Fix #74: Validate xsd:anyURI to be a valid IRI

Signed-off-by: James Leigh <james.leigh@ontotext.com>

Author: James Leigh

core/model/src/main/java/org/eclipse/rdf4j/model/datatypes/XMLDatatypeUtil.java

@@ -9,6 +9,7 @@
 
 import java.math.BigDecimal;
 import java.math.BigInteger;
+import java.net.URISyntaxException;
 import java.util.StringTokenizer;
 
 import javax.xml.datatype.DatatypeConfigurationException;
@@ -18,6 +19,7 @@
 import javax.xml.datatype.XMLGregorianCalendar;
 import javax.xml.namespace.QName;
 
+import org.eclipse.rdf4j.common.net.ParsedIRI;
 import org.eclipse.rdf4j.common.text.ASCIIUtil;
 import org.eclipse.rdf4j.model.IRI;
 import org.eclipse.rdf4j.model.vocabulary.XMLSchema;
@@ -269,6 +271,9 @@ else if (datatype.equals(XMLSchema.YEARMONTHDURATION)) {
 		else if (datatype.equals(XMLSchema.QNAME)) {
 			result = isValidQName(value);
 		}
+		else if (datatype.equals(XMLSchema.ANYURI)) {
+			result = isValidAnyURI(value);
+		}
 
 		return result;
 	}
@@ -652,6 +657,25 @@ public static boolean isValidQName(String value) {
 		return true;
 	}
 
+	/**
+	 * Determines if the supplied value is an Internationalized Resource Identifier Reference (IRI). An anyURI
+	 * value can be absolute or relative, and may have an optional fragment identifier (i.e., it may be an IRI
+	 * Reference). This type should be used when the value fulfills the role of an IRI, as defined in
+	 * [RFC 3987] or its successor(s) in the IETF Standards Track.
+	 *
+	 * @param value
+	 * @return <tt>true</tt> if a valid IRI, <tt>false</tt> otherwise
+	 */
+	public static boolean isValidAnyURI(String value) {
+		try {
+			new ParsedIRI(value.trim());
+			return true;
+		}
+		catch (URISyntaxException e) {
+			return false;
+		}
+	}
+
 	private static boolean isPrefixStartChar(int c) {
 		return ASCIIUtil.isLetter(c) || c >= 0x00C0 && c <= 0x00D6 || c >= 0x00D8 && c <= 0x00F6
 				|| c >= 0x00F8 && c <= 0x02FF || c >= 0x0370 && c <= 0x037D || c >= 0x037F && c <= 0x1FFF
@@ -758,6 +782,9 @@ else if (datatype.equals(XMLSchema.BOOLEAN)) {
 		else if (datatype.equals(XMLSchema.DATETIME)) {
 			result = normalizeDateTime(value);
 		}
+		else if (datatype.equals(XMLSchema.ANYURI)) {
+			result = collapseWhiteSpace(value);
+		}
 
 		return result;
 	}

core/model/src/test/java/org/eclipse/rdf4j/model/datatypes/XMLDatatypeUtilTest.java

@@ -198,9 +198,25 @@ public class XMLDatatypeUtilTest {
 			"⻐:⻘⻨" // random chinese chars, if this sequence happens to mean something, this is unintended
 	};
 
-	/** ivalid xsd:QName values */
+	/** invalid xsd:QName values */
 	private static final String[] INVALID_QNAMES = { "1:bar", "foo:1bar", "foo:bar:baz", "foo", "_:bar" };
 
+	/** http://www.datypic.com/sc/xsd/t-xsd_anyURI.html */
+	private static final String[] VALID_URI = {
+			"http://datypic.com",
+			"mailto:info@datypic.com",
+			"../%C3%A9dition.html",
+			"../édition.html",
+			"http://datypic.com/prod.html#shirt",
+			" http://datypic.com/prod.html#shirt ",
+			"../prod.html#shirt",
+			"urn:example:org",
+			"" };
+
+	private static final String[] INVALID_URI = {
+			"http://datypic.com#frag1#frag2",
+			"http://datypic.com#f% rag" };
+
 	@Test
 	public void testNormalize() {
 		assertEquals("-1.0E-1", XMLDatatypeUtil.normalize("-0.1", XMLSchema.DOUBLE));
@@ -251,6 +267,8 @@ public void testIsValidValue() {
 		testValidation(VALID_QNAMES, XMLSchema.QNAME, true);
 		testValidation(INVALID_QNAMES, XMLSchema.QNAME, false);
 
+		testValidation(VALID_URI, XMLSchema.ANYURI, true);
+		testValidation(INVALID_URI, XMLSchema.ANYURI, false);
 	}
 
 	private void testValidation(String[] values, IRI datatype, boolean validValues) {