C/C++ developers: Fill your XML toolbox
01 Sep 2001
It seems as if everywhere you look there is some new XML-related tool being released in source code form written in Java. Despite Java's apparent dominance in the XML arena, many C/C++ programmers do XML development, and there are a large assortment of XML tools for the C and C++ programmer. We'll confront XML library issues like validation, schemas, and API models. Next, we'll look at a collection of generic XML tools like IDEs and schema designers. Finally, we'll conclude with a list and discussion of libraries either usable from or actually written in C and/or C++.
In this article, I'll skip arguments for using XML; I'll assume that you already have good reasons for wanting to ramp up on this technology. Also, I'll leave the more detailed explanations of XML to the background sources in Resources. Suffice it to say that XML is a standard for data exchange (not just a file format). The data may be exchanged in the form of application files in XML format, or over network connections that exist only for a moment before being discarded at the conclusion of an online transaction or at the close of a network connection.
Also, this isn't a comparative review that rates tools. My goal is to explain the types of tools you'll probably need and to point you to likely candidates. You'll still need to research, test, and compare tool features against your project needs to assemble your ultimate toolbox.
To incorporate XML in your own software projects, you're going to want to have two sets of tools in your bag of tricks. The first set is a dialect designer (or more properly "schema designer"). The second set of tools includes software libraries that will add parsing and XML-generation features to your application.
An XML dialect is just a particular set of XML tags along with some rules for how the tags fit together. The two dominant ways for specifying or defining an XML dialect currently are through a Document Type Definition (DTD) or an XML Schema. I'll refer to both of these collectively as a schema.
Your project's domain may already have a particular schema designed for you. If not, you can hack up your own schema using a plain text editor. A more refined approach is to use an actual dialect designer that can check syntax. (An incorrect schema won't help when you try to use it later to validate your XML data.)
Nowadays, most programmers' text editors -- particularly the ones found in IDEs -- have decent macro and template support for features like syntax highlighting and autocompletion of a partially typed word or phrase. Therefore I am omitting from this discussion any XML editors that do only syntax highlighting and autocompletion. Microsoft Word or an Emacs macro can do that, so "XML editor" ought to mean something more.
The tools shown in Table 1 fall into three categories:
- The IDE (integrated development environment, the Swiss Army knife approach)
- The schema-sensitive XML editor (reinforces allowable tag structure and attributes found in a schema or DTD)
- Schema designer (for writing your own Schema or DTD)
Because all of the tools in Table 1 are running applications, they are suitable for all XML developers -- not just those using C++ (unlike the tools listed in the other tables in this article).Table 1. Dialect design tools for various platforms
|Turbo XML||TIBCO/Extensibility||XML Schema/DTD designer and IDE||commercial||Java, Win32|
|Komodo||Active State||XML editor and IDE||commercial||Linux, Win32, others|
|XML Spy||Altova||IDE for XML editing and schema design||commercial||Win32|
|XML Notepad||Microsoft||XML editor||free||Win32|
|Morphon XML||Lunatech Research||Schema-sensitive XML data editor||commercial||Java|
|XED||University of Edinburgh||Schema-sensitive XML data editor||noncommercial||Win32, Linux, Unix|
|Xeena||IBM alphaWorks||Schema-sensitive XML data editor||free trial/commercial||Mac, Unix, Win32|
|Visual XML||Pierlou||Schema-sensitive XML data editor||noncommercial||Java|
|Netpadd||Phillip Lenssen||Alternative to Microsoft's XML Notepad||noncommercial||Win32|
|XMetal||Softquad||DTD sensitive XML editor||commercial||Win32|
|Merlot||Channelpoint||Visual XML editor; supports plug-ins for DTDs||noncommercial||Java|
|XML Validator||ElCel Technologies||Command-line XML validation tool||noncommercial||Win32|
|XML Canon||ElCel Technologies||Produces canonical XML by merging XML data with DTD||noncommercial||Win32|
The rest of this article serves up the meat and potatoes of adding XML functionality for the C/C++ programmer through software libraries. You'll see in the next sections of this article that many more command-line utilities can be found included as test and/or sample programs in the software libraries. As an example, Transformiix can be used as a library, a Perl module, or a command-line tool.
Up until now I've tried to be frugal in my use of XML-related terms. Before you read further, though, if you are not familiar with XML basics, you might want to scan the terms defined in the XML terminology sidebar. The terms will assist you both in following the rest of this article and in sorting out the features of the tools and libraries mentioned as you explore them in greater depth on your own.
Once you have a DTD or schema and the XML document to go with it, you'll need a parser to read and interpret the XML document. Table 2 outlines parser libraries for C/C++ developers. Before you start poring over the grid in the table, though, you'll need a little bit of background.
XML parsers come in two forms: validating and nonvalidating. Which one do you need? If you are working without a formal DTD or schema, validation features won't be important to you. If you have or are planning to use a DTD or schema, you will probably prefer a validating parser. (In that case, I suggest that you also learn how to read and write a DTD/schema by hand so that when validation issues arise you will be able to deal with the errors. Sometimes the error is in the DTD/schema, so you may be debugging the DTD/schema files as well as your XML data.)
Two common API models exist for interfacing software with an XML parser: the document model and the event model. The document API model parses XML data to produce an object. The object abstracts the document's contents into a tree structure. The application operates on this tree-structure object. The event API model uses a callback mechanism to notify the application of the structure of the XML data. The events/callbacks usually occur at the time of parsing.
The generic parser API models have been further refined into specific API standards. The W3C has recommended DOM (levels 1 and 2, with level 3 in draft) as a standardized document API model. While not a W3C project, SAX has taken its place as the de facto standard event API model.
As you compare features in parsers and other XML tools, look for support for W3C recommendations and emerging specifications, such as namespaces, XPath, XLink, XInclude, and XInfoset. Keep in mind that XML technologies are maturing rapidly and that support for the first level of a specification, such as the DOM, may lack important functionality introduced in level two of that specification. If functionality in the most current form of a specification is important to your project, choose your tools accordingly.
In Table 2, the Event column specifies parsers that support a push or event model API, like SAX. The Doc column specifies parsers that support a pull or document model API, like DOM. As before, the table lists both commercial and noncommercial tools (see the sidebar Licensing for details about the software licenses.)Table 2. Parsers for C/C++ developers
|expat||James Clark/expat team||native & SAX||-||Very fast push model parser with a native API and SAX wrappers.||LGPL (free)|
|libxml||Gnome||SAX||DOM||Very robust; SAX & DOM wrappers; does DTD validation||LGPL (free)|
|MSXML||Microsoft||SAX||DOM||The Microsoft XML library for Win32||EULA (free)|
|Xerces||Apache Software Foundation||SAX||DOM||Does SAX plus DOM levels 1 and 2; DTD validation; incremental XML Schema||Apache (free)|
|XTL||Vivid Creations||SAX||DOM||STL-based XML toolkit with SAX and DOM||commercial|
|RXP||University of Edinburgh||-||native||Validating namespace-aware XML parser in C||GPL (free)|
|XML4C||IBM alphaWorks||SAX||DOM||IBM-sponsored variant of Xerces||Apache (free)|
|Oracle XDK 8i||Oracle||SAX||DOM||Oracle-sponsored XML toolkit for C++||noncommercial|
|Pull Parser||Extreme! Lab||-||native||Indiana University-sponsored lightweight XML toolkit for C++||noncommercial|
|XML Booster||PhiDaNi Software||-||native||Parser generator, generates C-source parser||commercial|
Open-source top three
The three most popular of the open-source XML libraries are expat, libxml, and Xerces. All three are cross-platform, and each serves as a basis for an XSLT library implementation, which give you a growth path once you've satisfied your basic XML needs.
- Expat is an open-source event-oriented XML parsing library originated by James Clark. He has transferred the project to a small team on SourceForge. A SAX wrapper is available. The expat parser can found in a number of projects, such as the open-source browser Mozilla, the XSLT processor Transformiix, and the RDF tool repat.
- Libxml offers a dual-mode API for both SAX and DOM-like operations. It supports validating against a DTD and is used in Gnome's XSLT processor, libxslt. Libxml was rewritten and released as libxml(2) but may also be referred to as libxml2. Users of this library should make sure they have the current version.
- Xerces is a very solid, well-documented library that serves as the basis for the IBM alphaWorks XML4C library. Xerces is also used in the Apache XSLT processor, Xalan. Xerces supports DOM, SAX, and validating against a DTD. The latest versions reads and interprets parts of the W3C XML Schema Recommendation (with complete XML Schema support targeted for the end of 2001).
Building libxml on Windows from scratch is an easy four steps:
- Download the source tarball.
- Use a program like Winzip to unzip the contents to a directory. Be sure you instruct your unzipping utility to preserve path names for any subdirectories that libxml may need.
- Locate the
libxml2.dswfile in the
./win32/dspsubfolder and open it from MS Developer Studio.
- Select Build All from the top menu in DevStudio. This builds all the sample and test programs along with the libxml DLLs needed to run them.
You can use the above steps to build Xerces on Windows. The only difference is to look for the
samples.dsw workspace file in the
Expat has started including DSP project makefiles. Look in the
For projects running on Linux or Unix, in most cases you can untar the source code to an empty directory, set some options, and type "make" to build a shared library. Solaris users: Don't forget to use the GNU untar utility. The following code worked for me from the bash shell under Slackware Linux:
MSXML, Microsoft's proprietary XML offering for the Windows family of operating systems, is implemented as a collection of scriptable COM objects, so it plays well in other language environments and is well documented. The library supports DOM along with a native document-oriented interface. SAX events are also supported.
As an alternative to MSXML, the Apache XML Project's Xerces library comes with a COM wrapper that will make it act as a drop-in replacement for MSXML in many cases. Vivid Creations offers COM wrappers for the SAX and DOM APIs to its XTL library that also serve as substitutes for MSXML.
The transform is the next step up the XML evolutionary ladder from merely processing XML data at the element and attribute level. An XML transform operates on incoming XML data to produce XML output. A transform can reorganize tag structure, add/remove tags and attributes, and filter to zoom in on select fragments of XML data.
The XQuery documentation refers to the transform process as a query but the meaning is the same.
XSLT and XQuery are XML dialects for specifying how to perform such actions on random XML data. You can write a script file with the changes expressed in XML as XSTL or XQuery instead of loading some XML data into a DOM and having to programmatically manipulate the DOM version to produce a desired result. This more generalized approach leads to greater flexibility and reduced development time. Now your Web developers who are not C/C++ programmers can write their own transforms as XML, which may free a C++ programmer for more complex work.Table 3. C/C++ Transform/Query libraries
|libxslt||Gnome||built on top of libxml||LGPL or X11-like (free)|
|Xalan||Apache||Built on top of the Xerces parser||Apache (free)|
|Transformiix||MITRE||XSLT processor built on expat||noncommercial|
|xsltc||Oliver Gerardin||XSLT compiler, produces C code||noncommercial|
|sablotron||Ginger Alliance||XSL engine||noncommercial|
For the purpose of this article, messaging refers to having two software agents communicate with one another. Such messaging is sometimes called message-oriented middleware. (This is not messaging like AOL, MSN, or ICQ, okay? There is an XML-based instant-messaging protocol effort underway called Jabber. I've included a link in Resources for your curiosity, but, again, that's not what I'm talking about here.)
Using XML for messaging has become popular enough to have produced these two alternatives: XML-RPC and SOAP. The most appealing feature of these protocols is that clients, servers, and peers can differ dramatically in terms of the developer's choice of tools for implementation. It's as though all developers get to use their favorite language, development kit, or software library and still work together.
(As a side note, Gregor Purdy has written an excellent critique of XML-RPC in the form of a proposed alternative (see Resources).
Table 4 includes a few libraries for use in the message-oriented middleware category. This is not an exhaustive list of the resources in this category, and there are new tools evolving quickly, but it's a good start.Table 4. C/C++ Messaging libraries
|4S4C SOAP services||Simon Fell||Open Source SOAP effort||noncommercial||Linux, Unix, Win32|
|SOAP client||SQL Data||C++ SOAP client toolkit||commercial||Win32|
|SOAP component||mozilla.org||Scriptable XPCOM component||noncommercial||numerous|
|XML-RPC for C/C++||First Peer||XML-RPC library in C||noncommercial||Linux, Unix, Win32|
|XML-RPC component||mozilla.org||Scriptable XPCOM component||noncommercial||numerous|
|XML-RPC for C/C++||Epinions||XML-RPC library in C||noncommercial||Linux, Unix, Win32|
These tools ought to give you a good start on your XML toolbox. If you want to suggest other C/C++ tools for XML that you have tried or to make any other comment, join the discussion attached to this article (use the link in Resources or click the Discuss icon at the top or bottom of the article page).
- Visit the W3C XML page, the host with the most in XML specifications.
- Explore this compilation of XML-related software.
- Catch up on XML basics with Zvon's general XML tutorial or Doug Tidwell's Intro to XML here on developerWorks and other offerings in the XML zone education area.
- Look at Howard Katz's introduction to XQuery.
- Review Gregor Purdy's excellent critique of XML-RPC.
- Gain an understanding of XSLT in Michael Kay's technical overview, What kind of language is XSLT?.
- Find out about that other kind of messaging (instant messaging) at Jabber.
- Get a handle on the XML developer skillset by reviewing the XML Certification guidelines for the IBM Certified Developer Program.
- XML Spec 1.0: The W3C core XML 1.0 specification.
- DOM Level 1.0: The W3C Document Object Model level 1 API Recommendation.
- DOM Level 2.0: The W3C Document Object Model level 2 API Recommendation.
- SAX/SAX2: The Simple API for XML event model de facto standard.
- Namespace: W3C Recommendation for handling XML namespaces.
- XML Schema: All about the W3C XML Schema Recommendation
Rick is a long-time programmer whose professional career consists of stalking stock options and defeating deadlines while giving gratuity enough to make any waitress blush. His one claim to fame is being known by first name at nearly every coffeehouse in town. He also enjoys speaking in a seminar setting on technical topics. A bit of a maverick on design, he is moving to more modern modeling methodologies like UML. A bumper sticker on his car reads: "I <br/> for XHTML." You can contact Rick at email@example.com.