将不同的 dynamodb 项（来自同一个 dynamodb 表）解组到多个 POJO

Question

Context: I've 3 java classes that have different attributes and the JSON representation that correspond to these classes vary in structure. Say the following are the 3 JSON structures

JSON representation for Class A

{
  "foo" : ["some string-a", "some string-b", "some string-c"],
  "bar" : 890,
  "maz" : {"x":-11, "y": 0, "z":89} 
}

JSON representation for Class B

{
  "abo" : [{"key": "val", "anotherKey": "anotherValue"}]
  "seq" : [4, 7, 1, 3]
}

JSON representation for Class C

{
  "zin" : "some nice text",
  "tag" : ["dynamodb", "unmarshalling", "different schemas", "same table"]
}

I'm storing these JSON documents in one table in dynamoDB - let's call it tblStructures.

Why am I storing these different classes in the same DynamoDB table? Because they belong to the same business concept. And having as many DynamoDB tables as the business concepts doesn't seem to be logical.

Question: How can I unmarshall (deserialize) these items from the same dynamodb table to multiple Java Objects?

Answer 1

Borrowing a tip from a module for Mongoose that provides an easy way to work with base schemas and differentiate different types of models in the same collection on a MongoDB database ('collections' are akin to 'tables' in dynamodb), you could add a _type property to each model's schema and use that as a hint to deserialize them with the right class.

For example:

JSON representation for Class A

{
  "foo" : ["some string-a", "some string-b", "some string-c"],
  "bar" : 890,
  "maz" : {"x":-11, "y": 0, "z":89},
  "_type" : "modelA"
}

JSON representation for Class B

{
  "abo" : [{"key": "val", "anotherKey": "anotherValue"}]
  "seq" : [4, 7, 1, 3],
  "_type" : "modelB"
}

I've used this approach in the past in conjunction with a routine to automatically treat all properties beginning with "_" as private fields and just stripping them when serialising to JSON (eg _password for password hashes in User objects, and for other metadata that I'd don't intent to directly expose).