<?php
/**
 * Discovery: XML sitemaps (with images), robots.txt, llms.txt for AI search, and IndexNow instant submission.
 */

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

class SXSEO_Sitemap {

	const PER_PAGE = 1000;

	public static function hook() {
		// Before WordPress registers its own sitemap rules (priority 10), so ours sit in front of them.
		add_action( 'init', array( __CLASS__, 'add_rewrites' ), 5 );
		add_filter( 'query_vars', function ( $v ) {
			$v[] = 'sxseo_out';
			$v[] = 'sxseo_kind';
			$v[] = 'sxseo_name';
			$v[] = 'sxseo_page';
			return $v;
		} );
		add_action( 'template_redirect', array( __CLASS__, 'serve' ), 0 );
		add_filter( 'robots_txt', array( __CLASS__, 'robots_txt' ), 99, 2 );
		add_filter( 'wp_sitemaps_enabled', '__return_false' );
		add_action( 'transition_post_status', array( __CLASS__, 'on_transition' ), 10, 3 );
		foreach ( array( 'save_post', 'deleted_post', 'edited_term', 'created_term', 'delete_term' ) as $h ) {
			add_action( $h, array( __CLASS__, 'purge' ) );
		}
	}

	public static function add_rewrites() {
		add_rewrite_rule( '^sitemap\.xml$', 'index.php?sxseo_out=index', 'top' );
		add_rewrite_rule( '^sitemap-([pt])-([a-z0-9_-]+?)-([0-9]+)\.xml$', 'index.php?sxseo_out=map&sxseo_kind=$matches[1]&sxseo_name=$matches[2]&sxseo_page=$matches[3]', 'top' );
		add_rewrite_rule( '^wp-sitemap[a-z0-9_-]*\.xml$', 'index.php?sxseo_out=legacy', 'top' );
		add_rewrite_rule( '^llms\.txt$', 'index.php?sxseo_out=llms', 'top' );
		$key = (string) sxseo_opt( 'indexnow_key' );
		if ( '' !== $key ) {
			add_rewrite_rule( '^' . preg_quote( $key, '/' ) . '\.txt$', 'index.php?sxseo_out=indexnow', 'top' );
		}
	}

	/**
	 * Invalidate every cached sitemap / llms.txt by bumping a version that is part of each cache key. (Deleting the
	 * rows from the options table would miss copies held by a persistent object cache such as Redis.)
	 */
	public static function purge() {
		update_option( 'sxseo_cache_ver', (string) microtime( true ), false );
	}

	// ------------------------------------------------------------------ serving

	public static function serve() {
		$out = get_query_var( 'sxseo_out' );
		if ( ! $out ) {
			return;
		}
		if ( 'legacy' === $out ) { // sitemaps WordPress used to publish, already submitted to Search Console
			wp_redirect( home_url( '/sitemap.xml' ), 301, 'Sentrixa SEO' ); // phpcs:ignore WordPress.Security.SafeRedirect
			exit;
		}
		if ( 'indexnow' === $out ) {
			header( 'Content-Type: text/plain; charset=utf-8' );
			echo esc_html( (string) sxseo_opt( 'indexnow_key' ) );
			exit;
		}
		if ( 'llms' === $out ) {
			if ( ! sxseo_opt( 'llms_txt' ) ) {
				self::not_found();
			}
			self::send( self::cached( 'llms', array( __CLASS__, 'build_llms' ) ), 'text/plain; charset=utf-8' );
		}
		if ( ! sxseo_opt( 'sitemap' ) || ! get_option( 'blog_public' ) ) {
			self::not_found();
		}
		if ( 'index' === $out ) {
			self::send( self::cached( 'index', array( __CLASS__, 'build_index' ) ), 'application/xml; charset=utf-8' );
		}
		$kind = get_query_var( 'sxseo_kind' );
		$name = sanitize_key( get_query_var( 'sxseo_name' ) );
		$page = max( 1, (int) get_query_var( 'sxseo_page' ) );
		$xml  = self::cached( "m_{$kind}_{$name}_{$page}", function () use ( $kind, $name, $page ) {
			return self::build_map( $kind, $name, $page );
		} );
		if ( '' === $xml ) {
			self::not_found();
		}
		self::send( $xml, 'application/xml; charset=utf-8' );
	}

	private static function not_found() {
		global $wp_query;
		$wp_query->set_404();
		status_header( 404 );
		nocache_headers();
		include get_404_template();
		exit;
	}

	private static function send( $body, $type ) {
		status_header( 200 );
		header( 'Content-Type: ' . $type );
		header( 'X-Robots-Tag: noindex, follow', true );
		header( 'Cache-Control: public, max-age=900' );
		echo $body; // phpcs:ignore WordPress.Security.EscapeOutput
		exit;
	}

	private static function cached( $key, $builder ) {
		$k = 'sxseo_' . substr( md5( $key . '|' . get_option( 'sxseo_cache_ver', '0' ) ), 0, 20 );
		$v = get_transient( $k );
		if ( false === $v ) {
			$v = call_user_func( $builder );
			set_transient( $k, $v, 6 * HOUR_IN_SECONDS );
		}
		return $v;
	}

	// ------------------------------------------------------------------- queries

	private static function base_args( $type ) {
		return array(
			'post_type'              => $type,
			'post_status'            => 'publish',
			'has_password'           => false,
			'no_found_rows'          => true,
			'update_post_term_cache' => false,
			'meta_query'             => array(
				'relation' => 'AND',
				array(
					'relation' => 'OR',
					array( 'key' => '_sxseo_noindex', 'compare' => 'NOT EXISTS' ),
					array( 'key' => '_sxseo_noindex', 'value' => '1', 'compare' => '!=' ),
				),
				array(
					'relation' => 'OR',
					array( 'key' => '_sxseo_exclude_sitemap', 'compare' => 'NOT EXISTS' ),
					array( 'key' => '_sxseo_exclude_sitemap', 'value' => '1', 'compare' => '!=' ),
				),
			),
		);
	}

	private static function count_posts( $type ) {
		$q = new WP_Query( array_merge( self::base_args( $type ), array( 'posts_per_page' => 1, 'fields' => 'ids', 'no_found_rows' => false ) ) );
		return (int) $q->found_posts;
	}

	private static function terms_for( $tax ) {
		$terms = get_terms( array( 'taxonomy' => $tax, 'hide_empty' => true, 'number' => 0 ) );
		if ( is_wp_error( $terms ) ) {
			return array();
		}
		return array_values( array_filter( $terms, function ( $t ) {
			return sxseo_term_indexable( $t ) && 'uncategorized' !== $t->slug;
		} ) );
	}

	// -------------------------------------------------------------------- builders

	public static function build_index() {
		$x = '<?xml version="1.0" encoding="UTF-8"?>' . "\n" . '<sitemapindex xmlns="http://www.sitemaps.org/schemas/sitemap/0.9">' . "\n";
		$latest = gmdate( 'c' );
		foreach ( sxseo_post_types() as $type ) {
			$n = self::count_posts( $type );
			if ( 0 === $n && ! ( 'page' === $type ) ) {
				continue;
			}
			$pages = max( 1, (int) ceil( $n / self::PER_PAGE ) );
			for ( $p = 1; $p <= $pages; $p++ ) {
				$x .= '<sitemap><loc>' . sxseo_x( home_url( "/sitemap-p-{$type}-{$p}.xml" ) ) . '</loc><lastmod>' . $latest . '</lastmod></sitemap>' . "\n";
			}
		}
		foreach ( sxseo_taxonomies() as $tax ) {
			if ( self::terms_for( $tax ) ) {
				$x .= '<sitemap><loc>' . sxseo_x( home_url( "/sitemap-t-{$tax}-1.xml" ) ) . '</loc><lastmod>' . $latest . '</lastmod></sitemap>' . "\n";
			}
		}
		return $x . '</sitemapindex>';
	}

	public static function build_map( $kind, $name, $page ) {
		$head = '<?xml version="1.0" encoding="UTF-8"?>' . "\n" . '<urlset xmlns="http://www.sitemaps.org/schemas/sitemap/0.9" xmlns:image="http://www.google.com/schemas/sitemap-image/1.1">' . "\n";
		$rows = '';
		if ( 'p' === $kind && in_array( $name, sxseo_post_types(), true ) ) {
			if ( 1 === $page && ( 'page' === $name ) && 'page' !== get_option( 'show_on_front' ) ) {
				$rows .= '<url><loc>' . sxseo_x( home_url( '/' ) ) . '</loc><changefreq>daily</changefreq><priority>1.0</priority></url>' . "\n";
			}
			$q = new WP_Query( array_merge( self::base_args( $name ), array( 'posts_per_page' => self::PER_PAGE, 'paged' => $page, 'orderby' => 'modified', 'order' => 'DESC' ) ) );
			$front = (int) get_option( 'page_on_front' );
			foreach ( $q->posts as $post ) {
				$loc  = get_permalink( $post );
				$prio = $post->ID === $front ? '1.0' : ( 'page' === $post->post_type ? '0.8' : '0.6' );
				$rows .= '<url><loc>' . sxseo_x( $loc ) . '</loc><lastmod>' . sxseo_w3c( $post->post_modified_gmt ) . '</lastmod><changefreq>' . ( $post->ID === $front ? 'daily' : 'weekly' ) . '</changefreq><priority>' . $prio . '</priority>';
				if ( sxseo_opt( 'sitemap_images' ) ) {
					foreach ( self::images_for( $post ) as $img ) {
						$rows .= '<image:image><image:loc>' . sxseo_x( $img['url'] ) . '</image:loc>' . ( $img['title'] ? '<image:title>' . sxseo_x( $img['title'] ) . '</image:title>' : '' ) . '</image:image>';
					}
				}
				$rows .= "</url>\n";
			}
			if ( ! $q->posts && 1 !== $page ) {
				return '';
			}
		} elseif ( 't' === $kind && in_array( $name, sxseo_taxonomies(), true ) ) {
			foreach ( self::terms_for( $name ) as $t ) {
				$link = get_term_link( $t );
				if ( ! is_wp_error( $link ) ) {
					$rows .= '<url><loc>' . sxseo_x( $link ) . '</loc><changefreq>weekly</changefreq><priority>0.4</priority></url>' . "\n";
				}
			}
		} else {
			return '';
		}
		return $head . $rows . '</urlset>';
	}

	private static function images_for( WP_Post $post ) {
		$imgs  = array();
		$thumb = get_post_thumbnail_id( $post );
		if ( $thumb ) {
			$u = wp_get_attachment_image_url( $thumb, 'full' );
			if ( $u ) {
				$imgs[ $u ] = array( 'url' => $u, 'title' => sxseo_title_text( $thumb ) );
			}
		}
		if ( preg_match_all( '/<img[^>]+src=["\']([^"\']+)["\'][^>]*>/i', $post->post_content, $m ) ) {
			foreach ( array_slice( $m[1], 0, 8 ) as $src ) {
				if ( 0 === strpos( $src, '//' ) ) {
					$src = 'https:' . $src;
				}
				if ( 0 === strpos( $src, '/' ) ) {
					$src = home_url( $src );
				}
				if ( 0 === strpos( $src, 'http' ) && ! isset( $imgs[ $src ] ) ) {
					$imgs[ $src ] = array( 'url' => $src, 'title' => '' );
				}
			}
		}
		return array_slice( array_values( $imgs ), 0, 10 );
	}

	/** llms.txt: a plain-language map of the site for AI assistants and answer engines. */
	public static function build_llms() {
		$name = get_bloginfo( 'name' );
		$out  = "# {$name}\n\n";
		$desc = (string) sxseo_opt( 'home_description' ) ?: get_bloginfo( 'description' );
		if ( $desc ) {
			$out .= '> ' . $desc . "\n\n";
		}
		$o = SXSEO_Settings::all();
		$bits = array_filter( array( $o['org_phone'], $o['org_email'], trim( implode( ', ', array_filter( array( $o['org_street'], $o['org_city'], $o['org_postcode'] ) ) ) ) ) );
		if ( $bits ) {
			$out .= 'Contact: ' . implode( ' | ', $bits ) . "\n\n";
		}
		$pages = get_posts( array( 'post_type' => 'page', 'post_status' => 'publish', 'numberposts' => 50, 'orderby' => 'menu_order title', 'order' => 'ASC', 'has_password' => false ) );
		if ( $pages ) {
			$out .= "## Pages\n\n";
			foreach ( $pages as $p ) {
				if ( '1' === get_post_meta( $p->ID, '_sxseo_noindex', true ) ) {
					continue;
				}
				$d    = SXSEO_Context::for_post( $p );
				$out .= '- [' . sxseo_title_text( $p ) . '](' . get_permalink( $p ) . '): ' . $d['desc'] . "\n";
			}
			$out .= "\n";
		}
		foreach ( array_diff( sxseo_post_types(), array( 'page' ) ) as $type ) {
			$posts = get_posts( array( 'post_type' => $type, 'post_status' => 'publish', 'numberposts' => 40, 'has_password' => false ) );
			if ( ! $posts ) {
				continue;
			}
			$obj  = get_post_type_object( $type );
			$out .= '## ' . ( $obj ? $obj->labels->name : $type ) . "\n\n";
			foreach ( $posts as $p ) {
				if ( '1' === get_post_meta( $p->ID, '_sxseo_noindex', true ) ) {
					continue;
				}
				$d    = SXSEO_Context::for_post( $p );
				$out .= '- [' . sxseo_title_text( $p ) . '](' . get_permalink( $p ) . '): ' . $d['desc'] . "\n";
			}
			$out .= "\n";
		}
		$out .= "## Sitemap\n\n- " . home_url( '/sitemap.xml' ) . "\n";
		return $out;
	}

	// --------------------------------------------------------------------- robots

	public static function robots_txt( $output, $public ) {
		if ( ! $public ) {
			return $output;
		}
		if ( sxseo_opt( 'robots_txt_extras' ) ) {
			$output = rtrim( $output ) . "\n";
			$output .= "Disallow: /?s=\nDisallow: /search/\nDisallow: /*?replytocom=\nDisallow: /*?add-to-cart=\n";
		}
		if ( 'block_training' === sxseo_opt( 'ai_policy' ) ) {
			foreach ( array( 'GPTBot', 'CCBot', 'Google-Extended', 'anthropic-ai', 'ClaudeBot', 'Bytespider', 'Applebot-Extended', 'meta-externalagent' ) as $bot ) {
				$output .= "\nUser-agent: {$bot}\nDisallow: /\n";
			}
		}
		if ( sxseo_opt( 'sitemap' ) ) {
			$output .= "\nSitemap: " . home_url( '/sitemap.xml' ) . "\n";
		}
		return $output;
	}

	// ------------------------------------------------------------------- IndexNow

	public static function on_transition( $new, $old, $post ) {
		if ( ! sxseo_opt( 'indexnow' ) || ! get_option( 'blog_public' ) || wp_is_post_revision( $post ) ) {
			return;
		}
		if ( ! in_array( $post->post_type, sxseo_post_types(), true ) ) {
			return;
		}
		if ( 'publish' !== $new && 'publish' !== $old ) {
			return;
		}
		if ( 'publish' === $new && '1' === get_post_meta( $post->ID, '_sxseo_noindex', true ) ) {
			return;
		}
		self::submit( array( get_permalink( $post ) ) );
	}

	public static function submit( array $urls ) {
		$host = sxseo_home_host();
		$key  = (string) sxseo_opt( 'indexnow_key' );
		if ( '' === $key || ! $host || in_array( $host, array( 'localhost', '127.0.0.1' ), true ) || substr( $host, -6 ) === '.local' ) {
			return;
		}
		$fresh = array();
		foreach ( $urls as $u ) {
			$t = 'sxseo_in_' . md5( $u );
			if ( false === get_transient( $t ) ) {
				set_transient( $t, 1, 10 * MINUTE_IN_SECONDS );
				$fresh[] = $u;
			}
		}
		if ( ! $fresh ) {
			return;
		}
		wp_remote_post(
			'https://api.indexnow.org/indexnow',
			array(
				'blocking' => false,
				'timeout'  => 3,
				'headers'  => array( 'Content-Type' => 'application/json; charset=utf-8' ),
				'body'     => wp_json_encode( array( 'host' => $host, 'key' => $key, 'keyLocation' => home_url( '/' . $key . '.txt' ), 'urlList' => $fresh ) ),
			)
		);
	}
}