- Feb 21, 2003
Extensible Markup Language (XML)
To say that XML has been popular with the industry press is an understatement. At the moment, every major vendor has announced support for XML in one form or another, and innovative uses for XML are emerging almost daily.
But what is exactly is XML? It is not a programming language like Java, C++, or C#; that is, it cannot be used to write applications per se. Rather, it is a meta-language that can be used to create self-describing, modular documents (data), programs, and even other languages, commonly referred to as XML grammars.1 These documents are often used for data exchange between otherwise incompatible systems.
XML is incredibly diverse and includes a host of other technologies including XPointer, XLink, and Resource Description Framework (RDF). Our intent here is not to give an in-depth discussion of XML, but to provide enough background information to help you understand the implications of how XML is being used in the context of other technologies such as SOAP, WSDL, and UDDIthe foundation of Web Services. Strictly speaking, Web Services can be implemented by using only XML, but for the purposes of our discussion, we are defining Web Services to be built on XML, SOAP, WSDL, and UDDI over a transport protocol such as HTTP. This definition will become clearer as our discussions progress.
The Worldwide Web Consortium (W3C), an international standards body, began working on XML in mid-1996 and released XML 1.0 in 1998. XML was heavily inspired by the Standard Generalized Markup Language (SGML), but, in many ways, XML is more readable and simpler. The real value of XML is not in its innovativeness as much as its industry acceptance as a common way of describing and exchanging data (and, as we will see later with WSDL and SOAP, XML can also be used to describe applications and invoke them as well).
As a markup language, XML uses tags to describe information (the tags are highlighted in bold in the following example).
<?xml version="1.0" encoding="UTF-8"?> <Order> <Customer> <name>John Doe</name> <street>1111 AnyStreet</street> <city>AnyTown</city> <state>GA</state> <zip>10000</zip> /Customer> </Order>
A tag, enclosed in brackets (<>), is a label or a description (e.g., street in our example) of the data that follows, which is called an element (the element for street in our example is 1111 AnyStreet). The element is delimited by a similar tag preceded by a slash (/), to indicate the end of the element. In our example, the element 1111 AnyStreet is terminated by the closing tag </street>.
The first line in our example is a convention used to signal the XML parser (the program that has to parse the XML document) that the incoming document is an XML document. Also, the Customer element has several child elements:
- John Doe
- 1111 AnyStreet
You may have already noticed one advantage of XMLsince it is a text-based language, XML is fairly verbose and therefore human readable. However, this advantage can also be a disadvantage: because they are verbose, XML documents can quickly become very large for complex data sets. There are other points worth noting about XML:
XML is extensible. Unlike HTML, which has a fixed number of tags, XML allows the developer to define any number of tagswhatever is necessary to solve the problem. In our example, the document represents an abstraction of a customer and includes fields to describe the customer.
XML is hierarchical. Elements can have subordinate elements under them. In the example, the Customer element contains several child elements.
XML is modular. By allowing documents to reference other documents, XML provides for modular designs and promotes reuse.
XML does not include built-in typing. This data enforcement and validation is provided through document type definitions (DTDs) and XML schemas, two concepts that will be discussed in further detail later.
XML does not make any assumptions about the presentation mechanism. This is unlike HTML, which does make these assumptions. In fact, XML has to be coupled with another technology (such as XSLT or Cascading Style Sheets) to be displayed. This separation stems from one of XML's primary goals of being a way of exchanging data; oftentimes data is exchanged between systems and hence may not need to be displayed at all.
XML is programming language independent. Since XML is not a programming language per se, it can be used as a common mechanism for data exchange between programming languages and, as we will see later, a common way of connecting applications as well (via SOAP).
XML provides validation mechanisms. Through the use of DTDs and XML schema, XML documents can be validated to determine whether the elements are correct (i.e., whether the values are within a specified range).
Some of the main XML concepts that are especially relevant to Web Services include parsers, namespaces, DTDs, and XML schemas.
Processing an XML document requires the use of an XML parser, a program that can decompose the XML document into its individual elements. There are two major categories of XML parsers: Document Object Model (DOM) and Simple API for XML (SAX).
DOM is a language-neutral API for accessing and modifying tree-based representations of documents such as HTML or XML documents. Developers can use language-specific DOM parsers to programmatically build and process XML documents.
DOM parsers have two major shortcomings:
The entire XML document is represented in memory; this can lead to performance issues if the XML document is exceedingly large.
Since the API is language independent, it is quite generic; therefore more steps are often required to process an XML document than would be the case if it were optimized for a particular implementation language. This has led to language-specific variants such as the JDOM parser, which is tuned for the Java language.
The SAX parser is an event-based parser and can be used only for reading an XML document. A SAX parser works from event registration. The developer registers event handlers, which are then invoked as the XML document is processed. Each event handler is a small block of code that performs a specific task. The main advantage of a SAX parser over a DOM parser is that the former does not require the entire document to be in memorythe XML document is processed as a stream of data, and the event handlers are invoked. While SAX is easier to work with than DOM, there are some disadvantages:
Once the XML document has been read, there is no internal representation of the document in memory. Thus, any additional processing requires the document to be parsed again.
A SAX parser cannot modify the XML document.
Thus it is important to understand the needs of the application before selecting an XML parser.
Well-Formed and Valid XMLs
XML documents must conform to a certain set of guidelines before they can be processed. This leads to two terms that are used to describe the state of a document: well formed and valid.
A well-formed XML document is one that follows the syntax of XML and that can be completely processed by an XML parser. If there are syntax errors in the document, then the parser rejects the entire document. As far as an XML parser is concerned, there is no such thing as a partially well-formed XML document.
A valid XML document is a well-formed document that can also be verified against a DTD, which defines constraints for the individual elementsthe order of the elements, the range of the values, and so on. A validating XML parser is one that can validate an XML document against a DTD or XML schema, which are described next.
DTDs and Schemas
XML offers two mechanisms for verifying whether or not a document is valid. A DTD is an external document that acts as a template against which an XML document is compared. The XML document references this DTD in its declaration, and the XML parser (assuming it is a validating parser) then validates the elements of the XML document with the DTD. A DTD can specify the order of the elements, the frequency at which elements can occur (for example, an order can contain 0n line items), etc.
While a powerful concept, DTDs have many shortcomings.
The concept of a DTD predates that of XML (it originated from SGML) and does not conform to XML syntax. This increases the learning curve and can lead to some confusion.
A DTD does not support data types; this means it is impossible to specify that a given element must be bound to a type. Using the order example under "XML Syntax" earlier in this chapter, there is no way to specify that the line item count needs to be a positive integer.
An XML document can reference only one DTD; this limits how much validation can occur.
A DTD cannot enforce data formats; i.e., there is no way to specify that a date must be of the mm/dd/yyyy format.
DTDs were invented before the standardization of namespaces and consequently do not support namespaces, which can lead to many element name collisions. For more on namespaces, see the next section.
Because of these limitations, applications that have to process XML documents include a lot of error checking functionality. Additionally, SOAP, one of the cornerstone technologies of Web Services, prohibits the use of DTDs in the document declarations.
To address the shortcoming of DTDs, the W3C produced the XML schema specifications. XML schemas provide the following advantages:
The XML schema grammar supports namespaces.
XML schemas include a predefine set of types including string, base64 binary, integer, positive integer, negative integer, date, and time, along with acceptable ranges and data formats.
XML schemas also allow for the creation of new types (simple and complex) by following a well-established set of rules.
An enterprise system consists of dozens if not hundreds of XML documents. As these XML documents are merged from other sources, inevitably there will be duplicate element names. This can cause problems because each element must have a unique name. XML resolves this name collision issue through the use of namespaces (Java provides a similar feature through packages). Each element is prefixed with a namespace and therefore has to be unique only for that given namespace rather than globally. In practice, the prefix is usually the name of the company, although any Uniform Resource Locator (URL) will do.2 Thus, an element name is composed of two parts: the namespace and the name of the element. By qualifying the name of each element with a qualifier, the likelihood of a name collision is greatly reduced. Consider the file system, for example. For a given directory, a filename must be unique. However, there can be multiple identical filenames as long as each exists in a different directory. In a sense, the directory provides the namespace and qualifies the filename to resolve filename conflicts.
Service-Oriented Access Protocol (SOAP)
One of the challenges of performing integration using traditional middleware is the lack of a universal protocol. By being XML based and not tied to any particular language, SOAP has evolved to become the primary de facto standard protocol for performing integration between multiple platforms and languages.
SOAP originally meant Simple Object Access Protocol, but the term has been unofficially redefined to mean Service-Oriented Access Protocol because SOAP is not simple and certainly not object oriented; the latter point is important because not all languages are object oriented.
This flexibility in the protocol allows a program that is written in one language and running on one operating system to communicate with a program written in another language running on a different operating system (i.e., a program written in perl running on Solaris can communicate with another program written in Java running on Windows 2000). There is at least one SOAP implementation for each of the popular programming languages including perl, Java, C++, C#, and Visual Basic.
Advantages of SOAP
Before discussing the characteristics of SOAP, let's examine why it has become so popular.
SOAP is a fairly lightweight protocol. Some of the earlier distributed computing protocols (CORBA, RMI, DCOM, etc.) contain fairly advanced features such as registering and locating objects. At its core, SOAP defines only how to connect systems and relies on additional technologies to provide registration features (UDDI) and location features (WSDL).
SOAP is language and operating system independent. In this respect, SOAP is unlike many other middleware technologies l (such as RMI, which works only with Java, and DCOM, which works only on Microsoft Windows and NT).
SOAP is XML based. Instead of relying on proprietary binary protocols (as is the case with CORBA and DCOM), SOAP is based on XML, a ubiquitous standard. As previously noted, XML is fairly readable.
SOAP can be used with multiple transport protocols. These include HTTP, Simple Mail Transfer Protocol (SMTP), file transfer protocol (FTP), and Java Message Service (JMS). Most of the examples in this book will focus on HTTP since it is the most commonly used protocol with SOAP-based systems.
SOAP can traverse firewalls. SOAP needs no additional modifications to do this. Contrast this with CORBA- or DCOM-based systems, which require that a port be opened on the firewall. This is a key requirement for building distributed systems that have to interact with external systems beyond the firewall. (This is also a disadvantage, as we will see later.)
SOAP is supported by many vendors. All major vendors including IBM, Microsoft, BEA, and Apache provide support for SOAP in the form of SOAP toolkits (the IBM and Apache SOAP toolkits are two of the most popular).
SOAP is extensible. The header values (specified in the Header element) in the XML document can be used to provide additional features such as authentication, versioning, and optimization. These features are discussed further in the next chapter.
Disadvantages of SOAP
On the down side, SOAP does have some disadvantages.
There are interoperability issues between the SOAP toolkits. It seems ironic that there would be interoperability issues with a technology that promotes interoperability, but this is mostly attributable to the infancy of the SOAP specifications. These have been identified and documented, and the various vendors have been quite cooperative in resolving these differences.
SOAP lacks many advanced features. Much has been written about the advantages of SOAP as a lightweight protocol, but there are a host of missing features such as guaranteed messaging and security policies.
Many of these issues can be addressed through third-party technologies such as Web Services networks, which are discussed in further detail in later chapters.
SOAP is built on a messaging concept of passing XML documents from a sender to a receiver (also called the endpoint). The XML document becomes known as a SOAP document and is composed of three sections: Envelope, Header, and Body. Figure 22 illustrates the structure of a SOAP document.
Figure 22 The structure of a SOAP document.
The SOAP standards define three major parameters:
Envelope body structure. The envelope contains information such as which methods to invoke, optional parameters, return values, and, where something did not execute successfully, optional exceptions (known as SOAP faults).
Data encoding rules. Since SOAP has to support multiple languages and operating systems, it has to define a universally accepted representation for different data types such as float, integer, and arrays. More complex data types (such as Customer) require custom coding, although some toolkits, such as GLUE, inherently provide this mapping.
Usage conventions. SOAP can be used in a multitude of ways, but they are all variations of the same actions: a sender sends an XML document, and the receiver, optionally, returns a response in the form of an XML document (this is the case of a two-way message exchange). As mentioned previously, the XML document may contains faults if errors occurred during processing.
By allowing receivers to be chained together, SOAP-based architectures can be quite sophisticated. Figure 23 shows five common architectures that are used with SOAP-based systemsFire and Forget, Request Response, Notification, Broadcast, and Workflow/Orchestration.
Figure 23 Common SOAP architectures.
Any link in the processing chain that is not the endpoint is referred to as an intermediary. The SOAP specifications allow an intermediary to process a SOAP message partially before passing it to the next link in the processing chain (which can be another intermediary or the endpoint). You will see an example of this in the discussion of the SOAP header later in the chapter.
Migrating from XML to SOAP
Migrating from XML to SOAP is a fairly straightforward procedure. The migration includes these steps:
Adding optional Header elements
Wrapping the body of the XML document in the SOAP body, which in turn is included in the SOAP envelope
Declaring the appropriate SOAP namespaces
Adding optional exception handling
Specifying the protocol that should be used
For example, converting our earlier XML document to SOAP involves adding the following parts (highlighted in bold; for the sake of simplicity, the HTTP fragment has been stripped away):
<?xml version="1.0" encoding="UTF-8"?> <SOAP-ENV:Envelope xmnls:SOAP-ENV="http://schemas.xmlsoap.org/soap/envelope" xmnls:xsi="http://www.w3.org/1999/XMLSchema-instance" xmnls:xsi="http://www.w3.org/1999/XMLSchema"> <SOAP-ENV:Header> . . . [optional header information] </SOAP-ENV:Header> <SOAP-ENV:Body> <Order> <Customer> <name>John Doe</name> <street>1111 AnyStreet<street> <city>AnyTown</city> <state>GA<state> <zip>10000</zip> </Customer> </Order> <SOAP-ENV:Body> <SOAP-ENV:Envelope>
The next sections describe these additional parts.
The SOAP envelope is the container for the other elements in the SOAP message. A server-side process called a SOAP handler can use the availability of the SOAP envelope (along with the http://schemas.xmlsoap.org/soap/envelope/ namespace declaration) to determine whether the incoming XML document is a SOAP message or not. The handler can be part of the application server, or it can be an external product such as Cape Clear CapeConnect. SOAP handlers are explained in more detail later in the section on adding SOAP support.
As part of their extensibility design goal, the architects of SOAP provided the Header element to allow SOAP messages to be extended generically while still conforming to the SOAP specifications. If a SOAP Header element is present (and there can be more than one Header element present), it has to be the first child of the Envelope element. Each Header element can in turn have child elements.
Two examples of using header information to provide extensibility include
embedding authentication information
specifying an account number for use with a pay-per-use SOAP service
/A SOAP intermediary can use this header information to determine whether the incoming message is properly authorized before forwarding it (to either another intermediary or the endpoint).
In cases where a SOAP handler cannot decipher a message, a SOAP fault is generated, identified by the Fault element. Its child element, faultcode, identifies the category of errors that can happen. SOAP 1.1 defines four values for the faultcode element:
VersionMismatch. The recipient of the message found an invalid namespace for the SOAP envelope element.
MustUnderstand. The recipient encountered a mandatory Header element it could not understand. Remember, header elements are optional.
Client. The fault was in the message being received. Possible causes: missing elements, malformed elements, and the like.
Server. The fault occurred on the recipient side, i.e., a server error.
Note that an application is free to extend these values using a (.) notation. For example, a value of Client.Login can be used to specify that there was a problem with a client login.
In addition to the faultcode element, there are two other elements that can be used to provide further clarification on the fault:
faultstring. This element provides a readable explanation on why the fault occurred.
detail. The value of the detail element indicates that the problem occurred while processing the body element. If the detail element is not present, then the fault occurred outside of the body of the message.
Adding SOAP Support
One of the advantages of adopting SOAP is that the support can be built on top of existing technologies. Figure 24 shows a typical J2EE Web-based architecture without support for SOAP.3 Adding SOAP support to such a system typically requires the addition of a SOAP handler (if the application server cannot support SOAP requests), which parses incoming SOAP requests and then calls the appropriate native method in the implementation language. Recall that SOAP is a protocol, not a programming language; hence, the request must be mapped to an entry point in an executing application. The entry point can be a method in a class (for object-oriented systems such as Java, C++, or C#) or a function name (for systems such as perl, which are not object oriented).
Figure 24 Typical J2EE deployment.
Common SOAP handlers include CapeConnect from Cape Clear, Iona's XMLBus (see Appendix D for a more detailed discussion), and Apache Axis. In summary, a system is said to be SOAP compliant if it can take an incoming SOAP request, forward it to the appropriate endpoint, and package the result back in a SOAP response.
Figure 25 Adding SOAP support to J2EE environment.
While SOAP provides many useful features, it is still incomplete because it does not address this issue: how does an endpoint unambiguously describe its services? Likewise, another outstanding issue: how does a requester locate the endpoint? These two features are provided by two other key technologiesWSDL and UDDI.
Web Services Definition Language (WSDL)
To enable a client to use a Web Service effectively, there first has to be a mechanism for describing that Web Service. At first glance, this may seem difficult, but the challenges are many. We can provide a description in prose format (such as a README file), but it would not be practical to describe all the different ways a Web Service can be used in prose. We can also list some examples of how the Web Service can be used effectively, but, again, that may not adequately describe all the combinations for which a Web Service can be invoked.
The problem of succinctly and unambiguously describing a Web Serv-ice is similar to the challenge faced by compiler writersthe conventional solution is to use a grammar tree to describe the syntax of a language. While quite effective, a grammar tree is rarely decipherable for those without a strong background in compiler theory.
WSDL was created in response to the need for unambiguously describing the various characteristics of a Web Service. As an XML grammar, WSDL is not easy to learn, but it is considerably less intimidating than a programming language (such as C or C++).
A WSDL document is a well-formed XML document that lists the following characteristics for one or more Web Services.
Publicly accessible functions.5 A WSDL lists all the operations a client can expect a Web Service to support.
Input and output parameters along with associated types. In order to invoke each operation, the client needs to know the expected parameters for each input operation and the expected output of each operation. Again, this is identical to a normal function declaration. In order to support portability between languages and operating systems, all data types are defined in XML schema format.
Binding and address information for each Web Service. To allow loose coupling between a requester and a Web Service, WSDL specifies where the service can be found (usually a URL) and the transport protocol that should be used to invoke the service (remember that a Web Service can be used with multiple protocols).
In essence, WSDL defines a contract that a provider is committed to supporting, and, in the spirit of separating implementation from interface, WSDL does not specify how each Web Service is implemented. As a point of comparison, WSDL can best be likened to CORBA's IDL.
Most of the existing toolkits (GLUE, IBM Web Services Toolkit [WSTK], BEA Web Services Workshop, Cape Clear CapeStudio, etc.) have built-in functionality to automatically parse and generate WSDL files (although, due to the immaturity of the tools, the generated files still require some manual tweaking). Even so, it is still worthwhile to understand the structure of a WSDL document.
The WSDL specifications list six major elements:
The definitions element is the root element containing the five remaining elements; it defines the name of the service and declares the namespaces used throughout the document.
The message element represents a single piece of data moving between the requester and the provider (or vice versa). It declares the name of the message along with zero or more part elements, each of which represents either a single parameter (if this is a request) or a single return value (if it is a response). If there is no part, then the request requires no parameter or there is no return value, depending on whether the message represents a request or a response. Note that each message element declares only the name (which is used by the operation element below), value(s), and the type of each value; it does not specify whether the message is for input or outputthat is the role of the next element.
The portType element represents a collection of one or more operations, each of which has an operation element. Each operation element has a name value and specifies which message (from the message element) is the input and which is the output. If an operation represents a request/response interaction (a method invocation with a return value), then the operation would include two messages. If an operation represents only a request with no response or a response with no request (e.g., an automatic notification from the provider with no request from the requester), it would include only a single message. In Java terms, a portType can best be thought of as an interface; an operation can best be thought of as a single method declaration; a message can best be thought of as a individual piece of an operation, with each message representing (if the operation is an input) a parameter name and the associated type or (if the operation is an output) return value name and the associated type.
The types element is used to declare all the types that are used between the requester and the provider for all the services declared in the WSDL document.
The binding element represents a particular portType implemented using a specific protocol such as SOAP. If a service supports more than one protocol (SOAP, CORBA, etc.), the WSDL document includes a listing for each.
The service element represents a collection of port elements, each of which represents the availability of a particular binding at a specified endpoint, usually specified as a URL where the service can be invoked.
Invoking Existing Web Services: A Sample
To invoke a Web Service, we can either write a SOAP client or use an existing generic one. The www.soapclient.com site provides a Web interface that allows us to enter the WSDL file and invoke the service. Before we can launch the service, we need to find the WSDL file. In this case, we can find some sample WSDL files at www.xmethods.net, a public repository of Web Services. For our example, we will invoke a Web Service that can print the traffic conditions of a specified California highway. Use the following instructions:
Visit the www.soapclient.com/soaptest.html site.
Type www.xmethods.net/sd/2001/CATrafficService.wsdl in the WSDL address field.
Select HTML instead of XML for the output.
Click Retrieve, which loads the WSDL file from across the Internet.
In the textfield, type 101 (for Highway 101) and click Invoke to invoke the service.
The resulting screen should print text that explains the current conditions for Highway 101.
This example illustrates how straightforward it is to invoke a Web Service from a browser. The user, in most cases, will not even be aware that Web Services are being used to return the values. Of course, the user can just as easily be a program, in which case the program would programmatically pass the appropriate parameters.