Introduction

This is a brief overview of the digester-rules-in-Java5 Annotations feature. Inspired by the basic idea behind the JPA, BeanValidation and JAXB's specifications, this feature lets you define Digester rules directly in target POJOs, instead of creating and initializing the Rules objects programmatically, which can become tedious.

Annotation Rules

A digester rule on a POJO is expressed through one or more annotations. An annotation is considered a digester rule definition if its retention policy contains RUNTIME and if the annotation itself is annotated with org.apache.commons.digester.annotations.DigesterRule.

The DigesterRule is defined by the combination of:

• the reflected Class<? extends org.apache.commons.digester.Rule> by the annotation;
• the org.apache.commons.digester.annotations.DigesterLoaderHandler class that has to be invoked during the target class traversal (if not specifyied, the annotation processor will supply the org.apache.commons.digester.annotations.handlers.DefaultLoaderHandler);
• the org.apache.commons.digester.annotations.AnnotationRuleProvider provider that produces the pattern, rule pair.

Digester annotations can target any of the following ElementTypes:

• FIELD for Digester rules concerning attributes;
• METHOD for Digester rules concerning methods calls;
• PARAMETER for Digester rules concerning methods parameters setting;
• TYPE for Digester rules concerning types creation.

While other ElementTypes are not forbidden, the Digester annotations processor does not have to recognize and process annotation rules placed on such types.

Every Digester rule annotation must define a pattern element of type String that represents the element matching path pattern.

@Documented
@Retention(RetentionPolicy.RUNTIME)
@Target(ElementType.TYPE)
@CreationRule
@DigesterRule(
        reflectsRule = ObjectCreateRule.class,
        providedBy = ObjectCreateRuleProvider.class
)
public @interface ObjectCreate {

    String pattern();

}

Applying multiple annotation rule of the same type

It is often useful to declare the same annotation rule more than once to the same target, with different properties. To support this requirement, the Digester annotation processor treats annotations annotated by @org.apache.commons.digester.annotations.DigesterRuleList whose value element has a return type of an array of rule annotations in a special way. Each element in the value array are processed by the Digester annotation processor as regular annotation rule annotations. This means that each Digester rule specified in the value element is applied to the target. The annotation must have retention RUNTIME and can be applied on a type, field, method or method parameter. It is recommended to use the same set of targets as the initial Digester annotation rule.

Note to designers: each Digester annotation rule should be coupled with its corresponding multi-valued annotation. It is recommended, though not mandated, the definition of an inner annotation named List.

@Documented
@Retention(RetentionPolicy.RUNTIME)
@Target(ElementType.TYPE)
@CreationRule
@DigesterRule(
    reflectsRule = ObjectCreateRule.class,
    providedBy = ObjectCreateRuleProvider.class
)
public @interface ObjectCreate {

    String pattern();

    @Documented
    @Retention(RetentionPolicy.RUNTIME)
    @Target(ElementType.TYPE)
    @DigesterRuleList
    @interface List {
        ObjectCreate[] value();
    }

}

Rule provider implementation

A Digester rule provider implementation performs the rule creation of a given annotation for a given annotated element. The implementation classes are specified by the providedBy element of the @DigesterRule annotation that decorates the rule annotation definition. The rule provider implementation implements the org.apache.commons.digester.annotations.AnnotationRuleProvider<A extends Annotation, E extends AnnotatedElement, R extends Rule> interface.

class ObjectCreateRuleProvider implements AnnotationRuleProvider<ObjectCreate, Class<?>, ObjectCreateRule> {

    private Class element) {
        this.clazz = element;
    }

    public ObjectCreateRule get() {
        return new ObjectCreateRule(this.clazz);
    }

}

Notes

A new instance of the provider will be created each time the Digester annotations processor will meet the relative rule that requests it.

To supply the missing AnnotatedElement for methods PARAMETERs, the Digester annotation processor come with the org.apache.commons.digester.annotations.reflect.MethodArgument class.

Digester loader handler implementation

The Digester loader handler is an AnnotatedElement interceptor invoked when meeting a particular Digester rule annotation while analyzing the target class.

By default, the Digester annotations processor, when meeting a Digester annotation rule, extracts the rule pattern and the relative rule provider to store it in the org.apache.commons.digester.annotations.FromAnnotationsRuleSet, an org.apache.commons.digester.RuleSet implementation.

If designers have the need of a more elaborate annottaion processing, they can specify the handledBy element of the @DigesterRule annotation that decorates the rule annotation definition. The Digester loader handler implementation implements the DigesterLoaderHandler<A extends Annotation, E extends AnnotatedElement> interface. Follows below an example:

class SetPropertiesLoaderHandler implements DigesterLoaderHandler<SetProperty, Field> {

    public void handle(SetProperty annotation, Field element, FromAnnotationsRuleSet ruleSet) {
        SetPropertiesRuleProvider ruleProvider = ruleSet.getProvider(annotation.pattern(), SetPropertiesRuleProvider.class);

        if (ruleProvider == null) {
            ruleProvider = new SetPropertiesRuleProvider();
            ruleSet.addRuleProvider(annotation.pattern(), ruleProvider);
        }

        ruleProvider.addAlias(annotation, element);
    }

}

Built-in Rules

All built-in annotation rules are in the package org.apache.commons.digester.annotations.rules. Here is the list of annotations and their usage.

TYPE annotations

FIELD annotations

METHOD annotations

PARAMETER annotations

Bootstrapping

The core of Digester annotations rules processor is the org.apache.commons.digester.annotations.DigesterLoader class.

A org.apache.commons.digester.annotations.DigesterLoader instance is able to analyze Class<?> graphs and builds the relative org.apache.commons.digester.RuleSet to create org.apache.commons.digester.Digester instances.

The bootstrap sequence has been designed to be as simple as possible, all that's needed is creating a new org.apache.commons.digester.annotations.DigesterLoaderBuilder instance, plugging the desired org.apache.commons.digester.annotations.spi.AnnotationRuleProviderFactory and org.apache.commons.digester.annotations.spi.DigesterLoaderHandlerFactory. using a chaining builders pattern.

An org.apache.commons.digester.annotations.spi.AnnotationRuleProviderFactory implementation performs the creation of org.apache.commons.digester.annotations.AnnotationRuleProvider<A extends Annotation, E extends AnnotatedElement, R extends Rule> instances; the default implementation is limited to create the provider by invoking the default empty constructor of the required class, but users are free to give their implementation if they need a more complex factory, i.e. providers requires components that could be injected from a context, etc. etc.

Note

It is strongly descouraged caching AnnotationRuleProvider instances!!!

Same thing for the org.apache.commons.digester.annotations.spi.DigesterLoaderHandlerFactory, which implementation performs the creation of DigesterLoaderHandler<A extends Annotation, E extends AnnotatedElement> instances; the default implementation is limited to create the handler by invoking the default empty constructor of the required class, but users are free to give their implementation if they need a more complex factory, i.e. providers requires components that could be injected from a context, etc. etc.

Said that, to obtain a fresh new org.apache.commons.digester.annotations.DigesterLoader instance with default factories, it is enough invoking the default empty constructor:

DigesterLoader digesterLoader = new DigesterLoaderBuilder()
                                    .useDefaultAnnotationRuleProviderFactory()
                                    .useDefaultDigesterLoaderHandlerFactory();

Otherwise, if users need specify theyr custom factories:

DigesterLoader digesterLoader = new DigesterLoaderBuilder()
                                    .useAnnotationRuleProviderFactory(new MyAnnotationRuleProviderFactory())
                                    .useDigesterLoaderHandlerFactory(new MyDigesterLoaderHandlerFactory());

Example: a simple RSS parser

Let's assume there is the need to parse the following (simplified) XML/RSS feed:

<rss version="2.0">
    <channel>

        <title>Apache</title>
        <link>http://www.apache.org</link>
        <description>The Apache Software Foundation</description>
        <language>en-US</language>
        <rating>(PICS-1.1 "http://www.rsac.org/ratingsv01.html"
            2 gen true comment "RSACi North America Server"
            for "http://www.rsac.org" on "1996.04.16T08:15-0500"
            r (n 0 s 0 v 0 l 0))</rating>

        <image>
            <title>Apache</title>
            <url>http://jakarta.apache.org/images/jakarta-logo.gif</url>
            <link>http://jakarta.apache.org</link>
            <width>505</width>
            <height>480</height>
            <description>The Jakarta project. Open source, serverside java.</description>
        </image>

        <item>
            <title>Commons Attributes 2.1 Released</title>
            <link>http://jakarta.apache.org/site/news/news-2004-2ndHalf.html#20040815.1</link>
            <description>The Apache Commons team is happy to announce the release of Commons Attributes 2.1.
            This is the first release of the new Commons-Attributes code.</description>
        </item>

        <item>
            <title>Cloudscape Becomes Apache Derby</title>
            <link>http://jakarta.apache.org/site/news/elsewhere-2004-2ndHalf.html#20040803.1</link>
            <description>IBM has submitted a proposal to the Apache DB project for a Java-based package to be called 'Derby'.</description>
        </item>

        <item>
            <title>Commons BeanUtils 1.7 Released</title>
            <link>http://jakarta.apache.org/site/news/news-2004-2ndHalf.html#20040802.1</link>
            <description/>
        </item>

        <item>
            <title>Commons JXPath 1.2 Released</title>
            <link>http://jakarta.apache.org/site/news/news-2004-2ndHalf.html#20040801.2</link>
            <description/>
        </item>
    </channel>
</rss>

So, let's define the Java entities and annotate them; first the Channel entity:

@ObjectCreate(pattern = "rss/channel")
class Channel {

    private final List<Item> items = new ArrayList<Item>();

    @BeanPropertySetter(pattern = "rss/channel/title")
    private String title;

    @BeanPropertySetter(pattern = "rss/channel/link")
    private String link;

    @BeanPropertySetter(pattern = "rss/channel/description")
    private String description;

    @BeanPropertySetter(pattern = "rss/channel/language")
    private String language;

    private Image image;

    // getters and setters

    @SetNext
    public void setImage(Image image) {
        this.image = image;
    }

    @SetNext
    public void addItem(Item item) {
        this.items.add(item);
    }

}

Then the Image entity:

@ObjectCreate(pattern = "rss/channel/image")
class Image {

    @BeanPropertySetter(pattern = "rss/channel/image/description")
    private String description;

    @BeanPropertySetter(pattern = "rss/channel/image/width")
    private int width;

    @BeanPropertySetter(pattern = "rss/channel/image/height")
    private int height;

    @BeanPropertySetter(pattern = "rss/channel/image/link")
    private String link;

    @BeanPropertySetter(pattern = "rss/channel/image/title")
    private String title;

    @BeanPropertySetter(pattern = "rss/channel/image/url")
    private String url;

    // getters and setters

}

and finally the Item entity:

@ObjectCreate(pattern = "rss/channel/item")
class Item {

    @BeanPropertySetter(pattern = "rss/channel/item/description")
    private String description;

    @BeanPropertySetter(pattern = "rss/channel/item/link")
    private String link;

    @BeanPropertySetter(pattern = "rss/channel/item/title")
    private String title;

    // getters and setters

}

It is now possible to create the Digester instance and parse the XML:

DigesterLoader digesterLoader = new DigesterLoaderBuilder()
                                    .useDefaultAnnotationRuleProviderFactory()
                                    .useDefaultDigesterLoaderHandlerFactory();
...
Digester digester = digesterLoader.createDigester(Channel.class);
try {
    Channel channel = (Channel) digester.parse(new URL("http://www.myfeedprovider.com/rss.xml").openStream());
} catch (Exception e) {
    // do something
}
    

Notes

If asking to the DigesterLoader instance more then twice the Digester for the same Class<?>, the DigesterLoader won't analize the target class for each request, but rather will reuse cached results.

The same DigesterLoader instance can be reused to create other Digester instances.