mirror of
https://github.com/jwtk/jjwt.git
synced 2025-02-10 18:24:42 +00:00
13906d3746
Jackson chooses the target type for JSON numbers based on their value,
while deserializing without correct typing information present.
This leads to a confusing behavior:
String token = Jwts.builder()
.claim("byte", (byte) 42)
.claim("short", (short) 42)
.claim("int", 42)
.claim("long_small", (long) 42)
.claim("long_big", ((long) Integer.MAX_VALUE) + 42)
.compact();
Claims claims = (Claims) Jwts.parser().parse(token).getBody();
claims.get("int", Integer.class); // => 42
claims.get("long_big", Long.class); // => ((long) Integer.MAX_VALUE) + 42
claims.get("long_small", Long.class); // throws RequiredTypeException: required=Long, found=Integer
claims.get("short", Short.class); // throws RequiredTypeException: required=Short, found=Integer
claims.get("byte", Byte.class); // throws RequiredTypeException: required=Byte, found=Integer
With this commit, `DefaultClaims.getClaim(String, Class<T>)` will
correctly handle cases when required type is `Long`, `Integer`, `Short`
or `Byte`: check that value fits in the required type and cast to it.
// ... setup is the same as above
claims.get("int", Integer.class); // => 42
claims.get("long_big", Long.class); // => ((long) Integer.MAX_VALUE) + 42
claims.get("long_small", Long.class); // => (long) 42
claims.get("short", Short.class); // => (short) 42
claims.get("byte", Byte.class); // => (byte) 42
Fixes #142.