I have a bugzilla install set up with English and Japanese templates to support our users in the US and Japan. My users are having a problem with attachments that have Japanese characters in the filename; English filenames are OK.
Say I upload a file called "Blah操作手.doc". When I go to download the file, I get a prompt asking me to save it as "=_UTF-8_Q_Blah=E6=93=8D=E4=BD=9C=E6=89=8B_=.doc". All of the data in the file is preserved, but the filename appears to not be decoded properly. Often, the filename ends up so long that it loses the original extension, and the user gets a page filled with garbage if they try to open the attachment in their browser. This is happening even though the database "filename" field displays the original characters correctly.
This is happening with all types of files that contain non-ASCII characters - doc, xls, pdf, png, txt, etc. I was able to reproduce this on the default templates for Bugzilla 3.6.3, regardless of regional settings or template language.
This is primarily a problem for Internet Explorer. Chrome and Safari have similar problems, but I don't really have any users that use either. Firefox works fine and returns the original filename as expected.
Is there a way I can fix this?
There's a very long answer to this question that somebody else may give you, but the basic problem is that there is no reasonable method of encoding UTF-8 filenames that all browsers support. I suspect that you're using IE, and it doesn't support the standard encoding method that we're using. More or less, this is something that can be worked around in Bugzilla, but we haven't done it yet. There's a bug filed for it at bugzilla.mozilla.org, but I'm not finding it at the moment.