preg_match in inlineLink() causes segfault

[clphillips]

Here's the culprit:

preg_match('/\[((?:[^][]|(?R))*)\]/', $remainder, $matches)

The following $EmRegex also suffers from this problem:

/^_((?:\\\\_|[^_]|__[^_]*__)+?)_(?!_)\b/us

These regular expressions trigger segfaults on an 11 KB input string of mostly JSON text (~80 chars per line).

See PHP issue #45735.

I believe the regexes could be improved to prevent segfault.

[ifeltsweet]

+1, happens to us often on large documents.

[perryflynn]

I've replaced the regex in my extensions as workaround:

https://github.com/perryflynn/parsedown-extra-extensions/blob/master/src/PerryFlynn/ParsedownExtraExtensions.php#L372

[CasimirEtHippolyte]

To prevent backtracking limit to be reached and to fail faster, the two patterns can be written like this:

preg_match('/\[([^][]*+(?:(?R)[^][]*)*+)]/', $remainder, $matches)

/^_([^\\\\_]*+(?:(?:\\\\.|__[^_]*__)[^\\\\_]*)*+_(?!_)/s

( (?!_) looked redundant with the word-boundary, but you can always replace (?!_) with \b if it's the behaviour you are looking for. In this case don't forget to add the u modifier, but if you keep (?!_) the u modifier is useless.)

[erusev]

Thanks @CasimirEtHippolyte, I'll review these as soon as I have time.