Description Thomas Becker 2012-01-24 16:42:57 EST

Build Identifier: 

There's two method implementations there:

1st is "UrlEncoded.decodeUtf8To(byte[] raw,int offset, int length, MultiMap map,Utf8StringBuilder buffer)" which does a:

                catch(NotUtf8Exception e)
                {
                    LOG.warn(e.toString());
                    LOG.debug(e);
                }

Thus invalid chars get replaced and the exception is ignored. 

2nd is "decodeUtf8To(InputStream in, MultiMap map, int maxLength, int maxKeys)" which doesn't catch the exception and thus throws it up the stack.

We should either reject URLs containing invalid utf8 chars or always replace them.

Reproducible: Always