dev1.1.0-issue-8556

Version dev1.1.0-issue-8556

1.1.0

• Era support for date and date time (2022-01-23T01:23:45.678-11:30 BC, 2021-1-1 BC)
• String numbers
• Infinity, -Infinity, NaN for avro double types
• Updated Jitpack Java version to 17.0.5

v1.0.1

Summary

• Fix Jitpack

v1.0.0

Fluent Constructor

The construction of JsonAvroConverter and JsonGenericRecordReader is now done with a builder.

JsonAvroConverter converter = JsonAvroConverter.builder()
    .setUnknownFieldListener(listener)
    .build();

Name Transformer

According to Avro spec, Avro names must start with [A-Za-z_] followed by [A-Za-z0-9_] (see here for details), while Json object property names can be more flexible. A name transformer can be set on JsonAvroConverter during the construction to convert a Json object property name to a legal Avro name.

Function<String, String> nameTransformer = String::toUpperCase;
JsonGenericRecordReader reader = JsonGenericRecordReader.builder()
    .setNameTransformer(nameTransformer)
    .build();

Enforced String Fields

When a Json array field has no items, the element of that array field may have any type. However, Avro requires that each array has a clear type specification. To solve this problem, the Json to Avro schema converter assigns string as the type of the elements for any untyped arrays. Accordingly, this Json to Avro object converter follows the string schema, and any non-null field value is forced to be a string.

For example, given the following Json schema:

{
  "type": "object",
  "properties": {
    "identifier": {
      "type": "array"
    }
  }
}

and Json object:

{
  "identifier": ["151", 152, true, { "id": 153 }, null]
}

the corresponding Avro schema is:

{
  "type": "record",
  "fields": [
    {
      "name": "identifier",
      "type": [
        "null",
        {
          "type": "array",
          "items": ["null", "string"]
        }
      ],
      "default": null
    }
  ]
}

and the Avro object is:

{
  "identifier": ["151", "152", "true", "{\"id\": 153}", null]
}

Note that every non-null element inside the identifier array field is converted to string.

This enforcement only applies to an array whose items can only be nullable strings. For arrays with union item types, no string enforcing will be carried out.

Additional Properties

Avro Additional Properties

A Json object can have additional properties of unknown types, which is not compatible with the Avro schema. To solve this problem during Json to Avro object conversion, we introduce a special field: _airbyte_additional_properties typed as a nullable map from string to string:

{
  "name": "_airbyte_additional_properties",
  "type": ["null", { "type": "map", "values": "string" }],
  "default": null
}

The name of this field is customizable:

JsonAvroConverter converter = JsonAvroConverter.builder()
    .setAvroAdditionalPropsFieldName("_additional_properties")
    .build();

When this field exists in the Avro schema for a record, any unknown fields will be serialized as a string be stored under this field.

Given the following Avro schema:

{
  "type": "record",
  "name": "simple_schema",
  "fields": [
    {
      "name": "username",
      "type": ["null", "string"],
      "default": null
    },
    {
      "name": "_airbyte_additional_properties",
      "type": ["null", { "type": "map", "values": "string" }],
      "default": null
    }
  ]
}

this Json object

{
  "username": "Thomas",
  "active": true,
  "age": 21,
  "auth": {
    "auth_type": "ssl",
    "api_key": "abcdefg/012345",
    "admin": false,
    "id": 1000
  }
}

will be converted to the following Avro object:

{
  "username": "Thomas",
  "_airbyte_additional_properties": {
    "active": "true",
    "age": "21",
    "auth": "{\"auth_type\":\"ssl\",\"api_key\":\"abcdefg/012345\",\"admin\":false,\"id\":1000}"
  }
}

Note that all fields other than the username is moved under __airbyte_additional_properties as a string.

Json Additional Properties

The input Json object may also have some fields with unknown types. For example, if the Json object already has an _airbyte_additional_properties field, it will follow the same rule:

• If the Avro schema defines an _airbyte_additional_properties field, all subfields inside this field will be kept in the Avro object, but they will all become string. Other unknown fields will also be stored under this field.
• If the Avro schema does not define such field, it will be completely ignored.

For example, with the same Avro schema, given this Json object:

{
  "username": "Thomas",
  "active": true,
  "_airbyte_additional_properties": {
    "age": 21,
    "auth": {
      "auth_type": "ssl",
      "api_key": "abcdefg/012345",
      "admin": false,
      "id": 1000
    }
  }
}

will be converted to the following Avro object:

{
  "username": "Thomas",
  "_airbyte_additional_properties": {
    "age": "21",
    "active": "true",
    "auth": "{\"auth_type\":\"ssl\",\"api_key\":\"abcdefg/012345\",\"admin\":false,\"id\":1000}"
  }
}

Note the undefined active field is moved into _airbyte_additional_properties. age and auth that are originally in _airbyte_additional_properties becomes strings.

The name of the additional properties field in the Json object is also customizable:

JsonAvroConverter converter = JsonAvroConverter.builder()
    .setJsonAdditionalPropsFieldNames(Set.of("_additional_properties"))
    .build();

By default, both _ab_additional_properties and _airbyte_additional_properties are the additional properties field names on the Json object.