Unable to map the avro data to the POJO

[dprajasekhar]

I am working on a requirement to develop a generic cdc consumer using kafka . The records are put into kafka using jdbc connector , so based on the source the avro schema may vary . For example if I use Idcol which is long data type in the java , the datatypes in the databases could be different

postgres :
 idcol bigint,
avro schema :
"name": "idcol",
      "type": [
        "null",
        "long"
      ]

sqlserver  :

numeric(19,0)

avro schema :
"name": "idcol",
      "type": [
        "null",
        {
          "type": "bytes",
          "scale": 0,
          "precision": 64,
          "connect.version": 1,
          "connect.parameters": {
            "scale": "0"
          },
          "connect.name": "org.apache.kafka.connect.data.Decimal",
          "logicalType": "decimal"
        }

oracle:

idcol NUMBER(19,0)

avroschema:
"name":"IDCOL",
         "type":{
            "type":"bytes",
            "scale":0,
            "precision":64,
            "connect.version":1,
            "connect.parameters":{
               "scale":"0"
            },
            "connect.name":"org.apache.kafka.connect.data.Decimal",
            "logicalType":"decimal"
         }

So when ever I tried to deserialize the data using avro mapper , the code works for postgres and fails for the oracle & mssql with exception.

com.fasterxml.jackson.databind.exc.MismatchedInputException: Cannot deserialize instance of long out of VALUE_EMBEDDED_OBJECT token

I can avoid this by adding a custom deserializer above the destination POJO , some thing like this

    @Id
    @Column(name = "id")
    @JsonProperty("idcol")
    @JsonDeserialize(using = LongDeserializer.class)
    private Long id;

But this is the same case for all the data types like int , boolean, short etc ..,

I am mapping the pojo using the following snippet

        AvroMapper mapper = new AvroMapper();
        mapper.configure(JsonGenerator.Feature.IGNORE_UNKNOWN, true);
        mapper.configure(MapperFeature.ACCEPT_CASE_INSENSITIVE_PROPERTIES, true);
        mapper.setSerializationInclusion(JsonInclude.Include.NON_NULL);
        
         GenericRecord avroRecord = (GenericRecord)consumerRecord.value();
        byte[] avroBytes = convertToBytes(avroRecord);
      
        Schema schema = avroRecord.getSchema();
        AvroSchema avroSchema = new AvroSchema(schema);
        
        JavaType javaType = TypeFactory.defaultInstance().constructType(pojoType);
        
        return avroMapper.readerFor(javaType).with(avroSchema)
                .readValue(avroBytes);

Let me know if I am missing some thing , any help is appreciated .

[cowtowncoder]

Would need a stand-alone reproduction (ideally: unit tests) here. Leaving open for now but may close without test.