XML Deserializer (XD)

Overview

XML Deserializer (XD) is a high-level, Kotlin JVM library for easily and efficiently converting an XML document into a sequence of objects.

The library uses a simple DSL to map element names (with or without reference to a namespace) to handler lambdas, using a context object that the lambdas populate, without exposing the low-level XML processing details.

The library is fast and memory efficient because it uses the Streaming API for XML API (StAX) for processing the XML.

Motivation

There are a number of different ways of processing XML documents in Java. This article has a good overview of the options: SAX, DOM, StAX and JAXB.

While each of them has their use, none of them provides a high-level API for processing an XML document. Instead, the user of the library must use and navigate various XML-specific data structures and configurations.

The goal of the XD library is to provide a high-level API that does not require knowledge of XML processing. The XML processing is hidden by the library and the use of the library is wholly programmatic, without any annotations or XML configuration. In addition, the processing does not require an XSD and can proceed piecemeal, only mapping the elements that are of interest.

Usage

The Maven coordinates for XD are:

<dependency>
    <groupId>dev.bombinating</groupId>
    <artifactId>xml-deserializer</artifactId>
    <version>1.1.0</version>
</dependency>

The Gradle coordinates are:

dependencies {
    implementation(group = "dev.bombinating", name = "xml-deserialzier", version = "1.1.0")
}

Details

The entry-point for the library is a Reader extension method called parse.

The parse method has one required argument, handlers, which is of type HandlerRegistrations.

An object of type HandlerRegistrations can be created using the top-level method handlers.

Within the lambda parameter of the method, handlers are defined using a String extension method that takes a function literal with receiver.

The receiver of the lambda is of type HandlerContext.

The HandlerContext provides the following functionality:

Ability to access the current XML element (element) via the XmlElement interface.
The context object being populated (obj) (the generic type of the HandlerContext)
The ability to create a new XmlContextParser object with a different HandlerContext for handling child elements (createParser<P>)

The XmlElement interface provides access to:

The text of the XML element (text)
The attribute values on the XML element (element["attr_name"])

The XmlContextParser interface specifies a parse method that returns a populated object, based on a HandlerRegistrations object.

Simple Example

In this example, the XML document contains a list of names.

<People>
    <Person>
        <FirstName>John</FirstName>
        <LastName>Smith</LastName>
    </Person>
    <Person>
        <FirstName>Mary</FirstName>
        <LastName>Jane</LastName>
    </Person>
</People>

The data is stored in a Person class. The properties of the object are populated as they are encountered in the XML document. Therefore, it is convenient (though not required) if there is a no-arg constructor, and the properties are mutable (var) and have default values. Note that on the JVM, if all of the constructor arguments have default values, the compiler will generate a no-arg constructor.

data class Person(
    var firstName: String? = null,
    var lastName: String? = null
)

The next step is to define how to map the XML elements to object properties using the handlers method. The generic type (<Person>) indicates that the context object (obj) the handlers are populating is of type Person. For both handlers, the property is simply set to the text of the element.

val handlers = handlers<Person> {
    "FirstName" { obj.firstName = text }
    "LastName" { obj.lastName = text }
}

Since the handlers are related to the Person class, it’s convenient to move the definition to the companion object of the class.

data class Person(var firstName: String? = null, var lastName: Int? = null) {
    companion object {
        val handlers = handlers<Person> {
            "FirstName" { obj.firstName = text }
            "LastName" { obj.lastName = text }
        }
    }
}

For this example, the XML is stored in a String and then converted into a Reader.

val reader = """
    |<People>
    |    <Person>
    |        <FirstName>John</FirstName>
    |        <LastName>Smith</LastName>
    |    </Person>
    |    <Person>
    |        <FirstName>Mary</FirstName>
    |        <LastName>Jane</LastName>
    |    </Person>
    |</People>
""".trimMargin().reader()

The final piece is to invoke the parse() method on the reader.

The parse method takes three parameters:

Top-level handlers
XML handler configuration
Lambda for creating objects of the type the parser emits and the handlers populate

For defining the top-level handlers, the handlers method is used. The parser emits Person objects so the generic type of the method is Person. The name of the XML element is Person and the use method tells the parser to use the handlers defined in the Person companion object.

val topLevelHandlers = handlers<Person> {
    "Person" { use(Person.handlers) }
}

The use method takes an optional lambda initializer for configuring the obj object before it is populated.

val topLevelHandlers = handlers<Person> {
    "Person" { use(Person.handlers) { firstName = "NFN" } }
}

The second parameter to the parse method is an optional object of type XmlContextParserConfig. This controls what happens when no handler is found for an XML element and what happens when a handler throws an exception. If not specified, the default is to do nothing when an XML element does not have a handler and to throw an exception of type ProcessingException when a handler throws an exception.

val config = XmlContextParserConfig(
    processingExceptionHandler = {
        println("Error parsing element '${element.name.name}': $cause")
    }
)

The third parameter to the parse method is an optional lambda for creating objects of the type the parser emits. If omitted, the parser will attempt to create objects using the no-arg constructor for the class.

val factory = { Person(lastName = "Not Specified") }

Putting all of this together, the parser call looks like this:

val config = XmlContextParserConfig(
    processingExceptionHandler = {
        logger.error { "Error parsing element '${element.name.name}': $cause" }
    }
)
val factory = { Person(lastName = "Not Specified") }
val people = reader.parse(handlers<Person> {
    "Person" { use(Person.handlers) }
}, config, factory)

Since there is a no-arg constructor for the Person class and we don’t need any error handling, the factory lambda and configuration can be omitted and the code can be simplified.

val people = reader.parse(handlers<Person> {
    "Person" { use(Person.handlers) }
})

Nested Example

Let’s extend the previous example to include an address for a person.

The XML looks like this:

<People>
    <Person>
        <FirstName>John</FirstName>
        <LastName>Smith</LastName>
        <Address>
            <Street>666 Park Ave</Street>
            <City>New York</City>
        </Address>
    </Person>
</People>

We’ll add a data class to store the address info. Note that because all of the constructor arguments are mutable and have default values, the compiler will generate a no-arg constructor for the class on the JVM.

data class Address(
    var street: String? = null,
    var city: String? = null
)

Next we’ll modify the Person class to reference the Address class.

data class Person(
    var firstName: String? = null,
    var lastName: String? = null,
    var address: Address? = null
)

The Address property handlers are easy to define since they just populated from the element text.

val handlers = handlers<Address> {
    "Street" { obj.street = text }
    "City" { obj.city = text }
}

As we did with the Person handlers, we’ll define the Address handlers in the companion object for the Address class.

data class Address(
    var street: String? = null,
    var city: String? = null
) {
    companion object {
        val handlers = handlers<Address> {
            "Street" { obj.street = text }
            "City" { obj.city = text }
        }
    }
}

Finally, we’ll define the handler for the Address XML element in the Person handlers. To do this, we use the parse method. The first parameter to the parse method is a HandlerRegistrations object. In this case, the one we created using the handlers method in the companion object of the Address class.

val handlers = handlers<Person> {
    // ...
    "Address" { obj.address = parse(Address.handlers) }
}

Putting it all together, it looks like this. Note that no changes are needed to the parse call.

data class Person(
    var firstName: String? = null,
    var lastName: String? = null,
    var address: Address? = null
) {
    companion object {
        val handlers = handlers<Person> {
            "FirstName" { obj.firstName = text }
            "LastName" { obj.lastName = text }
            "Address" { obj.address = parse(Address.handlers) }
        }
    }
}

Repeated Nested Example with Attributes

Let’s change the XML to allow a person to have multiple addresses. To distinguish the addresses, we’ll add a desc XML attribute to the Address element that describes the address.

<People>
    <Person>
        <FirstName>John</FirstName>
        <LastName>Smith</LastName>
        <Address desc="summer">
            <Street>666 Park Ave</Street>
            <City>New York</City>
        </Address>
        <Address desc="winter">
            <Street>6834 Hollywood Blvd</Street>
            <City>Los Angeles</City>
        </Address>
    </Person>
</People>

First, we’ll add the desc property to the Address class.

data class Address(
    var street: String? = null,
    var city: String? = null,
    var desc: String? = null
    // ...
}

We’ll modify the Person class to accommodate multiple addresses. Note that the addresses property is neither mutable nor nullable but that it does have a default value so the compiler will still generate a no-arg constructor for the class. Also, while the addresses property cannot be set, Address objects can be added to the list.

 class Person(
    var firstName: String? = null,
    var lastName: String? = null,
    val addresses: MutableList<Address> = mutableListOf()
)

We also need to change what happens when an <Address> element is encountered. Rather than setting the address property, we add the Address object to the addresses (mutable) list.

val handlers = handlers<Person> {
    // ...
    "Address" { obj.addresses.add(parse(Address.handlers)) }
}

The XML attribute can be dealt with using the optional second parameter to the parse method. The second parameter specifies a lambda for creating objects of the given type for the parser handlers to populate. The lambda has access to the XML element, which can be used to read the XML attributes.

val handlers = handlers<Person> {
    // ...
    "Address" { obj.addresses.add(
        parse(Address.handlers) { Address(desc = element["desc"]) })
    }
}

The full handlers mapping for the Person class is as follows.

data class Person(
    var firstName: String? = null,
    var lastName: String? = null,
    val addresses: MutableList<Address> = mutableListOf()
) {
    companion object {
        val handlers = handlers<Person> {
            "FirstName" { obj.firstName = text }
            "LastName" { obj.lastName = text }
            "Address" {
                obj.addresses.add(parse(Address.handlers) { Address(desc = element["desc"]) })
            }
        }
    }
}

As before, no changes are need to the parse call itself.

Non-String Properties

Because all of the handling is done in code, it is straightforward to convert an XML attribute or element text to a non-string value.

For this example, we’ll add a person’s age and an address type.

The XML looks like this:

<People>
    <Person>
        <FirstName>John</FirstName>
        <LastName>Smith</LastName>
        <Age>55</Age>
        <Address desc="summer" type="apartment">
            <Street>666 Park Ave</Street>
            <City>New York</City>
        </Address>
        <Address desc="winter" type="house">
            <Street>6834 Hollywood Blvd</Street>
            <City>Los Angeles</City>
        </Address>
    </Person>
</People>

We’ll create an enum for the <Address> type attribute:

enum class AddressType {
    Apartment,
    House
}

Let’s add a companion object method to convert from the XML attribute value to an AddressType enum.

enum class AddressType(val xmlValue: String) {
    Apartment("apartment"),
    House("house");
    companion object {
        private val map: Map<String, AddressType> = values().map { it.xmlValue to it }.toMap()
        operator fun get(xmlValue: String): AddressType? = map[xmlValue]
    }
}

And we’ll add a type property to the Address class:

data class Address(
    var street: String? = null,
    var city: String? = null,
    var desc: String? = null,
    var type: AddressType? = null
    // ...
}

Since the <Address> type is an attribute (as opposed to an element), we’ll handle it when the Address object is created in the Person mapping.

val handlers = handlers<Person> {
    // ...
    "Address" {
        obj.addresses.add(parse(Address.handlers) {
            Address(desc = element["desc"], type = AddressType[element["type"]])
        })
    }
}

For the person’s age, we’ll add an age property to the Person class.

data class Person(
    var firstName: String? = null,
    var lastName: String? = null,
    val addresses: MutableList<Address> = mutableListOf(),
    var age: Int? = null
) // ...

The last step is to specify how this property will be populated.

val handlers = handlers<Person> {
    "Age" { obj.age = text.toIntOrNull() }
    // ...
}

Here’s the final code. Note that the parse call hasn’t changed since the first example.

enum class AddressType(val xmlValue: String) {
    Apartment("apartment"),
    House("house");
    companion object {
        private val map: Map<String, AddressType> = values().map { it.xmlValue to it }.toMap()
        operator fun get(xmlValue: String): AddressType? = map[xmlValue]
    }
}

data class Address(
    var street: String? = null,
    var city: String? = null,
    var desc: String? = null,
    var type: AddressType? = null
) {
    companion object {
        val handlers = handlers<Person> {
            "Street" { obj.street = text }
            "City" { obj.city = text }
        }
    }
}

data class Person(
    var firstName: String? = null,
    var lastName: String? = null,
    val addresses: MutableList<Address> = mutableListOf(),
    var age: Int? = null
) {
    companion object {
        val handlers = handlers<Person> {
            "FirstName" { obj.firstName = text }
            "LastName" { obj.lastName = text }
            "Age" { obj.age = text.toIntOrNull() }
            "Address" {
                obj.addresses.add(parse(Address.handlers) {
                    Address(desc = element["desc"], type = AddressType[element["type"]])
                })
            }
        }
    }
}

val people = reader.parse(handlers<Person> {
    "Person" { use(Person.handlers) }
})

XML Namespace Example

The XD library can also be configured to take namespaces into account when mapping to handlers.

Let’s go back to the XML from the simple example and add a namespace.

val reader = """
    |<ns1:People xmlns:ns1="http://example.com">
    |    <ns1:Person>
    |        <ns1:FirstName>John</ns1:FirstName>
    |        <ns1:LastName>Smith</ns1:LastName>
    |    </ns1:Person>
    |    <ns1:Person>
    |        <ns1:FirstName>Mary</ns1:FirstName>
    |        <ns1:LastName>Jane</ns1:LastName>
    |    </ns1:Person>
    |</ns1:People>
""".trimMargin().reader()

By default, the namespace is not taken consideration when registering or looking up handlers so the original handler code works without any changes.

data class Person(var firstName: String? = null, var lastName: Int? = null) {
    companion object {
        val handlers = handlers<Person> {
            "FirstName" { obj.firstName = text }
            "LastName" { obj.lastName = text }
        }
    }
}

To take namespaces into consideration, two changes need to be made:

Change the handlers() method call to be namespace aware
Change how the element names are declared in the mapping to include the namespace info

data class Person(var firstName: String? = null, var lastName: Int? = null) {
    companion object {
        val ns = "http://example.com"
        val handlers = handlers<Person>(namespaceAware = true) { // (1)
            ns["FirstName"] { obj.firstName = text } // (2)
            ns["LastName"] { obj.lastName = text }
        }
    }
}

Indicate that the handler mapping includes any namespace
Include the namespace in the mapping

You can have as many namespaces as you want — or ignore simply ignore them in simpler cases.

Change Log

CHANGELOG.adoc

Name		Name	Last commit message	Last commit date
Latest commit History 35 Commits
.github/workflows		.github/workflows
buildSrc		buildSrc
gradle/wrapper		gradle/wrapper
src		src
.gitignore		.gitignore
CHANGELOG.adoc		CHANGELOG.adoc
LICENSE.txt		LICENSE.txt
README.adoc		README.adoc
build.gradle.kts		build.gradle.kts
gradle.properties		gradle.properties
gradlew		gradlew
gradlew.bat		gradlew.bat
settings.gradle.kts		settings.gradle.kts

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Repository files navigation

XML Deserializer (XD)

Overview

Motivation

Usage

Details

Simple Example

Nested Example

Repeated Nested Example with Attributes

Non-String Properties

XML Namespace Example

Change Log

About

Releases

Packages

Languages

License

bombinating/xml-deserializer

Folders and files

Latest commit

History

Repository files navigation

XML Deserializer (XD)

Overview

Motivation

Usage

Details

Simple Example

Nested Example

Repeated Nested Example with Attributes

Non-String Properties

XML Namespace Example

Change Log

About

Resources

License

Stars

Watchers

Forks

Releases

Packages 0

Languages

Packages