ) {
		if ( empty( $taxonomies ) ) {
			return array();
		}

		$terms = get_terms(
			array(
				'taxonomy'   => $taxonomies,
				'hide_empty' => true,
				'number'     => absint( $limit ),
				'orderby'    => 'count',
				'order'      => 'DESC',
			)
		);

		if ( is_wp_error( $terms ) || ! is_array( $terms ) ) {
			return array();
		}

		return $terms;
	}

	/**
	 * Determine if a post should be skipped.
	 *
	 * @param WP_Post $post Post object.
	 * @return bool
	 */
	public function should_skip_post( WP_Post $post ) {
		if ( post_password_required( $post ) ) {
			return true;
		}

		if ( 'attachment' === $post->post_type && ! $this->settings->get( 'include_attachments' ) ) {
			return true;
		}

		if ( $this->settings->get( 'respect_noindex' ) && $this->is_noindex( $post->ID ) ) {
			return true;
		}

		/**
		 * Filter whether a post should be skipped from llms.txt.
		 *
		 * @param bool    $skip Whether to skip.
		 * @param WP_Post $post Post object.
		 */
		return (bool) apply_filters( 'agizoai_agentic_skip_post', false, $post );
	}

	/**
	 * Basic noindex detection across common SEO plugin meta keys.
	 *
	 * @param int $post_id Post ID.
	 * @return bool
	 */
	public function is_noindex( $post_id ) {
		$post_id = absint( $post_id );

		$yoast = get_post_meta( $post_id, '_yoast_wpseo_meta-robots-noindex', true );
		if ( '1' === (string) $yoast || 'yes' === strtolower( (string) $yoast ) ) {
			return true;
		}

		$rank_math = get_post_meta( $post_id, 'rank_math_robots', true );
		if ( is_array( $rank_math ) && in_array( 'noindex', $rank_math, true ) ) {
			return true;
		}
		if ( is_string( $rank_math ) && false !== strpos( strtolower( $rank_math ), 'noindex' ) ) {
			return true;
		}

		$aioseo = get_post_meta( $post_id, '_aioseo_robots_noindex', true );
		if ( '1' === (string) $aioseo || 'on' === strtolower( (string) $aioseo ) ) {
			return true;
		}

		$seopress = get_post_meta( $post_id, '_seopress_robots_index', true );
		if ( 'yes' === strtolower( (string) $seopress ) || '1' === (string) $seopress ) {
			return true;
		}

		$rsm = get_post_meta( $post_id, '_rsm_robots_noindex', true );
		if ( '1' === (string) $rsm || 'yes' === strtolower( (string) $rsm ) ) {
			return true;
		}

		/**
		 * Filter noindex detection.
		 *
		 * @param bool $is_noindex Whether post is noindex.
		 * @param int  $post_id Post ID.
		 */
		return (bool) apply_filters( 'agizoai_agentic_is_noindex', false, $post_id );
	}

	/**
	 * Return sitemap URL.
	 *
	 * @return string
	 */
	private function get_sitemap_url() {
		if ( function_exists( 'wp_sitemaps_get_server' ) ) {
			return home_url( '/wp-sitemap.xml' );
		}

		return home_url( '/sitemap.xml' );
	}

	/**
	 * Get post summary.
	 *
	 * @param WP_Post $post Post object.
	 * @return string
	 */
	private function get_post_summary( WP_Post $post ) {
		if ( $this->integrations && $this->settings->get( 'integrations_enabled', 1 ) && $this->settings->get( 'integration_use_seo_descriptions_in_llms', 1 ) ) {
			$metadata = $this->integrations->get_post_metadata( $post->ID );
			if ( ! empty( $metadata['description'] ) ) {
				return trim( preg_replace( '/\s+/', ' ', wp_strip_all_tags( $metadata['description'], true ) ) );
			}
		}

		$excerpt = has_excerpt( $post ) ? get_the_excerpt( $post ) : '';

		if ( '' === trim( $excerpt ) ) {
			$content = strip_shortcodes( $post->post_content );
			$content = wp_strip_all_tags( $content, true );
			$excerpt = wp_trim_words( $content, 24, '…' );
		}

		return trim( preg_replace( '/\s+/', ' ', wp_strip_all_tags( $excerpt, true ) ) );
	}

	/**
	 * Escape line for Markdown/plain text.
	 *
	 * @param string $value Value.
	 * @return string
	 */
	private function escape_markdown_line( $value ) {
		$value = wp_strip_all_tags( (string) $value, true );
		$value = html_entity_decode( $value, ENT_QUOTES, get_bloginfo( 'charset' ) );
		$value = preg_replace( '/\s+/', ' ', $value );

		return trim( $value );
	}

	/**
	 * Escape link text.
	 *
	 * @param string $value Value.
	 * @return string
	 */
	private function escape_markdown_link_text( $value ) {
		$value = $this->escape_markdown_line( $value );
		$value = str_replace( array( '[', ']' ), array( '\\[', '\\]' ), $value );

		return $value;
	}

	/**
	 * Clean paragraph text.
	 *
	 * @param string $value Value.
	 * @return string
	 */
	private function paragraph( $value ) {
		$value = wp_strip_all_tags( (string) $value, true );
		$value = html_entity_decode( $value, ENT_QUOTES, get_bloginfo( 'charset' ) );
		$value = preg_replace( "/\n{3,}/", "\n\n", $value );

		return trim( $value );
	}

	/**
	 * Normalise endpoint text.
	 *
	 * @param string $text Text.
	 * @return string
	 */
	private function normalise_text( $text ) {
		$text = (string) $text;
		$text = str_replace( "\r\n", "\n", $text );
		$text = str_replace( "\r", "\n", $text );
		$text = trim( $text );

		return $text . "\n";
	}
}