mkht.php/mkht.php

#!/usr/bin/env php
<?php
if (php_sapi_name() !== 'cli')
	exit('Must be run from CLI.' . PHP_EOL);
umask(0022);
const LF = "\n";

define('ROOT', dirname(realpath($_SERVER['SCRIPT_FILENAME'])));

if (!extension_loaded('tidy'))
	echo 'PHP tidy extension unavailable. Feature disabled.' . PHP_EOL;

foreach ($argv as $arg) {
	if ($arg === '-f')
		$opt['force'] = true;
	else
		$args[] = $arg;
}
$opt['force'] ??= false;

$site = $args[1] ?? getcwd();
define('SITE', str_ends_with($site, '/') ? $site : $site . '/');

define('DESTINATION', $args[2] ?? 'dns');

if (file_exists(SITE . 'config.ini'))
	$config = parse_ini_file(SITE . 'config.ini');

$config['header'] ??= false;
$config['author'] ??= NULL;
$config['base-url'] ??= [];
$config['center-index'] ??= false;
$config['default-lang'] ??= NULL;
$config['announce-css'] ??= false;
$config['announce-feed'] ??= false;
$config['gzip'] ??= false;

if (!isset($config['id'])) {
	$config['id'] = bin2hex(random_bytes(32));
	file_put_contents(SITE . 'config.ini', 'id = "' . $config['id'] . '"' . LF, FILE_APPEND);
}

foreach (['pandoc', 'gzip', 'po4a'] as $command) {
	if ($command === 'po4a' && file_exists(SITE . 'po4a.cfg') === false)
		continue;
	if ($command === 'gzip' && !$config['gzip'])
		continue;
	exec('command -v ' . $command, result_code: $code);
	if ($code !== 0)
		exit($command . ' command not available.' . PHP_EOL);
}

// Determine whether links need to use Onion or DNS
function clearnetOrOnion($clearnet_url, $onion_url) {
	return (DESTINATION === 'onion') ? $onion_url : $clearnet_url;
}

$dirs_last_update = [];
$pages_langs = [];

$nodes_src = new RecursiveIteratorIterator(new RecursiveDirectoryIterator(SITE, RecursiveDirectoryIterator::SKIP_DOTS));

foreach($nodes_src as $node) {
	$node_info = new SplFileInfo($node->getPathName());
	$src = $node_info->getPathname();
	if (str_starts_with($src, SITE . 'target/') OR str_starts_with($src, SITE . 'cache/'))
		continue;
	$target = str_replace(SITE, SITE . 'cache/', $src);

	$path_parts_src = pathinfo($src);
	$path_parts_target = pathinfo($target);
	if (preg_match('#/\.(?!htaccess|well-known)#', $src) !== 0) // Skip hidden nodes other than .htaccess and .well-known
		continue;
	if ($node_info->getType() !== 'file')
		continue;
	if (in_array('draft', explode('.', $path_parts_target['basename']), true))
		continue;
	if (!file_exists($path_parts_target['dirname'])) // Create parent directory if needed
		mkdir($path_parts_target['dirname'], 0755, true);
	copy($src, $target);
}

if (file_exists(SITE . 'po4a.cfg'))
	exec('po4a --destdir ' . SITE . ' --srcdir ' . SITE . ' ' . SITE . 'po4a.cfg');

$nodes_cache = new RecursiveIteratorIterator(new RecursiveDirectoryIterator(SITE . 'cache/', RecursiveDirectoryIterator::SKIP_DOTS));

foreach($nodes_cache as $node) {
	$node_info = new SplFileInfo($node->getPathName());
	$src = $node_info->getPathname();
	if (str_starts_with($src, SITE . 'target/'))
		continue;
	$target = str_replace(SITE . 'cache/', SITE . 'target/', $src);

	$path_parts_src = pathinfo($src);
	$path_parts_target = pathinfo($target);
	if (preg_match('#/\.(?!htaccess|well-known)#', $src) !== 0) // Skip hidden nodes other than .htaccess and .well-known
		continue;
	if ($node_info->getType() !== 'file')
		continue;
	if (in_array('draft', explode('.', $path_parts_target['basename']), true))
		continue;
	if (!file_exists($path_parts_target['dirname'])) // Create parent directory if needed
		mkdir($path_parts_target['dirname'], 0755, true);
	copy($src, $target);
	if ($node_info->getExtension() !== 'md')
		continue;

	preg_match('/(?<pagename>[-\w]+)(?:\.(?<lang>[a-z]{2}))?\.md/', $path_parts_src['basename'], $matches);
	if ($matches['pagename'] === strtoupper($matches['pagename'])) // Ignore uppercase-only filenames (like README.md)
		continue;

	$pages[] = $node;

	$lang = $matches['lang'] ?? $config['default-lang'] ?? exit('no language found for ' . $src . LF);

	$pages_langs[$path_parts_src['dirname'] . '/' . $matches['pagename']][] = $lang;
	$files_langs[$src] = $lang;
	$page_names[$src] = $matches['pagename'];
}

foreach ($pages as $node) {
	$node_info = new SplFileInfo($node->getPathName());
	$src = $node_info->getPathname();
	$target = str_replace(SITE . 'cache/', SITE . 'target/', $src);
	$path_parts_src = pathinfo($src);
	$path_parts_target = pathinfo($target);

	$base_filepath_src = $path_parts_src['dirname'] . '/' . $path_parts_src['filename'];
	$base_filepath_target = $path_parts_target['dirname'] . '/' . $path_parts_target['filename'];

	$content = file_get_contents($src);
	$lang = $files_langs[$src];

	preg_match('/^# (?<title>.*)$/Dm', $content, $matches);
	$title = $matches['title'] ?? NULL;

	if (!file_exists($base_filepath_target . '.html') OR (filemtime($src) > filemtime($base_filepath_target . '.html')) OR $opt['force']) {
		echo 'Compiling ' . $src . ' ' . date("Y-m-d H:i:s", $node_info->getMTime()) . LF;

		// Execute PHP code
		ob_start();
		eval('?>' . $content);
		$content = ob_get_clean();

		// Convert Gemtext to Markdown
		if ($path_parts_target['extension'] === 'gmi' OR $path_parts_target['extension'] === 'md') {
			$content = preg_replace_callback(
				'/^=>\h*(?<addr>\S+)(:?\h+(?<title>\V+))?$/m',
				function ($matches) {
					if (!str_contains($matches['addr'], ':') AND str_ends_with($matches['addr'], '.gmi'))
						$matches['addr'] = substr($matches['addr'], 0, -3) . 'md';
					return '[' . ($matches['title'] ?? $matches['addr']) . '](' . $matches['addr'] . ')';
				},
				$content,
			);
		}

		// Compile Markdown to HTML
		$process = proc_open('pandoc --fail-if-warnings --section-divs -f markdown-citations-native_divs-native_spans+abbreviations+hard_line_breaks+lists_without_preceding_blankline+multiline_tables+fenced_divs+bracketed_spans+markdown_attribute -t html --wrap none', [
			0 => ['pipe', 'r'],
			1 => ['pipe', 'w'],
		], $pipes);
		if (is_resource($process) !== true)
			exit('Can\'t spawn pandoc.' . PHP_EOL);
		fwrite($pipes[0], $content);
		fclose($pipes[0]);
		$content = stream_get_contents($pipes[1]);
		fclose($pipes[1]);
		if (proc_close($process) !== 0)
			exit('pandoc failed.' . PHP_EOL);

		// .md > .html for local links
		$content = preg_replace('/ href="([^:"]+)\.md"/', ' href="$1.html"', $content);

		$relative_root_path = str_repeat('../', substr_count(str_replace(SITE . 'target', '', $path_parts_target['dirname']), '/'));

		ob_start();

?>
<!DOCTYPE html>
<html lang="<?= $lang ?>">
	<head>
		<meta charset="utf-8">
<?php
		if (isset($title) AND isset($config['title']))
			echo '<title>' . $title . ' · ' . $config['title'] . '</title>';
		else if (isset($title))
			echo '<title>' . $title . '</title>';
		else if (isset($config['title']))
			echo '<title>' . $config['title'] . '</title>';
?>
		<meta name="viewport" content="width=device-width, initial-scale=1">
		<meta name="referrer" content="no-referrer">
<?php
		if (isset($config['author']))
			echo '		<meta name="author" content="' . $config['author'] . '">';
?>
		<link rel="canonical" href="<?= ($page_names[$src] === 'index') ? './' : $page_names[$src] ?>">
<?php
		if ($config['announce-feed'])
			echo '		<link rel="alternate" type="application/atom+xml" href="' . $relative_root_path . 'feed.atom">' . LF;

		if ($config['announce-css']) {
			if (file_exists(SITE . 'style.css'))
				echo '		<link rel="stylesheet" media="screen" href="' . $relative_root_path . 'style.css">' . LF;
			echo '		<link rel="stylesheet" media="screen" href="' . $relative_root_path . 'mkht-php.css">' . LF;
		}

		if (file_exists(SITE . 'head.inc.html'))
			echo file_get_contents(SITE . 'head.inc.html');
?>
	</head>

	<body>
<?php

		$page_langs = $pages_langs[$path_parts_src['dirname'] . '/' . $page_names[$src]];
		if (count($page_langs) > 1) {
			sort($page_langs);
			$lang_txt = [
				'fr' => 'français',
				'en' => 'english',
			];
?>
		<aside>
			<nav>
				<ul>
			<?php foreach ($page_langs as $page_lang) { ?>
					<li lang="<?= $page_lang ?>">
						<small>
							<a href="<?= $page_names[$src] ?>.<?= $page_lang ?>" rel="alternate" hreflang="<?= $page_lang ?>">
								<?= $lang_txt[$page_lang] ?? $page_lang ?>
							</a>
						</small>
					</li>
			<?php } ?>
				</ul>
			</nav>
		</aside>
<?php
		}

		if (file_exists(SITE . 'header.inc.php'))
			eval('?>' . file_get_contents(SITE . 'header.inc.php'));

		if ($config['center-index'] AND $path_parts_target['filename'] === 'index')
			echo '<div class="centered">' . $content . '</div>';
		else
			echo '<main>' . $content . '</main>';
		if (file_exists(SITE . 'end.inc.html'))
			require SITE . 'end.inc.html';
		echo '</body></html>';

		$content = ob_get_clean();

		if (extension_loaded('tidy')) {
			$content = tidy_repair_string($content, [
				'indent' => true,
				'indent-spaces' => 4,
				'output-xhtml' => true,
				'wrap' => 0,
				'hide-comments' => true,
			]);
			$content = str_replace('    ', '	', $content);
		}

		file_put_contents($base_filepath_target . '.html', $content);

		if ($config['gzip'])
			exec('gzip --keep --fast --force ' . $base_filepath_target . '.html');
	}

	if ($page_names[$src] === 'index')
		continue;

	$relative_addr = str_replace(SITE . 'cache/', '', $base_filepath_src . '.html');

	// As of RFC 3151: A URN Namespace for Public Identifiers
	$public_id = 'urn:publicid:' . $config['id'] . str_replace('/', '%2F', $relative_addr);

	preg_match('#\<body\>(?<content>.*)\</body\>#s', file_get_contents($base_filepath_target . '.html'), $match);
	$atom_entry_content = $match['content'];

	// Make relative links absolute
	$atom_entry_content = preg_replace_callback('# href=\"(?<relative_url>[^:"]+)\"#', function ($matches) {
		global $config;
		global $path_parts_target;

		return ' href="' . ($config['base-url'][0] ?? '') . substr($path_parts_target['dirname'], strlen(SITE)) . '/' . $matches['relative_url'] . '"';
	}, $atom_entry_content);

	ob_start();
?>
	<entry>
		<title><?= $title ?></title>
		<id><?= $public_id ?></id>
		<updated><?= date('c', $node_info->getMTime()) ?></updated>
<?php
	foreach ($config['base-url'] as $base_url)
		echo '		<link rel="alternate" type="text/html" href="' . $base_url . $relative_addr . '"></link>' . LF;
?>
		<content type="html"><?= htmlspecialchars($atom_entry_content) ?></content>
	</entry>
<?php
	$atom_entry = ob_get_clean();

	foreach (explode('/', $relative_addr) as $level => $unused) {
		$dir = implode('/', array_slice(explode('/', $relative_addr), 0, $level)) . '/';
		$last_update = $node_info->getMTime();

		$dirs_entries_per_lang['all'][$dir] = ($dirs_entries_per_lang['all'][$dir] ?? '') . $atom_entry;
		$dirs_entries_per_lang[$lang][$dir] = ($dirs_entries_per_lang[$lang][$dir] ?? '') . $atom_entry;

		if (!isset($dirs_last_update['all'][$dir]) OR $last_update > $dirs_last_update['all'][$dir])
			$dirs_last_update['all'][$dir] = $last_update;
		if (!isset($dirs_last_update[$lang][$dir]) OR $last_update > $dirs_last_update[$lang][$dir])
			$dirs_last_update[$lang][$dir] = $last_update;
	}
}

foreach ($dirs_entries_per_lang as $lang => $dirs_entries) {
	foreach ($dirs_entries as $dir => $entries) {
		ob_start();
?>
<?xml version="1.0" encoding="UTF-8"?>
<feed xmlns="http://www.w3.org/2005/Atom">
	<title><?= ($config['title'] ?? '') . (($dir !== '/') ? ' ' . $dir : '') ?></title>
	<id>urn:publicid:<?= $config['id'] . str_replace('/', '%2F', $dir) ?></id>
<?php
	foreach ($config['base-url'] as $url)
		echo '	<link rel="self" type="application/atom+xml" href="' . $url . $dir . 'feed.' . $lang . '.atom"></link>' . LF;
?>
	<updated><?= date('c', $dirs_last_update[$lang][$dir]) ?></updated>
	<author>
		<name><?= $config['author'] ?? '' ?></name>
	</author>
<?php
		file_put_contents(SITE . 'target/' . $dir . 'feed.' . $lang . '.atom', ob_get_clean() . $entries . '</feed>' . LF);
	}
}

if ($config['announce-css'])
	copy(ROOT . '/style.css', SITE . 'target/mkht-php.css');