gson-comments/gson/src/main/java/com/google/gson/internal/bind/ReflectiveTypeAdapterFactory.java

Ignoring revisions in .git-blame-ignore-revs. Click here to bypass and see the normal blame view.

670 lines
26 KiB
Java
Raw Normal View History

Down to 22 failing tests. Consolidated all of the different code paths that we use to construct instances. We now have an ObjectConstructor class that knows what type it constructs; this means that we don't need to ever do reflection to lookup a constructor at construction time. Cleaned up some buggy type adapters, particularly around handling of null. Removed dead code for object graph navigation. Moved some classes into 'internal' so they are visible to the 'bind' subpackage. Turned some TypeAdapterFactory/TypeAdapter pairs inside out so that the TypeAdapter is now the inner class. This is necessary so that the factories can take parameters. Added an API to request the 'next' type adapter for a type. This allows type adapters to compose other type adapters. We're using this in two places: - where the user has excluded a type from serialization but not deserialization, we need to use the "default" deserialization but interpose null on serialization. We create a type adapter that delegates for one and returns null for the other. - similarly when a DOM type serializer is registered but no deserializer, or vice versa. This is the biggest change to the MiniGson core. For backwards compatibility, return null for the empty string. Simplify JsonSerializationContext/JsonDeserializationContext to simply call through to GSON. SerializeDefault is currently unsupported. More useful error messages when calling getAsBoolean on a JsonNull. Remove currently unused MemoryRefStack. We might need this back again, though wiring it back in will be much more difficult because we don't interject ourselves between the users' various type adapters.
2011-09-11 09:04:56 +02:00
/*
* Copyright (C) 2011 Google Inc.
*
* Licensed under the Apache License, Version 2.0 (the "License");
* you may not use this file except in compliance with the License.
* You may obtain a copy of the License at
*
* http://www.apache.org/licenses/LICENSE-2.0
*
* Unless required by applicable law or agreed to in writing, software
* distributed under the License is distributed on an "AS IS" BASIS,
* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
* See the License for the specific language governing permissions and
* limitations under the License.
*/
package com.google.gson.internal.bind;
import com.google.gson.FieldNamingStrategy;
import com.google.gson.Gson;
import com.google.gson.JsonIOException;
import com.google.gson.JsonParseException;
Down to 22 failing tests. Consolidated all of the different code paths that we use to construct instances. We now have an ObjectConstructor class that knows what type it constructs; this means that we don't need to ever do reflection to lookup a constructor at construction time. Cleaned up some buggy type adapters, particularly around handling of null. Removed dead code for object graph navigation. Moved some classes into 'internal' so they are visible to the 'bind' subpackage. Turned some TypeAdapterFactory/TypeAdapter pairs inside out so that the TypeAdapter is now the inner class. This is necessary so that the factories can take parameters. Added an API to request the 'next' type adapter for a type. This allows type adapters to compose other type adapters. We're using this in two places: - where the user has excluded a type from serialization but not deserialization, we need to use the "default" deserialization but interpose null on serialization. We create a type adapter that delegates for one and returns null for the other. - similarly when a DOM type serializer is registered but no deserializer, or vice versa. This is the biggest change to the MiniGson core. For backwards compatibility, return null for the empty string. Simplify JsonSerializationContext/JsonDeserializationContext to simply call through to GSON. SerializeDefault is currently unsupported. More useful error messages when calling getAsBoolean on a JsonNull. Remove currently unused MemoryRefStack. We might need this back again, though wiring it back in will be much more difficult because we don't interject ourselves between the users' various type adapters.
2011-09-11 09:04:56 +02:00
import com.google.gson.JsonSyntaxException;
import com.google.gson.ReflectionAccessFilter;
import com.google.gson.ReflectionAccessFilter.FilterResult;
import com.google.gson.TypeAdapter;
import com.google.gson.TypeAdapterFactory;
2014-03-26 18:59:54 +01:00
import com.google.gson.annotations.JsonAdapter;
import com.google.gson.annotations.SerializedName;
Down to 22 failing tests. Consolidated all of the different code paths that we use to construct instances. We now have an ObjectConstructor class that knows what type it constructs; this means that we don't need to ever do reflection to lookup a constructor at construction time. Cleaned up some buggy type adapters, particularly around handling of null. Removed dead code for object graph navigation. Moved some classes into 'internal' so they are visible to the 'bind' subpackage. Turned some TypeAdapterFactory/TypeAdapter pairs inside out so that the TypeAdapter is now the inner class. This is necessary so that the factories can take parameters. Added an API to request the 'next' type adapter for a type. This allows type adapters to compose other type adapters. We're using this in two places: - where the user has excluded a type from serialization but not deserialization, we need to use the "default" deserialization but interpose null on serialization. We create a type adapter that delegates for one and returns null for the other. - similarly when a DOM type serializer is registered but no deserializer, or vice versa. This is the biggest change to the MiniGson core. For backwards compatibility, return null for the empty string. Simplify JsonSerializationContext/JsonDeserializationContext to simply call through to GSON. SerializeDefault is currently unsupported. More useful error messages when calling getAsBoolean on a JsonNull. Remove currently unused MemoryRefStack. We might need this back again, though wiring it back in will be much more difficult because we don't interject ourselves between the users' various type adapters.
2011-09-11 09:04:56 +02:00
import com.google.gson.internal.$Gson$Types;
import com.google.gson.internal.ConstructorConstructor;
import com.google.gson.internal.Excluder;
Down to 22 failing tests. Consolidated all of the different code paths that we use to construct instances. We now have an ObjectConstructor class that knows what type it constructs; this means that we don't need to ever do reflection to lookup a constructor at construction time. Cleaned up some buggy type adapters, particularly around handling of null. Removed dead code for object graph navigation. Moved some classes into 'internal' so they are visible to the 'bind' subpackage. Turned some TypeAdapterFactory/TypeAdapter pairs inside out so that the TypeAdapter is now the inner class. This is necessary so that the factories can take parameters. Added an API to request the 'next' type adapter for a type. This allows type adapters to compose other type adapters. We're using this in two places: - where the user has excluded a type from serialization but not deserialization, we need to use the "default" deserialization but interpose null on serialization. We create a type adapter that delegates for one and returns null for the other. - similarly when a DOM type serializer is registered but no deserializer, or vice versa. This is the biggest change to the MiniGson core. For backwards compatibility, return null for the empty string. Simplify JsonSerializationContext/JsonDeserializationContext to simply call through to GSON. SerializeDefault is currently unsupported. More useful error messages when calling getAsBoolean on a JsonNull. Remove currently unused MemoryRefStack. We might need this back again, though wiring it back in will be much more difficult because we don't interject ourselves between the users' various type adapters.
2011-09-11 09:04:56 +02:00
import com.google.gson.internal.ObjectConstructor;
import com.google.gson.internal.Primitives;
import com.google.gson.internal.ReflectionAccessFilterHelper;
import com.google.gson.internal.TroubleshootingGuide;
import com.google.gson.internal.reflect.ReflectionHelper;
Down to 22 failing tests. Consolidated all of the different code paths that we use to construct instances. We now have an ObjectConstructor class that knows what type it constructs; this means that we don't need to ever do reflection to lookup a constructor at construction time. Cleaned up some buggy type adapters, particularly around handling of null. Removed dead code for object graph navigation. Moved some classes into 'internal' so they are visible to the 'bind' subpackage. Turned some TypeAdapterFactory/TypeAdapter pairs inside out so that the TypeAdapter is now the inner class. This is necessary so that the factories can take parameters. Added an API to request the 'next' type adapter for a type. This allows type adapters to compose other type adapters. We're using this in two places: - where the user has excluded a type from serialization but not deserialization, we need to use the "default" deserialization but interpose null on serialization. We create a type adapter that delegates for one and returns null for the other. - similarly when a DOM type serializer is registered but no deserializer, or vice versa. This is the biggest change to the MiniGson core. For backwards compatibility, return null for the empty string. Simplify JsonSerializationContext/JsonDeserializationContext to simply call through to GSON. SerializeDefault is currently unsupported. More useful error messages when calling getAsBoolean on a JsonNull. Remove currently unused MemoryRefStack. We might need this back again, though wiring it back in will be much more difficult because we don't interject ourselves between the users' various type adapters.
2011-09-11 09:04:56 +02:00
import com.google.gson.reflect.TypeToken;
import com.google.gson.stream.JsonReader;
import com.google.gson.stream.JsonToken;
import com.google.gson.stream.JsonWriter;
import java.io.IOException;
import java.lang.reflect.AccessibleObject;
import java.lang.reflect.Constructor;
import java.lang.reflect.Field;
import java.lang.reflect.InvocationTargetException;
import java.lang.reflect.Member;
import java.lang.reflect.Method;
import java.lang.reflect.Modifier;
import java.lang.reflect.Type;
import java.util.ArrayList;
import java.util.Arrays;
import java.util.Collections;
import java.util.HashMap;
import java.util.LinkedHashMap;
import java.util.List;
import java.util.Map;
Down to 22 failing tests. Consolidated all of the different code paths that we use to construct instances. We now have an ObjectConstructor class that knows what type it constructs; this means that we don't need to ever do reflection to lookup a constructor at construction time. Cleaned up some buggy type adapters, particularly around handling of null. Removed dead code for object graph navigation. Moved some classes into 'internal' so they are visible to the 'bind' subpackage. Turned some TypeAdapterFactory/TypeAdapter pairs inside out so that the TypeAdapter is now the inner class. This is necessary so that the factories can take parameters. Added an API to request the 'next' type adapter for a type. This allows type adapters to compose other type adapters. We're using this in two places: - where the user has excluded a type from serialization but not deserialization, we need to use the "default" deserialization but interpose null on serialization. We create a type adapter that delegates for one and returns null for the other. - similarly when a DOM type serializer is registered but no deserializer, or vice versa. This is the biggest change to the MiniGson core. For backwards compatibility, return null for the empty string. Simplify JsonSerializationContext/JsonDeserializationContext to simply call through to GSON. SerializeDefault is currently unsupported. More useful error messages when calling getAsBoolean on a JsonNull. Remove currently unused MemoryRefStack. We might need this back again, though wiring it back in will be much more difficult because we don't interject ourselves between the users' various type adapters.
2011-09-11 09:04:56 +02:00
/** Type adapter that reflects over the fields and methods of a class. */
public final class ReflectiveTypeAdapterFactory implements TypeAdapterFactory {
Down to 22 failing tests. Consolidated all of the different code paths that we use to construct instances. We now have an ObjectConstructor class that knows what type it constructs; this means that we don't need to ever do reflection to lookup a constructor at construction time. Cleaned up some buggy type adapters, particularly around handling of null. Removed dead code for object graph navigation. Moved some classes into 'internal' so they are visible to the 'bind' subpackage. Turned some TypeAdapterFactory/TypeAdapter pairs inside out so that the TypeAdapter is now the inner class. This is necessary so that the factories can take parameters. Added an API to request the 'next' type adapter for a type. This allows type adapters to compose other type adapters. We're using this in two places: - where the user has excluded a type from serialization but not deserialization, we need to use the "default" deserialization but interpose null on serialization. We create a type adapter that delegates for one and returns null for the other. - similarly when a DOM type serializer is registered but no deserializer, or vice versa. This is the biggest change to the MiniGson core. For backwards compatibility, return null for the empty string. Simplify JsonSerializationContext/JsonDeserializationContext to simply call through to GSON. SerializeDefault is currently unsupported. More useful error messages when calling getAsBoolean on a JsonNull. Remove currently unused MemoryRefStack. We might need this back again, though wiring it back in will be much more difficult because we don't interject ourselves between the users' various type adapters.
2011-09-11 09:04:56 +02:00
private final ConstructorConstructor constructorConstructor;
private final FieldNamingStrategy fieldNamingPolicy;
private final Excluder excluder;
private final JsonAdapterAnnotationTypeAdapterFactory jsonAdapterFactory;
private final List<ReflectionAccessFilter> reflectionFilters;
Down to 22 failing tests. Consolidated all of the different code paths that we use to construct instances. We now have an ObjectConstructor class that knows what type it constructs; this means that we don't need to ever do reflection to lookup a constructor at construction time. Cleaned up some buggy type adapters, particularly around handling of null. Removed dead code for object graph navigation. Moved some classes into 'internal' so they are visible to the 'bind' subpackage. Turned some TypeAdapterFactory/TypeAdapter pairs inside out so that the TypeAdapter is now the inner class. This is necessary so that the factories can take parameters. Added an API to request the 'next' type adapter for a type. This allows type adapters to compose other type adapters. We're using this in two places: - where the user has excluded a type from serialization but not deserialization, we need to use the "default" deserialization but interpose null on serialization. We create a type adapter that delegates for one and returns null for the other. - similarly when a DOM type serializer is registered but no deserializer, or vice versa. This is the biggest change to the MiniGson core. For backwards compatibility, return null for the empty string. Simplify JsonSerializationContext/JsonDeserializationContext to simply call through to GSON. SerializeDefault is currently unsupported. More useful error messages when calling getAsBoolean on a JsonNull. Remove currently unused MemoryRefStack. We might need this back again, though wiring it back in will be much more difficult because we don't interject ourselves between the users' various type adapters.
2011-09-11 09:04:56 +02:00
public ReflectiveTypeAdapterFactory(
ConstructorConstructor constructorConstructor,
FieldNamingStrategy fieldNamingPolicy,
Excluder excluder,
JsonAdapterAnnotationTypeAdapterFactory jsonAdapterFactory,
List<ReflectionAccessFilter> reflectionFilters) {
Down to 22 failing tests. Consolidated all of the different code paths that we use to construct instances. We now have an ObjectConstructor class that knows what type it constructs; this means that we don't need to ever do reflection to lookup a constructor at construction time. Cleaned up some buggy type adapters, particularly around handling of null. Removed dead code for object graph navigation. Moved some classes into 'internal' so they are visible to the 'bind' subpackage. Turned some TypeAdapterFactory/TypeAdapter pairs inside out so that the TypeAdapter is now the inner class. This is necessary so that the factories can take parameters. Added an API to request the 'next' type adapter for a type. This allows type adapters to compose other type adapters. We're using this in two places: - where the user has excluded a type from serialization but not deserialization, we need to use the "default" deserialization but interpose null on serialization. We create a type adapter that delegates for one and returns null for the other. - similarly when a DOM type serializer is registered but no deserializer, or vice versa. This is the biggest change to the MiniGson core. For backwards compatibility, return null for the empty string. Simplify JsonSerializationContext/JsonDeserializationContext to simply call through to GSON. SerializeDefault is currently unsupported. More useful error messages when calling getAsBoolean on a JsonNull. Remove currently unused MemoryRefStack. We might need this back again, though wiring it back in will be much more difficult because we don't interject ourselves between the users' various type adapters.
2011-09-11 09:04:56 +02:00
this.constructorConstructor = constructorConstructor;
this.fieldNamingPolicy = fieldNamingPolicy;
this.excluder = excluder;
this.jsonAdapterFactory = jsonAdapterFactory;
this.reflectionFilters = reflectionFilters;
Down to 22 failing tests. Consolidated all of the different code paths that we use to construct instances. We now have an ObjectConstructor class that knows what type it constructs; this means that we don't need to ever do reflection to lookup a constructor at construction time. Cleaned up some buggy type adapters, particularly around handling of null. Removed dead code for object graph navigation. Moved some classes into 'internal' so they are visible to the 'bind' subpackage. Turned some TypeAdapterFactory/TypeAdapter pairs inside out so that the TypeAdapter is now the inner class. This is necessary so that the factories can take parameters. Added an API to request the 'next' type adapter for a type. This allows type adapters to compose other type adapters. We're using this in two places: - where the user has excluded a type from serialization but not deserialization, we need to use the "default" deserialization but interpose null on serialization. We create a type adapter that delegates for one and returns null for the other. - similarly when a DOM type serializer is registered but no deserializer, or vice versa. This is the biggest change to the MiniGson core. For backwards compatibility, return null for the empty string. Simplify JsonSerializationContext/JsonDeserializationContext to simply call through to GSON. SerializeDefault is currently unsupported. More useful error messages when calling getAsBoolean on a JsonNull. Remove currently unused MemoryRefStack. We might need this back again, though wiring it back in will be much more difficult because we don't interject ourselves between the users' various type adapters.
2011-09-11 09:04:56 +02:00
}
private boolean includeField(Field f, boolean serialize) {
return !excluder.excludeField(f, serialize);
Down to 22 failing tests. Consolidated all of the different code paths that we use to construct instances. We now have an ObjectConstructor class that knows what type it constructs; this means that we don't need to ever do reflection to lookup a constructor at construction time. Cleaned up some buggy type adapters, particularly around handling of null. Removed dead code for object graph navigation. Moved some classes into 'internal' so they are visible to the 'bind' subpackage. Turned some TypeAdapterFactory/TypeAdapter pairs inside out so that the TypeAdapter is now the inner class. This is necessary so that the factories can take parameters. Added an API to request the 'next' type adapter for a type. This allows type adapters to compose other type adapters. We're using this in two places: - where the user has excluded a type from serialization but not deserialization, we need to use the "default" deserialization but interpose null on serialization. We create a type adapter that delegates for one and returns null for the other. - similarly when a DOM type serializer is registered but no deserializer, or vice versa. This is the biggest change to the MiniGson core. For backwards compatibility, return null for the empty string. Simplify JsonSerializationContext/JsonDeserializationContext to simply call through to GSON. SerializeDefault is currently unsupported. More useful error messages when calling getAsBoolean on a JsonNull. Remove currently unused MemoryRefStack. We might need this back again, though wiring it back in will be much more difficult because we don't interject ourselves between the users' various type adapters.
2011-09-11 09:04:56 +02:00
}
/** first element holds the default name */
Fix error prone warns (#2320) * Adds `@SuppressWarnings("NarrowingCompoundAssignment")` * Adds `@SuppressWarnings("TypeParameterUnusedInFormals")` * Adds `@SuppressWarnings("JavaUtilDate")` * Adds a limit to `String.split()` * Add `error_prone_annotations` to the `pom.xml` * Adds `@InlineMe(...)` to deprecated methods * Adds `@SuppressWarnings("ImmutableEnumChecker")` * Adds `@SuppressWarnings("ModifiedButNotUsed")` * Adds `@SuppressWarnings("MixedMutabilityReturnType")` * Removes an unused import * Adds `requires` to `module-info.java` * Adds ErrorProne `link` into `pom.xml` * Remove unused imports Removed from: - ParseBenchmark * Adds `@SuppressWarnings("EqualsGetClass")` * Excludes from `proto` just the generated code. Replaces `.*proto.*` with `.*/generated-test-sources/protobuf/.*` in such way will be excluded just the generated code and not the whole `proto` directory * Removes an unused variable Removes the `descriptor` variable because is unused. * Fixes the `BadImport` warn into `ProtosWithAnnotationsTest` * Fixes the `BadImport` warns into `ProtosWithAnnotationsTest` * Enables ErrorProne in `gson/src/test.*` Removes the `gson/src/test.*` path from the `-XepExcludedPaths` parameter of the ErrorProne plugin * Fixes `UnusedVariable` warns This commit fix all `UnusedVariable` warns given by ErrorProne in the `gson/src/test.*` path. Some field is annotated with `@Keep` that means is used by reflection. In some other case the record is annotated with `@SuppressWarnings("unused")` * Fixes `JavaUtilDate` warns This commit fix all `JavaUtilDate` warns given by ErrorProne in the `gson/src/test.*` path. Classes/Methods are annotated with `@SuppressWarnings("JavaUtilDate")` because it's not possible use differente Date API. * Fixes `EqualsGetClass` warns This commit fix all `EqualsGetClass` warns given by ErrorProne in the `gson/src/test.*` path. I have rewrite the `equals()` methods to use `instanceof` * Replaces pattern matching for instanceof with casts * Fixes `JdkObsolete` warns This commit fix all `JdkObsolete` warns given by ErrorProne in the `gson/src/test.*` path. In some cases I have replaced the obsolete JDK classes with the newest alternatives. In other cases, I have added the `@SuppressWarnings("JdkObsolete")` * Fixes `ClassCanBeStatic` warns This commit fix all `ClassCanBeStatic` warns given by ErrorProne in the `gson/src/test.*` path. I have added the `static` keyword, or I have added `@SuppressWarnings("ClassCanBeStatic")` * Fixes `UndefinedEquals` warns This commit fix all `UndefinedEquals` warns given by ErrorProne in the `gson/src/test.*` path. I have added `@SuppressWarnings("UndefinedEquals")` or fixed the asserts Note: In this commit I have also renamed a test from `testRawCollectionDeserializationNotAlllowed` to `testRawCollectionDeserializationNotAllowed` * Fixes `GetClassOnEnum` warns This commit fix all `GetClassOnEnum` warns given by ErrorProne in the `gson/src/test.*` path. I have replaced the `.getClass()` with `.getDeclaringClass()` * Fixes `ImmutableEnumChecker` warns This commit fix all `ImmutableEnumChecker` warns given by ErrorProne in the `gson/src/test.*` path. I have added the `final` keyword, or I have added the `@SuppressWarnings("ImmutableEnumChecker")` annotation * Fixes `StaticAssignmentOfThrowable` warns This commit fix all `StaticAssignmentOfThrowable` warns given by ErrorProne in the `gson/src/test.*` path. * Fixes `AssertionFailureIgnored` warns This commit fix all `AssertionFailureIgnored` warns given by ErrorProne in the `gson/src/test.*` path. I have added the `@SuppressWarnings("AssertionFailureIgnored")` annotation * Fixes `ModifiedButNotUsed` warns This commit fix all `ModifiedButNotUsed` warns given by ErrorProne in the `gson/src/test.*` path. I have added the `@SuppressWarnings("ModifiedButNotUsed")` annotation * Fixes `MissingSummary` warns This commit fix all `MissingSummary` warns given by ErrorProne in the `gson/src/test.*` path. I have remove the Javadoc `@author` * Fixes `FloatingPointLiteralPrecision` warns This commit fix all `FloatingPointLiteralPrecision` warns given by ErrorProne in the `gson/src/test.*` path. I have added the `@SuppressWarnings("FloatingPointLiteralPrecision")` annotation * Fixes `StringSplitter` warns This commit fix all `StringSplitter` warns given by ErrorProne in the `gson/src/test.*` path. I have replaced the `String.split(...)` with `Splitter` * Fixes `EmptyCatch` warns This commit fix all `EmptyCatch` warns given by ErrorProne in the `gson/src/test.*` path. * Fixes `UnicodeEscape` warns This commit fix all `UnicodeEscape` warns given by ErrorProne in the `gson/src/test.*` path. * Fixes `EmptyBlockTag` warns This commit fix all `EmptyBlockTag` warns given by ErrorProne in the `gson/src/test.*` path. * Fixes `LongFloatConversion` warns This commit fix all `LongFloatConversion` warns given by ErrorProne in the `gson/src/test.*` path. * Fixes `LongDoubleConversion` warns This commit fix all `LongDoubleConversion` warns given by ErrorProne in the `gson/src/test.*` path. * Fixes `TruthAssertExpected` warns This commit fix all `TruthAssertExpected` warns given by ErrorProne in the `gson/src/test.*` path. * Fixes `UnusedMethod` warns This commit fix all `UnusedMethod` warns given by ErrorProne in the `gson/src/test.*` path. * Fixes `UnusedTypeParameter` warns This commit fix all `UnusedTypeParameter` warns given by ErrorProne in the `gson/src/test.*` path. * Fixes `CatchFail` warns This commit fix all `CatchFail` warns given by ErrorProne in the `gson/src/test.*` path. * Fixes `MathAbsoluteNegative` warns This commit fix all `MathAbsoluteNegative` warns given by ErrorProne in the `gson/src/test.*` path. * Fixes `LoopOverCharArray` warns This commit fix all `LoopOverCharArray` warns given by ErrorProne in the `gson/src/test.*` path. `toCharArray` allocates a new array, using `charAt` is more efficient * Fixes `HidingField` warns This commit fix all `HidingField` warns given by ErrorProne in the `gson/src/test.*` path. * Implements code review feedback * Implements code review feedback This commit implements some other code review feedback * Enable ErrorProne in `extra` Thi commit removes the `.*extras/src/test.*` path from the `-XepExcludedPaths` parameter of ErrorProne. * Fix the `JavaUtilDate` warns This commit fix all `JavaUtilDate` warns given by ErrorProne in the `extras/src/test.*` path. * Implements code review feedback * Removes redundant new-line * Implements code review feedback * Adds `JDK11` to run test with `--release 11` * Revert "Adds `JDK11` to run test with `--release 11`" This reverts commit a7cca386098ae847a10a31c09c3ab9b11eee5920.
2023-03-01 23:23:27 +01:00
@SuppressWarnings("MixedMutabilityReturnType")
private List<String> getFieldNames(Field f) {
SerializedName annotation = f.getAnnotation(SerializedName.class);
if (annotation == null) {
String name = fieldNamingPolicy.translateName(f);
return Collections.singletonList(name);
}
String serializedName = annotation.value();
String[] alternates = annotation.alternate();
if (alternates.length == 0) {
return Collections.singletonList(serializedName);
}
List<String> fieldNames = new ArrayList<>(alternates.length + 1);
fieldNames.add(serializedName);
Collections.addAll(fieldNames, alternates);
return fieldNames;
Down to 22 failing tests. Consolidated all of the different code paths that we use to construct instances. We now have an ObjectConstructor class that knows what type it constructs; this means that we don't need to ever do reflection to lookup a constructor at construction time. Cleaned up some buggy type adapters, particularly around handling of null. Removed dead code for object graph navigation. Moved some classes into 'internal' so they are visible to the 'bind' subpackage. Turned some TypeAdapterFactory/TypeAdapter pairs inside out so that the TypeAdapter is now the inner class. This is necessary so that the factories can take parameters. Added an API to request the 'next' type adapter for a type. This allows type adapters to compose other type adapters. We're using this in two places: - where the user has excluded a type from serialization but not deserialization, we need to use the "default" deserialization but interpose null on serialization. We create a type adapter that delegates for one and returns null for the other. - similarly when a DOM type serializer is registered but no deserializer, or vice versa. This is the biggest change to the MiniGson core. For backwards compatibility, return null for the empty string. Simplify JsonSerializationContext/JsonDeserializationContext to simply call through to GSON. SerializeDefault is currently unsupported. More useful error messages when calling getAsBoolean on a JsonNull. Remove currently unused MemoryRefStack. We might need this back again, though wiring it back in will be much more difficult because we don't interject ourselves between the users' various type adapters.
2011-09-11 09:04:56 +02:00
}
@Override
public <T> TypeAdapter<T> create(Gson gson, final TypeToken<T> type) {
Down to 22 failing tests. Consolidated all of the different code paths that we use to construct instances. We now have an ObjectConstructor class that knows what type it constructs; this means that we don't need to ever do reflection to lookup a constructor at construction time. Cleaned up some buggy type adapters, particularly around handling of null. Removed dead code for object graph navigation. Moved some classes into 'internal' so they are visible to the 'bind' subpackage. Turned some TypeAdapterFactory/TypeAdapter pairs inside out so that the TypeAdapter is now the inner class. This is necessary so that the factories can take parameters. Added an API to request the 'next' type adapter for a type. This allows type adapters to compose other type adapters. We're using this in two places: - where the user has excluded a type from serialization but not deserialization, we need to use the "default" deserialization but interpose null on serialization. We create a type adapter that delegates for one and returns null for the other. - similarly when a DOM type serializer is registered but no deserializer, or vice versa. This is the biggest change to the MiniGson core. For backwards compatibility, return null for the empty string. Simplify JsonSerializationContext/JsonDeserializationContext to simply call through to GSON. SerializeDefault is currently unsupported. More useful error messages when calling getAsBoolean on a JsonNull. Remove currently unused MemoryRefStack. We might need this back again, though wiring it back in will be much more difficult because we don't interject ourselves between the users' various type adapters.
2011-09-11 09:04:56 +02:00
Class<? super T> raw = type.getRawType();
if (!Object.class.isAssignableFrom(raw)) {
return null; // it's a primitive!
}
// Don't allow using reflection on anonymous and local classes because synthetic fields for
// captured enclosing values make this unreliable
if (ReflectionHelper.isAnonymousOrNonStaticLocal(raw)) {
// This adapter just serializes and deserializes null, ignoring the actual values
// This is done for backward compatibility; troubleshooting-wise it might be better to throw
// exceptions
return new TypeAdapter<T>() {
@Override
public T read(JsonReader in) throws IOException {
in.skipValue();
return null;
}
@Override
public void write(JsonWriter out, T value) throws IOException {
out.nullValue();
}
@Override
public String toString() {
return "AnonymousOrNonStaticLocalClassAdapter";
}
};
}
FilterResult filterResult =
ReflectionAccessFilterHelper.getFilterResult(reflectionFilters, raw);
if (filterResult == FilterResult.BLOCK_ALL) {
throw new JsonIOException(
"ReflectionAccessFilter does not permit using reflection for "
+ raw
+ ". Register a TypeAdapter for this type or adjust the access filter.");
}
boolean blockInaccessible = filterResult == FilterResult.BLOCK_INACCESSIBLE;
// If the type is actually a Java Record, we need to use the RecordAdapter instead. This will
// always be false on JVMs that do not support records.
if (ReflectionHelper.isRecord(raw)) {
@SuppressWarnings("unchecked")
TypeAdapter<T> adapter =
(TypeAdapter<T>)
new RecordAdapter<>(
raw, getBoundFields(gson, type, raw, blockInaccessible, true), blockInaccessible);
return adapter;
}
ObjectConstructor<T> constructor = constructorConstructor.get(type);
return new FieldReflectionAdapter<>(
constructor, getBoundFields(gson, type, raw, blockInaccessible, false));
}
private static <M extends AccessibleObject & Member> void checkAccessible(
Object object, M member) {
if (!ReflectionAccessFilterHelper.canAccess(
member, Modifier.isStatic(member.getModifiers()) ? null : object)) {
String memberDescription = ReflectionHelper.getAccessibleObjectDescription(member, true);
throw new JsonIOException(
memberDescription
+ " is not accessible and ReflectionAccessFilter does not permit making it"
+ " accessible. Register a TypeAdapter for the declaring type, adjust the access"
+ " filter or increase the visibility of the element and its declaring type.");
}
Down to 22 failing tests. Consolidated all of the different code paths that we use to construct instances. We now have an ObjectConstructor class that knows what type it constructs; this means that we don't need to ever do reflection to lookup a constructor at construction time. Cleaned up some buggy type adapters, particularly around handling of null. Removed dead code for object graph navigation. Moved some classes into 'internal' so they are visible to the 'bind' subpackage. Turned some TypeAdapterFactory/TypeAdapter pairs inside out so that the TypeAdapter is now the inner class. This is necessary so that the factories can take parameters. Added an API to request the 'next' type adapter for a type. This allows type adapters to compose other type adapters. We're using this in two places: - where the user has excluded a type from serialization but not deserialization, we need to use the "default" deserialization but interpose null on serialization. We create a type adapter that delegates for one and returns null for the other. - similarly when a DOM type serializer is registered but no deserializer, or vice versa. This is the biggest change to the MiniGson core. For backwards compatibility, return null for the empty string. Simplify JsonSerializationContext/JsonDeserializationContext to simply call through to GSON. SerializeDefault is currently unsupported. More useful error messages when calling getAsBoolean on a JsonNull. Remove currently unused MemoryRefStack. We might need this back again, though wiring it back in will be much more difficult because we don't interject ourselves between the users' various type adapters.
2011-09-11 09:04:56 +02:00
}
private BoundField createBoundField(
final Gson context,
final Field field,
final Method accessor,
final String serializedName,
final TypeToken<?> fieldType,
final boolean serialize,
final boolean blockInaccessible) {
Down to 22 failing tests. Consolidated all of the different code paths that we use to construct instances. We now have an ObjectConstructor class that knows what type it constructs; this means that we don't need to ever do reflection to lookup a constructor at construction time. Cleaned up some buggy type adapters, particularly around handling of null. Removed dead code for object graph navigation. Moved some classes into 'internal' so they are visible to the 'bind' subpackage. Turned some TypeAdapterFactory/TypeAdapter pairs inside out so that the TypeAdapter is now the inner class. This is necessary so that the factories can take parameters. Added an API to request the 'next' type adapter for a type. This allows type adapters to compose other type adapters. We're using this in two places: - where the user has excluded a type from serialization but not deserialization, we need to use the "default" deserialization but interpose null on serialization. We create a type adapter that delegates for one and returns null for the other. - similarly when a DOM type serializer is registered but no deserializer, or vice versa. This is the biggest change to the MiniGson core. For backwards compatibility, return null for the empty string. Simplify JsonSerializationContext/JsonDeserializationContext to simply call through to GSON. SerializeDefault is currently unsupported. More useful error messages when calling getAsBoolean on a JsonNull. Remove currently unused MemoryRefStack. We might need this back again, though wiring it back in will be much more difficult because we don't interject ourselves between the users' various type adapters.
2011-09-11 09:04:56 +02:00
final boolean isPrimitive = Primitives.isPrimitive(fieldType.getRawType());
int modifiers = field.getModifiers();
final boolean isStaticFinalField = Modifier.isStatic(modifiers) && Modifier.isFinal(modifiers);
JsonAdapter annotation = field.getAnnotation(JsonAdapter.class);
TypeAdapter<?> mapped = null;
if (annotation != null) {
// This is not safe; requires that user has specified correct adapter class for @JsonAdapter
mapped =
jsonAdapterFactory.getTypeAdapter(
constructorConstructor, context, fieldType, annotation, false);
}
final boolean jsonAdapterPresent = mapped != null;
if (mapped == null) {
mapped = context.getAdapter(fieldType);
}
@SuppressWarnings("unchecked")
final TypeAdapter<Object> typeAdapter = (TypeAdapter<Object>) mapped;
final TypeAdapter<Object> writeTypeAdapter;
if (serialize) {
writeTypeAdapter =
jsonAdapterPresent
? typeAdapter
: new TypeAdapterRuntimeTypeWrapper<>(context, typeAdapter, fieldType.getType());
} else {
// Will never actually be used, but we set it to avoid confusing nullness-analysis tools
writeTypeAdapter = typeAdapter;
}
return new BoundField(serializedName, field) {
@Override
void write(JsonWriter writer, Object source) throws IOException, IllegalAccessException {
if (blockInaccessible) {
if (accessor == null) {
checkAccessible(source, field);
} else {
// Note: This check might actually be redundant because access check for canonical
// constructor should have failed already
checkAccessible(source, accessor);
}
}
Object fieldValue;
if (accessor != null) {
try {
fieldValue = accessor.invoke(source);
} catch (InvocationTargetException e) {
String accessorDescription =
ReflectionHelper.getAccessibleObjectDescription(accessor, false);
throw new JsonIOException(
"Accessor " + accessorDescription + " threw exception", e.getCause());
}
} else {
fieldValue = field.get(source);
}
if (fieldValue == source) {
// avoid direct recursion
return;
}
writer.name(serializedName);
writeTypeAdapter.write(writer, fieldValue);
Down to 22 failing tests. Consolidated all of the different code paths that we use to construct instances. We now have an ObjectConstructor class that knows what type it constructs; this means that we don't need to ever do reflection to lookup a constructor at construction time. Cleaned up some buggy type adapters, particularly around handling of null. Removed dead code for object graph navigation. Moved some classes into 'internal' so they are visible to the 'bind' subpackage. Turned some TypeAdapterFactory/TypeAdapter pairs inside out so that the TypeAdapter is now the inner class. This is necessary so that the factories can take parameters. Added an API to request the 'next' type adapter for a type. This allows type adapters to compose other type adapters. We're using this in two places: - where the user has excluded a type from serialization but not deserialization, we need to use the "default" deserialization but interpose null on serialization. We create a type adapter that delegates for one and returns null for the other. - similarly when a DOM type serializer is registered but no deserializer, or vice versa. This is the biggest change to the MiniGson core. For backwards compatibility, return null for the empty string. Simplify JsonSerializationContext/JsonDeserializationContext to simply call through to GSON. SerializeDefault is currently unsupported. More useful error messages when calling getAsBoolean on a JsonNull. Remove currently unused MemoryRefStack. We might need this back again, though wiring it back in will be much more difficult because we don't interject ourselves between the users' various type adapters.
2011-09-11 09:04:56 +02:00
}
@Override
void readIntoArray(JsonReader reader, int index, Object[] target)
throws IOException, JsonParseException {
Object fieldValue = typeAdapter.read(reader);
if (fieldValue == null && isPrimitive) {
throw new JsonParseException(
"null is not allowed as value for record component '"
+ fieldName
+ "' of primitive type; at path "
+ reader.getPath());
}
target[index] = fieldValue;
}
@Override
void readIntoField(JsonReader reader, Object target)
Down to 22 failing tests. Consolidated all of the different code paths that we use to construct instances. We now have an ObjectConstructor class that knows what type it constructs; this means that we don't need to ever do reflection to lookup a constructor at construction time. Cleaned up some buggy type adapters, particularly around handling of null. Removed dead code for object graph navigation. Moved some classes into 'internal' so they are visible to the 'bind' subpackage. Turned some TypeAdapterFactory/TypeAdapter pairs inside out so that the TypeAdapter is now the inner class. This is necessary so that the factories can take parameters. Added an API to request the 'next' type adapter for a type. This allows type adapters to compose other type adapters. We're using this in two places: - where the user has excluded a type from serialization but not deserialization, we need to use the "default" deserialization but interpose null on serialization. We create a type adapter that delegates for one and returns null for the other. - similarly when a DOM type serializer is registered but no deserializer, or vice versa. This is the biggest change to the MiniGson core. For backwards compatibility, return null for the empty string. Simplify JsonSerializationContext/JsonDeserializationContext to simply call through to GSON. SerializeDefault is currently unsupported. More useful error messages when calling getAsBoolean on a JsonNull. Remove currently unused MemoryRefStack. We might need this back again, though wiring it back in will be much more difficult because we don't interject ourselves between the users' various type adapters.
2011-09-11 09:04:56 +02:00
throws IOException, IllegalAccessException {
Object fieldValue = typeAdapter.read(reader);
if (fieldValue != null || !isPrimitive) {
if (blockInaccessible) {
checkAccessible(target, field);
} else if (isStaticFinalField) {
// Reflection does not permit setting value of `static final` field, even after calling
// `setAccessible`
// Handle this here to avoid causing IllegalAccessException when calling `Field.set`
String fieldDescription = ReflectionHelper.getAccessibleObjectDescription(field, false);
throw new JsonIOException("Cannot set value of 'static final' " + fieldDescription);
}
field.set(target, fieldValue);
Down to 22 failing tests. Consolidated all of the different code paths that we use to construct instances. We now have an ObjectConstructor class that knows what type it constructs; this means that we don't need to ever do reflection to lookup a constructor at construction time. Cleaned up some buggy type adapters, particularly around handling of null. Removed dead code for object graph navigation. Moved some classes into 'internal' so they are visible to the 'bind' subpackage. Turned some TypeAdapterFactory/TypeAdapter pairs inside out so that the TypeAdapter is now the inner class. This is necessary so that the factories can take parameters. Added an API to request the 'next' type adapter for a type. This allows type adapters to compose other type adapters. We're using this in two places: - where the user has excluded a type from serialization but not deserialization, we need to use the "default" deserialization but interpose null on serialization. We create a type adapter that delegates for one and returns null for the other. - similarly when a DOM type serializer is registered but no deserializer, or vice versa. This is the biggest change to the MiniGson core. For backwards compatibility, return null for the empty string. Simplify JsonSerializationContext/JsonDeserializationContext to simply call through to GSON. SerializeDefault is currently unsupported. More useful error messages when calling getAsBoolean on a JsonNull. Remove currently unused MemoryRefStack. We might need this back again, though wiring it back in will be much more difficult because we don't interject ourselves between the users' various type adapters.
2011-09-11 09:04:56 +02:00
}
}
};
}
private static class FieldsData {
public static final FieldsData EMPTY =
new FieldsData(
Collections.<String, BoundField>emptyMap(), Collections.<BoundField>emptyList());
/** Maps from JSON member name to field */
public final Map<String, BoundField> deserializedFields;
public final List<BoundField> serializedFields;
public FieldsData(
Map<String, BoundField> deserializedFields, List<BoundField> serializedFields) {
this.deserializedFields = deserializedFields;
this.serializedFields = serializedFields;
}
}
private static IllegalArgumentException createDuplicateFieldException(
Class<?> declaringType, String duplicateName, Field field1, Field field2) {
throw new IllegalArgumentException(
"Class "
+ declaringType.getName()
+ " declares multiple JSON fields named '"
+ duplicateName
+ "'; conflict is caused by fields "
+ ReflectionHelper.fieldToString(field1)
+ " and "
+ ReflectionHelper.fieldToString(field2)
+ "\nSee "
+ TroubleshootingGuide.createUrl("duplicate-fields"));
}
private FieldsData getBoundFields(
Gson context, TypeToken<?> type, Class<?> raw, boolean blockInaccessible, boolean isRecord) {
Down to 22 failing tests. Consolidated all of the different code paths that we use to construct instances. We now have an ObjectConstructor class that knows what type it constructs; this means that we don't need to ever do reflection to lookup a constructor at construction time. Cleaned up some buggy type adapters, particularly around handling of null. Removed dead code for object graph navigation. Moved some classes into 'internal' so they are visible to the 'bind' subpackage. Turned some TypeAdapterFactory/TypeAdapter pairs inside out so that the TypeAdapter is now the inner class. This is necessary so that the factories can take parameters. Added an API to request the 'next' type adapter for a type. This allows type adapters to compose other type adapters. We're using this in two places: - where the user has excluded a type from serialization but not deserialization, we need to use the "default" deserialization but interpose null on serialization. We create a type adapter that delegates for one and returns null for the other. - similarly when a DOM type serializer is registered but no deserializer, or vice versa. This is the biggest change to the MiniGson core. For backwards compatibility, return null for the empty string. Simplify JsonSerializationContext/JsonDeserializationContext to simply call through to GSON. SerializeDefault is currently unsupported. More useful error messages when calling getAsBoolean on a JsonNull. Remove currently unused MemoryRefStack. We might need this back again, though wiring it back in will be much more difficult because we don't interject ourselves between the users' various type adapters.
2011-09-11 09:04:56 +02:00
if (raw.isInterface()) {
return FieldsData.EMPTY;
Down to 22 failing tests. Consolidated all of the different code paths that we use to construct instances. We now have an ObjectConstructor class that knows what type it constructs; this means that we don't need to ever do reflection to lookup a constructor at construction time. Cleaned up some buggy type adapters, particularly around handling of null. Removed dead code for object graph navigation. Moved some classes into 'internal' so they are visible to the 'bind' subpackage. Turned some TypeAdapterFactory/TypeAdapter pairs inside out so that the TypeAdapter is now the inner class. This is necessary so that the factories can take parameters. Added an API to request the 'next' type adapter for a type. This allows type adapters to compose other type adapters. We're using this in two places: - where the user has excluded a type from serialization but not deserialization, we need to use the "default" deserialization but interpose null on serialization. We create a type adapter that delegates for one and returns null for the other. - similarly when a DOM type serializer is registered but no deserializer, or vice versa. This is the biggest change to the MiniGson core. For backwards compatibility, return null for the empty string. Simplify JsonSerializationContext/JsonDeserializationContext to simply call through to GSON. SerializeDefault is currently unsupported. More useful error messages when calling getAsBoolean on a JsonNull. Remove currently unused MemoryRefStack. We might need this back again, though wiring it back in will be much more difficult because we don't interject ourselves between the users' various type adapters.
2011-09-11 09:04:56 +02:00
}
Map<String, BoundField> deserializedFields = new LinkedHashMap<>();
// For serialized fields use a Map to track duplicate field names; otherwise this could be a
// List<BoundField> instead
Map<String, BoundField> serializedFields = new LinkedHashMap<>();
Class<?> originalRaw = raw;
Down to 22 failing tests. Consolidated all of the different code paths that we use to construct instances. We now have an ObjectConstructor class that knows what type it constructs; this means that we don't need to ever do reflection to lookup a constructor at construction time. Cleaned up some buggy type adapters, particularly around handling of null. Removed dead code for object graph navigation. Moved some classes into 'internal' so they are visible to the 'bind' subpackage. Turned some TypeAdapterFactory/TypeAdapter pairs inside out so that the TypeAdapter is now the inner class. This is necessary so that the factories can take parameters. Added an API to request the 'next' type adapter for a type. This allows type adapters to compose other type adapters. We're using this in two places: - where the user has excluded a type from serialization but not deserialization, we need to use the "default" deserialization but interpose null on serialization. We create a type adapter that delegates for one and returns null for the other. - similarly when a DOM type serializer is registered but no deserializer, or vice versa. This is the biggest change to the MiniGson core. For backwards compatibility, return null for the empty string. Simplify JsonSerializationContext/JsonDeserializationContext to simply call through to GSON. SerializeDefault is currently unsupported. More useful error messages when calling getAsBoolean on a JsonNull. Remove currently unused MemoryRefStack. We might need this back again, though wiring it back in will be much more difficult because we don't interject ourselves between the users' various type adapters.
2011-09-11 09:04:56 +02:00
while (raw != Object.class) {
Field[] fields = raw.getDeclaredFields();
// For inherited fields, check if access to their declaring class is allowed
if (raw != originalRaw && fields.length > 0) {
FilterResult filterResult =
ReflectionAccessFilterHelper.getFilterResult(reflectionFilters, raw);
if (filterResult == FilterResult.BLOCK_ALL) {
throw new JsonIOException(
"ReflectionAccessFilter does not permit using reflection for "
+ raw
+ " (supertype of "
+ originalRaw
+ "). Register a TypeAdapter for this type or adjust the access filter.");
}
blockInaccessible = filterResult == FilterResult.BLOCK_INACCESSIBLE;
}
Down to 22 failing tests. Consolidated all of the different code paths that we use to construct instances. We now have an ObjectConstructor class that knows what type it constructs; this means that we don't need to ever do reflection to lookup a constructor at construction time. Cleaned up some buggy type adapters, particularly around handling of null. Removed dead code for object graph navigation. Moved some classes into 'internal' so they are visible to the 'bind' subpackage. Turned some TypeAdapterFactory/TypeAdapter pairs inside out so that the TypeAdapter is now the inner class. This is necessary so that the factories can take parameters. Added an API to request the 'next' type adapter for a type. This allows type adapters to compose other type adapters. We're using this in two places: - where the user has excluded a type from serialization but not deserialization, we need to use the "default" deserialization but interpose null on serialization. We create a type adapter that delegates for one and returns null for the other. - similarly when a DOM type serializer is registered but no deserializer, or vice versa. This is the biggest change to the MiniGson core. For backwards compatibility, return null for the empty string. Simplify JsonSerializationContext/JsonDeserializationContext to simply call through to GSON. SerializeDefault is currently unsupported. More useful error messages when calling getAsBoolean on a JsonNull. Remove currently unused MemoryRefStack. We might need this back again, though wiring it back in will be much more difficult because we don't interject ourselves between the users' various type adapters.
2011-09-11 09:04:56 +02:00
for (Field field : fields) {
boolean serialize = includeField(field, true);
boolean deserialize = includeField(field, false);
Down to 22 failing tests. Consolidated all of the different code paths that we use to construct instances. We now have an ObjectConstructor class that knows what type it constructs; this means that we don't need to ever do reflection to lookup a constructor at construction time. Cleaned up some buggy type adapters, particularly around handling of null. Removed dead code for object graph navigation. Moved some classes into 'internal' so they are visible to the 'bind' subpackage. Turned some TypeAdapterFactory/TypeAdapter pairs inside out so that the TypeAdapter is now the inner class. This is necessary so that the factories can take parameters. Added an API to request the 'next' type adapter for a type. This allows type adapters to compose other type adapters. We're using this in two places: - where the user has excluded a type from serialization but not deserialization, we need to use the "default" deserialization but interpose null on serialization. We create a type adapter that delegates for one and returns null for the other. - similarly when a DOM type serializer is registered but no deserializer, or vice versa. This is the biggest change to the MiniGson core. For backwards compatibility, return null for the empty string. Simplify JsonSerializationContext/JsonDeserializationContext to simply call through to GSON. SerializeDefault is currently unsupported. More useful error messages when calling getAsBoolean on a JsonNull. Remove currently unused MemoryRefStack. We might need this back again, though wiring it back in will be much more difficult because we don't interject ourselves between the users' various type adapters.
2011-09-11 09:04:56 +02:00
if (!serialize && !deserialize) {
continue;
}
// The accessor method is only used for records. If the type is a record, we will read out
// values via its accessor method instead of via reflection. This way we will bypass the
// accessible restrictions
Method accessor = null;
if (isRecord) {
// If there is a static field on a record, there will not be an accessor. Instead we will
// use the default field serialization logic, but for deserialization the field is
// excluded for simplicity.
// Note that Gson ignores static fields by default, but
// GsonBuilder.excludeFieldsWithModifiers can overwrite this.
if (Modifier.isStatic(field.getModifiers())) {
deserialize = false;
} else {
accessor = ReflectionHelper.getAccessor(raw, field);
// If blockInaccessible, skip and perform access check later
if (!blockInaccessible) {
ReflectionHelper.makeAccessible(accessor);
}
// @SerializedName can be placed on accessor method, but it is not supported there
// If field and method have annotation it is not easily possible to determine if
// accessor method is implicit and has inherited annotation, or if it is explicitly
// declared with custom annotation
if (accessor.getAnnotation(SerializedName.class) != null
&& field.getAnnotation(SerializedName.class) == null) {
String methodDescription =
ReflectionHelper.getAccessibleObjectDescription(accessor, false);
throw new JsonIOException(
"@SerializedName on " + methodDescription + " is not supported");
}
}
}
// If blockInaccessible, skip and perform access check later
// For Records if the accessor method is used the field does not have to be made accessible
if (!blockInaccessible && accessor == null) {
ReflectionHelper.makeAccessible(field);
}
Down to 22 failing tests. Consolidated all of the different code paths that we use to construct instances. We now have an ObjectConstructor class that knows what type it constructs; this means that we don't need to ever do reflection to lookup a constructor at construction time. Cleaned up some buggy type adapters, particularly around handling of null. Removed dead code for object graph navigation. Moved some classes into 'internal' so they are visible to the 'bind' subpackage. Turned some TypeAdapterFactory/TypeAdapter pairs inside out so that the TypeAdapter is now the inner class. This is necessary so that the factories can take parameters. Added an API to request the 'next' type adapter for a type. This allows type adapters to compose other type adapters. We're using this in two places: - where the user has excluded a type from serialization but not deserialization, we need to use the "default" deserialization but interpose null on serialization. We create a type adapter that delegates for one and returns null for the other. - similarly when a DOM type serializer is registered but no deserializer, or vice versa. This is the biggest change to the MiniGson core. For backwards compatibility, return null for the empty string. Simplify JsonSerializationContext/JsonDeserializationContext to simply call through to GSON. SerializeDefault is currently unsupported. More useful error messages when calling getAsBoolean on a JsonNull. Remove currently unused MemoryRefStack. We might need this back again, though wiring it back in will be much more difficult because we don't interject ourselves between the users' various type adapters.
2011-09-11 09:04:56 +02:00
Type fieldType = $Gson$Types.resolve(type.getType(), raw, field.getGenericType());
List<String> fieldNames = getFieldNames(field);
String serializedName = fieldNames.get(0);
BoundField boundField =
createBoundField(
context,
field,
accessor,
serializedName,
TypeToken.get(fieldType),
serialize,
blockInaccessible);
if (deserialize) {
for (String name : fieldNames) {
BoundField replaced = deserializedFields.put(name, boundField);
if (replaced != null) {
throw createDuplicateFieldException(originalRaw, name, replaced.field, field);
}
}
}
if (serialize) {
BoundField replaced = serializedFields.put(serializedName, boundField);
if (replaced != null) {
throw createDuplicateFieldException(originalRaw, serializedName, replaced.field, field);
}
Down to 22 failing tests. Consolidated all of the different code paths that we use to construct instances. We now have an ObjectConstructor class that knows what type it constructs; this means that we don't need to ever do reflection to lookup a constructor at construction time. Cleaned up some buggy type adapters, particularly around handling of null. Removed dead code for object graph navigation. Moved some classes into 'internal' so they are visible to the 'bind' subpackage. Turned some TypeAdapterFactory/TypeAdapter pairs inside out so that the TypeAdapter is now the inner class. This is necessary so that the factories can take parameters. Added an API to request the 'next' type adapter for a type. This allows type adapters to compose other type adapters. We're using this in two places: - where the user has excluded a type from serialization but not deserialization, we need to use the "default" deserialization but interpose null on serialization. We create a type adapter that delegates for one and returns null for the other. - similarly when a DOM type serializer is registered but no deserializer, or vice versa. This is the biggest change to the MiniGson core. For backwards compatibility, return null for the empty string. Simplify JsonSerializationContext/JsonDeserializationContext to simply call through to GSON. SerializeDefault is currently unsupported. More useful error messages when calling getAsBoolean on a JsonNull. Remove currently unused MemoryRefStack. We might need this back again, though wiring it back in will be much more difficult because we don't interject ourselves between the users' various type adapters.
2011-09-11 09:04:56 +02:00
}
}
type = TypeToken.get($Gson$Types.resolve(type.getType(), raw, raw.getGenericSuperclass()));
raw = type.getRawType();
}
return new FieldsData(deserializedFields, new ArrayList<>(serializedFields.values()));
Down to 22 failing tests. Consolidated all of the different code paths that we use to construct instances. We now have an ObjectConstructor class that knows what type it constructs; this means that we don't need to ever do reflection to lookup a constructor at construction time. Cleaned up some buggy type adapters, particularly around handling of null. Removed dead code for object graph navigation. Moved some classes into 'internal' so they are visible to the 'bind' subpackage. Turned some TypeAdapterFactory/TypeAdapter pairs inside out so that the TypeAdapter is now the inner class. This is necessary so that the factories can take parameters. Added an API to request the 'next' type adapter for a type. This allows type adapters to compose other type adapters. We're using this in two places: - where the user has excluded a type from serialization but not deserialization, we need to use the "default" deserialization but interpose null on serialization. We create a type adapter that delegates for one and returns null for the other. - similarly when a DOM type serializer is registered but no deserializer, or vice versa. This is the biggest change to the MiniGson core. For backwards compatibility, return null for the empty string. Simplify JsonSerializationContext/JsonDeserializationContext to simply call through to GSON. SerializeDefault is currently unsupported. More useful error messages when calling getAsBoolean on a JsonNull. Remove currently unused MemoryRefStack. We might need this back again, though wiring it back in will be much more difficult because we don't interject ourselves between the users' various type adapters.
2011-09-11 09:04:56 +02:00
}
abstract static class BoundField {
/** Name used for serialization (but not for deserialization) */
final String serializedName;
final Field field;
/** Name of the underlying field */
final String fieldName;
Down to 22 failing tests. Consolidated all of the different code paths that we use to construct instances. We now have an ObjectConstructor class that knows what type it constructs; this means that we don't need to ever do reflection to lookup a constructor at construction time. Cleaned up some buggy type adapters, particularly around handling of null. Removed dead code for object graph navigation. Moved some classes into 'internal' so they are visible to the 'bind' subpackage. Turned some TypeAdapterFactory/TypeAdapter pairs inside out so that the TypeAdapter is now the inner class. This is necessary so that the factories can take parameters. Added an API to request the 'next' type adapter for a type. This allows type adapters to compose other type adapters. We're using this in two places: - where the user has excluded a type from serialization but not deserialization, we need to use the "default" deserialization but interpose null on serialization. We create a type adapter that delegates for one and returns null for the other. - similarly when a DOM type serializer is registered but no deserializer, or vice versa. This is the biggest change to the MiniGson core. For backwards compatibility, return null for the empty string. Simplify JsonSerializationContext/JsonDeserializationContext to simply call through to GSON. SerializeDefault is currently unsupported. More useful error messages when calling getAsBoolean on a JsonNull. Remove currently unused MemoryRefStack. We might need this back again, though wiring it back in will be much more difficult because we don't interject ourselves between the users' various type adapters.
2011-09-11 09:04:56 +02:00
protected BoundField(String serializedName, Field field) {
this.serializedName = serializedName;
this.field = field;
this.fieldName = field.getName();
Down to 22 failing tests. Consolidated all of the different code paths that we use to construct instances. We now have an ObjectConstructor class that knows what type it constructs; this means that we don't need to ever do reflection to lookup a constructor at construction time. Cleaned up some buggy type adapters, particularly around handling of null. Removed dead code for object graph navigation. Moved some classes into 'internal' so they are visible to the 'bind' subpackage. Turned some TypeAdapterFactory/TypeAdapter pairs inside out so that the TypeAdapter is now the inner class. This is necessary so that the factories can take parameters. Added an API to request the 'next' type adapter for a type. This allows type adapters to compose other type adapters. We're using this in two places: - where the user has excluded a type from serialization but not deserialization, we need to use the "default" deserialization but interpose null on serialization. We create a type adapter that delegates for one and returns null for the other. - similarly when a DOM type serializer is registered but no deserializer, or vice versa. This is the biggest change to the MiniGson core. For backwards compatibility, return null for the empty string. Simplify JsonSerializationContext/JsonDeserializationContext to simply call through to GSON. SerializeDefault is currently unsupported. More useful error messages when calling getAsBoolean on a JsonNull. Remove currently unused MemoryRefStack. We might need this back again, though wiring it back in will be much more difficult because we don't interject ourselves between the users' various type adapters.
2011-09-11 09:04:56 +02:00
}
/** Read this field value from the source, and append its JSON value to the writer */
abstract void write(JsonWriter writer, Object source)
throws IOException, IllegalAccessException;
/** Read the value into the target array, used to provide constructor arguments for records */
abstract void readIntoArray(JsonReader reader, int index, Object[] target)
throws IOException, JsonParseException;
/**
* Read the value from the reader, and set it on the corresponding field on target via
* reflection
*/
abstract void readIntoField(JsonReader reader, Object target)
throws IOException, IllegalAccessException;
Down to 22 failing tests. Consolidated all of the different code paths that we use to construct instances. We now have an ObjectConstructor class that knows what type it constructs; this means that we don't need to ever do reflection to lookup a constructor at construction time. Cleaned up some buggy type adapters, particularly around handling of null. Removed dead code for object graph navigation. Moved some classes into 'internal' so they are visible to the 'bind' subpackage. Turned some TypeAdapterFactory/TypeAdapter pairs inside out so that the TypeAdapter is now the inner class. This is necessary so that the factories can take parameters. Added an API to request the 'next' type adapter for a type. This allows type adapters to compose other type adapters. We're using this in two places: - where the user has excluded a type from serialization but not deserialization, we need to use the "default" deserialization but interpose null on serialization. We create a type adapter that delegates for one and returns null for the other. - similarly when a DOM type serializer is registered but no deserializer, or vice versa. This is the biggest change to the MiniGson core. For backwards compatibility, return null for the empty string. Simplify JsonSerializationContext/JsonDeserializationContext to simply call through to GSON. SerializeDefault is currently unsupported. More useful error messages when calling getAsBoolean on a JsonNull. Remove currently unused MemoryRefStack. We might need this back again, though wiring it back in will be much more difficult because we don't interject ourselves between the users' various type adapters.
2011-09-11 09:04:56 +02:00
}
/**
* Base class for Adapters produced by this factory.
*
* <p>The {@link RecordAdapter} is a special case to handle records for JVMs that support it, for
* all other types we use the {@link FieldReflectionAdapter}. This class encapsulates the common
* logic for serialization and deserialization. During deserialization, we construct an
* accumulator A, which we use to accumulate values from the source JSON. After the object has
* been read in full, the {@link #finalize(Object)} method is used to convert the accumulator to
* an instance of T.
*
* @param <T> type of objects that this Adapter creates.
* @param <A> type of accumulator used to build the deserialization result.
*/
// This class is public because external projects check for this class with `instanceof` (even
// though it is internal)
public abstract static class Adapter<T, A> extends TypeAdapter<T> {
private final FieldsData fieldsData;
Down to 22 failing tests. Consolidated all of the different code paths that we use to construct instances. We now have an ObjectConstructor class that knows what type it constructs; this means that we don't need to ever do reflection to lookup a constructor at construction time. Cleaned up some buggy type adapters, particularly around handling of null. Removed dead code for object graph navigation. Moved some classes into 'internal' so they are visible to the 'bind' subpackage. Turned some TypeAdapterFactory/TypeAdapter pairs inside out so that the TypeAdapter is now the inner class. This is necessary so that the factories can take parameters. Added an API to request the 'next' type adapter for a type. This allows type adapters to compose other type adapters. We're using this in two places: - where the user has excluded a type from serialization but not deserialization, we need to use the "default" deserialization but interpose null on serialization. We create a type adapter that delegates for one and returns null for the other. - similarly when a DOM type serializer is registered but no deserializer, or vice versa. This is the biggest change to the MiniGson core. For backwards compatibility, return null for the empty string. Simplify JsonSerializationContext/JsonDeserializationContext to simply call through to GSON. SerializeDefault is currently unsupported. More useful error messages when calling getAsBoolean on a JsonNull. Remove currently unused MemoryRefStack. We might need this back again, though wiring it back in will be much more difficult because we don't interject ourselves between the users' various type adapters.
2011-09-11 09:04:56 +02:00
Adapter(FieldsData fieldsData) {
this.fieldsData = fieldsData;
Down to 22 failing tests. Consolidated all of the different code paths that we use to construct instances. We now have an ObjectConstructor class that knows what type it constructs; this means that we don't need to ever do reflection to lookup a constructor at construction time. Cleaned up some buggy type adapters, particularly around handling of null. Removed dead code for object graph navigation. Moved some classes into 'internal' so they are visible to the 'bind' subpackage. Turned some TypeAdapterFactory/TypeAdapter pairs inside out so that the TypeAdapter is now the inner class. This is necessary so that the factories can take parameters. Added an API to request the 'next' type adapter for a type. This allows type adapters to compose other type adapters. We're using this in two places: - where the user has excluded a type from serialization but not deserialization, we need to use the "default" deserialization but interpose null on serialization. We create a type adapter that delegates for one and returns null for the other. - similarly when a DOM type serializer is registered but no deserializer, or vice versa. This is the biggest change to the MiniGson core. For backwards compatibility, return null for the empty string. Simplify JsonSerializationContext/JsonDeserializationContext to simply call through to GSON. SerializeDefault is currently unsupported. More useful error messages when calling getAsBoolean on a JsonNull. Remove currently unused MemoryRefStack. We might need this back again, though wiring it back in will be much more difficult because we don't interject ourselves between the users' various type adapters.
2011-09-11 09:04:56 +02:00
}
@Override
public void write(JsonWriter out, T value) throws IOException {
if (value == null) {
out.nullValue();
return;
}
out.beginObject();
try {
for (BoundField boundField : fieldsData.serializedFields) {
boundField.write(out, value);
}
} catch (IllegalAccessException e) {
throw ReflectionHelper.createExceptionForUnexpectedIllegalAccess(e);
}
out.endObject();
}
@Override
public T read(JsonReader in) throws IOException {
2011-12-03 20:46:25 +01:00
if (in.peek() == JsonToken.NULL) {
in.nextNull();
Down to 22 failing tests. Consolidated all of the different code paths that we use to construct instances. We now have an ObjectConstructor class that knows what type it constructs; this means that we don't need to ever do reflection to lookup a constructor at construction time. Cleaned up some buggy type adapters, particularly around handling of null. Removed dead code for object graph navigation. Moved some classes into 'internal' so they are visible to the 'bind' subpackage. Turned some TypeAdapterFactory/TypeAdapter pairs inside out so that the TypeAdapter is now the inner class. This is necessary so that the factories can take parameters. Added an API to request the 'next' type adapter for a type. This allows type adapters to compose other type adapters. We're using this in two places: - where the user has excluded a type from serialization but not deserialization, we need to use the "default" deserialization but interpose null on serialization. We create a type adapter that delegates for one and returns null for the other. - similarly when a DOM type serializer is registered but no deserializer, or vice versa. This is the biggest change to the MiniGson core. For backwards compatibility, return null for the empty string. Simplify JsonSerializationContext/JsonDeserializationContext to simply call through to GSON. SerializeDefault is currently unsupported. More useful error messages when calling getAsBoolean on a JsonNull. Remove currently unused MemoryRefStack. We might need this back again, though wiring it back in will be much more difficult because we don't interject ourselves between the users' various type adapters.
2011-09-11 09:04:56 +02:00
return null;
}
A accumulator = createAccumulator();
Map<String, BoundField> deserializedFields = fieldsData.deserializedFields;
Down to 22 failing tests. Consolidated all of the different code paths that we use to construct instances. We now have an ObjectConstructor class that knows what type it constructs; this means that we don't need to ever do reflection to lookup a constructor at construction time. Cleaned up some buggy type adapters, particularly around handling of null. Removed dead code for object graph navigation. Moved some classes into 'internal' so they are visible to the 'bind' subpackage. Turned some TypeAdapterFactory/TypeAdapter pairs inside out so that the TypeAdapter is now the inner class. This is necessary so that the factories can take parameters. Added an API to request the 'next' type adapter for a type. This allows type adapters to compose other type adapters. We're using this in two places: - where the user has excluded a type from serialization but not deserialization, we need to use the "default" deserialization but interpose null on serialization. We create a type adapter that delegates for one and returns null for the other. - similarly when a DOM type serializer is registered but no deserializer, or vice versa. This is the biggest change to the MiniGson core. For backwards compatibility, return null for the empty string. Simplify JsonSerializationContext/JsonDeserializationContext to simply call through to GSON. SerializeDefault is currently unsupported. More useful error messages when calling getAsBoolean on a JsonNull. Remove currently unused MemoryRefStack. We might need this back again, though wiring it back in will be much more difficult because we don't interject ourselves between the users' various type adapters.
2011-09-11 09:04:56 +02:00
try {
2011-12-03 20:46:25 +01:00
in.beginObject();
while (in.hasNext()) {
String name = in.nextName();
BoundField field = deserializedFields.get(name);
if (field == null) {
2011-12-03 20:46:25 +01:00
in.skipValue();
Down to 22 failing tests. Consolidated all of the different code paths that we use to construct instances. We now have an ObjectConstructor class that knows what type it constructs; this means that we don't need to ever do reflection to lookup a constructor at construction time. Cleaned up some buggy type adapters, particularly around handling of null. Removed dead code for object graph navigation. Moved some classes into 'internal' so they are visible to the 'bind' subpackage. Turned some TypeAdapterFactory/TypeAdapter pairs inside out so that the TypeAdapter is now the inner class. This is necessary so that the factories can take parameters. Added an API to request the 'next' type adapter for a type. This allows type adapters to compose other type adapters. We're using this in two places: - where the user has excluded a type from serialization but not deserialization, we need to use the "default" deserialization but interpose null on serialization. We create a type adapter that delegates for one and returns null for the other. - similarly when a DOM type serializer is registered but no deserializer, or vice versa. This is the biggest change to the MiniGson core. For backwards compatibility, return null for the empty string. Simplify JsonSerializationContext/JsonDeserializationContext to simply call through to GSON. SerializeDefault is currently unsupported. More useful error messages when calling getAsBoolean on a JsonNull. Remove currently unused MemoryRefStack. We might need this back again, though wiring it back in will be much more difficult because we don't interject ourselves between the users' various type adapters.
2011-09-11 09:04:56 +02:00
} else {
readField(accumulator, in, field);
Down to 22 failing tests. Consolidated all of the different code paths that we use to construct instances. We now have an ObjectConstructor class that knows what type it constructs; this means that we don't need to ever do reflection to lookup a constructor at construction time. Cleaned up some buggy type adapters, particularly around handling of null. Removed dead code for object graph navigation. Moved some classes into 'internal' so they are visible to the 'bind' subpackage. Turned some TypeAdapterFactory/TypeAdapter pairs inside out so that the TypeAdapter is now the inner class. This is necessary so that the factories can take parameters. Added an API to request the 'next' type adapter for a type. This allows type adapters to compose other type adapters. We're using this in two places: - where the user has excluded a type from serialization but not deserialization, we need to use the "default" deserialization but interpose null on serialization. We create a type adapter that delegates for one and returns null for the other. - similarly when a DOM type serializer is registered but no deserializer, or vice versa. This is the biggest change to the MiniGson core. For backwards compatibility, return null for the empty string. Simplify JsonSerializationContext/JsonDeserializationContext to simply call through to GSON. SerializeDefault is currently unsupported. More useful error messages when calling getAsBoolean on a JsonNull. Remove currently unused MemoryRefStack. We might need this back again, though wiring it back in will be much more difficult because we don't interject ourselves between the users' various type adapters.
2011-09-11 09:04:56 +02:00
}
}
} catch (IllegalStateException e) {
throw new JsonSyntaxException(e);
} catch (IllegalAccessException e) {
throw ReflectionHelper.createExceptionForUnexpectedIllegalAccess(e);
Down to 22 failing tests. Consolidated all of the different code paths that we use to construct instances. We now have an ObjectConstructor class that knows what type it constructs; this means that we don't need to ever do reflection to lookup a constructor at construction time. Cleaned up some buggy type adapters, particularly around handling of null. Removed dead code for object graph navigation. Moved some classes into 'internal' so they are visible to the 'bind' subpackage. Turned some TypeAdapterFactory/TypeAdapter pairs inside out so that the TypeAdapter is now the inner class. This is necessary so that the factories can take parameters. Added an API to request the 'next' type adapter for a type. This allows type adapters to compose other type adapters. We're using this in two places: - where the user has excluded a type from serialization but not deserialization, we need to use the "default" deserialization but interpose null on serialization. We create a type adapter that delegates for one and returns null for the other. - similarly when a DOM type serializer is registered but no deserializer, or vice versa. This is the biggest change to the MiniGson core. For backwards compatibility, return null for the empty string. Simplify JsonSerializationContext/JsonDeserializationContext to simply call through to GSON. SerializeDefault is currently unsupported. More useful error messages when calling getAsBoolean on a JsonNull. Remove currently unused MemoryRefStack. We might need this back again, though wiring it back in will be much more difficult because we don't interject ourselves between the users' various type adapters.
2011-09-11 09:04:56 +02:00
}
2011-12-03 20:46:25 +01:00
in.endObject();
return finalize(accumulator);
Down to 22 failing tests. Consolidated all of the different code paths that we use to construct instances. We now have an ObjectConstructor class that knows what type it constructs; this means that we don't need to ever do reflection to lookup a constructor at construction time. Cleaned up some buggy type adapters, particularly around handling of null. Removed dead code for object graph navigation. Moved some classes into 'internal' so they are visible to the 'bind' subpackage. Turned some TypeAdapterFactory/TypeAdapter pairs inside out so that the TypeAdapter is now the inner class. This is necessary so that the factories can take parameters. Added an API to request the 'next' type adapter for a type. This allows type adapters to compose other type adapters. We're using this in two places: - where the user has excluded a type from serialization but not deserialization, we need to use the "default" deserialization but interpose null on serialization. We create a type adapter that delegates for one and returns null for the other. - similarly when a DOM type serializer is registered but no deserializer, or vice versa. This is the biggest change to the MiniGson core. For backwards compatibility, return null for the empty string. Simplify JsonSerializationContext/JsonDeserializationContext to simply call through to GSON. SerializeDefault is currently unsupported. More useful error messages when calling getAsBoolean on a JsonNull. Remove currently unused MemoryRefStack. We might need this back again, though wiring it back in will be much more difficult because we don't interject ourselves between the users' various type adapters.
2011-09-11 09:04:56 +02:00
}
/** Create the Object that will be used to collect each field value */
abstract A createAccumulator();
/**
* Read a single BoundField into the accumulator. The JsonReader will be pointed at the start of
* the value for the BoundField to read from.
*/
abstract void readField(A accumulator, JsonReader in, BoundField field)
throws IllegalAccessException, IOException;
/** Convert the accumulator to a final instance of T. */
abstract T finalize(A accumulator);
}
private static final class FieldReflectionAdapter<T> extends Adapter<T, T> {
private final ObjectConstructor<T> constructor;
FieldReflectionAdapter(ObjectConstructor<T> constructor, FieldsData fieldsData) {
super(fieldsData);
this.constructor = constructor;
}
@Override
T createAccumulator() {
return constructor.construct();
}
@Override
void readField(T accumulator, JsonReader in, BoundField field)
throws IllegalAccessException, IOException {
field.readIntoField(in, accumulator);
}
@Override
T finalize(T accumulator) {
return accumulator;
}
}
private static final class RecordAdapter<T> extends Adapter<T, Object[]> {
static final Map<Class<?>, Object> PRIMITIVE_DEFAULTS = primitiveDefaults();
// The canonical constructor of the record
private final Constructor<T> constructor;
// Array of arguments to the constructor, initialized with default values for primitives
private final Object[] constructorArgsDefaults;
// Map from component names to index into the constructors arguments.
private final Map<String, Integer> componentIndices = new HashMap<>();
RecordAdapter(Class<T> raw, FieldsData fieldsData, boolean blockInaccessible) {
super(fieldsData);
constructor = ReflectionHelper.getCanonicalRecordConstructor(raw);
if (blockInaccessible) {
checkAccessible(null, constructor);
} else {
// Ensure the constructor is accessible
ReflectionHelper.makeAccessible(constructor);
}
String[] componentNames = ReflectionHelper.getRecordComponentNames(raw);
for (int i = 0; i < componentNames.length; i++) {
componentIndices.put(componentNames[i], i);
Down to 22 failing tests. Consolidated all of the different code paths that we use to construct instances. We now have an ObjectConstructor class that knows what type it constructs; this means that we don't need to ever do reflection to lookup a constructor at construction time. Cleaned up some buggy type adapters, particularly around handling of null. Removed dead code for object graph navigation. Moved some classes into 'internal' so they are visible to the 'bind' subpackage. Turned some TypeAdapterFactory/TypeAdapter pairs inside out so that the TypeAdapter is now the inner class. This is necessary so that the factories can take parameters. Added an API to request the 'next' type adapter for a type. This allows type adapters to compose other type adapters. We're using this in two places: - where the user has excluded a type from serialization but not deserialization, we need to use the "default" deserialization but interpose null on serialization. We create a type adapter that delegates for one and returns null for the other. - similarly when a DOM type serializer is registered but no deserializer, or vice versa. This is the biggest change to the MiniGson core. For backwards compatibility, return null for the empty string. Simplify JsonSerializationContext/JsonDeserializationContext to simply call through to GSON. SerializeDefault is currently unsupported. More useful error messages when calling getAsBoolean on a JsonNull. Remove currently unused MemoryRefStack. We might need this back again, though wiring it back in will be much more difficult because we don't interject ourselves between the users' various type adapters.
2011-09-11 09:04:56 +02:00
}
Class<?>[] parameterTypes = constructor.getParameterTypes();
Down to 22 failing tests. Consolidated all of the different code paths that we use to construct instances. We now have an ObjectConstructor class that knows what type it constructs; this means that we don't need to ever do reflection to lookup a constructor at construction time. Cleaned up some buggy type adapters, particularly around handling of null. Removed dead code for object graph navigation. Moved some classes into 'internal' so they are visible to the 'bind' subpackage. Turned some TypeAdapterFactory/TypeAdapter pairs inside out so that the TypeAdapter is now the inner class. This is necessary so that the factories can take parameters. Added an API to request the 'next' type adapter for a type. This allows type adapters to compose other type adapters. We're using this in two places: - where the user has excluded a type from serialization but not deserialization, we need to use the "default" deserialization but interpose null on serialization. We create a type adapter that delegates for one and returns null for the other. - similarly when a DOM type serializer is registered but no deserializer, or vice versa. This is the biggest change to the MiniGson core. For backwards compatibility, return null for the empty string. Simplify JsonSerializationContext/JsonDeserializationContext to simply call through to GSON. SerializeDefault is currently unsupported. More useful error messages when calling getAsBoolean on a JsonNull. Remove currently unused MemoryRefStack. We might need this back again, though wiring it back in will be much more difficult because we don't interject ourselves between the users' various type adapters.
2011-09-11 09:04:56 +02:00
// We need to ensure that we are passing non-null values to primitive fields in the
// constructor. To do this, we create an Object[] where all primitives are initialized to
// non-null values.
constructorArgsDefaults = new Object[parameterTypes.length];
for (int i = 0; i < parameterTypes.length; i++) {
// This will correctly be null for non-primitive types:
constructorArgsDefaults[i] = PRIMITIVE_DEFAULTS.get(parameterTypes[i]);
Down to 22 failing tests. Consolidated all of the different code paths that we use to construct instances. We now have an ObjectConstructor class that knows what type it constructs; this means that we don't need to ever do reflection to lookup a constructor at construction time. Cleaned up some buggy type adapters, particularly around handling of null. Removed dead code for object graph navigation. Moved some classes into 'internal' so they are visible to the 'bind' subpackage. Turned some TypeAdapterFactory/TypeAdapter pairs inside out so that the TypeAdapter is now the inner class. This is necessary so that the factories can take parameters. Added an API to request the 'next' type adapter for a type. This allows type adapters to compose other type adapters. We're using this in two places: - where the user has excluded a type from serialization but not deserialization, we need to use the "default" deserialization but interpose null on serialization. We create a type adapter that delegates for one and returns null for the other. - similarly when a DOM type serializer is registered but no deserializer, or vice versa. This is the biggest change to the MiniGson core. For backwards compatibility, return null for the empty string. Simplify JsonSerializationContext/JsonDeserializationContext to simply call through to GSON. SerializeDefault is currently unsupported. More useful error messages when calling getAsBoolean on a JsonNull. Remove currently unused MemoryRefStack. We might need this back again, though wiring it back in will be much more difficult because we don't interject ourselves between the users' various type adapters.
2011-09-11 09:04:56 +02:00
}
}
private static Map<Class<?>, Object> primitiveDefaults() {
Map<Class<?>, Object> zeroes = new HashMap<>();
zeroes.put(byte.class, (byte) 0);
zeroes.put(short.class, (short) 0);
zeroes.put(int.class, 0);
zeroes.put(long.class, 0L);
zeroes.put(float.class, 0F);
zeroes.put(double.class, 0D);
zeroes.put(char.class, '\0');
zeroes.put(boolean.class, false);
return zeroes;
}
@Override
Object[] createAccumulator() {
return constructorArgsDefaults.clone();
}
@Override
void readField(Object[] accumulator, JsonReader in, BoundField field) throws IOException {
// Obtain the component index from the name of the field backing it
Integer componentIndex = componentIndices.get(field.fieldName);
if (componentIndex == null) {
throw new IllegalStateException(
"Could not find the index in the constructor '"
+ ReflectionHelper.constructorToString(constructor)
+ "' for field with name '"
+ field.fieldName
+ "', unable to determine which argument in the constructor the field corresponds"
+ " to. This is unexpected behavior, as we expect the RecordComponents to have the"
+ " same names as the fields in the Java class, and that the order of the"
+ " RecordComponents is the same as the order of the canonical constructor"
+ " parameters.");
}
field.readIntoArray(in, componentIndex, accumulator);
}
@Override
T finalize(Object[] accumulator) {
try {
return constructor.newInstance(accumulator);
} catch (IllegalAccessException e) {
throw ReflectionHelper.createExceptionForUnexpectedIllegalAccess(e);
}
// Note: InstantiationException should be impossible because record class is not abstract;
// IllegalArgumentException should not be possible unless a bad adapter returns objects of
// the wrong type
catch (InstantiationException | IllegalArgumentException e) {
throw new RuntimeException(
"Failed to invoke constructor '"
+ ReflectionHelper.constructorToString(constructor)
+ "' with args "
+ Arrays.toString(accumulator),
e);
} catch (InvocationTargetException e) {
// TODO: JsonParseException ?
throw new RuntimeException(
"Failed to invoke constructor '"
+ ReflectionHelper.constructorToString(constructor)
+ "' with args "
+ Arrays.toString(accumulator),
e.getCause());
}
Down to 22 failing tests. Consolidated all of the different code paths that we use to construct instances. We now have an ObjectConstructor class that knows what type it constructs; this means that we don't need to ever do reflection to lookup a constructor at construction time. Cleaned up some buggy type adapters, particularly around handling of null. Removed dead code for object graph navigation. Moved some classes into 'internal' so they are visible to the 'bind' subpackage. Turned some TypeAdapterFactory/TypeAdapter pairs inside out so that the TypeAdapter is now the inner class. This is necessary so that the factories can take parameters. Added an API to request the 'next' type adapter for a type. This allows type adapters to compose other type adapters. We're using this in two places: - where the user has excluded a type from serialization but not deserialization, we need to use the "default" deserialization but interpose null on serialization. We create a type adapter that delegates for one and returns null for the other. - similarly when a DOM type serializer is registered but no deserializer, or vice versa. This is the biggest change to the MiniGson core. For backwards compatibility, return null for the empty string. Simplify JsonSerializationContext/JsonDeserializationContext to simply call through to GSON. SerializeDefault is currently unsupported. More useful error messages when calling getAsBoolean on a JsonNull. Remove currently unused MemoryRefStack. We might need this back again, though wiring it back in will be much more difficult because we don't interject ourselves between the users' various type adapters.
2011-09-11 09:04:56 +02:00
}
}
}