为什么Gson序列化列表中的运行时类型,而不是指定的编译时类型？

Question

为什么Gson在序列化时会忽略嵌套的泛型类型声明？

我试图让Gson使用我指定的编译时类型,而不是列表中的运行时类型的对象.我也使用抽象超类A.java,但下面的例子有同样的问题.

public class A {
    public String foo;
}

public class B extends A {
    public String bar;
}

public static void main( String[] args ) {
    Gson gson = new Gson();

    B b = new B();
    b.foo = "foo";
    b.bar = "bar";

    List<A> list = new ArrayList<A>();
    list.add(b);

    System.out.println(gson.toJson(b, new TypeToken<A>(){}.getType()));
    System.out.println(gson.toJson(b, new TypeToken<B>(){}.getType()));

    System.out.println(gson.toJson(list, new TypeToken<List<A>>(){}.getType()));
    System.out.println(gson.toJson(list, new TypeToken<List<B>>(){}.getType()));
}

输出:

{"foo":"foo"}
{"bar":"bar","foo":"foo"}
[{"bar":"bar","foo":"foo"}]
[{"bar":"bar","foo":"foo"}]

预期:

{"foo":"foo"}
{"bar":"bar","foo":"foo"}
[{"foo":"foo"}]
[{"bar":"bar","foo":"foo"}]

Answer 1

这是因为 Gson 默认如何序列化集合。

为什么会出现这种情况？

如果您不关心原因而只想修复，请滚动到底部。

Gson 的默认值CollectionTypeAdapterFactory将其元素类型适配器包装在称为TypeAdapterRuntimeTypeWrapper. 选择合适的适配器时，会优先考虑以下因素：

// Order of preference for choosing type adapters
// First preference: a type adapter registered for the runtime type
// Second preference: a type adapter registered for the declared type
// Third preference: reflective type adapter for the runtime type (if it is a sub class of the declared type)
// Fourth preference: reflective type adapter for the declared type

在这种情况下，第三个首选项是 的适配器B，第四个首选项是 A 的适配器。使用默认序列化器时这是不可避免的，因为中没有条件CollectionTypeAdapterFactory：

public Adapter(Gson context, Type elementType,
    TypeAdapter<E> elementTypeAdapter,
    ObjectConstructor<? extends Collection<E>> constructor) {
  this.elementTypeAdapter =
      new TypeAdapterRuntimeTypeWrapper<E>(context, elementTypeAdapter, elementType);
  this.constructor = constructor;
}

不使用 时，此包装器不存在CollectionTypeAdapterFactory，这就是为什么它不会在前两个示例中发生。

tl;dr 好的，那我该如何解决呢？

解决这个问题的唯一方法是注册一个自定义序列化器。在您的用例中，写一个就A可以解决问题：

public class ATypeAdapter extends TypeAdapter<A> {
  public A read(JsonReader reader) throws IOException {
    if (reader.peek() == JsonToken.NULL) {
      reader.nextNull();
      return null;
    }
    reader.beginObject();
    String name = reader.nextName();
    if(!"foo".equals(name)) throw new JsonSyntaxException("Expected field named foo");
    A a = new A();
    a.foo = reader.nextString();
    reader.endObject();
    return a;
  }

  public void write(JsonWriter writer, A value) throws IOException {
    if (value == null) {
      writer.nullValue();
      return;
    }
    writer.beginObject();
    writer.name("foo");
    writer.value(value.foo);
    writer.endObject();
  }
}

那么，如果你这样做：

public static void main( String[] args ) {
    GsonBuilder builder = new GsonBuilder();
    builder.registerTypeAdapter(new TypeToken<A>(){}.getType(), new ATypeAdapter());
    Gson gson = builder.create();

    B b = new B();
    b.foo = "foo";
    b.bar = "bar";

    List<A> list = new ArrayList<A>();
    list.add(b);

    System.out.println(gson.toJson(b, new TypeToken<A>(){}.getType()));
    System.out.println(gson.toJson(b, new TypeToken<B>(){}.getType()));

    System.out.println(gson.toJson(list, new TypeToken<List<A>>(){}.getType()));
    System.out.println(gson.toJson(list, new TypeToken<List<B>>(){}.getType()));
}

您将得到预期的输出：

// Order of preference for choosing type adapters
// First preference: a type adapter registered for the runtime type
// Second preference: a type adapter registered for the declared type
// Third preference: reflective type adapter for the runtime type (if it is a sub class of the declared type)
// Fourth preference: reflective type adapter for the declared type