Hi, There is a kafka producer converting decimals to bytes using this: import java.math.BigDecimal;
import java.math.RoundingMode;
import java.nio.ByteBuffer;
import org.apache.avro.Conversion;
import org.apache.avro.Conversions;
import org.apache.avro.LogicalType;
import org.apache.avro.LogicalTypes;
import org.mapstruct.Mapper;
@Mapper
public class BigDecimalMapper {
public ByteBuffer bigDecimalToByteBuffer(BigDecimal bigDecimal) {
if (bigDecimal == null) return null;
Conversion<BigDecimal> conversion = new Conversions.DecimalConversion();
LogicalType type = LogicalTypes.decimal(15, 4);
BigDecimal scaledBigDecimal = bigDecimal.setScale(4, RoundingMode.HALF_UP);
// Schema parameter is not used by Conversions so pass as null
return conversion.toBytes(scaledBigDecimal, null, type);
}
We are using the confluent schema registry and this is the schema: { "type": "record", "namespace": "event", "name": "Event", "fields": [ { "name": "maxApprovedAmount", "type": "bytes", "logicalType": "decimal", "precision": 4, "scale": 4 }] } In Nifi I am using ConsumeKafkaRecord_2_6. using AvroReader and AvroRecordSetWriter. In the reader I am using the schema from confluent. On the writer I am using Inherent Record Schema. When viewing the data in view flow file and selecting formatted, i see the value as: "maxApprovedAmount" : { "bytes" : "\u0001}x@" }, Trouble is how to get the actual value and not this. I have tried multiple changes to my avro writer, and I have read other posts on similar issues but not able to get this resolved. I believe it has to with the lack of support for Logical types. I am even up to write a script in jython to do the conversion if it will work. Any help would be greatly appreciated!
... View more
