Home

Awesome

Overview

Jackson jr is a compact alternative to full Jackson Databind component. It implements a subset of functionality, for example for cases where:

  1. Size of jar matters (jackson-jr-objects is bit over 100 kB)
  2. Startup time matters (jackson-jr has very low initialization overhead)

In addition to basic datatypes (core JDK types like Lists, Maps, wrapper types), package supports reading and writing of standard Java Beans (implementation that mimics standard JDK Bean Introspection): that is, subset of POJOs that define setters/getters and/or public fields. And starting with 2.11 there is even optional support for a subset of Jackson annotations via optional jackson-jr-annotation-support extension.

Jackson-jr also adds composer implementation that can be used to construct JSON output with builder-style API, but without necessarily having to build an in-memory representation: instead, it can directly use streaming-api for direct output. It is also possible to build actual in-memory JSON String or byte[] representation, if that is preferable.

Main Jackson-jr artifact (jackson-jr-objects) itself is currently about 120 kB in size, and only depends on Jackson Streaming API package. Combined size, for "all" jar, is bit over 500 kB (of which streaming API is about 350 kB), for use cases where a single jar is preferred over more modular approach. Finally, use of jar minimizers like ProGuard can bring the jar size down even further, by renaming and removing debug information.

License

Good old Apache License.

Packaging

Project is composed of multiple Maven sub-modules, each corresponding to a jar:

If you are not sure which package to use, the answer is usually jr-objects, and build system (maven, gradle) will fetch the dependency needed. jr-all jar is only used if the single-jar deployment (self-contained, no external dependencies) is needed.

Status

Build Status Maven Central Javadoc

Usage

Reading/writing Simple Objects, Beans, List/arrays thereof

Functionality of this package is contained in Java package com.fasterxml.jackson.jr.ob.

All functionality is accessed through main JSON Object; you can either used singleton JSON.std, or construct individual objects -- either way, JSON instances are ALWAYS immutable and hence thread-safe.

We can start by reading JSON

String INPUT = "{\"a\":[1,2,{\"b\":true},3],\"c\":3}";
Object ob = JSON.std.anyFrom(INPUT);
// or
Map<String,Object> map = JSON.std.mapFrom(INPUT);
// or
MyBean bean = JSON.std.beanFrom(MyBean.class, INPUT);

from any of the usual input sources (InputStream, Reader, String or byte[] that contains JSON, URL, JsonParser); and can write same Objects as JSON:

String json = JSON.std.asString(map);
JSON.std.write(ob, new File("/tmp/stuff.json");
// and with indentation; but skip writing of null properties
byte[] bytes = JSON.std
    .with(Feature.PRETTY_PRINT_OUTPUT)
    .without(Feature.WRITE_NULL_PROPERTIES)
    .asBytes(bean);

and may also read Lists and arrays of simple and Bean types:

List<MyType> beans = JSON.std.listOfFrom(MyType.class, INPUT);

(writing of Lists and arrays works without addition effort: just pass List/array as-is)

Reading "streaming JSON" (LD-JSON)

Version 2.10 added ability to read Streaming JSON content. See "Jackson 2.10 features" (section "Jackson-jr feature expansion") for full example, but basic reading is done using new ValueIterator abstraction:

File input = new File("json-stream.ldjson");
try (ValueIterator<Bean> it = JSON.std.beanSequenceFrom(Bean.class, input)) {
  while ((Bean bean = it.nextValue()) != null) {
    // do something with 'bean'
  }
}

Writing with composers

An alternative method exists for writing: "fluent" style output can be used as follows:

String json = JSON.std
  .with(JSON.Feature.PRETTY_PRINT_OUTPUT)
  .composeString()
  .startObject()
    .put("a", 1)
    .startArrayField("arr")
      .add(1).add(2).add(3)
    .end()
    .startObjectField("ob")
      .put("x", 3)
      .put("y", 4)
      .startArrayField("args").add("none").end()
    .end()
    .put("last", true)
  .end()
  .finish();

would produce (since pretty-printing is enabled)

{
  "a" : 1,
  "arr" : [1,2,3],
  "ob" : {
    "x" : 3,
    "y" : 4,
    "args" : ["none"]
  },
  "last" : true
}

Reading/writing JSON Trees

Jackson jr allows pluggable "tree models", and also provides one implementation, jr-stree. Usage for jr-stree is by configuring JSON with codec, and then using treeFrom and write methods like so:

JSON json = JSON.std.with(new JacksonJrsTreeCodec());
TreeNode root = json.treeFrom("{\"value\" : [1, 2, 3]}");
assertTrue(root.isObject());
TreeNode array = root.get("value");
assertTrue(array.isArray());
JrsNumber n = (JrsNumber) array.get(1);
assertEquals(2, n.getValue().intValue());

String json = json.asString(root);

Note that jr-stree implementation is a small minimalistic implementation with immutable nodes. It is most useful for simple reading use cases.

It is however possible to write your own TreeCodec implementations that integrate seamlessly, and in future other tree models may be offered as part of jackson-jr, or via other libraries.

Designing your Beans

To support readability and writability of your own types, your Java objects must either:

Note that although getters and setters need to be public (since JDK Bean Introspection does not find any other methods), constructors may have any access right, including private.

Starting with version 2.8, public fields may also be used (although their discovery may be disabled using JSON.Feature.USE_FIELDS) as an alternative: this is useful when limiting number of otherwise useless "getter" and "setter" methods.

NEW! Jackson-jr 2.11 introduce jackson-jr-annotation-support extension (see more below) which allows use of Jackson annotations like @JsonProperty, @JsonIgnore and even @JsonAutoDetect for even more granular control of inclusion, naming and renaming.

Customizing behavior with Features

There are many customizable features you can use with JSON object; see Full List of Features for details. But usage itself is via fluent methods like so:

String json = JSON.std
  .with(JSON.Feature.PRETTY_PRINT_OUTPUT)
  .without(JSON.Feature.FAIL_ON_DUPLICATE_MAP_KEYS)
  .asString(...);

Adding custom value readers, writers

Version 2.10 added ability to add custom ValueReaders and ValueWriters, to allow pluggable support for types beyond basic JDK types and Beans.

See section "Jackson-jr ValueReaders" of Jackson-jr 2.10 improvements for an explanation of how to add custom ValueReaders and ValueWriters

You can also check out unit test

jr-objects/src/test/java/com/fasterxml/jackson/jr/ob/impl/CustomValueReadersTest.java

for sample usage.

There is also a blog post Enable support for java.time.* with Jackson-jr which shows how to write custom readers/writers; in this case ones for Java 8 date/time types, but the concept is general.

Using (some of) Jackson annotations

Jackson 2.11 added a new extension (a JacksonJrExtension) -- jr-annotation-support -- that adds support for a subset of Jackson annotations. See jr-annotation-support/README.md for details of this extension, but basic usage is by registering extension:

import com.fasterxml.jackson.jr.annotationsupport.JacksonAnnotationExtension;

JSON json = JSON.builder()
    .register(JacksonAnnotationExtension.std)
    .build();

and then using JSON instance as usual.

Get it!

You can use Maven dependency like:

<dependency>
  <groupId>com.fasterxml.jackson.jr</groupId>
  <artifactId>jackson-jr-objects</artifactId>
  <version>2.16.1</version>
</dependency>

and then you can also download jars via Central Maven repository.

Or you can also clone the project and build it locally with mvn clean install.

Alternatively if you want a single jar deployment, you can use jackson-jr-all jar which embeds jackson-core (repackaged using Shade plug-in, so as not to conflict with "vanilla" jackson-core):

https://repo1.maven.org/maven2/com/fasterxml/jackson/jr/jackson-jr-all/

Performance

Initial performance testing using JVM Serializers benchmark suggests that it is almost as fast as full Jackson databind -- additional overhead for tests is 5-10% for both serialization and deserialization. So performance is practically identical.

In fact, when only handling Lists and Maps style content, speed jackson-jr speed fully matches jackson-databind performance (Bean/POJO case is where full databinding's extensive optimizations help more). So performance should be adequate, and choice should be more based on functionality, convenience and deployment factors.

About the only thing missing is that there is no equivalent to Afterburner, which can further speed up databind by 20-30%, for most performance-sensitive systems.