Fixed #226

release-notes/VERSION-2.x

@@ -9,6 +9,8 @@ Project: jackson-dataformat-xml
 #205: `XmlMapper`/`UntypedObjectDeserializer` swallows duplicated elements in
   XML documents
  (reported by joaovarandas@github)
+#226: XML to JSON - IOException when parsing XML with XMLMapper
+ (reported by dbories@github)
 #262: Make `ToXmlGenerator` non-final
  (requested by Dave J)
 #273: Input mismatch with case-insensitive properties

src/main/java/com/fasterxml/jackson/dataformat/xml/deser/FromXmlParser.java

@@ -630,18 +630,33 @@ public JsonToken nextToken() throws IOException
                     // we had an empty String (or all white space), and we are
                     // deserializing an array, we better hide the empty text.
                     // Also: must skip following END_ELEMENT
-                    _skipEndElement();
-                    if (_parsingContext.inArray()) {
-                        if (XmlTokenStream._allWs(_currText)) {
-                            // 06-Jan-2015, tatu: as per [dataformat-xml#180], need to
-                            //    expose as empty Object, not null (or, worse, as used to
-                            //    be done, by swallowing the token)
-                            _nextToken = JsonToken.END_OBJECT;
-                            _parsingContext = _parsingContext.createChildObjectContext(-1, -1);
-                            return (_currToken = JsonToken.START_OBJECT);
+                    // 05-Jun-2020, tatu: ... if there is one; we may actually alternatively
+                    //   get START_ELEMENT for "mixed content" case; if so, need to change to
+                    //   expose "XmlText" as separate property
+                    token = _nextToken();
+
+                    if (token == XmlTokenStream.XML_END_ELEMENT) {
+                        if (_parsingContext.inArray()) {
+                            if (XmlTokenStream._allWs(_currText)) {
+                                // 06-Jan-2015, tatu: as per [dataformat-xml#180], need to
+                                //    expose as empty Object, not null (or, worse, as used to
+                                //    be done, by swallowing the token)
+                                _nextToken = JsonToken.END_OBJECT;
+                                _parsingContext = _parsingContext.createChildObjectContext(-1, -1);
+                                return (_currToken = JsonToken.START_OBJECT);
+                            }
                         }
+                        return (_currToken = JsonToken.VALUE_STRING);
                     }
-                    return (_currToken = JsonToken.VALUE_STRING);
+                    if (token != XmlTokenStream.XML_START_ELEMENT) {
+                        throw new JsonParseException(this, String.format(
+"Internal error: Expected END_ELEMENT (%d) or START_ELEMENT (%d), got event of type %d",
+XmlTokenStream.XML_END_ELEMENT, XmlTokenStream.XML_START_ELEMENT, token));
+                    }
+                    // fall-through, except must create new context AND push back
+                    // START_ELEMENT we just saw:
+                    _xmlTokens.pushbackCurrentToken();
+                    _parsingContext = _parsingContext.createChildObjectContext(-1, -1);
                 }
                 // [dataformat-xml#177]: empty text may also need to be skipped
                 // but... [dataformat-xml#191]: looks like we can't short-cut, must

src/main/java/com/fasterxml/jackson/dataformat/xml/deser/XmlTokenStream.java

@@ -97,7 +97,16 @@ public class XmlTokenStream
     protected String _namespaceURI;
 
     protected String _textValue;
-
+
+    /**
+     * Marker flag set if caller wants to "push back" current token so
+     * that next call to {@link #next()} should simply be given what was
+     * already read.
+     *
+     * @since 2.12
+     */
+    protected boolean _repeatCurrentToken;
+
     /*
     /**********************************************************************
     /* State for handling virtual wrapping
@@ -203,6 +212,10 @@ private String _loc() {
 //    public int next0() throws XMLStreamException
     public int next() throws XMLStreamException
     {
+        if (_repeatCurrentToken) {
+            _repeatCurrentToken = false;
+            return _currentState;
+        }
         if (_repeatElement != 0) {
             return (_currentState = _handleRepeatElement());
         }
@@ -292,6 +305,17 @@ protected void repeatStartElement()
         _repeatElement = REPLAY_START_DUP;
     }
 
+    /**
+     * Method that can be called to ask stream to literally just return current token
+     * with the next call to {@link #next()}, without more work.
+     *
+     * @since 2.12
+     */
+    protected void pushbackCurrentToken()
+    {
+        _repeatCurrentToken = true;
+    }
+
     /**
      * Method called to skip any attributes current START_ELEMENT may have,
      * so that they are not returned as token.

src/test/java/com/fasterxml/jackson/dataformat/xml/deser/JsonNodeMixedContent403Test.java

@@ -38,4 +38,16 @@ public void testMultipleMixedContent() throws Exception
                 a2q("{'':['first','second','third'],'a':'1','b':'2'}")),
                 XML_MAPPER.readTree("<root>first<a>1</a>second<b>2</b>third</root>"));
     }
+
+    // [dataformat-xml#226]
+    public void testMixed226() throws Exception
+    {
+        final String XML = "<root>\n"
+                +"<a>mixed1 <b>leaf</b>"
+                +" mixed2</a>\n"
+                +"</root>";
+        assertEquals(JSON_MAPPER.readTree(
+                a2q("{'a':{'':['mixed1 ',' mixed2'],'b':'leaf'}}")),
+                XML_MAPPER.readTree(XML));
+    }
 }