log4j prints wrong characters

问题:

Somebody reports me that the program I gave him that uses log4j doesn't correctly print characters. He tells me that "é" gets printed in the file as "é" (for example: "Vidéo" becomes "Vidéo").

It's probably some encoding issue, but I like to reproduce problems to prove that it's fixed. I was unable to find good (and short) documentation on the subject so:

1. What causes this problem (and how does log4j chose the encoding?)?
2. Can it be fixed by simply using "log4j.appender.myappender.encoding=UTF-8" ?

Thank you for the help!

回答1:

WriterAppender (which is the base class for FileAppender and its variants. Has a setEcoding method. So yes: using log4j.appender.myappender.encoding=UTF-8 should simply work.

Note, however, that "Vidéo" becoming "Vidéo" looks like it is writing UTF-8, but whatever you use to view the file interprets it as some other encoding (usually that's some ISO-8859-* encoding or one of the ISO-derivatives).

Ã is U+00C3 and © is U+00A9. They are encoded as 0xC3 and 0xA9 in ISO-8859-1.

é is U+00E9 which is encoded as 0xC3 0xA9 in UTF-8.

回答2:

If your user is viewing the log files over SSH, then they need to tell their SSH client to use UTF-8 too.