Java Integer, Long, BigInteger or BigDecimal generation from NUMBER(X,Y) JSON Schema type based on X and Y

Currently jsonschema2pojo always generates Java BigDecimal for NUMBER(X,Y) type JSON Schema even if Y=0.

We suggest following more fine-tuned generation of Java types for NUMBER(X,Y):

• For NUMBER(X,0) where X <= 9 generate Java Integer
• For NUMBER(X,0) where X >= 10 and X <= 18 generate Java Long
• For NUMBER(X,0) where X >= 19 generate Java BigInteger
• For NUMBER(X,Y) where Y != 0 generate Java BigDecimal

PS: We have JSON schema from third party with lots of NUMBER(X,0) definitions and we were not able to persuade them to use integer. This results in lots of Java BigDecimal in our code with redundant conversions there and back which makes us unhappy.

Many thanks!

[joelittlejohn]

@marcelmatula The less magic we have in defining how these types should be generated the better. This idea of using int/long for number when multipleOf is a whole number is interesting but this really isn’t how you are supposed to represent whole numbers. Yes type ‘number’ allows for any number, but if you require that a number is an integer then you should use "type": "integer".

I appreciate this would be useful for you. However I think more people would be annoyed by this as would be pleased by it, and I feel it’s an odd choice to change the type. I’m going to close this as it isn’t really something I’m interested to add into jsonschema2pojo.

[eirnym]

“number” type in json schema by definition means it’s decimal (float, double and BigDecimal)

If you want integers (int or long in Java), “integer” type is for disposition.