-
-
Notifications
You must be signed in to change notification settings - Fork 142
Description
Search before asking
- I searched in the issues and found nothing similar.
When using Jackson ObjectMapper to convert Scala objects to Map[String, Any], numeric fields defined as BigInt are converted to Double instead of BigInteger or remaining as BigInt, causing precision loss.
This issue is not about regression across versions, but about undesired numeric coercion from a large integer to a Double.
We were using jackson 2.13.2 and in that Scala BigInt is converted to Long while deserializing at this line
https://github.com/FasterXML/jackson-databind/blame/87014ed5f3dfc1dcc21b704c2ce117266780c166/src/main/java/com/fasterxml/jackson/databind/util/TokenBuffer.java#L1849
In Jackson 2.15.2, this has beed changed and the BigInt value is converted to Double type. The change is done here:
https://github.com/FasterXML/jackson-databind/blame/74a57b2f6cd3b6e0fe2a61ed16734f15ba2abc95/src/main/java/com/fasterxml/jackson/databind/util/TokenBuffer.java#L1936
The serialisation part is same in both the version. But diff is while deserialisations. I wanted to understand if this is intentional or not? How can I fix this?
Version Information
2.15.2
Reproduction
case class A(b: BigInt = BigInt("9223372036854775808"))
val obj = A()
val map = objectMapper.convertValue(obj, classOf[Map[String, Any]])
println(map("b")) // 9.223372036854776E18 (Double) instead of BigInt or BigInteger
Expected behavior
BigInt in Scala should be treated a BigInteger in Java.
Additional context
No response