Support mixed 'strong' and 'italic' form in the Markdown processor

[Jiab77]

Hello and sorry to disturbing.

If I've understood correctly the code from file /lib/markdown.php, to process strong and italic you're counting the recurrence of '_' and '*' characters but you can't mix them, I mean use **_word_** and get <strong><em>word</em></strong> in return.

So is there a way to support the mixed form?

It took me times to understand the regex used in the code with the website regex101 but I'm lacking skills to support the "mixed" way...

[xfra35]

What you're saying is that, even though the forms ***word*** and ___word___ are supported, the mixed forms *__word__*, **_word_**, _**word**_ and __*word*__ should also be supported. Right?

Then the regex should be modified so that, instead of having \1 matching the first group in direct order, it should match it in reverse order.

Here's a fix proposal:

$str=preg_replace_callback(
    '/(?<=\s|^)(?<!\\\\)([*_])([*_]?)([*_]?)(.*?)(?!\\\\)\3\2\1(?=[\s[:punct:]]|$)/',
    function($expr) {
        if ($expr[3])
            return '<strong><em>'.$expr[4].'</em></strong>';
        if ($expr[2])
            return '<strong>'.$expr[4].'</strong>';
        return '<em>'.$expr[4].'</em>';
    },
    preg_replace(
        '/(?<!\\\\)~~(.*?)(?!\\\\)~~(?=[\s[:punct:]]|$)/',
        '<del>\1</del>',
        $tmp=$str
    )
);

[Jiab77]

@xfra35 yes it's exactly the issue I've tried to describe, glad you understood because sometimes I'm not clear enough... 👍 but I can't say "it should be supported" as it's to the devs to decide what should be supported but it might be really useful for those who use a client side library that don't give the possibility to specify what characters to use as 'strong' and 'italic'.

And this still respect the original Markdown spec that I know you devs want's to follow.

[Jiab77]

@xfra35 I've tested your regex on regex101, it took me times to understand how it works but it does the job ! 🤘 well done!

[exodus4d]

Hint: I had some similar issues recently (parsing MarkDown e.g. from GitHub with F3s Markdown parser).
I replaced the parsing function with a API call to GitHubs API: https://developer.github.com/v3/markdown/ . Maybe this API response could also be used in F3 for validation or testing the F3 parser.

The final result can be seen here https://www.pathfinder-w.space/ . Click the version number in the header.