Fix generating invalid XML tag names

This commit introduces the `PROCESS_ESCAPED_MALFORMED_TAGS` and
`ESCAPE_MALFORMED_TAGS` features that control whether invalid
tag names will be escaped with an attribute.

fixes #523
fixes #524

Author: mensinda

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

@@ -19,6 +19,7 @@
 
 import com.fasterxml.jackson.dataformat.xml.PackageVersion;
 import com.fasterxml.jackson.dataformat.xml.XmlMapper;
+import com.fasterxml.jackson.dataformat.xml.ser.ToXmlGenerator;
 import com.fasterxml.jackson.dataformat.xml.util.CaseInsensitiveNameSet;
 import com.fasterxml.jackson.dataformat.xml.util.StaxUtil;
 
@@ -82,6 +83,15 @@ public enum Feature implements FormatFeature
          */
         PROCESS_XSI_NIL(true),
 
+        /**
+         * Feature that controls whether the escaping mechanism from the
+         * {@code ESCAPE_MALFORMED_TAGS} from {@link ToXmlGenerator.Feature}
+         * is reversed.
+         *
+         * @since 2.14
+         */
+        PROCESS_ESCAPED_MALFORMED_TAGS(true)
+
         // 16-Nov-2020, tatu: would have been nice to add in 2.12 but is not
         //    trivial to implement... so leaving out for now
 

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

@@ -1,10 +1,13 @@
 package com.fasterxml.jackson.dataformat.xml.deser;
 
 import java.io.IOException;
+import java.util.Objects;
 
 import javax.xml.XMLConstants;
+import javax.xml.namespace.QName;
 import javax.xml.stream.*;
 
+import com.fasterxml.jackson.dataformat.xml.ser.ToXmlGenerator;
 import org.codehaus.stax2.XMLStreamLocation2;
 import org.codehaus.stax2.XMLStreamReader2;
 import org.codehaus.stax2.ri.Stax2ReaderAdapter;
@@ -73,6 +76,8 @@ public class XmlTokenStream
 
     protected boolean _cfgProcessXsiNil;
 
+    protected boolean _cfgProcessEscapedTag;
+
     /*
     /**********************************************************************
     /* Parsing state
@@ -158,6 +163,7 @@ public XmlTokenStream(XMLStreamReader xmlReader, ContentReference sourceRef,
         _sourceReference = sourceRef;
         _formatFeatures = formatFeatures;
         _cfgProcessXsiNil = FromXmlParser.Feature.PROCESS_XSI_NIL.enabledIn(_formatFeatures);
+        _cfgProcessEscapedTag = FromXmlParser.Feature.PROCESS_ESCAPED_MALFORMED_TAGS.enabledIn(_formatFeatures);
         _xmlReader = Stax2ReaderAdapter.wrapIfNecessary(xmlReader);
     }
 
@@ -176,7 +182,7 @@ public int initialize() throws XMLStreamException
         _localName = _xmlReader.getLocalName();
         _namespaceURI = _xmlReader.getNamespaceURI();
 
-        _checkXsiAttributes(); // sets _attributeCount, _nextAttributeIndex
+        _checkMagicAttributes(); // sets _attributeCount, _nextAttributeIndex
 
         // 02-Jul-2020, tatu: Two choices: if child elements OR attributes, expose
         //    as Object value; otherwise expose as Text
@@ -220,6 +226,7 @@ public XMLStreamReader2 getXmlReader() {
     protected void setFormatFeatures(int f) {
         _formatFeatures = f;
         _cfgProcessXsiNil = FromXmlParser.Feature.PROCESS_XSI_NIL.enabledIn(f);
+        _cfgProcessEscapedTag = FromXmlParser.Feature.PROCESS_ESCAPED_MALFORMED_TAGS.enabledIn(f);
     }
 
     /*
@@ -618,17 +625,14 @@ private final String _getText(XMLStreamReader2 r) throws XMLStreamException
 
     private final int _initStartElement() throws XMLStreamException
     {
-        final String ns = _xmlReader.getNamespaceURI();
-        final String localName = _xmlReader.getLocalName();
-
-        _checkXsiAttributes();
+        final QName qName = _checkMagicAttributes();
 
         // Support for virtual wrapping: in wrapping, may either create a new
         // wrapper scope (if in sub-tree, or matches wrapper element itself),
         // or implicitly close existing scope.
 
         if (_currentWrapper != null) {
-            if (_currentWrapper.matchesWrapper(localName, ns)) {
+            if (_currentWrapper.matchesWrapper(qName.getLocalPart(), qName.getNamespaceURI())) {
                 _currentWrapper = _currentWrapper.intermediateWrapper();
 //System.out.println(" _initStartElement(): START_ELEMENT ("+localName+") DOES match ["+_currentWrapper+"]: leave/add intermediate");
             } else {
@@ -638,23 +642,28 @@ private final int _initStartElement() throws XMLStreamException
                 _namespaceURI = _currentWrapper.getWrapperNamespace();
                 _currentWrapper = _currentWrapper.getParent();
                 // Important! We also need to restore the START_ELEMENT, so:
-                _nextLocalName = localName;
-                _nextNamespaceURI = ns;
+                _nextLocalName = qName.getLocalPart();
+                _nextNamespaceURI = qName.getNamespaceURI();
                 _repeatElement = REPLAY_START_DELAYED;
                 return (_currentState = XML_END_ELEMENT);
             }
         }
-        _localName = localName;
-        _namespaceURI = ns;
+        _localName = qName.getLocalPart();
+        _namespaceURI = qName.getNamespaceURI();
         return (_currentState = XML_START_ELEMENT);
     }
 
     /**
      * @since 2.10
      */
-    private final void _checkXsiAttributes() {
+    protected QName _checkMagicAttributes() {
         int count = _xmlReader.getAttributeCount();
         _attributeCount = count;
+        _nextAttributeIndex = 0;
+        _xsiNilFound = false;
+
+        String ns = _xmlReader.getNamespaceURI();
+        String localName = _xmlReader.getLocalName();
 
         // [dataformat-xml#354]: xsi:nul handling; at first only if first attribute
         if (count >= 1) {
@@ -666,13 +675,21 @@ private final void _checkXsiAttributes() {
                     _nextAttributeIndex = 1;
                     // but only mark as nil marker if enabled
                     _xsiNilFound = "true".equals(_xmlReader.getAttributeValue(0));
-                    return;
                 }
             }
         }
 
-        _nextAttributeIndex = 0;
-        _xsiNilFound = false;
+        // Handle PROCESS_ESCAPED_MALFORMED_TAGS
+        if (count > _nextAttributeIndex && _cfgProcessEscapedTag && localName.equals(ToXmlGenerator.ESCAPED_TAG_NAME)) {
+            final String attr = _xmlReader.getAttributeLocalName(_nextAttributeIndex);
+            final String attrNS = _xmlReader.getAttributeNamespace(_nextAttributeIndex);
+            if (Objects.equals(attrNS, ToXmlGenerator.ESCAPED_ATTR_NS) && Objects.equals(attr, ToXmlGenerator.ESCAPED_ATTR_NAME)) {
+                localName = _xmlReader.getAttributeValue(_nextAttributeIndex);
+                _nextAttributeIndex += 1;
+            }
+        }
+
+        return new QName(ns, localName);
     }
 
     /**