Fix #78

Author: cowtowncoder

pom.xml

@@ -4,7 +4,7 @@
   <parent>
     <groupId>com.fasterxml</groupId>
     <artifactId>oss-parent</artifactId>
-    <version>41</version>
+    <version>44</version>
   </parent>
 
   <artifactId>aalto-xml</artifactId>

release-notes/CREDITS

@@ -33,3 +33,7 @@ Claude Mamo (otcdlink-simpleuser@github)
 Erik Fäßler (khituras@github)
  * Contributed #75: Fixing a bug when multi-byte characters were split
   (1.3.1)
+
+Jamie Phelps (jphelp32@github)
+ * Reported #78: Async parsing turns " inside element content into apostrophe
+  (1.3.2)

release-notes/VERSION

@@ -4,6 +4,11 @@ Project: aalto-xml
 = Releases
 ------------------------------------------------------------------------
 
+1.3.2 (not yet released)
+
+#78: Async parsing turns " inside element content into apostrophe
+ (reported by Jamie P)
+
 1.3.1 (14-Jan-2022)
 
 #75: Fixing a bug when multi-byte characters were split

src/main/java/com/fasterxml/aalto/async/AsyncByteArrayScanner.java

@@ -2393,7 +2393,7 @@ protected int handleEntityInCharacters() throws XMLStreamException
                         && _inputBuffer[ptr+2] == BYTE_t
                         && _inputBuffer[ptr+3] == BYTE_SEMICOLON) {
                     _inputPtr = ptr + 4;
-                    return INT_APOS;
+                    return INT_QUOTE;
                 }
             }
         }
@@ -2949,7 +2949,7 @@ private int skipEntityInCharacters() throws XMLStreamException
                         && _inputBuffer[ptr+2] == BYTE_t
                         && _inputBuffer[ptr+3] == BYTE_SEMICOLON) {
                     _inputPtr = ptr + 4;
-                    return INT_APOS;
+                    return INT_QUOTE;
                 }
             }
         }

src/main/java/com/fasterxml/aalto/async/AsyncByteBufferScanner.java

@@ -2398,7 +2398,7 @@ protected int handleEntityInCharacters() throws XMLStreamException
                         && _inputBuffer.get(ptr+2) == BYTE_t
                         && _inputBuffer.get(ptr+3) == BYTE_SEMICOLON) {
                     _inputPtr = ptr + 4;
-                    return INT_APOS;
+                    return INT_QUOTE;
                 }
             }
         }
@@ -2954,7 +2954,7 @@ private int skipEntityInCharacters() throws XMLStreamException
                         && _inputBuffer.get(ptr+2) == BYTE_t
                         && _inputBuffer.get(ptr+3) == BYTE_SEMICOLON) {
                     _inputPtr = ptr + 4;
-                    return INT_APOS;
+                    return INT_QUOTE;
                 }
             }
         }

src/main/java/com/fasterxml/aalto/async/AsyncByteScanner.java

@@ -372,7 +372,7 @@ public void endOfInput() {
     protected void _releaseBuffers()
     {
         super._releaseBuffers();
-        if (_symbols.maybeDirty()) {
+        if ((_symbols != null) && _symbols.maybeDirty()) {
             _config.updateBBSymbols(_symbols);
         }
     }

src/test/java/async/TestCharactersParsing.java

@@ -141,7 +141,17 @@ private void _testLinefeeds(final AsyncXMLStreamReader<?> sr, final AsyncReaderW
 
     private void _testTextWithEntities(final int chunkSize, final boolean checkValues, final String SPC) throws Exception
     {
-        final String XML = SPC + "<root>a&lt;b\rMOT</root>";
+        _testTextWithEntities(chunkSize, checkValues, SPC, "&lt", "<");
+        _testTextWithEntities(chunkSize, checkValues, SPC, "&gt", ">");
+        _testTextWithEntities(chunkSize, checkValues, SPC, "&apos", "'");
+        // for [aalto-xml#78]
+        _testTextWithEntities(chunkSize, checkValues, SPC, "&quot", "\"");
+    }
+
+    private void _testTextWithEntities(final int chunkSize, final boolean checkValues, final String SPC,
+            final String entity, final String entityExpanded) throws Exception
+    {
+        final String XML = SPC + "<root>a"+entity+";b\rMOT</root>";
 
         final AsyncXMLInputFactory f = new InputFactoryImpl();
 
@@ -150,7 +160,7 @@ private void _testTextWithEntities(final int chunkSize, final boolean checkValue
         try {
             sr_array = f.createAsyncForByteArray();
             final AsyncReaderWrapperForByteArray reader_array = new AsyncReaderWrapperForByteArray(sr_array, chunkSize, XML);
-            _testTextWithEntities(sr_array, reader_array, checkValues);
+            _testTextWithEntities(sr_array, reader_array, checkValues, entityExpanded);
         } finally {
             if (sr_array != null) {
                 sr_array.close();
@@ -162,15 +172,17 @@ private void _testTextWithEntities(final int chunkSize, final boolean checkValue
         try {
             sr_buffer = f.createAsyncForByteBuffer();
             final AsyncReaderWrapperForByteBuffer reader_buffer = new AsyncReaderWrapperForByteBuffer(sr_buffer, chunkSize, XML);
-            _testTextWithEntities(sr_buffer, reader_buffer, checkValues);
+            _testTextWithEntities(sr_buffer, reader_buffer, checkValues, entityExpanded);
         } finally {
             if (sr_buffer != null) {
                 sr_buffer.close();
             }
         }
     }
 
-    private void _testTextWithEntities(final AsyncXMLStreamReader<?> sr, final AsyncReaderWrapper reader, final boolean checkValues) throws Exception
+    private void _testTextWithEntities(final AsyncXMLStreamReader<?> sr, final AsyncReaderWrapper reader,
+            final boolean checkValues,
+            final String entityExpanded) throws Exception
     {
         // should start with START_DOCUMENT, but for now skip
         int t = verifyStart(reader);
@@ -182,7 +194,7 @@ private void _testTextWithEntities(final AsyncXMLStreamReader<?> sr, final Async
         assertTokenType(CHARACTERS, reader.nextToken());
         if (checkValues) {
             String str = collectAsyncText(reader, CHARACTERS); // moves to end-element
-            assertEquals("a<b\nMOT", str);
+            assertEquals("a"+entityExpanded+"b\nMOT", str);
         } else {
             reader.nextToken();
         }