XOM API for Processing XML with Java

[Wolfgang Hoschek <whoschek AT lbl.gov>]

Thanks for all who responded to the issue.
Turns out the problem was on my side, not in XOM or the parser.

I have test cases for weird corner cases, and was printing debug info 
with System.err.println(doc.toXML()).
System.err.println(str) is fragile, since the PrintStream used by 
System.err uses the platform's default encoding by default, which is 
typically not UTF-8 on MacOSX, but rather "MacRoman".

This ugly work-around retain surrogate chars correctly:

System.setErr(new PrintStream(new FileOutputStream("test.out"), true, 
"UTF-8"));
System.err.println(doc.toXML());

All is fine now :-)
Wolfgang.

On Nov 4, 2004, at 4:10 AM, Elliotte Harold wrote:

> Wolfgang Hoschek wrote:
>
> > Hi,
> > Some characters such as "񐀀" and "'[[[񐀀]]]'" seem to 
> > get turned by XOM to "?" characters.
> > I've seen similar things before when using byte to char conversions, 
> > both with the nio converters and the java.lang.String converters. Any 
> > ideas on what could be done about it, or what I'm doing wrong here?
>
> I've walked through this with a debugger and I'm pretty sure that 
> what's going on here is that you're doing System.out.println or some 
> equivalent and that is turning the surrogate pairs into question 
> marks. Neither XOM nor the XML parser is doing this. I've only tested 
> the toXML case but I assume the serializer is the same until proven 
> otherwise.
>
> If you're using an OutputStreamWriter anywhere that might be doing it. 
> According to the JavaDoc for OutputStreamWriter, "If the character 
> represented by a surrogate pair cannot be encoded by a given charset 
> then a charset-dependent substitution sequence is written to the 
> output stream." Normally this substitution char is a question mark.
>
>
> --
> Elliotte Rusty Harold  elharo AT metalab.unc.edu
> XML in a Nutshell 3rd Edition Just Published!
> http://www.cafeconleche.org/books/xian3/
> http://www.amazon.com/exec/obidos/ISBN=0596007647/cafeaulaitA/ref=nosim