This is an automated email from the ASF dual-hosted git repository.
chaokunyang pushed a commit to branch main
in repository https://gitbox.apache.org/repos/asf/fury.git
The following commit(s) were added to refs/heads/main by this push:
new 5bd3de5f feat(java): add option to treat enum as string (#1892)
5bd3de5f is described below
commit 5bd3de5fb0e87274879e19a795bebc1148949af8
Author: orisgarno <[email protected]>
AuthorDate: Wed Oct 23 15:04:45 2024 +0700
feat(java): add option to treat enum as string (#1892)
## What does this PR do?
### Current implementation
For now fury read/write enum with ordinal approach by default. So when
serializer serialize as 1, deserializer will return the 2nd enum (index
1)
#### For example :
non updated library on serializer:
```
enum SearchMode {
CHEAPEST, EARLIEST
}
```
updated library on deserializer:
```
enum SearchMode {
RECOMMENDED, CHEAPEST, EARLIEST
}
```
- if serializer serialize CHEAPEST, deserializer will return
RECOMMENDED.
___
### New Option
This PR will allow fury to treat enum as string so it can have more
flexibility like JSON serializer.
#### For example:
non updated library on serializer:
```
enum SearchMode {
CHEAPEST, EARLIEST
}
```
updated library on deserializer:
```
enum SearchMode {
RECOMMENDED, CHEAPEST, EARLIEST
}
```
updated library with different class on deserializer:
```
enum SearchModeV2 {
RECOMMENDED, CHEAPEST, EARLIEST
}
```
- if serializer serialize CHEAPEST, deserializer will return CHEAPEST.
- if serializer serialize CHEAPEST, deserializer for different class
will still return CHEAPEST.
## Related issues
<!--
Is there any related issue? Please attach here.
- #xxxx0
- #xxxx1
- #xxxx2
-->
## Does this PR introduce any user-facing change?
<!--
If any user-facing interface changes, please [open an
issue](https://github.com/apache/fury/issues/new/choose) describing the
need to do so and update the document if necessary.
-->
- [x] Does this PR introduce any public API change?
- [ ] Does this PR introduce any binary protocol compatibility change?
## Benchmark
<!--
When the PR has an impact on performance (if you don't know whether the
PR will have an impact on performance, you can submit the PR first, and
if it will have impact on performance, the code reviewer will explain
it), be sure to attach a benchmark data here.
-->
---
docs/guide/java_serialization_guide.md | 1 +
.../main/java/org/apache/fury/config/Config.java | 7 +
.../java/org/apache/fury/config/FuryBuilder.java | 7 +
.../org/apache/fury/resolver/MetaStringBytes.java | 2 +-
.../apache/fury/resolver/MetaStringResolver.java | 4 +-
.../org/apache/fury/serializer/EnumSerializer.java | 65 ++++++++-
.../apache/fury/serializer/EnumSerializerTest.java | 145 +++++++++++++++++++++
7 files changed, 222 insertions(+), 9 deletions(-)
diff --git a/docs/guide/java_serialization_guide.md
b/docs/guide/java_serialization_guide.md
index 6d4a0d15..4715b378 100644
--- a/docs/guide/java_serialization_guide.md
+++ b/docs/guide/java_serialization_guide.md
@@ -116,6 +116,7 @@ public class Example {
| `asyncCompilationEnabled` | If enabled, serialization uses
interpreter mode first and switches to JIT serialization after async serializer
JIT for a class is finished.
[...]
| `scalaOptimizationEnabled` | Enables or disables Scala-specific
serialization optimization.
[...]
| `copyRef` | When disabled, the copy performance
will be better. But fury deep copy will ignore circular and shared reference.
Same reference of an object graph will be copied into different objects in one
`Fury#copy`.
[...]
+| `serializeEnumByName` | When Enabled, fury serialize enum by
name instead of ordinal.
[...]
## Advanced Usage
diff --git a/java/fury-core/src/main/java/org/apache/fury/config/Config.java
b/java/fury-core/src/main/java/org/apache/fury/config/Config.java
index 98422d73..98301413 100644
--- a/java/fury-core/src/main/java/org/apache/fury/config/Config.java
+++ b/java/fury-core/src/main/java/org/apache/fury/config/Config.java
@@ -60,6 +60,7 @@ public class Config implements Serializable {
private final boolean scalaOptimizationEnabled;
private transient int configHash;
private final boolean deserializeNonexistentEnumValueAsNull;
+ private final boolean serializeEnumByName;
public Config(FuryBuilder builder) {
name = builder.name;
@@ -93,6 +94,7 @@ public class Config implements Serializable {
asyncCompilationEnabled = builder.asyncCompilationEnabled;
scalaOptimizationEnabled = builder.scalaOptimizationEnabled;
deserializeNonexistentEnumValueAsNull =
builder.deserializeNonexistentEnumValueAsNull;
+ serializeEnumByName = builder.serializeEnumByName;
}
/** Returns the name for Fury serialization. */
@@ -133,6 +135,11 @@ public class Config implements Serializable {
return deserializeNonexistentEnumValueAsNull;
}
+ /** deserialize and serialize enum by name. */
+ public boolean serializeEnumByName() {
+ return serializeEnumByName;
+ }
+
/**
* Whether ignore reference tracking of all time types registered in {@link
TimeSerializers} and
* subclasses of those types when ref tracking is enabled.
diff --git
a/java/fury-core/src/main/java/org/apache/fury/config/FuryBuilder.java
b/java/fury-core/src/main/java/org/apache/fury/config/FuryBuilder.java
index 7e26d722..8e183d66 100644
--- a/java/fury-core/src/main/java/org/apache/fury/config/FuryBuilder.java
+++ b/java/fury-core/src/main/java/org/apache/fury/config/FuryBuilder.java
@@ -82,6 +82,7 @@ public final class FuryBuilder {
boolean scalaOptimizationEnabled = false;
boolean suppressClassRegistrationWarnings = true;
boolean deserializeNonexistentEnumValueAsNull = false;
+ boolean serializeEnumByName = false;
MetaCompressor metaCompressor = new DeflaterMetaCompressor();
public FuryBuilder() {}
@@ -133,6 +134,12 @@ public final class FuryBuilder {
return this;
}
+ /** deserialize and serialize enum by name. */
+ public FuryBuilder serializeEnumByName(boolean serializeEnumByName) {
+ this.serializeEnumByName = serializeEnumByName;
+ return this;
+ }
+
/**
* Whether ignore reference tracking of all time types registered in {@link
TimeSerializers} when
* ref tracking is enabled.
diff --git
a/java/fury-core/src/main/java/org/apache/fury/resolver/MetaStringBytes.java
b/java/fury-core/src/main/java/org/apache/fury/resolver/MetaStringBytes.java
index f14cbce8..56301645 100644
--- a/java/fury-core/src/main/java/org/apache/fury/resolver/MetaStringBytes.java
+++ b/java/fury-core/src/main/java/org/apache/fury/resolver/MetaStringBytes.java
@@ -26,7 +26,7 @@ import org.apache.fury.meta.MetaStringDecoder;
import org.apache.fury.util.MurmurHash3;
@Internal
-final class MetaStringBytes {
+public final class MetaStringBytes {
static final short DEFAULT_DYNAMIC_WRITE_STRING_ID = -1;
private static final int HEADER_MASK = 0xff;
diff --git
a/java/fury-core/src/main/java/org/apache/fury/resolver/MetaStringResolver.java
b/java/fury-core/src/main/java/org/apache/fury/resolver/MetaStringResolver.java
index 30ede252..c3322d16 100644
---
a/java/fury-core/src/main/java/org/apache/fury/resolver/MetaStringResolver.java
+++
b/java/fury-core/src/main/java/org/apache/fury/resolver/MetaStringResolver.java
@@ -54,7 +54,7 @@ public final class MetaStringResolver {
dynamicReadStringId = 0;
}
- MetaStringBytes getOrCreateMetaStringBytes(MetaString str) {
+ public MetaStringBytes getOrCreateMetaStringBytes(MetaString str) {
MetaStringBytes metaStringBytes = metaString2BytesMap.get(str);
if (metaStringBytes == null) {
metaStringBytes = new MetaStringBytes(str);
@@ -147,7 +147,7 @@ public final class MetaStringResolver {
}
}
- MetaStringBytes readMetaStringBytes(MemoryBuffer buffer) {
+ public MetaStringBytes readMetaStringBytes(MemoryBuffer buffer) {
int header = buffer.readVarUint32Small7();
int len = header >>> 1;
if ((header & 0b1) == 0) {
diff --git
a/java/fury-core/src/main/java/org/apache/fury/serializer/EnumSerializer.java
b/java/fury-core/src/main/java/org/apache/fury/serializer/EnumSerializer.java
index 1b23f8da..7a0f78dc 100644
---
a/java/fury-core/src/main/java/org/apache/fury/serializer/EnumSerializer.java
+++
b/java/fury-core/src/main/java/org/apache/fury/serializer/EnumSerializer.java
@@ -21,15 +21,24 @@ package org.apache.fury.serializer;
import java.util.Arrays;
import org.apache.fury.Fury;
+import org.apache.fury.collection.FuryObjectMap;
import org.apache.fury.memory.MemoryBuffer;
+import org.apache.fury.meta.Encoders;
+import org.apache.fury.meta.MetaString;
+import org.apache.fury.resolver.MetaStringBytes;
+import org.apache.fury.resolver.MetaStringResolver;
import org.apache.fury.util.Preconditions;
@SuppressWarnings("rawtypes")
-public final class EnumSerializer extends ImmutableSerializer<Enum> {
+public class EnumSerializer extends ImmutableSerializer<Enum> {
+ private final MetaStringResolver metaStringResolver;
private final Enum[] enumConstants;
+ private final FuryObjectMap<MetaStringBytes, Enum>
metaStringtoEnumRepresentation;
+ private final MetaStringBytes[] metaStringBytesArrByEnumOrdinal;
public EnumSerializer(Fury fury, Class<Enum> cls) {
super(fury, cls, false);
+ metaStringResolver = fury.getMetaStringResolver();
if (cls.isEnum()) {
enumConstants = cls.getEnumConstants();
} else {
@@ -40,20 +49,55 @@ public final class EnumSerializer extends
ImmutableSerializer<Enum> {
Preconditions.checkArgument(enclosingClass.isEnum());
enumConstants = enclosingClass.getEnumConstants();
}
+
+ metaStringBytesArrByEnumOrdinal = new
MetaStringBytes[enumConstants.length];
+
+ if (fury.getConfig().serializeEnumByName()) {
+ // as we know the size of enum is fixed, initialize the size of map with
that value
+ int initialCapacity = (int) Math.ceil(enumConstants.length / 0.5f);
+
+ metaStringtoEnumRepresentation = new FuryObjectMap<>(initialCapacity,
0.5f);
+
+ for (Enum enumConstant : enumConstants) {
+ if (enumConstant != null) {
+ MetaString ms = Encoders.GENERIC_ENCODER.encode(enumConstant.name());
+ MetaStringBytes msb =
metaStringResolver.getOrCreateMetaStringBytes(ms);
+ metaStringtoEnumRepresentation.put(msb, enumConstant);
+ metaStringBytesArrByEnumOrdinal[enumConstant.ordinal()] = msb;
+ }
+ }
+
+ } else {
+ metaStringtoEnumRepresentation = null;
+ }
}
@Override
public void write(MemoryBuffer buffer, Enum value) {
- buffer.writeVarUint32Small7(value.ordinal());
+ if (fury.getConfig().serializeEnumByName()) {
+ MetaStringBytes metaStringBytes =
metaStringBytesArrByEnumOrdinal[value.ordinal()];
+ metaStringResolver.writeMetaStringBytes(buffer, metaStringBytes);
+ } else {
+ buffer.writeVarUint32Small7(value.ordinal());
+ }
}
@Override
public Enum read(MemoryBuffer buffer) {
- int value = buffer.readVarUint32Small7();
- if (value >= enumConstants.length) {
- return handleNonexistentEnumValue(value);
+ if (fury.getConfig().serializeEnumByName()) {
+ MetaStringBytes metaStringBytes =
metaStringResolver.readMetaStringBytes(buffer);
+ Enum e = metaStringtoEnumRepresentation.get(metaStringBytes);
+ if (e != null) {
+ return e;
+ }
+ return
handleNonexistentEnumValue(metaStringBytes.decode(Encoders.GENERIC_DECODER));
+ } else {
+ int value = buffer.readVarUint32Small7();
+ if (value >= enumConstants.length) {
+ return handleNonexistentEnumValue(value);
+ }
+ return enumConstants[value];
}
- return enumConstants[value];
}
private Enum handleNonexistentEnumValue(int value) {
@@ -64,4 +108,13 @@ public final class EnumSerializer extends
ImmutableSerializer<Enum> {
String.format("Enum ordinal %s not in %s", value,
Arrays.toString(enumConstants)));
}
}
+
+ private Enum handleNonexistentEnumValue(String value) {
+ if (fury.getConfig().deserializeNonexistentEnumValueAsNull()) {
+ return null;
+ } else {
+ throw new IllegalArgumentException(
+ String.format("Enum string %s not in %s", value,
Arrays.toString(enumConstants)));
+ }
+ }
}
diff --git
a/java/fury-core/src/test/java/org/apache/fury/serializer/EnumSerializerTest.java
b/java/fury-core/src/test/java/org/apache/fury/serializer/EnumSerializerTest.java
index 8cdbc061..bec23235 100644
---
a/java/fury-core/src/test/java/org/apache/fury/serializer/EnumSerializerTest.java
+++
b/java/fury-core/src/test/java/org/apache/fury/serializer/EnumSerializerTest.java
@@ -29,6 +29,7 @@ import org.apache.fury.codegen.JaninoUtils;
import org.apache.fury.config.CompatibleMode;
import org.apache.fury.config.FuryBuilder;
import org.apache.fury.config.Language;
+import org.apache.fury.exception.DeserializationException;
import org.testng.annotations.Test;
public class EnumSerializerTest extends FuryTestBase {
@@ -101,6 +102,150 @@ public class EnumSerializerTest extends FuryTestBase {
Fury furySerialization = builderSerialization.build();
byte[] bytes = furySerialization.serialize(cls1.getEnumConstants()[1]);
Object data = furyDeserialize.deserialize(bytes);
+ assertNull(data);
+ }
+
+ @Test
+ public void testEnumSerializationAsString() {
+ String enumCode1 = "enum TestEnum1 {" + " A, B;" + "}";
+ String enumCode2 = "enum TestEnum1 {" + " B;" + "}";
+ Class<?> cls1 =
+ JaninoUtils.compileClass(getClass().getClassLoader(), "", "TestEnum1",
enumCode1);
+ Class<?> cls2 =
+ JaninoUtils.compileClass(getClass().getClassLoader(), "", "TestEnum1",
enumCode2);
+
+ Fury furyDeserialize =
+ Fury.builder()
+ .withLanguage(Language.JAVA)
+ .withCompatibleMode(CompatibleMode.COMPATIBLE)
+ .withRefTracking(true)
+ .requireClassRegistration(false)
+ .serializeEnumByName(true)
+ .withAsyncCompilation(false)
+ .build();
+ Fury furySerialization =
+ Fury.builder()
+ .withLanguage(Language.JAVA)
+ .withCompatibleMode(CompatibleMode.COMPATIBLE)
+ .withRefTracking(true)
+ .requireClassRegistration(false)
+ .serializeEnumByName(true)
+ .withAsyncCompilation(false)
+ .build();
+
+ // serialize enum "B"
+ byte[] bytes =
furySerialization.serializeJavaObject(cls1.getEnumConstants()[1]);
+
+ Object data = furyDeserialize.deserializeJavaObject(bytes, cls2);
+ assertEquals(cls2.getEnumConstants()[0], data);
+ }
+
+ @Test
+ public void testEnumSerializationAsString_differentClass() {
+ String enumCode1 = "enum TestEnum1 {" + " A, B;" + "}";
+ String enumCode2 = "enum TestEnum2 {" + " B;" + "}";
+ Class<?> cls1 =
+ JaninoUtils.compileClass(getClass().getClassLoader(), "", "TestEnum1",
enumCode1);
+ Class<?> cls2 =
+ JaninoUtils.compileClass(getClass().getClassLoader(), "", "TestEnum2",
enumCode2);
+
+ Fury furyDeserialize =
+ Fury.builder()
+ .withLanguage(Language.JAVA)
+ .withCompatibleMode(CompatibleMode.COMPATIBLE)
+ .withRefTracking(true)
+ .requireClassRegistration(false)
+ .serializeEnumByName(true)
+ .withAsyncCompilation(false)
+ .build();
+ Fury furySerialization =
+ Fury.builder()
+ .withLanguage(Language.JAVA)
+ .withCompatibleMode(CompatibleMode.COMPATIBLE)
+ .withRefTracking(true)
+ .requireClassRegistration(false)
+ .serializeEnumByName(true)
+ .withAsyncCompilation(false)
+ .build();
+
+ // serialize enum "B"
+ byte[] bytes =
furySerialization.serializeJavaObject(cls1.getEnumConstants()[1]);
+
+ Object data = furyDeserialize.deserializeJavaObject(bytes, cls2);
+ assertEquals(cls2.getEnumConstants()[0], data);
+ }
+
+ @Test
+ public void testEnumSerializationAsString_invalidEnum() {
+ String enumCode1 = "enum TestEnum1 {" + " A;" + "}";
+ String enumCode2 = "enum TestEnum2 {" + " B;" + "}";
+ Class<?> cls1 =
+ JaninoUtils.compileClass(getClass().getClassLoader(), "", "TestEnum1",
enumCode1);
+ Class<?> cls2 =
+ JaninoUtils.compileClass(getClass().getClassLoader(), "", "TestEnum2",
enumCode2);
+
+ Fury furyDeserialize =
+ Fury.builder()
+ .withLanguage(Language.JAVA)
+ .withCompatibleMode(CompatibleMode.COMPATIBLE)
+ .withRefTracking(true)
+ .requireClassRegistration(false)
+ .serializeEnumByName(true)
+ .withAsyncCompilation(false)
+ .build();
+ Fury furySerialization =
+ Fury.builder()
+ .withLanguage(Language.JAVA)
+ .withCompatibleMode(CompatibleMode.COMPATIBLE)
+ .withRefTracking(true)
+ .requireClassRegistration(false)
+ .serializeEnumByName(true)
+ .withAsyncCompilation(false)
+ .build();
+
+ byte[] bytes =
furySerialization.serializeJavaObject(cls1.getEnumConstants()[0]);
+
+ try {
+ furyDeserialize.deserializeJavaObject(bytes, cls2);
+ fail("expected to throw exception");
+ } catch (Exception e) {
+ assertTrue(e instanceof DeserializationException);
+ assertTrue(e.getCause() instanceof IllegalArgumentException);
+ }
+ }
+
+ @Test
+ public void testEnumSerializationAsString_nullValue() {
+ String enumCode1 = "enum TestEnum1 {" + " A;" + "}";
+ String enumCode2 = "enum TestEnum2 {" + " B;" + "}";
+ Class<?> cls1 =
+ JaninoUtils.compileClass(getClass().getClassLoader(), "", "TestEnum1",
enumCode1);
+ Class<?> cls2 =
+ JaninoUtils.compileClass(getClass().getClassLoader(), "", "TestEnum2",
enumCode2);
+
+ Fury furyDeserialize =
+ Fury.builder()
+ .withLanguage(Language.JAVA)
+ .withCompatibleMode(CompatibleMode.COMPATIBLE)
+ .withRefTracking(true)
+ .requireClassRegistration(false)
+ .serializeEnumByName(true)
+ .withAsyncCompilation(false)
+ .build();
+ Fury furySerialization =
+ Fury.builder()
+ .withLanguage(Language.JAVA)
+ .withCompatibleMode(CompatibleMode.COMPATIBLE)
+ .withRefTracking(true)
+ .requireClassRegistration(false)
+ .serializeEnumByName(true)
+ .withAsyncCompilation(false)
+ .build();
+
+ byte[] bytes = furySerialization.serializeJavaObject(null);
+
+ Object data = furyDeserialize.deserializeJavaObject(bytes, cls2);
+ assertNull(data);
}
@Data
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]