Does Gson have a way to read in non-standard JSON files?
Instead of a typical file like:
[{obj1},{objN}]
I have a file like this:
{obj1}
{objN}
Where there are no square brackets or commas and each object is separated by a newline character.
Yes, it has. Gson supports lenient reading. For example, the following JSON document (non-standard.json):
{
"foo": 1
}
{
"bar": 1
}
you can use the following reading way:
private static final Gson gson = new Gson();
private static final TypeAdapter<JsonElement> jsonElementTypeAdapter = gson.getAdapter(JsonElement.class);
public static void main(final String... args)
throws IOException {
try ( final Reader reader = getPackageResourceReader(Q43528208.class, "non-standard.json") ) {
final JsonReader jsonReader = new JsonReader(reader);
jsonReader.setLenient(true); // this makes it work
while ( jsonReader.peek() != END_DOCUMENT ) {
final JsonElement jsonElement = jsonElementTypeAdapter.read(jsonReader);
System.out.println(jsonElement);
}
}
}
Output:
{"foo":1}
{"bar":1}
I'm not sure if you can write a robust deserializer this way though.
Update
In order to simplify the Gson support, we can implement a few convenient reading methods:
// A shortcut method for the below implementation: aggregates the whole result into a single list
private static <T> List<T> parseToListLenient(final JsonReader jsonReader, final IMapper<? super JsonReader, ? extends T> mapper)
throws IOException {
final List<T> list = new ArrayList<>();
parseLenient(jsonReader, in -> list.add(mapper.map(in)));
return list;
}
// A convenient strategy-accepting method to configure a JsonReader instance to make it lenient and do read
// The consumer defines the strategy what to do with the current JsonReader token
private static void parseLenient(final JsonReader jsonReader, final IConsumer<? super JsonReader> consumer)
throws IOException {
final boolean isLenient = jsonReader.isLenient();
try {
jsonReader.setLenient(true);
while ( jsonReader.peek() != END_DOCUMENT ) {
consumer.accept(jsonReader);
}
} finally {
jsonReader.setLenient(isLenient);
}
}
// Since Java 8 Consumer inteface does not allow checked exceptions to be rethrown
private interface IConsumer<T> {
void accept(T value)
throws IOException;
}
private interface IMapper<T, R> {
R map(T value)
throws IOException;
}
Then simple reading is really simple, and we can just use the methods above:
final Gson gson = new Gson();
final TypeToken<Map<String, Integer>> typeToken = new TypeToken<Map<String, Integer>>() {
};
final TypeAdapter<Map<String, Integer>> typeAdapter = gson.getAdapter(typeToken);
try ( final JsonReader jsonReader = getPackageResourceJsonReader(Q43528208.class, "non-standard.json") ) {
final List<Map<String, Integer>> maps = parseToListLenient(jsonReader, typeAdapter::read);
System.out.println(maps);
}
Deserialization via Gson directly would require more complicated implementation:
// This is just a marker not meant to be instantiated but to create a sort of "gateway" to dispatch types in Gson
#SuppressWarnings("unused")
private static final class LenientListMarker<T> {
private LenientListMarker() {
throw new AssertionError("must not be instantiated");
}
}
private static void doDeserialize()
throws IOException {
final Gson gson = new GsonBuilder()
.registerTypeAdapterFactory(new TypeAdapterFactory() {
#Override
public <T> TypeAdapter<T> create(final Gson gson, final TypeToken<T> typeToken) {
// Check if the given type is the lenient list marker class
if ( !LenientListMarker.class.isAssignableFrom(typeToken.getRawType()) ) {
// Not the case? Just delegate the job to Gson
return null;
}
final Type listElementType = getTypeParameter0(typeToken.getType());
final TypeAdapter<?> listElementAdapter = gson.getAdapter(TypeToken.get(listElementType));
#SuppressWarnings("unchecked")
final TypeToken<List<?>> listTypeToken = (TypeToken<List<?>>) TypeToken.getParameterized(List.class, listElementType);
final TypeAdapter<List<?>> listAdapter = gson.getAdapter(listTypeToken);
final TypeAdapter<List<?>> typeAdapter = new TypeAdapter<List<?>>() {
#Override
public void write(final JsonWriter out, final List<?> value)
throws IOException {
// Always write a well-formed list
listAdapter.write(out, value);
}
#Override
public List<?> read(final JsonReader in)
throws IOException {
// Delegate the job to the reading method - we only have to tell how to obtain the list values
return parseToListLenient(in, listElementAdapter::read);
}
};
#SuppressWarnings("unchecked")
final TypeAdapter<T> castTypeAdapter = (TypeAdapter<T>) typeAdapter;
return castTypeAdapter;
}
// A simple method to resolve actual type parameter
private Type getTypeParameter0(final Type type) {
if ( !(type instanceof ParameterizedType) ) {
// List or List<?>
return Object.class;
}
return ((ParameterizedType) type).getActualTypeArguments()[0];
}
})
.create();
// This type declares a marker specialization to be used during deserialization
final Type type = new TypeToken<LenientListMarker<Map<String, Integer>>>() {
}.getType();
try ( final JsonReader jsonReader = getPackageResourceJsonReader(Q43528208.class, "non-standard.json") ) {
// This is where we're a sort of cheating:
// We tell Gson to deserialize LenientListMarker<Map<String, Integer>> but the type adapter above will return a list
final List<Map<String, Integer>> maps = gson.fromJson(jsonReader, type);
System.out.println(maps);
}
}
The output is now for Map<String, Integer>s, not JsonElements:
[{foo=1}, {bar=1}]
Update 2
TypeToken.getParameterized workaround:
#SuppressWarnings("unchecked")
final TypeToken<List<?>> listTypeToken = (TypeToken<List<?>>) TypeToken.get(new ParameterizedType() {
#Override
public Type getRawType() {
return List.class;
}
#Override
public Type[] getActualTypeArguments() {
return new Type[]{ listElementType };
}
#Override
public Type getOwnerType() {
return null;
}
});
We can have one more program to introduce comma(,) and construct a well formed JSON
With spark 2, we can add multiline as read option.
spark.df.option("multiline","true").json("data.json")
Related
Gson doesnt have direct support for serializing #AutoValue classes or for Optional<> fields, but com.ryanharter.auto.value adds #AutoValue and net.dongliu:gson-java8-datatype adds Optional<> and other java8 types.
However, they dont work together AFAICT.
Test code:
public class TestOptionalWithAutoValue {
private static final Gson gson = new GsonBuilder().serializeNulls()
// doesnt matter which order these are registered in
.registerTypeAdapterFactory(new GsonJava8TypeAdapterFactory())
.registerTypeAdapterFactory(AutoValueGsonTypeAdapterFactory.create())
.create();
#Test
public void testAutoValueOptionalEmpty() {
AvoTestClass subject = AvoTestClass.create(Optional.empty());
String json = gson.toJson(subject, AvoTestClass.class);
System.out.printf("Json produced = %s%n", json);
AvoTestClass back = gson.fromJson(json, new TypeToken<AvoTestClass>() {}.getType());
assertThat(back).isEqualTo(subject);
}
#Test
public void testAutoValueOptionalFull() {
AvoTestClass subject = AvoTestClass.create(Optional.of("ok"));
String json = gson.toJson(subject, AvoTestClass.class);
System.out.printf("Json produced = '%s'%n", json);
AvoTestClass back = gson.fromJson(json, new TypeToken<AvoTestClass>() {}.getType());
assertThat(back).isEqualTo(subject);
}
}
#AutoValue
public abstract class AvoTestClass {
abstract Optional<String> sval();
public static AvoTestClass create(Optional<String> sval) {
return new AutoValue_AvoTestClass(sval);
}
public static TypeAdapter<AvoTestClass> typeAdapter(Gson gson) {
return new AutoValue_AvoTestClass.GsonTypeAdapter(gson);
}
}
#GsonTypeAdapterFactory
public abstract class AutoValueGsonTypeAdapterFactory implements TypeAdapterFactory {
public static TypeAdapterFactory create() {
return new AutoValueGson_AutoValueGsonTypeAdapterFactory();
}
}
gradle dependencies:
annotationProcessor "com.google.auto.value:auto-value:1.7.4"
annotationProcessor("com.ryanharter.auto.value:auto-value-gson-extension:1.3.1")
implementation("com.ryanharter.auto.value:auto-value-gson-runtime:1.3.1")
annotationProcessor("com.ryanharter.auto.value:auto-value-gson-factory:1.3.1")
implementation 'net.dongliu:gson-java8-datatype:1.1.0'
Fails with:
Json produced = {"sval":null}
...
java.lang.NullPointerException: Null sval
...
net.dongliu.gson.OptionalAdapter is called on serialization, but not deserialization.
Im wondering if theres a workaround, or if the answer is that Gson needs to have direct support for Optional<> ?
Glad to see you've updated your question by adding much more information and even by adding a test! :) That really makes it clear!
I'm not sure, but the generated type adapter has no mention for the default value for sval:
jsonReader.beginObject();
// [NOTE] This is where it is initialized with null, so I guess it will definitely fail if the `sval` property is not even present in the deserialized JSON object
Optional<String> sval = null;
while (jsonReader.hasNext()) {
String _name = jsonReader.nextName();
// [NOTE] This is where it skips `null` value so it even does not reach to the `OptionalAdapter` run
if (jsonReader.peek() == JsonToken.NULL) {
jsonReader.nextNull();
continue;
}
switch (_name) {
default: {
if ("sval".equals(_name)) {
TypeAdapter<Optional<String>> optional__string_adapter = this.optional__string_adapter;
if (optional__string_adapter == null) {
this.optional__string_adapter = optional__string_adapter = (TypeAdapter<Optional<String>>) gson.getAdapter(TypeToken.getParameterized(Optional.class, String.class));
}
sval = optional__string_adapter.read(jsonReader);
continue;
}
jsonReader.skipValue();
}
}
}
jsonReader.endObject();
return new AutoValue_AvoTestClass(sval);
I have no idea if there is a way to configure the default values in AutoValue or other generators you mentioned, but it looks like a bug.
If there is no any way to work around it (say, library development abandoned; it takes too much time to wait for a fix; whatever), you can always implement it yourself, however with some runtime cost (basically this how Gson works under the hood for data bag objects).
The idea is delegating the job to the built-in RuntimeTypeAdapterFactory so that it could deal with a concrete class, not an abstract one, and set all fields according to the registered type adapters (so that the Java 8 types are supported as well).
The cost here is reflection, thus that adapter may work slower than generated type adapters.
Another thing is that if a JSON property does not even encounter in the JSON object, the corresponding field will remain null.
This requires another post-deserialization type adapter.
final class SubstitutionTypeAdapterFactory
implements TypeAdapterFactory {
private final Function<? super Type, ? extends Type> substitute;
private SubstitutionTypeAdapterFactory(final Function<? super Type, ? extends Type> substitute) {
this.substitute = substitute;
}
static TypeAdapterFactory create(final Function<? super Type, ? extends Type> substitute) {
return new SubstitutionTypeAdapterFactory(substitute);
}
#Override
#Nullable
public <T> TypeAdapter<T> create(final Gson gson, final TypeToken<T> typeToken) {
#Nullable
final Type substitution = substitute.apply(typeToken.getType());
if ( substitution == null ) {
return null;
}
#SuppressWarnings("unchecked")
final TypeAdapter<T> delegateTypeAdapter = (TypeAdapter<T>) gson.getDelegateAdapter(this, TypeToken.get(substitution));
return delegateTypeAdapter;
}
}
final class DefaultsTypeAdapterFactory
implements TypeAdapterFactory {
private final Function<? super Type, ? extends Type> substitute;
private final LoadingCache<Class<?>, Collection<Map.Entry<Field, ?>>> fieldsCache;
private DefaultsTypeAdapterFactory(final Function<? super Type, ? extends Type> substitute, final Function<? super Type, ?> toDefault) {
this.substitute = substitute;
fieldsCache = CacheBuilder.newBuilder()
// TODO tweak the cache
.build(new CacheLoader<Class<?>, Collection<Map.Entry<Field, ?>>>() {
#Override
public Collection<Map.Entry<Field, ?>> load(final Class<?> clazz) {
// TODO walk hieararchy
return Stream.of(clazz.getDeclaredFields())
.map(field -> {
#Nullable
final Object defaultValue = toDefault.apply(field.getGenericType());
if ( defaultValue == null ) {
return null;
}
field.setAccessible(true);
return new AbstractMap.SimpleImmutableEntry<>(field, defaultValue);
})
.filter(Objects::nonNull)
.collect(Collectors.toList());
}
});
}
static TypeAdapterFactory create(final Function<? super Type, ? extends Type> substitute, final Function<? super Type, ?> toDefault) {
return new DefaultsTypeAdapterFactory(substitute, toDefault);
}
#Override
#Nullable
public <T> TypeAdapter<T> create(final Gson gson, final TypeToken<T> typeToken) {
#Nullable
final Type substitution = substitute.apply(typeToken.getType());
if ( substitution == null ) {
return null;
}
if ( !(substitution instanceof Class) ) {
return null;
}
final Collection<Map.Entry<Field, ?>> fieldsToPatch = fieldsCache.getUnchecked((Class<?>) substitution);
if ( fieldsToPatch.isEmpty() ) {
return null;
}
final TypeAdapter<T> delegateTypeAdapter = gson.getDelegateAdapter(this, typeToken);
return new TypeAdapter<T>() {
#Override
public void write(final JsonWriter out, final T value)
throws IOException {
delegateTypeAdapter.write(out, value);
}
#Override
public T read(final JsonReader in)
throws IOException {
final T value = delegateTypeAdapter.read(in);
for ( final Map.Entry<Field, ?> e : fieldsToPatch ) {
final Field field = e.getKey();
final Object defaultValue = e.getValue();
try {
if ( field.get(value) == null ) {
field.set(value, defaultValue);
}
} catch ( final IllegalAccessException ex ) {
throw new RuntimeException(ex);
}
}
return value;
}
};
}
}
#AutoValue
abstract class AvoTestClass {
abstract Optional<String> sval();
static AvoTestClass create(final Optional<String> sval) {
return new AutoValue_AvoTestClass(sval);
}
static Class<? extends AvoTestClass> type() {
return AutoValue_AvoTestClass.class;
}
}
public final class OptionalWithAutoValueTest {
private static final Map<Type, Type> autoValueClasses = ImmutableMap.<Type, Type>builder()
.put(AvoTestClass.class, AvoTestClass.type())
.build();
private static final Map<Class<?>, ?> defaultValues = ImmutableMap.<Class<?>, Object>builder()
.put(Optional.class, Optional.empty())
.build();
private static final Gson gson = new GsonBuilder()
.registerTypeAdapterFactory(new GsonJava8TypeAdapterFactory())
.registerTypeAdapterFactory(SubstitutionTypeAdapterFactory.create(autoValueClasses::get))
.registerTypeAdapterFactory(DefaultsTypeAdapterFactory.create(autoValueClasses::get, type -> {
if ( type instanceof Class ) {
return defaultValues.get(type);
}
if ( type instanceof ParameterizedType ) {
return defaultValues.get(((ParameterizedType) type).getRawType());
}
return null;
}))
.create();
#SuppressWarnings("unused")
private static Stream<Optional<String>> test() {
return Stream.of(
Optional.of("ok"),
Optional.empty()
);
}
#ParameterizedTest
#MethodSource
public void test(final Optional<String> optional) {
final AvoTestClass before = AvoTestClass.create(optional);
final String json = gson.toJson(before, AvoTestClass.class);
final AvoTestClass after = gson.fromJson(json, AvoTestClass.class);
Assert.assertEquals(before, after);
}
}
This solution is reflection-based heavily, but it's just a work-around if the generators cannot do the job (again, not sure if they can be configured so that there are no such issues).
I have the following class
private static class ClassWithGenericType<T> {
Set<T> values;
}
If I initialize now the class with a Set of Enum-values, serialize and deserialize the object by using gson, the Set of the deserialized object does not contain the Enum-values, but the values as String.
I think this is because the generic type is thrown away through the serialization. I saw, that I could use new TypeToken<...>(){}.getType();, but the problem is, that the class above is part of a bigger object, so I cannot call gson.fromJson(classWithGenericType, typeToken) directly.
Is there a smart way of solving this problem? I thought of a TypeAdapter, which does not serialize only the values of the Set, but also it's type.
I found now a solution and created a TypeAdapter.
public class SetTypeAdapterFactory implements TypeAdapterFactory {
#Override
public <T> TypeAdapter<T> create(Gson gson, #NonNull TypeToken<T> type) {
if (!Set.class.isAssignableFrom(type.getRawType())) {
return null;
}
return (TypeAdapter<T>) new SetTypeAdapter(gson);
}
}
public class SetTypeAdapter extends TypeAdapter<Set<?>> {
public static final String TYPE = "#type";
public static final String DATA = "#data";
private final Gson gson;
public SetTypeAdapter(#NonNull Gson gson) {
this.gson = gson;
}
#Override
public void write(final JsonWriter out, final Set<?> set
) throws IOException {
out.beginArray();
for (Object item : set) {
out.beginObject();
out.name(TYPE).value(item.getClass().getName());
out.name(DATA).jsonValue(gson.toJson(item));
out.endObject();
}
out.endArray();
}
#Override
public Set<?> read(final JsonReader in) throws IOException {
final Set<Object> set = Sets.newHashSet();
in.beginArray();
while (in.hasNext()) {
in.beginObject();
set.add(readNextObject(in));
in.endObject();
}
in.endArray();
return set;
}
private Object readNextObject(JsonReader in) throws IOException {
try {
checkNextName(in, TYPE);
Class<?> cls = Class.forName(in.nextString());
checkNextName(in, DATA);
return gson.fromJson(in, cls);
} catch (ClassNotFoundException exception) {
throw new IOException(exception);
}
}
private void checkNextName(JsonReader in, String name) throws IOException {
if (!in.nextName().equals(name)) {
throw new IOException("Name was not: " + name);
}
}
}
We can add the factory to the GsonBuilder and afterwards we are capable of serializing a Set with generic types.
var gsonBuilder = new GsonBuilder();
gsonBuilder.registerTypeAdapterFactory(new SetTypeAdapterFactory());
var gson = gsonBuilder.create();
The serialized Set has then the following structure:
[
{
"#type":<class_name_first_element>,
"#data":<first_element_as_json>
},
...
]
Is there a way to configure Gson so that it treats any failed field parse as null instead of throwing a parse exception? Ideally we could catch and log the exception -- but we want the option to keep going with the program even if some fields (or subfields) do not parse as expected.
Example:
Malformed JSON:
{
"dog": []
}
With classes:
class Farm {
public Dog dog;
}
class Dog {
public String name;
}
Gson gson = new Gson();
Farm oldMcdonald = gson.fromJson(json, Farm.class); // should not throw exception
assertNull(oldMcdonald.dog); // should pass
In Gson, it can be implemented pretty easy.
Despite the following solution, I guess, seems not to work in any case (for example, primitives), it can be enhanced if necessary.
final class JsonFailSafeTypeAdapterFactory
implements TypeAdapterFactory {
private static final TypeAdapterFactory instance = new JsonFailSafeTypeAdapterFactory();
private JsonFailSafeTypeAdapterFactory() {
}
static TypeAdapterFactory get() {
return instance;
}
#Override
public <T> TypeAdapter<T> create(final Gson gson, final TypeToken<T> typeToken) {
// We can support non-primitive types only
if ( typeToken.getRawType().isPrimitive() ) {
return null;
}
final TypeAdapter<T> delegateTypeAdapter = gson.getAdapter(typeToken);
return new JsonFailSafeTypeAdapter<>(delegateTypeAdapter);
}
private static final class JsonFailSafeTypeAdapter<T>
extends TypeAdapter<T> {
private final TypeAdapter<T> delegateTypeAdapter;
private JsonFailSafeTypeAdapter(final TypeAdapter<T> delegateTypeAdapter) {
this.delegateTypeAdapter = delegateTypeAdapter;
}
#Override
public void write(final JsonWriter out, final T value)
throws IOException {
delegateTypeAdapter.write(out, value);
}
#Override
public T read(final JsonReader in)
throws IOException {
try {
return delegateTypeAdapter.read(in);
} catch ( final MalformedJsonException | RuntimeException ignored ) {
// Once we get into unexpected JSON token, let's *always* consider a fallback to the default value
// Well, the default is always `null` anyway, but we'll do more work
return fallback(in);
}
}
private static <T> T fallback(final JsonReader in)
throws IOException {
final JsonToken jsonToken = in.peek();
switch ( jsonToken ) {
case BEGIN_ARRAY:
case BEGIN_OBJECT:
case NAME:
case STRING:
case NUMBER:
case BOOLEAN:
case NULL:
// Assume we're at the beginning of a complex JSON value or a JSON primitive
in.skipValue();
break;
case END_ARRAY:
// Not sure if we skipValue() can fast-forward this one
in.endArray();
break;
case END_OBJECT:
// The same
in.endObject();
break;
case END_DOCUMENT:
// do nothing
break;
default:
throw new AssertionError(jsonToken);
}
// Just return null (at least at the moment)
return null;
}
}
}
Now just register the above type factory to handle all types (except java.lang.Object if I'm not mistaken).
private static final Gson gson = new GsonBuilder()
.registerTypeAdapterFactory(JsonFailSafeTypeAdapterFactory.get())
.create();
public static void main(final String... args)
throws IOException {
try ( final JsonReader jsonReader = Resources.getPackageResourceJsonReader(Q50002961.class, "farm.json") ) {
final Farm oldMcdonald = gson.fromJson(jsonReader, Farm.class);
if ( oldMcdonald.dog != null ) {
throw new AssertionError();
}
System.out.println(oldMcdonald);
}
}
Example output:
q50002961.Farm#626b2d4a
Another option is also specifying target fields if there is no need to register the factory globally. For instance:
final class Farm {
#JsonAdapter(JsonFailSafeTypeAdapterFactory.class)
final Dog dog = null;
}
I will post a solution for your problem but it would still require you to change the code on your side. For example if you have configured a property as an object and you receive an array - there is no way to map that properly. So I would suggest to change everything in your code to List and write a custom mapper that creates a list with one element when an object is received. This way you will be flexible to what you receive but you will also need to add some logic to handle problems when you have more than one objects to the array. For your example what would you do if you get 2 dogs? What is the correct behavior?
So I would do it like that:
public class MainClass {
public static <T> void main(String[] args) throws IOException {
Gson gson = new GsonBuilder().registerTypeAdapterFactory(new ArrayAdapterFactory()).create();
// Here I do the opposite - add one dog but expect a collection
String json = "{ \"dog\": {name=\"Snoopy\"} }";
Farm oldMcdonald = gson.fromJson(json, Farm.class); // should not throw exception
System.out.println("Dog:"+oldMcdonald.dog.get(0).name); //Works properly
}
}
class Farm {
#Expose
public List<Dog> dog; //All such properties become a list. You handle the situation when there are more than one values
}
class Dog {
#Expose
public String name;
}
class ArrayAdapter<T> extends TypeAdapter<List<T>> {
private Class<T> adapterclass;
public ArrayAdapter(Class<T> adapterclass) {
this.adapterclass = adapterclass;
}
public List<T> read(JsonReader reader) throws IOException {
List<T> list = new ArrayList<T>();
Gson gson = new GsonBuilder()
.registerTypeAdapterFactory(new ArrayAdapterFactory())
.create();
if (reader.peek() == JsonToken.BEGIN_OBJECT) {
T inning = gson.fromJson(reader, adapterclass);
list.add(inning);
// return null; here if you want to return null instead of list with one element
} else if (reader.peek() == JsonToken.BEGIN_ARRAY) {
reader.beginArray();
while (reader.hasNext()) {
T inning = gson.fromJson(reader, adapterclass);
list.add(inning);
}
reader.endArray();
}
return list;
}
public void write(JsonWriter writer, List<T> value) throws IOException {
}
}
class ArrayAdapterFactory implements TypeAdapterFactory {
#SuppressWarnings({ "unchecked", "rawtypes" })
#Override
public <T> TypeAdapter<T> create(final Gson gson, final TypeToken<T> type) {
TypeAdapter<T> typeAdapter = null;
try {
if (type.getRawType() == List.class)
typeAdapter = new ArrayAdapter(
(Class) ((ParameterizedType) type.getType())
.getActualTypeArguments()[0]);
} catch (Exception e) {
e.printStackTrace();
}
return typeAdapter;
}
}
Thanks to http://sachinpatil.com/blog/2012/07/03/gson/ for the idea
Assuming the following enum in a Java class:
enum AccessMode {
READ_WRITE,
READ_ONLY,
WRITE_ONLY
};
JSON deserialization works fine with Gson as long as the JSON contains a valid value for the enum field, e.g.:
"access": "READ_WRITE"
Unfortunately, fromJson() does seem to detect invalid enum values in the JSON, such as:
"access": "READ_XXX"
How can I add enum value checking when deserializing a JSON file using Gson?
As of version 2.8.2, Gson does not support such a use case.
I believe it's worthy to be submitted as a suggestion to the Gson development team as a special GsonBuilder configuration method.
The most you can do now is writing a custom enum type adapter that almost duplicates com.google.gson.internal.bind.EnumTypeAdapter functionality but adds the name check.
final class StrictEnumTypeAdapterFactory
implements TypeAdapterFactory {
private static final TypeAdapterFactory allStrictEnumTypeAdapterFactory = new StrictEnumTypeAdapterFactory(enumClass -> true);
private final Predicate<? super Class<? extends Enum<?>>> isStrictEnumClass;
private StrictEnumTypeAdapterFactory(final Predicate<? super Class<? extends Enum<?>>> isStrictEnumClass) {
this.isStrictEnumClass = isStrictEnumClass;
}
static TypeAdapterFactory get(final Predicate<? super Class<? extends Enum<?>>> isStrictEnumClass) {
return new StrictEnumTypeAdapterFactory(isStrictEnumClass);
}
static TypeAdapterFactory get() {
return allStrictEnumTypeAdapterFactory;
}
#Override
public <T> TypeAdapter<T> create(final Gson gson, final TypeToken<T> typeToken) {
final Class<? super T> rawType = typeToken.getRawType();
// Skip non-enums
if ( !Enum.class.isAssignableFrom(rawType) ) {
return null;
}
// Check if the enum is supported by the "strict" policy
#SuppressWarnings("unchecked")
final Class<? extends Enum<?>> enumRawType = (Class<? extends Enum<?>>) rawType;
if ( !isStrictEnumClass.test(enumRawType) ) {
return null;
}
// Trivial rawtypes/unchecked casts
#SuppressWarnings({ "rawtypes", "unchecked" })
final TypeAdapter<? extends Enum<?>> strictEnumTypeAdapter = StrictEnumTypeAdapter.get((Class) enumRawType);
#SuppressWarnings("unchecked")
final TypeAdapter<T> castTypeAdapter = (TypeAdapter<T>) strictEnumTypeAdapter;
return castTypeAdapter;
}
private static final class StrictEnumTypeAdapter<E extends Enum<E>>
extends TypeAdapter<E> {
private final Class<E> enumClass;
private final Map<String, E> nameToEnumConstant;
private final Map<E, String> enumConstantToName;
private StrictEnumTypeAdapter(final Class<E> enumClass, final Map<String, E> nameToEnumConstant, final Map<E, String> enumConstantToName) {
this.enumClass = enumClass;
this.nameToEnumConstant = nameToEnumConstant;
this.enumConstantToName = enumConstantToName;
}
private static <E extends Enum<E>> TypeAdapter<E> get(final Class<E> enumClass) {
final Map<String, E> nameToEnumConstant = new HashMap<>();
final Map<E, String> enumConstantToName = new HashMap<>();
final Map<String, E> enumNameToEnumConstant = Stream.of(enumClass.getEnumConstants())
.collect(Collectors.toMap(Enum::name, Function.identity()));
Stream.of(enumClass.getFields())
// It can be either a simple enum constant, or an enum constant that overrides
.filter(field -> enumClass.isAssignableFrom(field.getType()))
.forEach(field -> {
final E enumConstant = enumNameToEnumConstant.get(field.getName());
// For compatibility with the original type adapter, we have to respect the #SeriaizedName annotation
final SerializedName serializedName = field.getAnnotation(SerializedName.class);
if ( serializedName == null ) {
nameToEnumConstant.put(field.getName(), enumConstant);
enumConstantToName.put(enumConstant, field.getName());
} else {
nameToEnumConstant.put(serializedName.value(), enumConstant);
enumConstantToName.put(enumConstant, serializedName.value());
for ( final String alternate : serializedName.alternate() ) {
nameToEnumConstant.put(alternate, enumConstant);
}
}
});
return new StrictEnumTypeAdapter<>(enumClass, nameToEnumConstant, enumConstantToName)
.nullSafe(); // A convenient method to handle nulls
}
#Override
public void write(final JsonWriter out, final E value)
throws IOException {
out.value(enumConstantToName.get(value));
}
#Override
public E read(final JsonReader in)
throws IOException {
final String key = in.nextString();
// This is what the original type adapter probably misses
if ( !nameToEnumConstant.containsKey(key) ) {
throw new JsonParseException(enumClass + " does not have an enum named " + key + " at " + in);
}
return nameToEnumConstant.get(key);
}
}
}
Simple test:
private static final Gson gson = new GsonBuilder()
.registerTypeAdapterFactory(StrictEnumTypeAdapterFactory.get())
.create();
public static void main(final String... args)
throws IOException {
try ( final JsonReader jsonReader = Resources.getPackageResourceJsonReader(Q49572505.class, "good.json") ) {
System.out.println(gson.<Status>fromJson(jsonReader, Status.class).access);
}
try ( final JsonReader jsonReader = Resources.getPackageResourceJsonReader(Q49572505.class, "bad.json") ) {
try {
gson.<Status>fromJson(jsonReader, Status.class);
throw new AssertionError();
} catch ( final JsonParseException ex ) {
System.out.println(ex.getMessage());
}
}
}
Output:
READ_WRITE
class q49572505.AccessMode does not have an enum named READ_XXX at JsonReader at line 2 column 22 path $.access
You could look # Moshi. I have found it a suitable and straightforward replacement for GSON and it already supports this behavior.
#Lyubomyr_Shaydarlv's solution works, but if you don't want to duplicate GSON's internal code, you can use it as a delegate in a custom TypeAdapterFactory. Run the adapter, and if it returns null, you know the value was invalid. This has the advantage that it inherits and changes to default enum converter.
class StrictEnumTypeAdapterFactory implements TypeAdapterFactory {
public <T> TypeAdapter<T> create(Gson gson, TypeToken<T> type) {
Class<T> rawType = (Class<T>) type.getRawType();
if (!rawType.isEnum()) {
return null;
}
return newStrictEnumAdapter(gson.getDelegateAdapter(this, type));
}
private <T> TypeAdapter<T> newStrictEnumAdapter(
final TypeAdapter<T> delegateAdapter) {
return new TypeAdapter<T>() {
#Override
public void write(JsonWriter out, T value) throws IOException {
delegateAdapter.write(out, value);
}
#Override
public T read(JsonReader in) throws IOException {
// Peek at the next value and save it for the error message
// if you don't need the offending value's actual name
String enumValue = in.nextString();
JsonReader delegateReader = new JsonReader(new StringReader('"' + enumValue + '"'));
T value = delegateAdapter.read(delegateReader);
delegateReader.close();
if (value == null) throw new IllegalStateException("Invalid enum value - " + enumValue);
return value;
}
};
}
}
A client has decided to start sending empty JSON objects at the end of arrays (to help their caching facepalm) but this has caused a whole bunch of unexpected behaviour in my app.
For example this is the data I am being sent...
[{object}, {object}, {}, {object}]
I currently use GSON to deserialize data sent from the server. I have been looking into type adapters to filter out these empty objects, however I am not fully sure how to implement the filtering. Any ideas on how to skip empty objects using GSON?
You can try this solution from here
class CollectionAdapter implements JsonSerializer<Collection<?>> {
#Override
public JsonElement serialize(Collection<?> src, Type typeOfSrc, JsonSerializationContext context) {
if (src == null || src.isEmpty()) // exclusion is made here
return null;
JsonArray array = new JsonArray();
for (Object child : src) {
JsonElement element = context.serialize(child);
array.add(element);
}
return array;
}
}
Then register it
Gson gson = new GsonBuilder().registerTypeHierarchyAdapter(Collection.class, new CollectionAdapter()).create();
I solved this issue ... I had to make a TypeAdapterFactory that set empty objects to null then filtered out the nulls from the resulting list.
Here is my TypeAdapterFactory
private static class EmptyCheckTypeAdapterFactory implements TypeAdapterFactory {
#Override
public <T> TypeAdapter<T> create(final Gson gson, final TypeToken<T> type) {
// We filter out the EmptyCheckTypeAdapter as we need to check this for emptiness!
if (Story.class.isAssignableFrom(type.getRawType())) {
final TypeAdapter<T> delegate = gson.getDelegateAdapter(this, type);
final TypeAdapter<JsonElement> elementAdapter = gson.getAdapter(JsonElement.class);
return new EmptyCheckTypeAdapter<>(delegate, elementAdapter).nullSafe();
}
return null;
}
public class EmptyCheckTypeAdapter<T> extends TypeAdapter<T> {
private final TypeAdapter<T> delegate;
private final TypeAdapter<JsonElement> elementAdapter;
public EmptyCheckTypeAdapter(final TypeAdapter<T> delegate,
final TypeAdapter<JsonElement> elementAdapter) {
this.delegate = delegate;
this.elementAdapter = elementAdapter;
}
#Override
public void write(final JsonWriter out, final T value) throws IOException {
this.delegate.write(out, value);
}
#Override
public T read(final JsonReader in) throws IOException {
final JsonObject asJsonObject = elementAdapter.read(in).getAsJsonObject();
if (asJsonObject.entrySet().isEmpty()) {
return null;
}
return this.delegate.fromJsonTree(asJsonObject);
}
}
}
Finally, filtered out the nulls using the following code
myDto.stories.removeAll(Collections.singleton(null));