XML Deserialization : keep encoded html characters

[BOO35]

Hi !!

I need to keep HTML characters in XML Source.
Unfortunately, those are decoded, even before deserialization (p.getText() in deserialization method returns "Hello World!" instead of "Hello World!").

Example :
Main class :

public class XmlNoHtmlEntityDecoder {

    public static void main(String[] args) throws IOException {
        
        XmlMapper xmlMapper = new XmlMapper();

        String xmlContent = "<ADI><message>&quot;Hello World!&quot;</message></ADI>";

        AdiXmlSource result = xmlMapper.readValue(xmlContent, AdiXmlSource.class);
        System.out.println(result.getMessage());
    }

}

Destination Object :

public class AdiXmlSource {

    public AdiXmlSource() {
    }

    @JsonDeserialize(using = NoDecodeHtmlEntitiesDeserializer.class)
    private String message;

    public String getMessage(){
        return this.message;
    }

}

Custom deserializer :

class NoDecodeHtmlEntitiesDeserializer extends JsonDeserializer<String> {
    @Override
    public String deserialize(JsonParser p, DeserializationContext ctxt) throws IOException {
        return p.getText();
    }
}

Could you help me to find a solution to keep encoded HTML characters in XML Source ?

[pjfanning]

Have a look at https://github.com/FasterXML/jackson-dataformat-xml?tab=readme-ov-file#usage

There is an example of how to set up the XmlMapper to wrap a Woodstox WstxInputFactory.

You can configure the WstxInputFactory (also shown in the readme - the example sets P_MAX_ATTRIBUTE_SIZE).
You could set P_TREAT_CHAR_REFS_AS_ENTS instead. The value is true or false.

https://cowtowncoder.medium.com/configuring-woodstox-xml-parser-woodstox-specific-properties-1ce5030a5173

I'm not guaranteeing that this config will get you what you want but you could try setting different Woodstox configs.

[pjfanning]

quot is mandated by the XML spec while custom entities can be defined in DTDs. Maybe, Woodstox treats qout differently from custom entities. I suggest that you look around for an XML parser that supports not unescaping entities like quot. It is an unusual request to turn off entity unescaping and many parsers won't support it.

[cowtowncoder]

No proper XML parser can leave character entities unprocessed -- processing is mandatory part of XML spec.
And to pass underlying content -- should a parser provide that -- would need to be outside of standard Java XML APIs (i.e. cannot be done via SAX or StAX) which Jackson won't support.

So basically I doubt you can get it done: this is not how XML processing is designed to work anywhere. Character entities resolved on reading, re-encoded on writing.

[cowtowncoder]

NOTE: as @pjfanning suggested, https://fasterxml.github.io/woodstox/javadoc/5.0/com/ctc/wstx/api/WstxInputProperties.html#P_TREAT_CHAR_REFS_AS_ENTS would theoretically retain character entities by reporting them as separate events (and not as CHARACTERS).

But the problem is that deserializers do not really deal with Stax token stream and cannot make much use of this.
Custom deserializer could, I suppose, get there (by forcing access to XMLStreamReader) but would need to model that in some way.

In the end tho it is unclear to me WHY you would want to retain character entities. So maybe expanding on use case instead of focusing on assumed solution would make sense.

[BOO35]

Thank you very much !!
We're gonna adapt our solution to the process.
B.

[cowtowncoder]

Best of luck @BOO35 !