This project is a JSON to Avro conversion tool designed to make migration to Avro easier. It includes a simple command line validator.
Apache Avro ships with some very advanced and efficient tools for reading and writing binary Avro but their support for JSON to Avro conversion is unfortunately limited and requires wrapping fields with type declarations if you have some optional fields in your schema. This tool is supposed to help with migrating projects from using JSON to Avro without having to modify JSON data if it conforms to the JSON schema.
- conversion of binary JSON to binary Avro
- conversion of binary JSON to GenericData.Record
- conversion of binary JSON to Avro generated Java classes
- conversion of binary Avro to binary JSON
- optional field support (unions do not require wrapping)
- unknown fields that are not declared in schema are ignored
dependencies {
compile group: 'tech.allegro.schema.json2avro', name: 'converter', version: '0.2.13'
}
import tech.allegro.schema.json2avro.converter.AvroConversionException;
import tech.allegro.schema.json2avro.converter.JsonAvroConverter;
import org.apache.avro.generic.GenericData;
import org.apache.avro.Schema;
// Avro schema with one string field: username
String schema =
"{" +
" \"type\" : \"record\"," +
" \"name\" : \"Acme\"," +
" \"fields\" : [{ \"name\" : \"username\", \"type\" : \"string\" }]" +
"}";
String json = "{ \"username\": \"mike\" }";
JsonAvroConverter converter = new JsonAvroConverter();
// conversion to binary Avro
byte[] avro = converter.convertToAvro(json.getBytes(), schema);
// conversion to GenericData.Record
GenericData.Record record = converter.convertToGenericDataRecord(json.getBytes(), new Schema.Parser().parse(schema));
// conversion from binary Avro to JSON
byte[] binaryJson = converter.convertToJson(avro, schema);
// exception handling
String invalidJson = "{ \"username\": 8 }";
try {
converter.convertToAvro(invalidJson.getBytes(), schema);
} catch (AvroConversionException ex) {
System.err.println("Caught exception: " + ex.getMessage());
}
If some avro types are not managed by the library, you can extend it by adding your own AvroTypeConverter. An AvroTypeConverter read a json value and convert it to an avro value. This can be useful when some logical-types are missing. The AvroTypeConverter can also be used to define a customer converter for a specific path.
public class CustomFieldConverter implements AvroTypeConverter {
@Override
public Object convert(Schema.Field field, Schema schema, Object jsonValue, Deque<String> path, boolean silently) {
return "custom-" + jsonValue;
}
@Override
public boolean canManage(Schema schema, Deque<String> path) {
return "customField".equals(path.getLast());
}
}
To use the converter you should add it to the JsonAvroConverter
, to do that you should build it like that
new JsonAvroConverter(new CompositeJsonToAvroReader(new CustomFieldConverter()))
A command line tool for validating your JSON/Avro documents against a schema.
To bundle the tool into single executable JAR:
./gradlew :validator:shadowJar
java -jar validator/build/libs/json2avro-validator-{version}.jar --help
Running Validator with --help
option will print a help message listing all possible arguments.
Sample Avro schema and messages can be found in:
- schema:
validator/src/test/resources/user.avcs
- JSON message:
validator/src/test/resources/user.json
- Avro message:
validator/src/test/resources/user.avro
You can validate your JSON to Avro conversion:
java -jar json2avro-validator.jar -s user.avcs -i user.json
If everything processes correctly, the process will end with zero status code.
You can convert the Avro binary data into JSON by setting mode -m avro2json
option:
java -jar json2avro-validator.jar -s user.avcs -i user.avro -m avro2json
If you would like to know how messages will look like after encoding and decoding invoke:
java -jar json2avro-validator.jar -s user.avcs -i user.json -m json2avro2json
json-avro-converter is published under Apache License 2.0.