Parsedown class false positive for short tags

[danyj]

I am just reopening this since the post here https://github.com/WordPress/theme-check/issues/64 is over a year old .

@Otto42 how big of a change would it be on theme check side? Minor as in parsedown or ? Asking because parsedown dev refereed https://github.com/erusev/parsedown/issues/469#issuecomment-275927418 me back here and I hate to change their original class because of maintenance issues.

So if either of you guys could make the change it would be great.

Thank you!

[grappler]

~~Please can you explain why you can just the normal tags?~~~

Edit: I see now that the reason is from https://github.com/erusev/parsedown/issues/469 We are working on a new update of the Theme Check plugin which we hope to release within the next 5 months. It is not likely we will be making any changes to the old code. You can test the new version here https://github.com/ernilambar/ns-theme-check. It already contains the short tags test which should be better than before. Let me know that goes.

From the official documentation https://secure.php.net/manual/en/language.basic-syntax.phptags.php

PHP also allows for short open tag <? (which is discouraged since it is only available if enabled using the short_open_tag php.ini configuration file directive, or if PHP was configured with the --enable-short-tags option).

Your code would break if you use short tags and they were disabled on a server.

[danyj]

Your code would break if you use short tags and they were disabled on a server.

no one is using short tags here , this is false positive for regex

tested theme with ns theme check no mention of Parsedown for now

[aidantwoods]

It occurs to me that the simplest (and correct) solution is to just check whether an unclosed <?php tag exists in the document before reporting on the short open tag that the check in question returns? (if there were a short tag inside code that is already PHP, the error would be syntactic, and so the file wouldn't run anyway (worse than a standards error)).

Surely parsing characters in a string like a language construct is a pretty major bug?

[grappler]

Sorry, I came across a bit harsh in my previous comment as I did not initially realize it was a false positive.

As we are planning of releasing a complete rewrite of this plugin it does not make sense to continue to work on the current code.