This project contains the general-purpose data-binding functionality and tree-model for Jackson Data Processor It builds on core streaming parser/generator package, and uses Jackson Annotations for configuration. More low-level details can be found from Javadocs.
While the original use case for Jackson was JSON data-binding, it can now be used for other data formats as well, as long as parser and generator implementations exist. Naming of classes uses word 'JSON' in many places even though there is no actual hard dependency to JSON format.
Project contains versions 2.0 and above: source code for earlier (1.x) versions is available from Codehaus SVN repository Main differences compared to 1.0 "mapper" jar are:
- Maven build instead of Ant
- Java package is now
com.fasterxml.jackson.databind(instead oforg.codehaus.jackson.map)
Functionality of this package is contained in Java package com.fasterxml.jackson.databind, and can be used using following Maven dependency:
<dependency>
<groupId>com.fasterxml.jackson.core</groupId>
<artifactId>jackson-databind</artifactId>
<version>2.0.6</version>
</dependency>
Since package also depends on '''jackson-core''' and '''jackson-databind''' packages, you will need to download these if not using Maven; and you may also want to add them as Maven dependency to ensure that compatible versions are used. If so, also add:
<dependency>
<groupId>com.fasterxml.jackson.core</groupId>
<artifactId>jackson-annotations</artifactId>
<version>2.0.6</version>
</dependency>
<dependency>
<groupId>com.fasterxml.jackson.core</groupId>
<artifactId>jackson-core</artifactId>
<version>2.0.6</version>
</dependency>
For non-Maven use cases, you download jars from Central Maven repository or Download page.
Databind jar is also a functional OSGi bundle, with proper import/export declarations, so it can be use on OSGi container as is.
While wiki contains more documentation, here are brief introductionary tutorials, in recommended order of reading.
The most common usage is to take piece of JSON, and construct a Plain Old Java Object ("POJO") out of it. So let's start there.
All data binding starts with a com.fasterxml.jackson.databind.ObjectMapper instance, so let's construct one:
ObjectMapper mapper = new ObjectMapper(); // create once, reuse
The default instance is fine for our use -- we will learn later on how to configure mapper instance if necessary. Usage is simple:
MyValue value = mapper.readValue(new File("data.json"), MyValue.class);
// or:
value = mapper.readValue(new URL("http://some.com/api/entry.json"), MyValue.class);
// or:
value = mapper.readValue("{\"name\":\"Bob\", \"age\":13}", MyValue.class);
And if we want to write JSON, we do the reverse:
mapper.writeValue(new File("result.json"), myResultObject);
// or:
byte[] jsonBytes = mapper.writeValueAsBytes(myResultObject);
// or:
String jsonString = mapper.writeValueAsString(myResultObject);
So far so good?
Beyond dealing with simple Bean-style POJOs, you can also handle JDK Lists, Maps:
Map<String, Integer> scoreByName = mapper.readValue(jsonSource, Map.class);
List<String> names = mapper.readValue(jsonSource, List.class);
// and can obviously write out as well
mapper.writeValue(new File("names.json"), names);
as long as JSON structure matches, and types are simple.
If you have POJO values, you need to indicate actual type (note: this is NOT needed for POJO properties with List etc types):
Map<String, ResultValue> results = mapper.readValue(jsonSource,
new TypeReference<String, ResultValue>() { } );
// why extra work? Java Type Erasure will prevent type detection otherwise
(note: no extra effort needed for serialization, regardless of generic types)
But wait! There is more!
While dealing with Maps, Lists and other "simple" Object types (Strings, Numbers, Booleans) can be simple, Object traversal can be cumbersome.
This is where Jackson's Tree model can come in handy:
// can be read as generic JsonNode, if it can be Object or Array; or,
// if known to be Object, as ObjectNode, if array, ArrayNode etc:
ObjectNode root = mapper.readTree("stuff.json");
String name = root.get("name").asText();
int age = root.get("age").asInt();
// can modify as well: this adds child Object as property 'other', set property 'type'
root.with("other").put("type", "student");
String json = mapper.writeValueAsString(root);
// with above, we end up with something like as 'json' String:
// {
// "name" : "Bob", "age" : 13,
// "other" : {
// "type" : "student"
// {
// }
Tree Model can be more convenient than data-binding, especially in cases where structure is highly dynamic, or does not map nicely to Java classes.
As convenient as data-binding (to/from POJOs) can be; and as flexible as Tree model can be, there is one more canonical processing model available: incremental (aka "streaming") model. It is the underlying processing model that data-binding and Tree Model both build upon, but it is also exposed to users who want ultimate performance and/or control over parsing or generation details.
For in-depth explanation, look at Jackson Core component. But let's look at a simple teaser to whet your appetite:
(TO BE COMPLETED)
There are two entry-level configuration mechanisms you are likely to use: Features and Annotations.
Here are examples of configuration features that you are most likely to need to know about.
Let's start with higher-level data-binding configuration.
// SerializationFeature for changing how JSON is written
// to enable standard indentation ("pretty-printing"):
mapper.enable(SerializationFeature.INDENT_OUTPUT);
// to allow serialization of "empty" POJOs (no properties to serialize)
// (without this setting, an exception is thrown in those cases)
mapper.disable(SerializationFeature.FAIL_ON_EMPTY_BEANS);
// to write java.util.Date, Calendar as number (timestamp):
mapper.disable(SerializationFeature.WRITE_DATES_AS_TIMESTAMPS);
// DeserializationFeature for changing how JSON is read as POJOs:
// to prevent exception when encountering unknown property:
mapper.disable(DeserializationFeature.FAIL_ON_UNKNOWN_PROPERTIES);
// to allow coercion of JSON empty String ("") to null Object value:
mapper.enable(DeserializationFeature.ACCEPT_EMPTY_STRING_AS_NULL_OBJECT);
In addition, you may need to change some of low-level JSON parsing, generation details:
// JsonParser.Feature for configuring parsing settings:
// to allow C/C++ style comments in JSON (non-standard, disabled by default)
mapper.enable(JsonParser.Feature.ALLOW_COMMENTS);
// to allow (non-standard) unquoted field names in JSON:
mapper.enable(JsonParser.Feature.ALLOW_UNQUOTED_FIELD_NAMES);
// to allow use of apostrophes (single quotes), non standard
mapper.enable(JsonParser.Feature.ALLOW_SINGLE_QUOTES);
// JsonGenerator.Feature for configuring low-level JSON generation:
// to force escaping of non-ASCII characters:
mapper.enable(JsonGenerator.Feature.ESCAPE_NON_ASCII);
Full set of features are explained on Jackson Features page.
The simplest annotation-based approach is to use @JsonProperty annotation like so:
public class MyBean {
private String _name;
// without annotation, we'd get "theName", but we want "name":
@JsonProperty("name")
public String getTheName() { return _name; }
// note: it is enough to add annotation on just getter OR setter;
// so we can omit it here
public void setTheName(String n) { _name = n; }
}
There are other mechanisms to use for systematic naming changes: see Custom Naming Convention for details.
Note, too, that you can use Mix-in Annotations to associate all annotations.
There are two main annotations that can be used to to ignore properties: @JsonIgnore for individual properties; and @JsonIgnoreProperties for per-class definition
// means that if we see "foo" or "bar" in JSON, they will be quietly skipped
// regardless of whether POJO has such properties
@JsonIgnoreProperties({ "foo", "bar" })
public class MyBean
{
// will not be written as JSON; nor assigned from JSON:
@JsonIgnore
public String internal;
// no annotation, public field is read/written normally
public String external;
@JsonIgnore
public void setCode(int c) { _code = c; }
// note: will also be ignored because setter has annotation!
public int getCode() { return _code; }
}
As with renaming, note that annotations are "shared" between matching fields, getters and setters: if only one has @JsonIgnore, it affects others.
But it is also possible to use "split" annotations, to for example:
public class ReadButDontWriteProps {
private String _name;
@JsonProperty public void setName(String n) { _name = n; }
@JsonIgnore public String getName() { return _name; }
}
in this case, no "name" property would be written out (since 'getter' is ignored); but if "name" property was found from JSON, it would be assigned to POJO property!
For a more complete explanation of all possible ways of ignoring properties when writing out JSON, check "Filtering properties" article.
Unlike many other data-binding packages, Jackson does not require you to define "default constructor" (constructor that does not take arguments). While it will use one if nothing else is available, you can easily define that an argument-taking constructor is used:
public class CtorBean
{
public final String name;
public final int age;
@JsonCreator // constructor can be public, private, whatever
private CtorBean(@JsonProperty("name") String name,
@JsonProperty("age") int age)
{
this.name = name;
this.age = age;
}
}
Constructors are especially useful in supporting use of Immutable objects.
Alternatively, you can also define "factory methods":
public class FactoryBean
{
// fields etc omitted for brewity
@JsonCreator
public static FactoryBean create(@JsonProperty("name") String name) {
// construct and return an instance
}
}
Note that use of a "creator method" does not preclude use of setters: you can mix and match properties from constructor/factory method with ones that are set via setters or directly using fields.
One useful (but not very widely known) feature of Jackson is its ability to do arbitrary POJO-to-POJO conversions. Conceptually you can think of conversions as sequence of 2 steps: first, writing a POJO as JSON, and second, binding that JSON into another kind of POJO. Implementation just skips actual generation of JSON, and uses more efficient intermediate representation.
Conversations work between any compatible types, and invocation is as simple as:
ResultType result = mapper.convertValue(sourceObject, ResultType.class);
and as long as source and result types are compatible -- that is, if to-JSON, from-JSON sequence would succeed -- things will "just work". But here are couple of potentially useful use cases:
// Convert from int[] to List<Integer>
List<Integer> sourceList = ...;
int[] ints = mapper.convertValue(sourceList, int[].class);
// Convert a POJO into Map!
Map<String,Object> propertyMap = mapper.convertValue(pojoValue, Map.class);
// ... and back
PojoType pojo = mapper.convertValue(propertyMap, PojoType.class);
// decode Base64! (default byte[] representation is base64-encoded String)
String base64 = "TWFuIGlzIGRpc3Rpbmd1aXNoZWQsIG5vdCBvbmx5IGJ5IGhpcyByZWFzb24sIGJ1dCBieSB0aGlz";
byte[] binary = mapper.convertValue(base64, byte[].class);
Basically, Jackson can work as a replacement for many Apache Commons components, for tasks like base64 encoding/decoding, and handling of "dyna beans" (Maps to/from POJOs).
Project-specific documentation:
Related:
- Core annotations package defines annotations commonly used for configuring databinding details
- Core parser/generator package defines low-level incremental/streaming parsers, generators
- Jackson Project Home has additional documentation (although much of it for Jackson 1.x)