Markdown URLs for LLMs

/**

 * After enabling this snippet, go to Settings > Permalinks and click "Save Changes"

 * Configure the $enabled_post_types array to specify which post types should support .md URLs

 */

$enabled_post_types = [ 'post', 'page' ];

// Add rewrite rule to catch .md URLs

add_action( 'init', function () {

	add_rewrite_rule(

		'(.+?)\.md$',

		'index.php?markdown_url=$matches[1]',

		'top'

	);

} );

// Add custom query variable

add_filter( 'query_vars', function ( $vars ) {

	$vars[] = 'markdown_url';

	return $vars;

} );

// Handle the markdown request

add_action( 'template_redirect', function () use ( $enabled_post_types ) {

	$markdown_url = get_query_var( 'markdown_url' );

	if ( ! $markdown_url ) {

		return;

	}

	// Try to find the post by URL

	$post = ( function ( $url ) use ( $enabled_post_types ) {

		// Remove leading slash if present

		$url = ltrim( $url, '/' );

		// Try to get post by URL path

		$post_id = url_to_postid( '/' . $url );

		if ( $post_id ) {

			return get_post( $post_id );

		}

		// If that doesn't work, try by post name/slug

		foreach ( $enabled_post_types as $post_type ) {

			// Extract just the slug (last part after last slash)

			$slug = basename( $url );

			$posts = get_posts( [

				'name'        => $slug,

				'post_type'   => $post_type,

				'post_status' => 'publish',

				'numberposts' => 1

			] );

			if ( ! empty( $posts ) ) {

				return $posts[0];

			}

		}

		return null;

	} )( $markdown_url );

	if ( ! $post || ! in_array( $post->post_type, $enabled_post_types ) ) {

		status_header( 404 );

		nocache_headers();

		echo "Post not found or markdown not enabled for this post type.";

		exit;

	}

	// Generate markdown content

	$markdown_content = '';

	// Add title

	$markdown_content .= '# ' . get_the_title( $post ) . "\n\n";

	// Add metadata

	$markdown_content .= '**Published:** ' . get_the_date( 'F j, Y', $post ) . "\n";

	$markdown_content .= '**Author:** ' . get_the_author_meta( 'display_name', $post->post_author ) . "\n";

	// Add categories for posts

	if ( $post->post_type === 'post' ) {

		$categories = get_the_category( $post->ID );

		if ( ! empty( $categories ) ) {

			$cat_names        = array_map( function ( $cat ) {

				return $cat->name;

			}, $categories );

			$markdown_content .= '**Categories:** ' . implode( ', ', $cat_names ) . "\n";

		}

		// Add tags for posts

		$tags = get_the_tags( $post->ID );

		if ( ! empty( $tags ) ) {

			$tag_names        = array_map( function ( $tag ) {

				return $tag->name;

			}, $tags );

			$markdown_content .= '**Tags:** ' . implode( ', ', $tag_names ) . "\n";

		}

	}

	$markdown_content .= "\n---\n\n";

	// Convert HTML content to markdown-friendly format

	$post_content = apply_filters( 'the_content', $post->post_content );

	// Basic HTML to Markdown conversion

	$post_content     = ( function ( $html ) {

		$html = trim( preg_replace( '/\s+/', ' ', $html ) );

		$html = preg_replace( '/<h1[^>]*>(.*?)<\/h1>/i', "\n# $1\n", $html );

		$html = preg_replace( '/<h2[^>]*>(.*?)<\/h2>/i', "\n## $1\n", $html );

		$html = preg_replace( '/<h3[^>]*>(.*?)<\/h3>/i', "\n### $1\n", $html );

		$html = preg_replace( '/<h4[^>]*>(.*?)<\/h4>/i', "\n#### $1\n", $html );

		$html = preg_replace( '/<h5[^>]*>(.*?)<\/h5>/i', "\n##### $1\n", $html );

		$html = preg_replace( '/<h6[^>]*>(.*?)<\/h6>/i', "\n###### $1\n", $html );

		$html = preg_replace( '/<p[^>]*>(.*?)<\/p>/i', "$1\n\n", $html );

		$html = str_replace( [ '<br>', '<br/>', '<br />' ], "\n", $html );

		$html = preg_replace( '/<(strong|b)[^>]*>(.*?)<\/\1>/i', "**$2**", $html );

		$html = preg_replace( '/<(em|i)[^>]*>(.*?)<\/\1>/i', "*$2*", $html );

		$html = preg_replace( '/<a[^>]*href=["\']([^"\']*)["\'][^>]*>(.*?)<\/a>/i', "[$2]($1)", $html );

		$html = preg_replace( '/<img[^>]*src=["\']([^"\']*)["\'][^>]*alt=["\']([^"\']*)["\'][^>]*\/?>/i', "![$2]($1)", $html );

		$html = preg_replace( '/<img[^>]*alt=["\']([^"\']*)["\'][^>]*src=["\']([^"\']*)["\'][^>]*\/?>/i', "![$1]($2)", $html );

		$html = preg_replace( '/<img[^>]*src=["\']([^"\']*)["\'][^>]*\/?>/i', "![]($1)", $html );

		$html = preg_replace( '/<ul[^>]*>/i', "", $html );

		$html = preg_replace( '/<\/ul>/i', "\n", $html );

		$html = preg_replace( '/<ol[^>]*>/i', "", $html );

		$html = preg_replace( '/<\/ol>/i', "\n", $html );

		$html = preg_replace( '/<li[^>]*>(.*?)<\/li>/i', "- $1\n", $html );

		$html = preg_replace_callback( '/<blockquote[^>]*>(.*?)<\/blockquote>/is', function ( $matches ) {

			$lines  = explode( "\n", trim( $matches[1] ) );

			$quoted = array_map( function ( $line ) {

				return '> ' . trim( $line );

			}, $lines );

			return "\n" . implode( "\n", $quoted ) . "\n\n";

		}, $html );

		$html = preg_replace_callback( '/<pre[^>]*><code[^>]*>(.*?)<\/code><\/pre>/is', function ( $matches ) {

			return "\n```\n" . html_entity_decode( strip_tags( $matches[1] ) ) . "\n```\n\n";

		}, $html );

		$html = preg_replace( '/<code[^>]*>(.*?)<\/code>/i', "`$1`", $html );

		$html = strip_tags( $html );

		$html = html_entity_decode( $html, ENT_QUOTES, 'UTF-8' );

		$html = preg_replace( '/\n\s*\n\s*\n/', "\n\n", $html );

		return trim( $html );

	} )( $post_content );

	$markdown_content .= $post_content;

	// Add permalink at the end

	$markdown_content .= "\n\n---\n\n";

	$markdown_content .= '**Original URL:** ' . get_permalink( $post ) . "\n";

	// Set proper headers

	nocache_headers();

	header( 'Content-Type: text/plain; charset=UTF-8' );

	header( 'Content-Disposition: inline; filename="' . sanitize_file_name( $post->post_name ) . '.md"' );

	// Output the markdown content

	echo $markdown_content;

	// Stop WordPress from processing further

	exit;

} );