W3C Schema Validation does not cater for xs:unique constraints

[Jens-Dittrich]

Consider the following XSD (called idc2.xsd):

<xsd:schema xmlns:xsd="http://www.w3.org/2001/XMLSchema"
            xmlns="idc2.xsd"
            xmlns:idc="idc2.xsd"
            targetNamespace="idc2.xsd"
            elementFormDefault="qualified"
            version="1.0"
            >
  <xsd:element name="itemList">
	<xsd:complexType>
	  <xsd:sequence>
	    <xsd:element name="item" maxOccurs="unbounded" type="xsd:decimal" />
	  </xsd:sequence>
	</xsd:complexType>
	<xsd:unique name="itemAttr">
	  <xsd:selector xpath="idc:item"/>
	  <xsd:field    xpath="."/>
	</xsd:unique>
  </xsd:element>
</xsd:schema>

And the corresponding (invalid) XML (called idc2.xml):

<?xml version="1.0"?>
<itemList xmlns="idc2.xsd"
          xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance"
          xsi:schemaLocation="idc2.xsd idc2.xsd">
   <item>1</item>
   <item>1</item>
   <item>2</item>
</itemList>

The following pom.xml dependencies I used:

<dependency>
    <groupId>com.fasterxml.woodstox</groupId>
    <artifactId>woodstox-core</artifactId>
    <version>5.0.3</version>
</dependency>
<dependency>
    <groupId>msv</groupId>
    <artifactId>msv</artifactId>
    <version>20050913</version>
</dependency>
<dependency>
    <groupId>relaxngDatatype</groupId>
    <artifactId>relaxngDatatype</artifactId>
    <version>20020414</version>
</dependency>
<dependency>
    <groupId>com.sun.msv.datatype.xsd</groupId>
    <artifactId>xsdlib</artifactId>
    <version>2013.2</version>
</dependency>

And the following example program (TestXSD.java):

package jdi.test.xsdvalidation;

import java.io.InputStream;

import javax.xml.stream.XMLInputFactory;

import org.codehaus.stax2.XMLInputFactory2;
import org.codehaus.stax2.XMLStreamReader2;
import org.codehaus.stax2.validation.XMLValidationException;
import org.codehaus.stax2.validation.XMLValidationSchema;
import org.codehaus.stax2.validation.XMLValidationSchemaFactory;

public class TestXSD {
	
	public static void main(final String[] args) throws Exception {
		final String xmlFileName = "idc2.xml";
		final String xsdFileName = "idc2.xsd";
		
		//load Schema
		final XMLValidationSchemaFactory xmlValidationSchemaFactory = XMLValidationSchemaFactory.newInstance(XMLValidationSchema.SCHEMA_ID_W3C_SCHEMA);
		
		final InputStream schemaInputStream = TestXSD.class.getResourceAsStream(xsdFileName);
		final XMLValidationSchema xmlValidationSchema = xmlValidationSchemaFactory.createSchema(schemaInputStream);
		
		//load (invalid) XML file
		final InputStream xmlInputStream = TestXSD.class.getResourceAsStream(xmlFileName);
		final XMLInputFactory2 xmlInputFactory2 = (XMLInputFactory2)XMLInputFactory.newInstance();
		final XMLStreamReader2 xmlStreamReader =(XMLStreamReader2) xmlInputFactory2.createXMLStreamReader(xmlInputStream);
		try{
			//validate the XML file
			xmlStreamReader.validateAgainst(xmlValidationSchema);
			//traverse the streaming document
			while(xmlStreamReader.hasNext()){
				xmlStreamReader.next();
			}
		} catch(final XMLValidationException e){
			//catch validation exception
			System.err.println("XML file: " + xmlFileName + " failed to validatate against: " + relaxNgFileName);
			return ;
		}
		System.out.println("XML file: " + xmlFileName + " successfully validated against: " + relaxNgFileName);
	}

}

produces the not expected output:

XML file: idc2.xml successfully validated against: idc2.xsd

Note:

• Xerces-J 2.11 and Eclipse validate this file correctly.

Apparently I suspect that the class GenericMsvValidator is not doing any of Unique, Key, KeyRef validations. Only ID and IDREF seems to be supported (not sure whether this coincides with Key / KeyRef respectively). I have not found out whether this is supposed to be the case or just an issue. I furthermore discovered the field mVGM.grammer.topLevel.element.identityConstraints did contain the unique constraint. The obvious suggestion to use Xerces-J instead does not solve my problem either, as https://issues.apache.org/jira/browse/XERCESJ-1276 is doing unique constraint validation in O(n^2) instead of O(n log(n)) as expected.

[cowtowncoder]

Thank you for reporting this. I do not know for sure whether this is supportable: since actual validation logic is provided by MSV (multi-schema validator), that would give the ultimate truth. If MSV supports this, it should be supportable by Woodstox.

But I suspect the problem is that use of xpath expression for definition might require access to full tree model (dom), and if so, it would not be supportable with streaming(-only) parser.

[cowtowncoder]

For what it is worth, Woodstox is using the latest release of MSV, 2013.6.1:

http://mvnrepository.com/artifact/net.java.dev.msv/msv-core

otherwise only found this:

https://java.net/jira/browse/MSV-24

which could suggest that validation is not supported. On the other hand, limitations spelled out in:

https://github.com/kohsuke/msv/blob/master/msv/doc/commandline.html

do not mention that this is missing, unless I misread it.

So... not sure whether msv provides for it. But even if it does, it is quite possible Woodstox is not feeding all information necessary if so.