Fix #358: allow skipping 'xsi` attributes (#837)

Author: cowtowncoder

release-notes/CREDITS

@@ -47,6 +47,9 @@ Tobias M (@lightbringer)
 Eduard Wirch (@ewirch)
  * Reported #306: Can not use `@JacksonXmlText` for Creator property (creator parameter)
   (3.2.0)
+ * Requested #358: Allow skipping attributes from the `http://www.w3.org/2001/XMLSchema-instance`
+   namespace (with new `XmlReadFeature.SKIP_UNKNOWN_XSI_ATTRIBUTES`)
+  (3.2.0)
 
 Ruven Chu (@Chubacca)
  * Reported #334: `@JacksonXmlElementWrapper(useWrapping=false)` fails when using

release-notes/VERSION

@@ -33,6 +33,10 @@ Version: 3.x (for earlier see VERSION-2.x)
   a `DeserializerModifier` that delegates
  (reported by Ruven C)
  (fix by @cowtowncoder, w/ Claude code)
+#358: Allow skipping attributes from the `http://www.w3.org/2001/XMLSchema-instance`
+  namespace (with new `XmlReadFeature.SKIP_UNKNOWN_XSI_ATTRIBUTES`)
+ (requested by Eduard W)
+ (fix by @cowtowncoder, w/ Claude code)
 #426: `InvalidTypeIdException` when parsing XML to POJO containing nested `List<>`,
   custom `TypeIdResolver`
  (reported by @ankagar)

src/main/java/tools/jackson/dataformat/xml/XmlReadFeature.java

@@ -55,6 +55,24 @@ public enum XmlReadFeature implements FormatFeature
      */
     PROCESS_XSI_NIL(true),
 
+    /**
+     * Feature that controls whether XML Schema Instance (XSI) namespace attributes
+     * other than {@code xsi:nil} and {@code xsi:type} (which have their own handling
+     * via {@link #PROCESS_XSI_NIL} and {@link #AUTO_DETECT_XSI_TYPE}) are silently
+     * skipped during deserialization. Attributes affected include {@code xsi:schemaLocation}
+     * and {@code xsi:noNamespaceSchemaLocation}.
+     *<p>
+     * When enabled, these attributes are ignored and will not cause
+     * {@code UnrecognizedPropertyException}. When disabled (default), they are
+     * exposed as regular attributes and may require matching POJO properties
+     * or {@code DeserializationFeature.FAIL_ON_UNKNOWN_PROPERTIES} to be disabled.
+     *<p>
+     * Default setting is {@code false}.
+     *
+     * @since 3.2
+     */
+    SKIP_UNKNOWN_XSI_ATTRIBUTES(false),
+
     ;
 
     private final boolean _defaultState;

src/main/java/tools/jackson/dataformat/xml/deser/XmlTokenStream.java

@@ -79,6 +79,9 @@ public class XmlTokenStream
 
     protected final boolean _cfgProcessXsiType;
 
+    // [dataformat-xml#358]
+    protected final boolean _cfgSkipUnknownXsiAttributes;
+
     protected XmlNameProcessor _nameProcessor;
 
     /*
@@ -183,6 +186,7 @@ public XmlTokenStream(XMLStreamReader xmlReader, ContentReference sourceRef,
         _formatFeatures = formatFeatures;
         _cfgProcessXsiNil = XmlReadFeature.PROCESS_XSI_NIL.enabledIn(_formatFeatures);
         _cfgProcessXsiType = XmlReadFeature.AUTO_DETECT_XSI_TYPE.enabledIn(_formatFeatures);
+        _cfgSkipUnknownXsiAttributes = XmlReadFeature.SKIP_UNKNOWN_XSI_ATTRIBUTES.enabledIn(_formatFeatures);
         // 04-Dec-2023, tatu: [dataformat-xml#618] Need further customized adapter:
         _xmlReader = Stax2JacksonReaderAdapter.wrapIfNecessary(xmlReader);
         _nameProcessor = nameProcessor;
@@ -487,6 +491,23 @@ private final int _next() throws XMLStreamException
 //System.out.println(" XmlTokenStream._next(): Got xsi:nil, skipping element");
                 return _handleEndElement();
             }
+            // [dataformat-xml#358]: Optionally skip XSI namespace attributes other
+            //   than "type" (handled by _decodeAttributeName) and "nil" (when xsi:nil
+            //   processing is disabled, it should be exposed as a regular attribute)
+            if (_cfgSkipUnknownXsiAttributes) {
+                while (_nextAttributeIndex < _attributeCount) {
+                    final String attrNs = _xmlReader.getAttributeNamespace(_nextAttributeIndex);
+                    if (XSI_NAMESPACE.equals(attrNs)) {
+                        final String localName = _xmlReader.getAttributeLocalName(_nextAttributeIndex);
+                        if (!"type".equals(localName)
+                                && !(!_cfgProcessXsiNil && "nil".equals(localName))) {
+                            ++_nextAttributeIndex;
+                            continue;
+                        }
+                    }
+                    break;
+                }
+            }
             if (_nextAttributeIndex < _attributeCount) {
 //System.out.println(" XmlTokenStream._next(): Got attr(s)!");
                 _decodeAttributeName(_xmlReader.getAttributeNamespace(_nextAttributeIndex),

src/test/java/tools/jackson/dataformat/xml/deser/XsiSchemaLocationTest.java

@@ -0,0 +1,90 @@
+package tools.jackson.dataformat.xml.deser;
+
+import org.junit.jupiter.api.Test;
+
+import tools.jackson.dataformat.xml.*;
+
+import static org.junit.jupiter.api.Assertions.*;
+
+// [dataformat-xml#358]: XSI namespace attributes like schemaLocation
+// should be silently ignored when SKIP_UNKNOWN_XSI_ATTRIBUTES is enabled
+public class XsiSchemaLocationTest extends XmlTestUtil
+{
+    static class Dto {
+        public String value;
+    }
+
+    static class DtoWithAttr {
+        public String value;
+        public int count;
+    }
+
+    static class DtoWithSchemaLocation {
+        public String value;
+        public String schemaLocation;
+    }
+
+    private final XmlMapper MAPPER_SKIP = mapperBuilder()
+            .enable(XmlReadFeature.SKIP_UNKNOWN_XSI_ATTRIBUTES)
+            .build();
+
+    private final XmlMapper MAPPER_DEFAULT = mapperBuilder().build();
+
+    // Basic case from issue #358: xsi:schemaLocation should be ignored when enabled
+    @Test
+    public void testXsiSchemaLocationIgnored() throws Exception
+    {
+        Dto result = MAPPER_SKIP.readValue(
+                "<dto xmlns=\"http://SomeNamespace\""
+                + " xmlns:xsi=\"http://www.w3.org/2001/XMLSchema-instance\""
+                + " xsi:schemaLocation=\"http://SomeNamespace"
+                + " http://SomeNamespace/schema.xsd\">"
+                + "<value>hello</value>"
+                + "</dto>",
+                Dto.class);
+        assertEquals("hello", result.value);
+    }
+
+    // Also test xsi:noNamespaceSchemaLocation
+    @Test
+    public void testXsiNoNamespaceSchemaLocationIgnored() throws Exception
+    {
+        Dto result = MAPPER_SKIP.readValue(
+                "<dto xmlns:xsi=\"http://www.w3.org/2001/XMLSchema-instance\""
+                + " xsi:noNamespaceSchemaLocation=\"schema.xsd\">"
+                + "<value>world</value>"
+                + "</dto>",
+                Dto.class);
+        assertEquals("world", result.value);
+    }
+
+    // Ensure regular attributes still work alongside XSI attributes
+    @Test
+    public void testXsiSchemaLocationWithRegularAttributes() throws Exception
+    {
+        DtoWithAttr result = MAPPER_SKIP.readValue(
+                "<DtoWithAttr xmlns:xsi=\"http://www.w3.org/2001/XMLSchema-instance\""
+                + " xsi:schemaLocation=\"http://SomeNamespace schema.xsd\""
+                + " count=\"42\">"
+                + "<value>test</value>"
+                + "</DtoWithAttr>",
+                DtoWithAttr.class);
+        assertEquals("test", result.value);
+        assertEquals(42, result.count);
+    }
+
+    // When feature is disabled (default), XSI attributes should be exposed
+    // and bindable to POJO properties
+    @Test
+    public void testXsiSchemaLocationExposedByDefault() throws Exception
+    {
+        DtoWithSchemaLocation result = MAPPER_DEFAULT.readValue(
+                "<DtoWithSchemaLocation xmlns:xsi=\"http://www.w3.org/2001/XMLSchema-instance\""
+                + " xsi:schemaLocation=\"http://SomeNamespace schema.xsd\">"
+                + "<value>bound</value>"
+                + "</DtoWithSchemaLocation>",
+                DtoWithSchemaLocation.class);
+        assertEquals("bound", result.value);
+        assertEquals("http://SomeNamespace schema.xsd", result.schemaLocation);
+    }
+}