Ty Clifford
240 lines
9.0 KiB
PHP
240 lines
9.0 KiB
PHP
<?php
|
|
declare(strict_types=1);
|
|
|
|
namespace NeonBlog;
|
|
|
|
final class Markdown
|
|
{
|
|
public static function toHtml(string $markdown, bool $executePhp = true): string
|
|
{
|
|
$markdown = str_replace(["\r\n", "\r"], "\n", $markdown);
|
|
if ($executePhp) {
|
|
$markdown = self::executePhp($markdown);
|
|
}
|
|
$markdown = Shortcodes::render($markdown);
|
|
$markdown = trim($markdown);
|
|
if ($markdown === '') {
|
|
return '';
|
|
}
|
|
|
|
$lines = explode("\n", $markdown);
|
|
$html = [];
|
|
$count = count($lines);
|
|
$i = 0;
|
|
|
|
while ($i < $count) {
|
|
$line = rtrim($lines[$i]);
|
|
if (trim($line) === '') {
|
|
$i++;
|
|
continue;
|
|
}
|
|
|
|
if (self::startsHtmlBlock($line)) {
|
|
[$block, $i] = self::consumeHtmlBlock($lines, $i, $count);
|
|
$html[] = $block;
|
|
continue;
|
|
}
|
|
|
|
if (preg_match('/^```([A-Za-z0-9_-]*)\s*$/', $line, $matches)) {
|
|
$language = $matches[1] !== '' ? ' class="language-' . self::e($matches[1]) . '"' : '';
|
|
$code = [];
|
|
$i++;
|
|
while ($i < $count && !preg_match('/^```\s*$/', rtrim($lines[$i]))) {
|
|
$code[] = $lines[$i];
|
|
$i++;
|
|
}
|
|
$html[] = '<pre><code' . $language . '>' . self::e(implode("\n", $code)) . '</code></pre>';
|
|
$i++;
|
|
continue;
|
|
}
|
|
|
|
if (preg_match('/^(#{1,6})\s+(.+)$/', $line, $matches)) {
|
|
$level = strlen($matches[1]);
|
|
$html[] = '<h' . $level . '>' . self::inline($matches[2]) . '</h' . $level . '>';
|
|
$i++;
|
|
continue;
|
|
}
|
|
|
|
if (preg_match('/^(-{3,}|\*{3,}|_{3,})$/', trim($line))) {
|
|
$html[] = '<hr>';
|
|
$i++;
|
|
continue;
|
|
}
|
|
|
|
if (preg_match('/^\s*>\s?(.*)$/', $line)) {
|
|
$quote = [];
|
|
while ($i < $count && preg_match('/^\s*>\s?(.*)$/', rtrim($lines[$i]), $matches)) {
|
|
$quote[] = $matches[1];
|
|
$i++;
|
|
}
|
|
$html[] = '<blockquote>' . self::toHtml(implode("\n", $quote), false) . '</blockquote>';
|
|
continue;
|
|
}
|
|
|
|
if (preg_match('/^\s*[-*+]\s+(.+)$/', $line)) {
|
|
$items = [];
|
|
while ($i < $count && preg_match('/^\s*[-*+]\s+(.+)$/', rtrim($lines[$i]), $matches)) {
|
|
$items[] = '<li>' . self::inline($matches[1]) . '</li>';
|
|
$i++;
|
|
}
|
|
$html[] = '<ul>' . implode('', $items) . '</ul>';
|
|
continue;
|
|
}
|
|
|
|
if (preg_match('/^\s*\d+\.\s+(.+)$/', $line)) {
|
|
$items = [];
|
|
while ($i < $count && preg_match('/^\s*\d+\.\s+(.+)$/', rtrim($lines[$i]), $matches)) {
|
|
$items[] = '<li>' . self::inline($matches[1]) . '</li>';
|
|
$i++;
|
|
}
|
|
$html[] = '<ol>' . implode('', $items) . '</ol>';
|
|
continue;
|
|
}
|
|
|
|
$paragraph = [$line];
|
|
$i++;
|
|
while ($i < $count && trim($lines[$i]) !== '' && !self::startsBlock(rtrim($lines[$i]))) {
|
|
$paragraph[] = trim($lines[$i]);
|
|
$i++;
|
|
}
|
|
$html[] = '<p>' . self::inline(implode(' ', $paragraph)) . '</p>';
|
|
}
|
|
|
|
return implode("\n", $html);
|
|
}
|
|
|
|
public static function excerpt(string $markdown, int $words = 28): string
|
|
{
|
|
$markdown = preg_replace('/<\?(?:php|=)?[\s\S]*?\?>/i', ' ', $markdown) ?? $markdown;
|
|
$html = self::toHtml($markdown, false);
|
|
$html = preg_replace('#<(script|style)\b[^>]*>.*?</\1>#is', ' ', $html) ?? $html;
|
|
$text = trim(preg_replace('/\s+/', ' ', strip_tags($html)) ?? '');
|
|
if ($text === '') {
|
|
return '';
|
|
}
|
|
|
|
$parts = preg_split('/\s+/', $text) ?: [];
|
|
if (count($parts) <= $words) {
|
|
return $text;
|
|
}
|
|
|
|
return implode(' ', array_slice($parts, 0, $words)) . '...';
|
|
}
|
|
|
|
private static function startsBlock(string $line): bool
|
|
{
|
|
return self::startsHtmlBlock($line)
|
|
|| (bool) preg_match('/^(#{1,6}\s+|```|^\s*[-*+]\s+|^\s*\d+\.\s+|^\s*>\s?|(-{3,}|\*{3,}|_{3,})$)/', $line);
|
|
}
|
|
|
|
private static function inline(string $text): string
|
|
{
|
|
$placeholders = [];
|
|
$text = preg_replace_callback('/`([^`]+)`/', static function (array $matches) use (&$placeholders): string {
|
|
$key = "\x1A" . count($placeholders) . "\x1A";
|
|
$placeholders[$key] = '<code>' . self::e($matches[1]) . '</code>';
|
|
return $key;
|
|
}, $text) ?? $text;
|
|
|
|
$text = preg_replace_callback('/<\/?[A-Za-z][A-Za-z0-9:-]*(?:\s+[^<>]*)?>|<!--.*?-->|<![A-Z][^>]*>/s', static function (array $matches) use (&$placeholders): string {
|
|
$key = "\x1A" . count($placeholders) . "\x1A";
|
|
$placeholders[$key] = $matches[0];
|
|
return $key;
|
|
}, $text) ?? $text;
|
|
|
|
$text = self::e($text);
|
|
$text = preg_replace_callback('/!\[([^\]]*)\]\(([^)\s]+)(?:\s+"([^"]+)")?\)/', static function (array $matches): string {
|
|
$title = isset($matches[3]) ? ' title="' . self::e($matches[3]) . '"' : '';
|
|
return '<img src="' . self::e($matches[2]) . '" alt="' . self::e($matches[1]) . '"' . $title . '>';
|
|
}, $text) ?? $text;
|
|
$text = preg_replace_callback('/\[([^\]]+)\]\(([^)\s]+)(?:\s+"([^"]+)")?\)/', static function (array $matches): string {
|
|
$title = isset($matches[3]) ? ' title="' . self::e($matches[3]) . '"' : '';
|
|
return '<a href="' . self::e($matches[2]) . '"' . $title . '>' . $matches[1] . '</a>';
|
|
}, $text) ?? $text;
|
|
$text = preg_replace('/\*\*([^*]+)\*\*/', '<strong>$1</strong>', $text) ?? $text;
|
|
$text = preg_replace('/\*([^*]+)\*/', '<em>$1</em>', $text) ?? $text;
|
|
|
|
return strtr($text, $placeholders);
|
|
}
|
|
|
|
private static function executePhp(string $source): string
|
|
{
|
|
if (!str_contains($source, '<?')) {
|
|
return $source;
|
|
}
|
|
|
|
$placeholders = [];
|
|
$source = preg_replace_callback('/```[\s\S]*?```/', static function (array $matches) use (&$placeholders): string {
|
|
$key = "\x1BPHPBLOCK" . count($placeholders) . "\x1B";
|
|
$placeholders[$key] = $matches[0];
|
|
return $key;
|
|
}, $source) ?? $source;
|
|
$source = preg_replace_callback('/`[^`\n]*<\?[^`\n]*`/', static function (array $matches) use (&$placeholders): string {
|
|
$key = "\x1BPHPBLOCK" . count($placeholders) . "\x1B";
|
|
$placeholders[$key] = $matches[0];
|
|
return $key;
|
|
}, $source) ?? $source;
|
|
|
|
ob_start();
|
|
try {
|
|
eval('?>' . $source);
|
|
return strtr((string) ob_get_clean(), $placeholders);
|
|
} catch (\Throwable $error) {
|
|
ob_end_clean();
|
|
return strtr($source, $placeholders) . "\n\n<!-- PHP content error: " . self::e($error->getMessage()) . " -->";
|
|
}
|
|
}
|
|
|
|
private static function startsHtmlBlock(string $line): bool
|
|
{
|
|
$line = trim($line);
|
|
return (bool) preg_match('/^(<!--|<![A-Z]+|<\/?[A-Za-z][A-Za-z0-9:-]*(\s|>|\/>))/i', $line);
|
|
}
|
|
|
|
/** @param array<int, string> $lines @return array{0: string, 1: int} */
|
|
private static function consumeHtmlBlock(array $lines, int $i, int $count): array
|
|
{
|
|
$block = [rtrim($lines[$i])];
|
|
$trimmed = trim($lines[$i]);
|
|
$i++;
|
|
|
|
if (str_starts_with($trimmed, '<!--')) {
|
|
while ($i < $count && !str_contains(end($block), '-->')) {
|
|
$block[] = rtrim($lines[$i]);
|
|
$i++;
|
|
}
|
|
return [implode("\n", $block), $i];
|
|
}
|
|
|
|
if (!preg_match('/^<([A-Za-z][A-Za-z0-9:-]*)\b/i', $trimmed, $matches)) {
|
|
return [implode("\n", $block), $i];
|
|
}
|
|
|
|
$tag = strtolower($matches[1]);
|
|
$voidTags = ['area', 'base', 'br', 'col', 'embed', 'hr', 'img', 'input', 'link', 'meta', 'param', 'source', 'track', 'wbr'];
|
|
if (in_array($tag, $voidTags, true) || str_ends_with($trimmed, '/>') || stripos($trimmed, '</' . $tag . '>') !== false) {
|
|
return [implode("\n", $block), $i];
|
|
}
|
|
|
|
while ($i < $count) {
|
|
$block[] = rtrim($lines[$i]);
|
|
if (stripos((string) end($block), '</' . $tag . '>') !== false) {
|
|
$i++;
|
|
break;
|
|
}
|
|
if (trim($lines[$i]) === '' && !in_array($tag, ['article', 'aside', 'div', 'footer', 'form', 'header', 'iframe', 'main', 'nav', 'pre', 'script', 'section', 'style', 'table', 'textarea'], true)) {
|
|
$i++;
|
|
break;
|
|
}
|
|
$i++;
|
|
}
|
|
|
|
return [implode("\n", $block), $i];
|
|
}
|
|
|
|
private static function e(string $value): string
|
|
{
|
|
return htmlspecialchars($value, ENT_QUOTES | ENT_SUBSTITUTE, 'UTF-8');
|
|
}
|
|
}
|