See the attached comparison of original and instrumented code:

unicode_decoding_error.png

A problem is as follows:

• Clover has a UnicodeDecodingReader class which translates 6-letter character sequences "\uNNNN" into a single UTF16 character
• this is used to correctly tokenize code identifiers (IDENT) written in non-ascii characters (a variable with German umlaut for instance)

In this specific example, some "\uNNNN" sequences were translated inside comments, and some characters were "swallowed". So for instance a fragment:

{@code \u}

ends with "java: illegal unicode escape".

Related issues:

• CLOV-1305 - syntax highlighting problem
• CLOV-1131 - Eclipse/IDEA editors get out of sync