Invalid attributes producing extra null characters during DTD validation

[ChrisTrenkamp]

Use the catalog here to reproduce this issue: #104 (comment)

test.xml

<?xml version="1.0" encoding="UTF-8"?>
<!DOCTYPE concept PUBLIC "-//OASIS//DTD DITA Concept//EN" "concept.dtd">
<concept id="foo">
   <title>foo</title>
   <conbody>
      <note type="note" verbose="yes">
         <p>The verbose attribute is invalid, and it causes wstx to produce a
         null character when writing the XML.</p>
      </note>
   </conbody>
</concept>

WstxValidatorTest.java

import com.ctc.wstx.stax.WstxInputFactory;
import com.ctc.wstx.stax.WstxOutputFactory;
import java.io.ByteArrayOutputStream;
import java.io.File;
import java.nio.file.Paths;
import javax.xml.catalog.CatalogFeatures;
import javax.xml.catalog.CatalogManager;
import javax.xml.stream.XMLInputFactory;

public class WstxValidatorTest {
  public static void main(String[] args) throws Exception {
    var catalog =
        CatalogManager.catalogResolver(
            CatalogFeatures.defaults(), Paths.get("path/to/catalog.xml").toUri());
    var factory = new WstxInputFactory();
    factory.setProperty(XMLInputFactory.IS_VALIDATING, true);
    factory.setProperty(XMLInputFactory.RESOLVER, catalog);

    var streamReader = factory.createXMLStreamReader(new File("path/to/test.xml"));
    streamReader.setValidationProblemHandler(p -> System.err.println(p.getMessage()));
    var reader = factory.createXMLEventReader(streamReader);
    var buffer = new ByteArrayOutputStream();
    var writer = new WstxOutputFactory().createXMLEventWriter(buffer);

    while (reader.hasNext()) {
      var event = reader.nextEvent();

      if (event.isStartElement()) {
        var attrs = event.asStartElement().getAttributes();

        while (attrs.hasNext()) {
          var attr = attrs.next();

          System.out.println(
              attr.getName()
                  + " = '"
                  + attr.getValue()
                  + "' ( "
                  + hex(attr.getValue().getBytes())
                  + ")");
        }
      }
      writer.add(event);
    }

    var xmlResult = new String(buffer.toByteArray(), "UTF-8");
    System.out.println(xmlResult);
  }

  static String hex(byte[] bytes) {
    var buffer = new StringBuilder();

    for (var i : bytes) {
      buffer.append(String.format("0x%02x ", i));
    }

    return buffer.toString();
  }
}

Output:

...
Element <note> has no attribute "verbose"
type = 'note' ( 0x6e 0x6f 0x74 0x65 )
verbose = 'yes    ' ( 0x79 0x65 0x73 0x00 0x00 0x00 0x00 )
class = '- topic/note ' ( 0x2d 0x20 0x74 0x6f 0x70 0x69 0x63 0x2f 0x6e 0x6f 0x74 0x65 0x20 )
Exception in thread "main" com.ctc.wstx.exc.WstxIOException: Invalid null character in text to output
	at com.ctc.wstx.sw.BaseNsStreamWriter.doWriteAttr(BaseNsStreamWriter.java:531)
	at com.ctc.wstx.sw.SimpleNsStreamWriter.writeAttribute(SimpleNsStreamWriter.java:90)
	at org.codehaus.stax2.ri.Stax2EventWriterImpl.add(Stax2EventWriterImpl.java:61)
	at org.codehaus.stax2.ri.Stax2EventWriterImpl.add(Stax2EventWriterImpl.java:108)
	at WstxValidatorTest.main(WstxValidatorTest.java:46)
Caused by: java.io.IOException: Invalid null character in text to output
	at com.ctc.wstx.api.InvalidCharHandler$FailingHandler.convertInvalidChar(InvalidCharHandler.java:52)
	at com.ctc.wstx.sw.XmlWriter.handleInvalidChar(XmlWriter.java:628)
	at com.ctc.wstx.sw.BufferingXmlWriter.writeAttrValue(BufferingXmlWriter.java:1082)
	at com.ctc.wstx.sw.BufferingXmlWriter.writeAttribute(BufferingXmlWriter.java:925)
	at com.ctc.wstx.sw.BaseNsStreamWriter.doWriteAttr(BaseNsStreamWriter.java:528)
	... 4 more

Notice the output for the verbose attribute. The attribute value is being padded with extra null characters.

[ChrisTrenkamp]

This issue seems to be causing other problems, such as StringIndexOutOfBoundsException being thrown here:

https://github.com/FasterXML/woodstox/blob/master/src/main/java/com/ctc/wstx/dtd/DTDValidator.java#L257

[ChrisTrenkamp]

Created a test case demonstrating the issue:

https://github.com/ChrisTrenkamp/woodstox/commit/6e73a8ded29735c71860e74fe45e5dd5eaca908f

[cowtowncoder]

Fixed; will be included in 6.2.6 to be released shortly.