Fixed #442 (missing START_OBJECT for leading "mixed text")

release-notes/VERSION-2.x

@@ -6,7 +6,8 @@ Project: jackson-dataformat-xml
 
 2.13.0 (not yet released)
 
-No changes since 2.12
+#442: Missing `START_OBJECT` token in complex element starting with text
+ (reported by richardsonwk@github)
 
 2.12.3 (not yet released)
 

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

@@ -153,7 +153,7 @@ private Feature(boolean defaultState) {
      */
     protected boolean _closed;
 
-    final protected IOContext _ioContext;
+    protected final IOContext _ioContext;
 
     /*
     /**********************************************************
@@ -179,6 +179,16 @@ private Feature(boolean defaultState) {
 
     protected String _currText;
 
+    /**
+     * Additional flag that is strictly needed when exposing "mixed" leading
+     * String value as "anonymous" property/string pair. If so, code returns
+     * START_OBJECT first, sets {@code _nextToken} to be {@code FIELD_NAME}
+     * and sets this flag to indicate use of "anonymous" marker.
+     *
+     * @since 2.13
+     */
+    protected boolean _nextIsLeadingMixed;
+
     /*
     /**********************************************************
     /* Parsing state, parsed values
@@ -668,7 +678,15 @@ public JsonToken nextToken() throws IOException
                 _parsingContext = _parsingContext.getParent();
                 break;
             case FIELD_NAME:
-                _parsingContext.setCurrentName(_xmlTokens.getLocalName());
+                // 29-Mar-2021, tatu: [dataformat-xml#442]: special case of leading
+                //    mixed text added
+                if (_nextIsLeadingMixed) {
+                    _nextIsLeadingMixed = false;
+                    _parsingContext.setCurrentName(_cfgNameForTextElement);
+                    _nextToken = JsonToken.VALUE_STRING;
+                } else {
+                    _parsingContext.setCurrentName(_xmlTokens.getLocalName());
+                }
                 break;
             default: // VALUE_STRING, VALUE_NULL
                 // 13-May-2020, tatu: [dataformat-xml#397]: advance `index` anyway; not
@@ -756,7 +774,7 @@ public JsonToken nextToken() throws IOException
                 _currText = _xmlTokens.getText();
                 if (_mayBeLeaf) {
                     _mayBeLeaf = false;
-                    // One more refinement (pronunced like "hack") is that if
+                    // One more refinement (pronounced like "hack") is that if
                     // 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
@@ -792,7 +810,14 @@ public JsonToken nextToken() throws IOException
                 // but... [dataformat-xml#191]: looks like we can't short-cut, must
                 // loop over again
                 if (_parsingContext.inObject()) {
-                    if ((_currToken != JsonToken.FIELD_NAME) && XmlTokenStream._allWs(_currText)) {
+                    if (_currToken == JsonToken.FIELD_NAME) {
+                        // 29-Mar-2021, tatu: [dataformat-xml#442]: need special handling for
+                        //    leading mixed content; requires 3-token sequence for which _nextToken
+                        //    along is not enough.
+                        _nextIsLeadingMixed = true;
+                        _nextToken = JsonToken.FIELD_NAME;
+                        return (_currToken = JsonToken.START_OBJECT);
+                    } else if (XmlTokenStream._allWs(_currText)) {
                         token = _nextToken();
                         continue;
                     }
@@ -802,6 +827,11 @@ public JsonToken nextToken() throws IOException
                         token = _nextToken();
                         continue;
                     }
+                    // 29-Mar-2021, tatu: This seems like an error condition...
+                    //   How should we indicate it? As of 2.13, report as unexpected state
+                    throw _constructError(
+"Unexpected non-whitespace text ('"+_currText+"' in Array context: should not occur (or should be handled)"
+);
                 }
 
                 // If not a leaf (or otherwise ignorable), need to transform into property...

src/test/java/com/fasterxml/jackson/dataformat/xml/failing/XmlParser442Test.java → src/test/java/com/fasterxml/jackson/dataformat/xml/stream/XmlParser442Test.java

@@ -1,4 +1,4 @@
-package com.fasterxml.jackson.dataformat.xml.failing;
+package com.fasterxml.jackson.dataformat.xml.stream;
 
 import com.fasterxml.jackson.core.JsonToken;
 
@@ -11,26 +11,26 @@ public class XmlParser442Test extends XmlTestBase
     private final XmlMapper MAPPER = newMapper();
 
     // For [dataformat-xml#442]
-    public void testMixedContentAfter() throws Exception
+    public void testMixedContentBeforeElement442() throws Exception
     {
-        try (FromXmlParser xp = (FromXmlParser) MAPPER.createParser(
-                "<root>\n" // START_OBJECT
-                +"  <branch>\n" // *Missing* START_OBJECT
-                +"     text\n"
-                +"    <leaf>stuff</leaf>\n"
-                +"  </branch>\n" // END_OBJECT
-                +"</root>\n" // END_OBJECT
-
-                /*
-"<SomeXml>\n" // START_OBJECT
-+"  <ParentElement>\n" // *Missing* START_OBJECT
-+"     text\n"
-+"     <ChildElement someAttribute=\"value\"/>\n" // START_OBJECT/END_OBJECT
-+"     further text\n"
-+"  </ParentElement>\n" // END_OBJECT
-+"</SomeXml>\n" // END_OBJECT
-*/
-)) {
+        final String XML =
+            "<root>\n" // START_OBJECT
+            +"  <branch>\n" // *Missing* START_OBJECT
+            +"     text\n"
+            +"    <leaf>stuff</leaf>\n"
+            +"  </branch>\n" // END_OBJECT
+            +"</root>\n" // END_OBJECT
+            ;
+
+        // Should get equivalent of:
+        //
+        // { "branch" : {
+        //      "" : "  text  ",
+        //      "leaf" : "stuff"
+        //    }
+        // }
+
+        try (FromXmlParser xp = (FromXmlParser) MAPPER.createParser(XML)) {
             assertToken(JsonToken.START_OBJECT, xp.nextToken());
             assertToken(JsonToken.FIELD_NAME, xp.nextToken());
             assertEquals("branch", xp.currentName());