Question

Yesterday Oracle decided to take down java.sun.com for a while. This screwed things up for me because xalan tried to validate some XML but couldn’t retrieve the properties.dtd.

I’m using xalan 2.7.1 to run some XSL transforms, and I don’t want it to validate anything.
so tried loading up the XSL like this:

SAXParserFactory spf = SAXParserFactory.newInstance();
spf.setNamespaceAware(true);
spf.setValidating(false);
XMLReader rdr = spf.newSAXParser().getXMLReader();      
Source xsl = new SAXSource(rdr, new InputSource(xslFilePath));  
Templates cachedXSLT  = factory.newTemplates(xsl);
Transformer transformer = cachedXSLT.newTransformer();         
transformer.transform(xmlSource, result);  

in the XSL itself, I do something like this:

  <xsl:variable name="entry" select="document(concat($prefix, $locale_part, $suffix))/properties/entry[@key=$key]"/>

The XML this code retrieves has the following definition at the top:

<!DOCTYPE properties SYSTEM "http://java.sun.com/dtd/properties.dtd">
<properties>
<entry key="...

Despite the java code above instructing the parser to NOT VALIDATE, it still sends a request to java.sun.com. While java.sun.com is unavailable, this makes the transform fail with the message:

 Can not load requested doc: http://java.sun.com/dtd/properties.dtd

How do I get xalan to stop trying to validate the XML loaded from the “document” function?

Answer 1

The documentation mentions that the parser may read the DTDs even if not validating, as it may become necessary to use the DTD to resolve (expand) entities.

Since I don’t have control over the XML documents, nont’s option of modifying the XML was not available to me.

I managed to shut down attempts to pull in DTD documents by sabotaging the resolver, as follows.

My code uses a DocumentBuilder to return a Document (= DOM) but the XMLReader as per the OP’s example also has a method setEntityResolver so the same technique should work with that.

DocumentBuilderFactory factory = DocumentBuilderFactory.newInstance();
factory.setValidating(false); // turns off validation
factory.setSchema(null);      // turns off use of schema
                              // but that's *still* not enough!
builder = factory.newDocumentBuilder();
builder.setEntityResolver(new NullEntityResolver()); // swap in a dummy resolver
return builder().parse(xmlFile); 

Here, now, is my fake resolver: It returns an empty InputStream no matter what’s asked of it.

/** my resolver that doesn't */
private static class NullEntityResolver implements EntityResolver {

    public InputSource resolveEntity(String publicId, String systemId) 
    throws SAXException, IOException {
        // Message only for debugging / if you care
        System.out.println("I'm asked to resolve: " + publicId + " / " + systemId);
        return new InputSource(new ByteArrayInputStream(new byte[0]));
    }

}

---

Alternatively, your fake resolver could return streams of actual documents read as local resources or whatever.