|
Replies:
2
-
Last Post:
Jul 19, 2009 2:56 AM
by: bratwurst
|
|
|
|
|
|
|
JAXB can generate Illegal XML characters
Posted:
Mar 17, 2009 7:06 PM
|
|
|
Apparently there are some XML characters that are illegal to have in the content area of an XML document, and there is no way to "escape" them to make them legal. For example, 0x1F is an illegal character. So if a java object happens to contain that character (which is legal in a Java string), JAXB will quietly generate invalid XML!
Then when we try to unmarshal that XML back into an object, an exception is thrown.
This means that the ability of JAXB to correctly marshal and unmarshal an object depends on what data happens to be in the fields of the object! That doesn't seem like a very safe or general solution. It means, for example, that if we marshal an object and store it in a database, then months later when we go to read the object, we will suddenly find out that it is unreadable! Yikes!
One obvious solution to this is that JAXB should be throwing an exception when the object is marshalled, i.e., it should detect that it cannot generate XML to represent the object.
Another, more powerful option would be for JAXB to use its own escaping mechanism to escape any characters that would be illegal in XML. Then when unmarshalling, it would look for such escaped characters in strings and "unescape" them (using the same mechanism), so that it could guarantee that any Java string can be preserved through the marshalling/unmarshalling process.
We could, of course, use our own wrapper to do a similar thing, but it would seem much safer for JAXB to be able to do this.
Is there a bug logged on this? Is there a standard workaround for it?
|
|
|
|
|
|
|
Re: JAXB can generate Illegal XML characters
Posted:
Mar 18, 2009 8:47 AM
in response to: ranboii
|
|
|
I found another one a few days ago. JAXB uses a BigDecimal for 'xsd:decimal' types and it uses BigDecimal#toString() to marshal to XML. However, the #toString() method allows exponents which is not supported by the W3C XML Schema Spec (http://www.w3.org/TR/1999/WD-xmlschema-2-19990924/#decimal). I couldn't find anywhere that this had been raised as a fault.
I got round it using a javax.xml.bind.annotation.adapters.XmlAdapter which used BigDecimal#toPlainString.
|
|
|
|
|
|
|
|
|
Re: JAXB can generate Illegal XML characters
Posted:
Jul 19, 2009 2:56 AM
in response to: ranboii
|
|
|
|
|
Firstly I agree with you that this behaviour is strange... It would be nice if Kohsuke would comment on it.
There is a way to supply you own CharacterEscaper thou. Implement the interface
com.sun.xml.bind.marshaller.CharacterEscapeHandler and pass it to the marshaller with
marshaller.setProperty("com.sun.xml.bind.characterEscapeHandler", myEscaperObject);
I've attached my implementation snatched from Kohsuke orginal (which didn't escape the ' and " characters properly).
|
|
|
|
|
java.net RSS
