It was found that the filter for the trimming of strings though, and declares "Unicode aware trim method", but not that he is "aware trim". If it gets a string ending with the Cyrillic uppercase "R", it deteriorates. Probably because its representation in UTF-8 ends at A0.
Specifically, a problem emerges here in the regular season:
preg_replace('/^[\\s]*|[\\s]*$/sSD', ", $value)
If you add the u modifier, the result is generally NULL.
No such problem is encountered?