Replace HTMLPurifier with a maintained dependency

[asmecher]

HTMLPurifier is not maintained anymore. Find another replacement toolset for "safe" HTML filtering.

It does not support HTML5: https://forum.pkp.sfu.ca/t/tinymce-stripping-anchor-tags-within-or-around-divs/72749/12

We currently use HTMLPurifier in the PKPString::stripUnsafeHtml function (lib/pkp/classes/core/PKPString.inc.php). This takes a configuration of allowed elements/attributes from config.inc.php.

PRs
pkp-lib --> #9257
ojs --> pkp/ojs#4025 [TEST ONLY]
omp --> pkp/omp#1581 [TEST ONLY]
ops --> pkp/ops#696 [TEST ONLY]
jatsTemplate --> pkp/jatsTemplate#30
oaiJats --> pkp/oaiJats#36

Update
Based on following reasons we have decided to revert back the Symfony HTML Sanitizer and keep using the HTMLPurifier

• Bug in the Symfony html sanitizer
  • [Html Sanitizer] Sanitization remove single < (less than) char which is not part of any tag([Html Sanitizer] Sanitization remove single < (less than) char which is not part of any tag symfony/symfony#57597)
  • Bug actually in the https://github.com/Masterminds/html5-php parser
  • Parser remove the single < (less than) character from given html string
• Also the use of strip_tags will cause problem
  • use of strip_tags omits anything after <=
  • we can overcome it by not using strip_tags and that better
  • Not using the strip_tags will removed any non standard tags and it's inner content, list of w3c standard tags/elements/attributes list based on w3c sanitization standard .

Reverting Back HTMLPurifier
pkp-lib --> #10197
ojs --> pkp/ojs#4365 [TEST ONLY]
omp --> pkp/omp#1634 [TEST ONLY]
ops --> pkp/ops#729 [TEST ONLY]
jatsTemplate --> pkp/jatsTemplate#46
oaiJats --> pkp/oaiJats#46

[touhidurabir]

There are few alternative packages/libraries to choose from as HTMLPurifier is still the most common and widely used one even if it's not maintained properly and there has been no major upgrade of it . Most the of popular ones are also kind of based on that one which make those more like a wrapper over it . Few available options are (based on initial R&D)

1. xemlock/htmlpurifier-html5 . This one also mentioned in the pkp forum at https://forum.pkp.sfu.ca/t/tinymce-stripping-anchor-tags-within-or-around-divs/72749/12 . As this allow compatibility of HTML5, it's still a wrapper over the not maintained HTMLPurifier . However one big advantage of this one is that this does not require much modification of current implementation also any modification or option addition to the config options .

2. tgalopin/html-sanitizer . This one seems relatively new comparing to HTMLPurifier (first release at 2018) and seems to be maintained regularly . It has much better options and compatibility for HTML5 also have better implementation. This one also available as Symfony Bundle so it may have high chance to be maintained properly . The only downside of this one is to make it work with our current config option implementation which may need more parsing before using with this package or we may need to add more config option in the config file . Need more research .

3. symfony/html-sanitizer . This one maintained by Symfony Team and only released with Symfony 6 . This one has a very fluent API to work with and high chance of maintained properly by Symfony Team. But even the initial released version is locked to PHP 8.1+ so it will need to be used with a custom forked version to make it work for the 3.4 release target which aims to work with PHP 8.0+ . And even with that , there may be additional workload the maintain the forked version up to date with future release .

@asmecher @NateWr few stuff we need to figure out first

1. what is the backward compatibility requirement for this enhancement/modification ?
2. up to which PHP version ?
3. How much of config option modification/addition we may allow ? Or should we try to keep the config option exact as same for this enhancement ?

personally I would like to go with option 2 or 3 if we target for a long term supported package but if we just need to make it HTML5 compatible , option 1 may work for now . However , as option 1 is based on old HTMLPurifier , it may not be maintained properly and security issue may arise . Need more research .

[asmecher]

Thanks, @touhidurabir! To summarize our convo today:

• [xemlock/htmlpurifier-html5](https://github.com/xemlock/htmlpurifier-html5) is a fork and I wouldn't normally recommend that approach but if we have a long-term solution that requires a stopgap (e.g. because of too high a PHP baseline) it's worth considering. (I don't see a formal statement anywhere that HTMLPurifier is not being maintained, but the release schedule sure doesn't look good.)
• [tgalopin/html-sanitizer](https://github.com/tgalopin/html-sanitizer) looks good but is a personal repo (does not seem institutionally supported). Not my first choice, though it has promising uptake.
• [symfony/html-sanitizer](https://github.com/symfony/html-sanitizer) -- we both agreed that something adopted/maintained by a big project like Symfony seemed like the best approach. However, as you noted, the PHP8.1 requirement is currently steep.

Decisions

• We don't have a lot of calls for HTMLPurifier to be replaced -- it seems safe/mature enough to continue using, and it's still getting PHP compatibility fixes merged into Github. We can cautiously continue to use it for the time being.
• @touhidurabir will take a quick look to see if we can determine the maintenance roadmap for [symfony/html-sanitizer](https://github.com/symfony/html-sanitizer). We could probably bump our PHP requirements for 3.6 (or possibly even 3.5) to PHP8.1 -- but if [symfony/html-sanitizer](https://github.com/symfony/html-sanitizer) is already on to a newer, more aggressive requirement by then, it may force our community into a difficult position.
• If [symfony/html-sanitizer](https://github.com/symfony/html-sanitizer)'s maintenance roadmap is compatible enough with ours, I'd recommend we consider adopting it later (3.5 or 3.6). This would mean waiting to implement it until that branch exists.

[touhidurabir]

@asmecher if this is not a higher priority right now , we can get back to this one later when we caught up with PHP 8.1+ . Another option is to use a forked version of symfony/html-sanitizer that we make compatible with PHP 8.0+ but even so this package by the Symfony Team is still in BETA state and that why it's still not included in their available component list at https://symfony.com/components . And maintaining a custom forked version may turn into a way more work that it should be.

Now there is no such info about how the future version will integrate newer version of PHP . So we are not sure if this package will have an aggressive PHP version upgrade policy but we can always use a lower release version. For example, if we look at https://symfony.com/releases/6.0 which support 8.0+ but the https://symfony.com/releases/6.1 will require 8.1+ which seems pretty aggressive . Some more general details can be found at https://symfony.com/doc/current/contributing/community/releases.html .

Also current LTS version of Symfony is 5.4 which will be supported for bug fix till 2024 and security fix till 2025 so there is a high chance we may not get a new LTS before 2024. So basically it's almost impossible to determine if we can have this symfony/html-sanitizer package for latest version with PHP version compatible for our next release so we may have to use a lower release version then.

few more details at https://symfony.com/blog/new-in-symfony-6-1-htmlsanitizer-component