SOAP defines a set of serialization rules for encoding datatypes in XML. All data is serialized as elements rather than attributes. Attributes are only used for structural metadata; for example, when references are needed.

For simple types such as strings, numbers, dates, and so forth, the datatypes defined in "XML Schema Part II-Datatypes" are used. For types such as classes or structures, each field in the type is serialized using an element with the same name as the field. For array types, each array element is typically serialized using an element with the same name as the type, although other element names may be used. In both cases, if the field being serialized is itself a structure or an array, then nested elements are used. The top-level element in both the structure case and the array case is namespace qualified. Descendant elements should be unqualified.

The serialization rules apply to children of the Header element as well as children of the Body element. Such children are serialized types just like any other type. A request and any associated response are also treated as types, and are serialized according to the same rules.

Examples
For an example of serialization of a structured Java or VB type, see Listing 1.

For an example of serialization of a Java or VB array, see Listing 2.

10.4.1 Serialization of Simple Structured Data
Serializing data structures, when each field is referred to exactly once, is straightforward. Each field is serialized as an embedded element, a descendant element of the Body element, not as an immediate child. Such an element is called a single-reference accessor, and it provides access to the data in the field at a single location in the message. The element name used to contain the data is the same as the field name used in the programmatic type.

For an example of serializing structured data, see Listing 3, which shows the Java and VB definitions for a method call taking a structured type representing a Person as a single parameter, followed by the SOAP message representing a request to execute such a method.

10.4.2 Serialization of Structured Data with Multiple References
In cases when a field in a data structure is referred to in several places in that data structure (for example, in a doubly linked list), then the field is serialized as an independent element, an immediate child element of Body, and must have an id attribute of type ID. Such elements are called multireference accessors. They provide access to the data in the field from multiple locations in the message. Each reference to the field in the data structure is serialized as an empty element with an href attribute of type IDREF, where the value of the attribute contains the identifier specified in the id attribute on the multireference accessor preceded by a fragment identifier, #.

Listing 4 shows the Java definition for a method call taking two parameters both of type Person, followed by the SOAP message representing a request to execute such a method where both parameters refer to the same instance of Person.

10.4.3 Dealing with Null References in Complex Data Structures
In certain cases when reference types exist in a programmatic data structure there is a need to represent a null reference. Such references are modeled in SOAP messages using the nil attribute in the www.w3.org/2001/XMLSchema-instance namespace. Setting the value of the attribute to 1 indicates that the accessor on which it appears represents a null reference. See Listing 5 is an example of null references, a Java class definition for a simple linked list. The end of the list is indicated by a null reference in the next field. A list of three items is passed in the request message.

10.4.4 Serializing Dynamically Typed Data
SOAP provides for serialization of dynamically typed data; that is, data typed at runtime, through a polymorphic accessor. Such accessors look like normal accessors apart from the presence of a type in the 'www.w3.org/2001/XMLSchema-instance' namespace. This attribute indicates the type the accessor actually holds. The value of this attribute may well vary from message to message. Listing 6, an example of dynamically typed data, shows the Java and VB definitions for a method call taking a dynamically typed parameter followed by several SOAP messages representing a request to execute such a method. The first SOAP message passes a parameter of type long whereas the second passes a parameter of type Person.

10.4.5 Arrays
SOAP provides comprehensive array support. Single and multidimensional arrays are supported, along with sparse and jagged arrays and partial transmission. Arrays in SOAP are always of type Array in the http://schemas.xml-soap.org/soap/encoding/ namespace, or a type derived by restriction from that type. If they are of the Array type, they are encoded using an Array element also in the http://schemas.xmlsoap.org/soap/encoding/ namespace. If they are of a derived type, then any element name may be used. In either case, an arrayType attribute in the http://schemas.xmlsoap.org/soap/encoding/ namespace is mandatory. The type of this attribute is string, but it in fact indicates the type of the array along with dimension information. Each dimension appears in square brackets after the QName for the type, separated by commas. Each array item is serialized as an element. The name of this element can be the type name or some arbitrary name. Listing 7, a simple array example, shows a response message containing an array of five long values. Note the value of the arrayType attribute indicating the size of the array.

10.4.6 Multidimensional Arrays
Multidimensional arrays can be encoded by specifying multiple dimensions separated by commas inside the square brackets in the arrayType attribute. Any number of dimensions may be specified. Listing 8, a multidimensional array example, shows a request message containing a two-dimensional array of strings. Note the value of the arrayType attribute indicating the type and dimensions of the array.

10.4.7 Partial Transmission of Arrays
In certain scenarios an array of a certain size may need to be transmitted, but only a subset of the items needs to be sent. For such arrays the array element is annotated with an offset attribute in the http://schemas.xml-soap.org/soap/encoding/ namespace. The value of the offset attribute indicates the zero-based offset of the first element. The value appears in square brackets. Listed items are assumed to appear at contiguous locations in the array. Items may be omitted from the end of the array. Listing 9, an example of partial array transmission, shows a request message that transmits the third, fourth, and fifth items in a nine-item array

10.4.8 Sparse Arrays
Sparse arrays, those in which noncontiguous items need to be transmitted, are also supported. Each serialized array item is annotated with a position attribute in the http://schemas.xmlsoap.org/soap/encoding/ namespace. The value of the position attribute is a zero-based offset of the position of the item in the array, enclosed in square brackets. Listing 10, an example of sparse arrays, shows a request message that transmits the second, fourth, and eighth items in a nine-item array.

10.4.9 Jagged Arrays
SOAP supports jagged arrays, also known as arrays of arrays. The arrayType attribute contains a type that includes empty square brackets, as many as necessary to indicate how many dimensions each array has, followed by the dimensions of the array of arrays in square brackets as normal. The inner array elements are also annotated with the appropriate arrayType attribute. Listing 11, an example of jagged arrays with single-reference accessors, shows a request message that transmits an array of arrays of strings. Each array is encoded using a single-reference accessor. Listing 12, an example of jagged arrays with multireference accessors, shows a request message that transmits an array of arrays of strings. Each array is encoded using a multireference accessor.

SIDEBAR

Essential XML Quick Reference
A Programmer's Reference to XML, XPath, XSLT, XML Schema, SOAP and More by Aaron Skonnard and Martin Gudgin

Published by: Addison-Wesley Pub Co.
Pages: 402
ISBN: 0201740958
List Price: $24.99

• Source Code