Hi,
I’m investigating a text editing issue in Unreal Engine and would like to understand whether this behavior is expected according to Unicode/ICU rules or if it might be an engine issue.
Environment
- Unreal Engine 5.2.1
- EditableText / chat input field
- Windows
Tested code points
| Character | Code Point | General Category |
|---|---|---|
| | U+0600 | Cf |
| | U+0601 | Cf |
| | U+0602 | Cf |
| | U+0603 | Cf |
| | U+0604 | Cf |
| | U+0605 | Cf |
| | U+06DD | Po |
Observed behavior
When entering multiple instances of these characters:
Plain Text
or
Plain Text
the caret cannot be placed between individual characters.
Backspace also treats the sequence as a single editing unit rather than allowing what appears to be normal character-by-character deletion.
Comparison tests
ZWJ (U+200D)
Plain Text
AAAAA
Results:
- Caret movement works as expected
- Backspace works as expected
- The character counter includes U+200D
For example:
Plain Text
AAAAA
is counted as 9 characters rather than 5.
Combining Mark (U+0301)
Plain Text
ÁÁÁÁÁ
Results:
- Caret movement works as expected
- Backspace works as expected
Emoji
Plain Text
![]()
Results:
- Appears to be counted as two UTF-16 code units
- Editing behavior appears normal
Notes
Based on the tests above:
- U+200D (Cf) behaves normally
- Combining character sequences behave normally
- Only U+0600–U+0605 and U+06DD exhibit this editing behavior
Because of this, I suspect the behavior may be related to:
- ICU BreakIterator
- Character Boundary detection
- Grapheme Cluster generation
- Slate / EditableText text editing logic
rather than being a general issue with all Cf characters.
Question
Is this behavior expected according to Unicode/ICU text boundary rules?
Has anyone encountered similar behavior with U+0600–U+0605 or U+06DD in Unreal Engine, or knows which part of the text handling pipeline would be responsible for this?
I am not very familiar with Unicode or ICU internals, so any explanation or pointers would be greatly appreciated.
Thanks!