Native Bid Response could not be parsed

[bschev]

Hi,

until now I used version 0.8.1 without problems, with all higher versions I get an exception while reading the following native response:

{
  "id": "123456",
  "seatbid": [
    {
      "seat": "987654",
      "bid": [
        {
          "id": "1",
          "impid": "1",
          "price": 0.345,
          "nurl": "http://someurl.com/win/123456?price=${AUCTION_PRICE}",
          "adm": "{\"native\": {\"ver\":1, \"assets\": [{\"id\": 1,\"title\": {\"text\": \"title text\"}},{\"id\": 2,\"img\": {\"url\": \"http://mainimage.com\", \"w\":300, \"h\":400}},{\"id\": 3,\"img\": {\"url\": \"http://logoimage.com\", \"w\":100, \"h\":100}},{\"id\": 4,\"data\": {\"value\": \"caption text\"}},{\"id\": 5,\"data\": {\"value\": \"brand name\"}}], \"link\" : {\"url\":\"http://destination.com\", \"clicktrackers\":[\"http://clktracker.com/clk\"]}, \"imptrackers\" : [\"http://url.com/123\", \"http://url.com/345\"]}}"
        }
      ]
    }
  ]
}

Exception:

com.google.protobuf.UninitializedMessageException: Message missing required fields: link
    at com.google.protobuf.AbstractMessage$Builder.newUninitializedMessageException(AbstractMessage.java:372)
    at com.google.openrtb.OpenRtbNative$NativeResponse$Builder.build(OpenRtbNative.java:14273)
    at com.google.openrtb.json.OpenRtbNativeJsonReader.readNativeResponse(OpenRtbNativeJsonReader.java:310)
    at com.google.openrtb.json.OpenRtbNativeJsonReader.readNativeResponse(OpenRtbNativeJsonReader.java:303)

The adm value seems to be the problem, expected is {"ver":1, "assets":[...]} and not {"native": {"ver":1, "assets":[...]}}.

If a response is created with the OpenRtb.BidResponse.Builder the native attribute is also not created.

Is that the desired behaviour?

If I understand the OpenRTB Native Ads Specification correctly, there should be a native attribute in the adm JSON. The example Bid Response JSON in the specification also includes the native. Or do I get it wrong?

[opinali]

This is not the desired behavior, it's a bug, it seems the library never supported the root native field. I think the reason this kind-of-worked before is that the parser would ignore things it didn't understood in the start of the stream, but this won't happen anymore in the current parser which was improved to scan the document more reliably in several ways. But it's a good thing that this uncovered an old bug. Expect a fix and release with this fix early next week...

[opinali]

Please test the current code just committed to master.

[bschev]

I tested the new code, it works.
Thank you for the fast fix!