-
Bug
-
Resolution: Fixed
-
Major
-
None
-
None
-
None
The ZipArchiveSerializer doesn't support UTF-8 encoded names: when an entry's name has 2-bytes characters (outside ISO-8859 charset, such as accented characters), they are incorrectly encoded.
When opening a zip archive created by this serializer with Ubuntu's file roller archiver, the file inside is named "test document accentué.pdf".
Apparently, Sun's JDK classes ZipOutputStream/ZipEntry are responsible for this behavior.
Apache commons compress library defines classes replacing these ones (ZipArchiveOutputStream and ZipArchiveEntry) which partly resolve this issue (full documentation at http://commons.apache.org/compress/zip.html, section "Encoding").
Here's a replacement proposal for ZipArchiveSerializer, which makes use of commons compress' classes instead of default JDK ones.
It resolves the problem with Ubuntu's file roller and 7-zip file manager on Windows XP, but not with Win XP's "zip as folder".