Fix #179

Author: cowtowncoder

avro/src/main/java/com/fasterxml/jackson/dataformat/avro/AvroGenerator.java

@@ -4,7 +4,6 @@
 import java.io.OutputStream;
 import java.math.BigDecimal;
 import java.math.BigInteger;
-import java.nio.ByteBuffer;
 
 import org.apache.avro.io.BinaryEncoder;
 
@@ -234,10 +233,10 @@ public boolean canUseSchema(FormatSchema schema) {
         return (schema instanceof AvroSchema);
     }
 
-    // 10-Sep-2019, Tatu: Should implement wrt [dataformats-binary#179], but...
-    //    can't. Not yet, will just break things. Wait until 2.11
-//    @Override
-//    public boolean canWriteBinaryNatively() { return true; }
+    // 26-Nov-2019, tatu: [dataformats-binary#179] needed this; could
+    //   only add in 2.11
+    @Override // since 2.11
+    public boolean canWriteBinaryNatively() { return true; }
 
     /*
     /**********************************************************
@@ -511,7 +510,7 @@ public void writeBinary(Base64Variant b64variant, byte[] data, int offset, int l
             writeNull();
             return;
         }
-        _avroContext.writeValue(ByteBuffer.wrap(data, offset, len));
+        _avroContext.writeBinary(data, offset, len);
     }
 
     /*

avro/src/main/java/com/fasterxml/jackson/dataformat/avro/schema/AvroSchemaHelper.java

@@ -88,7 +88,10 @@ public static boolean isStringable(AnnotatedClass type) {
     }
 
     protected static String getNamespace(JavaType type) {
-        Class<?> cls = type.getRawClass();
+        return getNamespace(type.getRawClass());
+    }
+
+    protected static String getNamespace(Class<?> cls) {
         // 16-Feb-2017, tatu: Fixed as suggested by `baharclerode@github`;
         //   NOTE: was reverted in 2.8.8, but is enabled for Jackson 2.9.
         Class<?> enclosing = cls.getEnclosingClass();
@@ -100,7 +103,11 @@ protected static String getNamespace(JavaType type) {
     }
 
     protected static String getName(JavaType type) {
-        String name = type.getRawClass().getSimpleName();
+        return getName(type.getRawClass());
+    }
+
+    protected static String getName(Class<?> cls) {
+        String name = cls.getSimpleName();
         // Alas, some characters not accepted...
         while (name.indexOf("[]") >= 0) {
             name = name.replace("[]", "Array");
@@ -142,6 +149,10 @@ public static Schema simpleSchema(JsonFormatTypes type, JavaType hint)
             }
             return Schema.create(Schema.Type.DOUBLE);
         case STRING:
+            // 26-Nov-2019, tatu: [dataformats-binary#179] UUIDs are special
+            if ((hint != null) && hint.hasRawClass(java.util.UUID.class)) {
+                return createUUIDSchema();
+            }
             return Schema.create(Schema.Type.STRING);
         case ARRAY:
         case OBJECT:
@@ -249,6 +260,17 @@ public static Schema createEnumSchema(BeanDescription bean, List<String> values)
         ), bean);
     }
 
+    /**
+     * Helper method to enclose details of expressing best Avro Schema for
+     * {@link java.util.UUID}: 16-byte fixed-length binary (alternative would
+     * be basic variable length "bytes").
+     *
+     * @since 2.11
+     */
+    public static Schema createUUIDSchema() {
+        return Schema.createFixed("UUID", "", "java.util", 16);
+    }
+
     /**
      * Looks for {@link AvroAlias @AvroAlias} on {@code bean} and adds it to {@code schema} if it exists
      * @param schema Schema to which the alias should be added

avro/src/main/java/com/fasterxml/jackson/dataformat/avro/schema/StringVisitor.java

@@ -44,6 +44,11 @@ public Schema builtAvroSchema() {
         if (_type.hasRawClass(char.class) || _type.hasRawClass(Character.class)) {
             return AvroSchemaHelper.numericAvroSchema(NumberType.INT, TypeFactory.defaultInstance().constructType(Character.class));
         }
+        // [dataformats-binary#179]: need special help with UUIDs, to coerce into Binary
+        //   (could actually be 
+        if (_type.hasRawClass(java.util.UUID.class)) {
+            return AvroSchemaHelper.createUUIDSchema();
+        }
         BeanDescription bean = _provider.getConfig().introspectClassAnnotations(_type);
         if (_enums != null) {
             Schema s = AvroSchemaHelper.createEnumSchema(bean, new ArrayList<>(_enums));

avro/src/main/java/com/fasterxml/jackson/dataformat/avro/ser/AvroWriteContext.java

@@ -2,6 +2,7 @@
 
 import java.io.IOException;
 import java.math.BigDecimal;
+import java.util.Arrays;
 import java.util.List;
 import java.util.Map;
 
@@ -120,6 +121,14 @@ public boolean writeFieldName(String name) throws IOException {
 
     public abstract void writeValue(Object value) throws IOException;
 
+    public void writeBinary(byte[] data, int offset, int len) throws IOException {
+        // 26-Nov-2019, tatu: Let's defer coercion, just need to remove fluff
+        if ((offset != 0) || (len != data.length)) {
+            data = Arrays.copyOfRange(data, offset, offset+len);
+        }
+        writeValue(data);
+    }
+
     /**
      * @since 2.5
      */

avro/src/main/java/com/fasterxml/jackson/dataformat/avro/ser/NonBSGenericDatumWriter.java

@@ -3,6 +3,7 @@
 import java.io.IOException;
 import java.math.BigDecimal;
 import java.math.BigInteger;
+import java.nio.ByteBuffer;
 import java.util.ArrayList;
 
 import org.apache.avro.Schema;
@@ -91,6 +92,27 @@ protected void write(Schema schema, Object datum, Encoder out) throws IOExceptio
                 }
             }
             break;
+        case BYTES:
+            if (datum instanceof byte[]) {
+                super.writeWithoutConversion(schema, ByteBuffer.wrap((byte[]) datum), out);
+                return;
+            }
+            break;
+        case FIXED:
+            // One more mismatch to fix
+            /*
+            if (datum instanceof ByteBuffer) {
+                byte[] buf = ((ByteBuffer) datum).array();
+                super.writeWithoutConversion(schema, new GenericData.Fixed(schema, buf), out);
+                return;
+            }
+            */
+            if (datum instanceof byte[]) {
+                super.writeWithoutConversion(schema, new GenericData.Fixed(schema, (byte[]) datum), out);
+                return;
+            }
+            break;
+
         default:
         }
         // EncodedDatum are already in an avro-encoded format and can be written out directly to the underlying encoder

avro/src/main/java/com/fasterxml/jackson/dataformat/avro/ser/ObjectWriteContext.java

@@ -78,17 +78,25 @@ public final boolean writeFieldName(String name)
     public void writeValue(Object value) throws JsonMappingException {
         _verifyValueWrite();
         if (_nextField != null) {
+            // 26-Nov-2019, tatu: Should not be needed any more, handled at a later
+            //    point in `NonBSGenericDatumWriter`
+            /*
             Schema schema = _nextField.schema();
-            if (schema.getType() == Schema.Type.FIXED && value instanceof ByteBuffer) {
-                // 13-Nov-2014 josh: AvroGenerator wraps all binary values in ByteBuffers,
-                // but avro wants FIXED, so rewrap the array, copying if necessary
-                ByteBuffer bb = (ByteBuffer) value;
-                byte[] bytes = bb.array();
-                if (bb.arrayOffset() != 0 || bb.remaining() != bytes.length) {
-                    bytes = Arrays.copyOfRange(bytes, bb.arrayOffset(), bb.remaining());
+            if (schema.getType() == Schema.Type.FIXED) {
+                if (value instanceof ByteBuffer) {
+                    // 13-Nov-2014 josh: AvroGenerator wraps all binary values in ByteBuffers,
+                    // but avro wants FIXED, so rewrap the array, copying if necessary
+                    ByteBuffer bb = (ByteBuffer) value;
+                    byte[] bytes = bb.array();
+                    if (bb.arrayOffset() != 0 || bb.remaining() != bytes.length) {
+                        bytes = Arrays.copyOfRange(bytes, bb.arrayOffset(), bb.remaining());
+                    }
+                    value = new GenericData.Fixed(schema, bytes);
+                } else if (value instanceof byte[]) {
+                    value = new GenericData.Fixed(schema, (byte[]) value);
                 }
-                value = new GenericData.Fixed(schema, bytes);
             }
+            */
             _record.put(_nextField.pos(), value);
         }
     }

avro/src/test/java/com/fasterxml/jackson/dataformat/avro/UUIDTest.java

@@ -20,7 +20,6 @@ protected UUIDWrapper() { }
     public void testUUIDRoundtrip() throws Exception
     {
         final AvroSchema schema = MAPPER.schemaFor(UUIDWrapper.class);
-//System.err.println("DEBUG: schema -> "+schema.getAvroSchema());
         UUIDWrapper input = new UUIDWrapper(UUID.nameUUIDFromBytes("BOGUS".getBytes("UTF-8")));
         byte[] avro = MAPPER.writer(schema).writeValueAsBytes(input);
 

release-notes/VERSION-2.x

@@ -10,13 +10,15 @@ Project: jackson-datatypes-binaryModules:
 
 2.11.0 (not yet released)
 
+#179: (avro) Add `AvroGenerator.canWriteBinaryNatively()` to support binary writes,
+  fix `java.util.UUID` representation
 - `AvroGenerator` overrides `getOutputContext()` properly
 
 2.10.1 (09-Nov-2019)
 
-#185: Internal parsing of tagged arrays can lead to stack overflow
+#185: (cbor) Internal parsing of tagged arrays can lead to stack overflow
  (reported by Paul A)
-#188: Unexpected `MismatchedInputException` for `byte[]` value bound to `String`
+#188: (cbor) Unexpected `MismatchedInputException` for `byte[]` value bound to `String`
   in collection/array (actual fix in `jackson-databind`)
  (reported by Yanming Z)