This is a refactoring for a part of the ReactStateXMLParser. I wanted to use functions that are more generic and not just handle a list of edge cases. So I encapsulated the operation that was done in this part of the code to a function `findNodesAndRemoveTheirParentNodes` which is more generic and could be used for different operations.
internal issue: https://2u-internal.atlassian.net/browse/TNL-11311
This fixes a bug where the editor was deleting the OLX tag when editing in the simple editor and then saving.
Also the description was being converted to em for the simple editor, but then not converted back. However, the xblock renders label and then em in reverse order for some reason.
To fix it, the em gets converted back to description now, but not for every em tag (added a class "olx_description" for the tags that should be converted).
The max attempts setting for a problem xblock should be set to null for
course default max attempt setting to take effect. This makes sure that
xblock setting is updated if course default is updated.
This commit reverts to advanced editor when partial_credit attribute is
added to multichoice, single select and numerical problems. Without this
change, the partial_credit attribute is removed from the problem on the
next edit.
The issue had to do with how Firefox handles pasting newlines inside a <pre
contenteditable> tag (TinyMCE's editor works via contenteditable) and
fast-xml-parser's parsing. In Firefox, newlines are converted to <br> when
pasted, while Chrome preserves them. The parser by default trims spaces in text
nodes. In Firefox, the parser creates individual text nodes between the <br>
elements, and those have leading spaces in the example. In Chrome, there are no
<br> elements and the entire content is a single text node as-is. Setting
trimValues to false disables the trimming and resolves this issue in Firefox.
While investigating this, I noticed the builder also mishandles <br /> tags
emitted by the editor, converting them to <br></br>. The unpairedTags option in
the builder ensures they are output correctly as a single tag, and setting
suppressUnpairedNode to false ensures that single tag is <br/> rather than <br>
to remain XML compatible.
While trying to resolve this, I was looking into the paste plugin in TinyMCE.
It changes the behavior of pasting, making it more consistent between Chrome
and Firefox (i.e. both emit <br>) and is incorporated into TinyMCE 6 core.
Unfortunately, it seems to mangle pasting inside a <pre> tag by inserting
redundant nbsp characters (tinymce/tinymce#9017). TinyMCE 6 also outputs <br>
rather than <br /> - adding unpairedTags to the parser options is meant to
handle this, but it does not seem to behave entirely correct
(NaturalIntelligence/fast-xml-parser#609). These should be kept in mind if/when
upgrading to TinyMCE 6 (the different behaviors can be seen easily at
https://fiddle.tiny.cloud).
this creates the ability for explanations to work if they are:
Not a child of a response type
Use an h2 tag for the explanation title
do not use the "Explanation" title.
This change increases the frequency at which odd tags (like shuffle) on multiple choice problems don't go to the visual editor.
Instead, they will now divert to the advanced editor. There might be other places where we need to follow a similar pattern, but we don't know those yet.
