Apache Avro™ is a data serialization system.
Learn more about Avro, please visit our website at:
- 1.10.0.5p - latest version, based on jackson 2.9+, aims to be api compatible with avro 1.10.X
- 1.8.1.50p - latest version, based on jackson 1.9
This fork, contains numerous fixes, and features not implement (but reported) in the original project. This fork focuses only on the java implementation. For non-java part please see official fork or other forks.
Starting with version 1.10.0.4p the additional logical types: any, big-integer, duration, instant, json*, regexp, schema temporal url, uri have moved to a separate library, which you will need to additionally include:
<dependency>
<groupId>org.spf4j</groupId>
<artifactId>avro-logical-types-fork</artifactId>
<version>1.1</version>
<scope>test</scope>
</dependency>you can try using avro-logical-types-official for using those logical types with the official avro library.
Join the discussion on Gitter:
This fork is published to github packages, and you can use it by adding the repositories to your pom.xml (see for more info):
<repositories>
<repository>
<id>github</id>
<url>https://maven.pkg.github.com/zolyfarkas/*</url>
<snapshots>
<enabled>true</enabled>
</snapshots>
</repository>
</repositories>
<pluginRepositories>
<pluginRepository>
<id>github</id>
<url>https://maven.pkg.github.com/zolyfarkas/*</url>
<snapshots>
<enabled>true</enabled>
</snapshots>
</pluginRepository>
</pluginRepositories>
and reference the avro lib as:
<dependency>
<groupId>org.apache.avro</groupId>
<artifactId>avro</artifactId>
<version>1.10.0.1p</version>
<type>pom</type>
</dependency>
previous versions have been published to bintray, and should be available there (if I interpret the deco text correctly) until Feb 1st, 2022.
1) ExtendedJsonDecoder/Encoder - encodes unions of null and a single type as a more normal key=value rather than key={type=value}.
ExtendedJsonDecoder will infer missing fields using the default values from the write'r schema and as such need not be present in the serialized json. The Extended[Generic|Specific]DatumWriter can be used to omit serializing fields that equal the default values defined in the schema. This can make the json payload smaller than the binary payload in some use cases.
for
union {null, string} field = null;
instead of:
"field" : {"string" : "value"}
the ExtendedJsonEncoder will serialize:
"field" : "value"
2) Evolvable enums, with custom string values, where additionally you can rename enum values to maintain backward compatibility. Example:
@symbolAliases({"DIAMONDS" : ["Bling Bling", "ROCKS"]}) // you can correct bad names backward compatibly.
@stringSymbols({"SPADES" : "S P A D E S"}) //you can have stringValues (toString()) that are not restricted to identifiers
enum Suit {
UNKNOWN, SPADES, DIAMONDS, CLUBS, HEARTS
} = "UNKNOWN";
3) Generated java classes are annotated with @Nullable or @Nonnull as appropriate. As such compiling in conjuction with tools like spotbugs or google error-prone will result in better quality code.
4) @Immutable record support. (setters will not be generated). This encourages use of builders and constructors to build the records.
6) Enhanced decimal logical type support, (custom binary encoding, or official controllable via @format("official") or feature flag) where JSON serialization is a proper number. (not bytes/string)
for
@logicalType("decimal") bytes field = "";
instead of:
"field" : "bytesvalue"
the ExtendedJsonEncoder will serialize:
"field" : 123.45
for
@logicalType("json_record") string field = "{}";
Avro JsonEncoder will serialize:
"field" : "{ \"field\" : \"value \"}"
the ExtendedJsonEncoder will serialize:
"field" : {"field" : "value"}
@logicalType("any")
record AnyType {
string avsc;
bytes content;
}
using this type like:
...
AnyType field;
...
will serialize in json to:
"field" : {
"avsc" : ... { the avro schema } ...
"content" : ... { the json image } ...
}