Molienda» would // glue into one sentence. $html = preg_replace( '#]*>#i', ' ', strip_shortcodes( $post->post_content ) ); $text = trim( preg_replace( '/\s+/', ' ', wp_strip_all_tags( $html ) ) ); if ( '' === $text ) { return ''; } if ( preg_match( '/^(.{40,200}?[.!?])(\s|$)/u', $text, $m ) ) { return genscore_one_line( $m[1] ); } return ''; } /** * Whether a page still holds WordPress's own template text (the privacy * policy guide), which says nothing about this site. * * @param string $content Post content. * @return bool */ function genscore_is_sample_content( $content ) { foreach ( array( 'Texto sugerido', 'Suggested text', 'wp-policy-help', 'privacy-policy-tutorial' ) as $needle ) { if ( false !== strpos( (string) $content, $needle ) ) { return true; } } return false; } /** * Pages that only work for a shopper with a session (WooCommerce cart, * checkout and account): nothing there for an AI to read. * * @return int[] */ function genscore_llms_skipped_pages() { $ids = array( (int) get_option( 'page_on_front' ), (int) get_option( 'wp_page_for_privacy_policy' ) ); if ( function_exists( 'wc_get_page_id' ) ) { foreach ( array( 'cart', 'checkout', 'myaccount' ) as $wc ) { $ids[] = (int) wc_get_page_id( $wc ); } } return array_values( array_filter( $ids, function ( $id ) { return $id > 0; } ) ); } /** * One line, at most 200 characters, no markdown brackets. * * @param string $s Text. * @return string */ function genscore_one_line( $s ) { $s = trim( str_replace( array( '[', ']', "\n", "\r" ), array( '(', ')', ' ', ' ' ), $s ) ); if ( function_exists( 'mb_strlen' ) && mb_strlen( $s ) > 200 ) { $s = rtrim( mb_substr( $s, 0, 199 ) ) . '…'; } return $s; } /** * A markdown list line. * * @param WP_Post $post Post. * @return string */ function genscore_llms_line( $post ) { $title = genscore_one_line( get_the_title( $post ) ); if ( '' === $title ) { return ''; } $desc = genscore_post_description( $post ); return '- [' . $title . '](' . get_permalink( $post ) . ')' . ( '' !== $desc ? ': ' . $desc : '' ); } /** * The generated llms.txt. * * @return string */ function genscore_llms_generate() { $out = array( '# ' . genscore_one_line( get_bloginfo( 'name' ) ) ); $tag = genscore_one_line( get_bloginfo( 'description' ) ); if ( '' !== $tag ) { $out[] = ''; $out[] = '> ' . $tag; } $skip = genscore_llms_skipped_pages(); $pages = get_posts( array( 'post_type' => 'page', 'post_status' => 'publish', 'posts_per_page' => GENSCORE_LLMS_PAGES, 'orderby' => array( 'menu_order' => 'ASC', 'title' => 'ASC' ), 'has_password' => false, ) ); $lines = array(); foreach ( $pages as $p ) { if ( in_array( (int) $p->ID, $skip, true ) ) { continue; } $l = genscore_llms_line( $p ); if ( '' !== $l ) { $lines[] = $l; } } if ( $lines ) { array_push( $out, '', '## ' . __( 'Páginas', 'genscore' ), '' ); $out = array_merge( $out, $lines ); } if ( post_type_exists( 'product' ) ) { $products = get_posts( array( 'post_type' => 'product', 'post_status' => 'publish', 'posts_per_page' => GENSCORE_LLMS_PRODUCTS, 'has_password' => false ) ); $lines = array_filter( array_map( 'genscore_llms_line', $products ) ); if ( $lines ) { array_push( $out, '', '## ' . __( 'Productos', 'genscore' ), '' ); $out = array_merge( $out, $lines ); } } $posts = get_posts( array( 'post_type' => 'post', 'post_status' => 'publish', 'posts_per_page' => GENSCORE_LLMS_POSTS, 'has_password' => false ) ); $lines = array_filter( array_map( 'genscore_llms_line', $posts ) ); if ( $lines ) { array_push( $out, '', '## ' . __( 'Artículos', 'genscore' ), '' ); $out = array_merge( $out, $lines ); } $privacy = (int) get_option( 'wp_page_for_privacy_policy' ); if ( $privacy && 'publish' === get_post_status( $privacy ) ) { array_push( $out, '', '## Optional', '', genscore_llms_line( get_post( $privacy ) ) ); } return implode( "\n", $out ) . "\n"; } /** * What /llms.txt serves: the person's own edited text, or the generated one. * * @return string */ function genscore_llms_content() { $custom = get_option( 'genscore_llms_custom', '' ); return is_string( $custom ) && '' !== trim( $custom ) ? $custom : genscore_llms_generate(); } /** * Whether GenScore is the one answering /llms.txt right now. * * @return bool */ function genscore_llms_serving() { return (bool) get_option( 'genscore_llms_enabled', false ) && ! genscore_llms_other_provider() && ! genscore_llms_physical_file(); } /** * Answers /llms.txt before WordPress looks for a post. */ function genscore_llms_maybe_serve() { if ( ! genscore_llms_serving() ) { return; } $path = wp_parse_url( isset( $_SERVER['REQUEST_URI'] ) ? esc_url_raw( wp_unslash( $_SERVER['REQUEST_URI'] ) ) : '', PHP_URL_PATH ); $home = wp_parse_url( home_url( '/' ), PHP_URL_PATH ); if ( ! is_string( $path ) || trailingslashit( (string) $home ) . 'llms.txt' !== $path ) { return; } nocache_headers(); header( 'Content-Type: text/plain; charset=utf-8' ); header( 'X-Robots-Tag: noindex' ); echo genscore_llms_content(); // phpcs:ignore WordPress.Security.EscapeOutput.OutputNotEscaped -- plain text file, not HTML. exit; } add_action( 'parse_request', 'genscore_llms_maybe_serve', 0 );