On Wed, Oct 31, 2018 at 12:25 PM Rick Ley <[email protected]> wrote:
>
> Thank you for your response. Given that information, we are investigating the 
> possibility of writing a custom deserializer for this type. To get me 
> started, it looks like we should be extending StdDeserializer<Item>? Is that 
> correct?

Yes.

Writing custom deserializers may be tricky with XML, but that  is the
useful base class here.

-+ Tatu +-

>
> On Tuesday, October 30, 2018 at 8:04:45 PM UTC-7, Tatu Saloranta wrote:
>>
>> On Tue, Oct 30, 2018 at 7:26 PM Rick Ley <[email protected]> wrote:
>> >
>> > Hello!
>> >
>> > We are working on client libraries for a service. One of the service APIs 
>> > can return an xml list with heterogeneous types ordered by name. There are 
>> > two possible types in this list. A sample response might be:
>> >
>> > <Items>
>> >     <TypeA>
>> >         <Name1>
>> >     </TypeA>
>> >     <TypeB>
>> >         <Name2>
>> >     </TypeB>
>> >     <TypeA>
>> >         <Name3>
>> >     </TypeA>
>> >     <TypeA>
>> >         <Name4>
>> >     </TypeA>
>> >     <TypeB>
>> >         <Name5>
>> >     </TypeB>
>> >     <TypeB>
>> >         <Name6>
>> >     </TypeB>
>> >     <TypeA>
>> >         <Name7>
>> >     </TypeA>
>> >     <TypeB>
>> >         <Name8>
>> >     </TypeB>
>> >     <TypeB>
>> >         <Name9>
>> >     </TypeB>
>> > </Item>
>> >
>> > The goal is to deserialize this list into two separate lists that each 
>> > hold all of one of the types. So with the given example, we would hope to 
>> > see (with the Name here being shorthand for a fully deserialized object): 
>> > listOfTypeA={Name1, Name3, Name4, Name7} and listOfTypeB = {Name2, Name5, 
>> > Name6, Name8, Name9}. Instead, we are seeing listOfTypeA = {Name7} and 
>> > listOfTypeB={Name8, Name9}. In other words, only the last contiguous set 
>> > of names is persisted and returned.
>> >
>> > I have stepped through the code and confirmed that this is because each 
>> > time an element of a certain type is encountered, a new list is created 
>> > and eventually overwrites any other list that was previously on the root 
>> > object rather than appending to an existing list. More specifically, each 
>> > time CollectionDeserializer.deserialize is called, a new list is 
>> > instantiated.
>> >
>> > Our object model is the parent type of name Item, which has an 
>> > ArrayList<TypeA> and ArrayList<TypeB>. Each is annoted with 
>> > @JsonProperty() and the respective type name. We are using Jackson version 
>> > 2.8.11.
>> >
>> > Is there a way we can configure Jackson to yield the behavior we want? Or 
>> > some sort of workaround you can suggest? Or should I make a feature 
>> > request/issue on the repo?
>> >
>> > Thank you for your help!
>>
>> To give a simple answer, no, Jackson can not be configured to do that.
>> I also do not see this as something that would be possible to support
>> in suggested form.
>>
>> Problem is two-fold:
>>
>> 1. Polymorphic type handling: it might be possible to make work, but
>> lack of wrapping for list would present a problem.
>> 2. Splitting of contents of one logical container at data format level
>> into 2 Java Collections: this is not supported for any format.
>>
>> If (1) was resolved (that is, you could get one List populated, from
>> polymorphic types), (2) could probably be handled by defining setter
>> method that takes list, and then splitting that up in code.
>>
>> -+ Tatu +-
>
> --
> You received this message because you are subscribed to the Google Groups 
> "jackson-user" group.
> To unsubscribe from this group and stop receiving emails from it, send an 
> email to [email protected].
> To post to this group, send email to [email protected].
> For more options, visit https://groups.google.com/d/optout.

-- 
You received this message because you are subscribed to the Google Groups 
"jackson-user" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to [email protected].
To post to this group, send email to [email protected].
For more options, visit https://groups.google.com/d/optout.

Reply via email to