Expected:
When we compare Грещките and Грешките we should get the following HTML code for the deleted part:
Гре<del>щ</del>ките
However, we get:
Гре�<del>�</del>ките
WP_Text_Diff_Renderer_inline::_splitOnWords() uses the following regular expression to split words: /([^\w])/. \w in this case matches [a-zA-Z0-9_] and everything else is outside of a word. This both isn't a good definition of a word and allows a word to end in the middle of a UTF-8 character, which is the case above.
The solution is to make the regular expression work on a UTF-8 string, using the /u modifier (available from PHP 4.1.0).
Patch attached.
