Deserialization from "{}" to java.lang.Object causes "out of END_OBJECT token" error

[pianovwork]

Hi, I've faced with a problem that is too similar this one #941. I expect that "{}" will be parsed correctly to empty Map when I'm using @JsonCreator

I've found that this case is invalid https://github.com/FasterXML/jackson-databind/blob/jackson-databind-2.6.3/src/main/java/com/fasterxml/jackson/databind/deser/std/UntypedObjectDeserializer.java#L272, but why?

Here is the minimum code to reproduce:

import java.io.IOException;
import com.fasterxml.jackson.annotation.JsonCreator;
import com.fasterxml.jackson.databind.ObjectMapper;

public class Main {
    public static void main(String[] args) throws IOException {
        ObjectMapper mapper = new ObjectMapper();
        mapper.readValue("[]", SomeObjectThatCanBeAggregated.class);
        mapper.readValue("[{}]", SomeObjectThatCanBeAggregated.class);
        mapper.readValue("{\"key\":null}", SomeObjectThatCanBeAggregated.class);
        mapper.readValue("{}", SomeObjectThatCanBeAggregated.class);
    }
}
class SomeObjectThatCanBeAggregated {

    @JsonCreator
    public SomeObjectThatCanBeAggregated(Object obj) {
        System.out.println(obj + " //" + obj.getClass());
    }
}

Output:

[] //class java.util.ArrayList
[{}] //class java.util.ArrayList
{key=null} //class java.util.LinkedHashMap
Exception in thread "main" com.fasterxml.jackson.databind.JsonMappingException: Can not deserialize instance of java.lang.Object out of END_OBJECT token
 at [Source: {}; line: 1, column: 2]
...

[cowtowncoder]

With Jackson 2.6, or some other version?

[pianovwork]

I've updated the first message with a link to source, yes, I'm using 2.6.3

[cowtowncoder]

Thanks! I'll have a look.

[pianovwork]

If someone need a hot fix, currently I've found this solution, but it's a little complicated with an extra custom deserializator

import java.io.IOException;
import com.fasterxml.jackson.annotation.JsonIgnore;
import com.fasterxml.jackson.annotation.JsonValue;
import com.fasterxml.jackson.core.JsonParser;
import com.fasterxml.jackson.databind.DeserializationContext;
import com.fasterxml.jackson.databind.JsonDeserializer;
import com.fasterxml.jackson.databind.ObjectMapper;
import com.fasterxml.jackson.databind.annotation.JsonDeserialize;

public class Main {
    public static void main(String[] args) throws IOException {
        ObjectMapper mapper = new ObjectMapper();
        mapper.readValue("[]", SomeObjectThatCanBeAggregated.class);
        mapper.readValue("[{}]", SomeObjectThatCanBeAggregated.class);
        mapper.readValue("{\"key\":null}", SomeObjectThatCanBeAggregated.class);
        mapper.readValue("{}", SomeObjectThatCanBeAggregated.class);
    }
}

@JsonDeserialize(using = MyDeserializer.class)
class SomeObjectThatCanBeAggregated {
    @JsonIgnore
    private Object data;
    //    @JsonCreator
    //    using custom deserialization instead of com.fasterxml.jackson.databind.deser.std.UntypedObjectDeserializer
    public SomeObjectThatCanBeAggregated(Object obj) {
        this.data = obj;
        System.out.println(obj + " //" + obj.getClass());
    }
    @JsonValue
    public Object getData() {
        return data;
    }
}
class MyDeserializer extends JsonDeserializer {
    @Override
    public SomeObjectThatCanBeAggregated deserialize(JsonParser jp, DeserializationContext ctxt) throws IOException {
        Object data = jp.readValueAs(Object.class);
        return new SomeObjectThatCanBeAggregated(data);
    }
}

[cowtowncoder]

Hmmh. This sounds quite familiar actually... like #941.

[pianovwork]

Yes, it's look like the same, but ObjectNode expects just "{...}" some object structure, and it doesn't accept any array or just text value.

[cowtowncoder]

What I meant is that the underlying problem is the same: due to buffering needed by @JsonCreator, token seen by delegate deserializer does not have current token JsonToken.START_OBJECT or JsonToken.FIELD_NAME. This needs special handling by deserializers that handle JSON Objects. And that was missing from JsonNode deserializer, was fixed, but now needs similar fix for "untyped" deserializer.

[cowtowncoder]

Thank you for reporting this! Just fixed, will be included in 2.6.4, 2.7.0

[pianovwork]

@cowtowncoder Thank you too!

[sgmiller]

I believe I'm still having this issue, though with the SmileFactory.