.
*
*/
namespace Friendica\Content\Text;
use Friendica\Core\System;
use Friendica\DI;
use Friendica\Model\Contact;
/**
* Friendica-specific usage of Markdown
*/
class Markdown
{
/**
* Converts a Markdown string into HTML. The hardwrap parameter maximizes
* compatibility with Diaspora in spite of the Markdown standard.
*
* @param string $text
* @param bool $hardwrap
* @return string
* @throws \Exception
*/
public static function convert($text, $hardwrap = true) {
$stamp1 = microtime(true);
$MarkdownParser = new MarkdownParser();
$MarkdownParser->code_class_prefix = 'language-';
$MarkdownParser->hard_wrap = $hardwrap;
$MarkdownParser->hashtag_protection = true;
$MarkdownParser->url_filter_func = function ($url) {
if (strpos($url, '#') === 0) {
$url = ltrim($_SERVER['REQUEST_URI'], '/') . $url;
}
return $url;
};
$text = self::convertDiasporaMentionsToHtml($text);
$html = $MarkdownParser->transform($text);
DI::profiler()->saveTimestamp($stamp1, "parser", System::callstack());
return $html;
}
/**
* Replace Diaspora-style mentions in a text since they trip the Markdown parser autolinker.
*
* @param string $text
* @return string
*/
private static function convertDiasporaMentionsToHtml(string $text)
{
return preg_replace_callback(
'/([@!]){(?:([^}]+?); ?)?([^} ]+)}/',
/*
* Matching values for the callback
* [1] = mention type (@ or !)
* [2] = name (optional)
* [3] = profile URL
*/
function ($matches) {
if ($matches[3] == '') {
return '';
}
$data = Contact::getDetailsByAddr($matches[3]);
if (empty($data)) {
return '';
}
$name = $matches[2];
if ($name == '') {
$name = $data['name'];
}
return $matches[1] . '' . $name . '';
},
$text
);
}
/*
* we don't want to support a bbcode specific markdown interpreter
* and the markdown library we have is pretty good, but provides HTML output.
* So we'll use that to convert to HTML, then convert the HTML back to bbcode,
* and then clean up a few Diaspora specific constructs.
*/
public static function toBBCode($s)
{
$s = html_entity_decode($s, ENT_COMPAT, 'UTF-8');
// The parser cannot handle paragraphs correctly
$s = str_replace(['', '', '
'], ['
', '
', '
'], $s);
// Escaping hashtags that could be titles
$s = preg_replace('/^\#([^\s\#])/im', '\#$1', $s);
$s = self::convert($s);
$s = HTML::toBBCode($s);
// protect the recycle symbol from turning into a tag, but without unescaping angles and naked ampersands
$s = str_replace('♲', html_entity_decode('♲', ENT_QUOTES, 'UTF-8'), $s);
// Convert everything that looks like a link to a link
$s = preg_replace('/([^\]=]|^)(https?\:\/\/)([a-zA-Z0-9:\/\-?&;.=_~#%$!+,@]+(?