Mahara

DOMDocument::loadHTML() expecting ';'

Series 23.04
Bug #1997291

Bug #1997291 reported by Gold on 2022-11-22

This bug affects 1 person

Affects		Status	Importance	Assigned to	Milestone
	Mahara
	23.04	In Progress	High	Unassigned	Mahara 24.04.0

Bug Description

While running /lib/cron.php I noticed a lot of PHP Warnings on the next page load. These are also in the error log.

My current suspicion is that these are triggered when trying to send e-mail about forum activity.

The actual error:

DOMDocument::loadHTML(): htmlParseEntityRef: expecting ';' in Entity.

While this is a PHP Warning it isn't causing crashes. However, it will be filling up error logs and may be causing unexpected behaviour in other places.

This error is occurring whenever html2text() is called. This is calling HtmltoText which calls DOMDocument and this is where the error happens. When DOMDocument::loadHTML() is called the errors are thrown whenever a non-encoded ampersand is found in the document. i.e. & rather than &

Showing the error in an interactive shell:

php > # Example 1:
php > $s = '<head><meta http-equiv="Content-Type" content="text/html; charset=utf-8"/></head>Forum topic<img width="1024" height="" style="" alt="body_fire.jpg" src="https://dev.mahara.local/artefact/file/download.php?file=193&embedded=1&group=1&topic=1&post=1">';
php >
php > # Example 2 is to demonstrate a working version of the string:
php > $t = '<head><meta http-equiv="Content-Type" content="text/html; charset=utf-8"/></head>Forum topic<img width="1024" height="" style="" alt="body_fire.jpg" src="https://dev.mahara.local/artefact/file/download.php?file=193&embedded=1&group=1&topic=1&post=1">';
php >
php > $doc = new DOMDocument;
php > $doc->loadHTML($s);
PHP Warning: DOMDocument::loadHTML(): htmlParseEntityRef: expecting ';' in Entity, line: 2 in php shell code on line 1
PHP Warning: DOMDocument::loadHTML(): htmlParseEntityRef: expecting ';' in Entity, line: 2 in php shell code on line 1
PHP Warning: DOMDocument::loadHTML(): htmlParseEntityRef: expecting ';' in Entity, line: 2 in php shell code on line 1
PHP Warning: DOMDocument::loadHTML(): htmlParseEntityRef: expecting ';' in Entity, line: 2 in php shell code on line 1
php > $doc->loadHTML($t);
php >

The examples I've been finding are in interaction_forum_post's with images in them.

The specific code that causes this to come about looks to be in in prepare_post_body() in htdocs/interaction/forum/lib.php. This is explicitly stripping out & and leaving just the & character in any tags that have a call to download.php when a post is saved.

Revision history for this message

Gold (gold.catalyst) wrote on 2022-11-22:

Example of the error when presented on screen. Edit (154.9 KiB, image/png)

Gold (gold.catalyst) on 2022-11-22

Changed in mahara:
milestone:	none → 23.04.0
milestone:	23.04.0 → 22.10.1

Revision history for this message

Mahara Bot (dev-mahara) wrote on 2023-02-21: A patch has been submitted for review

Patch for "main" branch: https://reviews.mahara.org/c/mahara/+/13797

Revision history for this message

Kristina Hoeppner (kris-hoeppner) wrote on 2023-03-13:

To replicate the issue:

1. Set up a group and add at least two members.
2. Make sure that everyone is subscribed to the standard forum. That should be the default setting.
3. Create a forum post that includes an embedded image. You can also attach an image to verify that this part is still working. However, since the & is involved, it's the image appearing in TinyMCE that is in question.
4. Run the cron and then refresh the page.

Results:
- Expected: The cron runs through without warnings and an email notification is sent to the subscribed people.
- Error messages referring to the DOM are displayed. The email is sent but doesn't display the image nor an active link to it, just a reference. The notification in the Mahara inbox has something like '![filename]'

Robert Lyon (robertl-9) on 2023-04-30