Type id not exposed for JsonTypeInfo.As.EXTERNAL_PROPERTY even when visible set to true

[libetl]

Jackson : 2.8.0.rc2

Hello,

I have these following jsons to deserialize, with a type property ("form_of_payment") located outside a polymorphic class ("payment_details") :

    {"form_of_payment":"INDIVIDUAL_CREDIT_CARD","payment_details":{"card_holder_first_name":"John", "card_holder_last_name":"Doe",  "number":"XXXXXXXXXXXXXXXX", "expiryDate":"MM/YY"}

    {"form_of_payment":"INSTRUMENTED_CREDIT_CARD","payment_details":{"payment_instrument_id":"00000000-0000-0000-0000-000000000000", "name":"Mr John Doe encrypted credit card"}

I have to use the @JsonTypeInfo annotation on the PaymentDetails class, to inform Jackson that there are several kinds of PaymentDetails.

The include must be a include = JsonTypeInfo.As.EXTERNAL_PROPERTY.

Once the TypeResolver has done his job, the JsonTypeInfo is removed from the feed, which annoys me a lot (the "form_of_payment" also must be read for business purposes). If it is an EXTERNAL_PROPERTY, there is no way to restore it (it works with PROPERTY, but the JsonTypeInfo must be inside the same object)

Can you add a spare visibility property to the @JsonTypeInfo to find it in the parent object ?
Here is a link to where it happens : https://github.com/FasterXML/jackson-databind/blob/master/src/main/java/com/fasterxml/jackson/databind/jsontype/impl/AsArrayTypeDeserializer.java#L99

Thanks.

(I have seen the #408 issue BTW)

[cowtowncoder]

Did you check the javadocs? If so, wouldn't visible:

http://fasterxml.github.io/jackson-annotations/javadoc/2.7/com/fasterxml/jackson/annotation/JsonTypeInfo.html#visible()

do what you ask? If not, why?

[libetl]

visible does not work with external_property

[cowtowncoder]

@libetl issue #408 claims it does, so I'd need a unit test to prove otherwise to see when and how it does not work as expected, reproducible against either 2.8.0 or 2.7.5 (earlier branches are closed for development).

[libetl]

with pleasure. Give me some time to try that (going back home for the moment)

[libetl]

Please have a look at this failing test (I have cloned HEAD today) :
https://gist.github.com/libetl/9c6ec1820444370c3c55670058f57426

[cowtowncoder]

@libetl thank you for reproduction. I hope to have a look at it soon.

[libetl]

I found out that this issue is closely linked to the builder deserialization.
I have updated my gist, with two versions of the same unmarshalling.
The first one (without builder) works, the second one fails.

[libetl]

NOW, I know.
The Jira JACKSON-831 correction on the BeanDeserializer was not applied on the BuilderBasedDeserializer (which appears to be a copy/paste).
Do you want me to submit a PR ?