ColdFusion MX and XML: Creating XML Part 2

Jun. 28, 2002 12:00 AM

I'll plead guilty to having tortured more than my share of bytes into completely unnatural configurations.

In the BX years (Before XML), if you had to share data with your business partners you might have represented the data in comma-delimited text files. If the data was too complex to fit one record to a line, you might have created a proprietary design to hold the information.

Today you'd probably consider XML. Its consistency makes it possible to create and parse text-based data files with standardized programming tools. When deciding how to transfer your data to XML you'd have a choice of programming approaches. Some would be hard, others easier, but until now none would have been native ColdFusion.

That's changed with the release of ColdFusion MX. This is the second of three articles describing the new XML development tools in CFMX. In the first article (CFDJ, Vol. 4, issue 6) I described methods for parsing, extracting, and searching for data from XML. This month I'll describe the new tools for turning your data into XML.

Toto, I Don't Think We're in WDDX Anymore
When I speak of XML, I'm not just talking about Web Distributed Data Exchange (WDDX), which has been supported in ColdFusion since version 4. While WDDX is a wonderful XML language, it hasn't been adopted widely beyond the world of Macromedia. The rest of the computing world uses other variants of XML, and there are too many different flavors to count. You need to be able to publish XML in any format, not just WDDX. That's what these tools provide.

Imagine that I have a database table containing employee information that I want to move to XML. In this example I'll use the employees table from the CFSnippets database that's installed with CFMX.

CFMX offers two methods for creating and publishing an XML document:

The "Text" method uses the new <CFXML> tag and allows you to model the XML tree as text, much as you use <CFOutput> to model HTML output.

The "Object" method uses CFMX's new XML functions to build an object model.

The Text method is considerably easier to use, but I'll show both approaches and you can decide for yourself.

Modeling XML as Text
The Text method builds the XML structure as text, then converts it to an XML document object variable. First, create a model of the XML structure inside a pair of <CFXML> tags. Use placeholders to indicate where the variable data will go:

<CFXML variable="xmlDoc">
<Employees>
<Employee ID="XXX">
...
Child Elements
...
</Employee>
</Employee>
</CFXML>

The <CFXML> tag creates a new XML document object variable named xmlDoc. As I described in the last article, you can view the structure of the resulting XML document object with the <CFDump> command:

See Figure 1 for the dumped structure of the XML document before being populated with real data.

Adding Your Data
Assuming your content is in a database, add a <CFQuery> before the call to <CFXML>. To add variable content to the XML document, wrap the employee element with a <CFOutput> tag pair that loops over the query contents. This is just like looping over a query to create multiple list items or table rows in HTML:

<CFOutput query="qEmps">
<Employee>
...
</Employee>
</CFOutput>

To place the data where you want it, replace the placeholders with the data from the query. If there might be any reserved characters in a value, such as ampersands (&), quotes ("), or tag symbols ("<" or ">"), wrap the value with the XML-Format() function. This replaces any such characters with their equivalent entities and ensures that your XML is well formed:

<Employee ID="#qEmps.Emp_ID#">
<FirstName>#XMLFormat(qEmps.FirstName)
#</FirstName>
</Employee>

Notice that the element tags are wrapped around the value without any additional white space (spaces, tabs, line feeds). Adding such space for readability is fine when creating HTML, since the Web browser usually normalizes the resulting output, removing extra spaces. In XML you'd be changing the data.

Your XML document object has been created. Once again, you can view the structure of the document with <CFDump>. See Listing 1 for the complete code, and Listing 2 for the contents of the XML file it creates.

The toString() Function
Right now your XML document is a structured object that you can read programmatically using the techniques described in last month's article. To share the data, you now need to convert the document to text. For this, wrap the XML document object variable in the toString() function. For instance, to publish the XML document to a text file, use this code:

<CFFile action="WRITE"
file="#ExpandPath('.')#\employees.xml"
output="#toString(xmlDoc)#">

The toString() function adds an XML declaration at the beginning of the output and always sets the declaration's "encoding" attribute to UTF-8.

Modeling XML as Objects
The Object method of creating XML files uses two new XML functions:

XMLNew(): Creates a new XML document
XMLElemNew(): Creates a new XML element

Start with XMLNew(). This function returns an instance of an empty XML document:

The document initially isn't well formed - that is, it doesn't contain any elements. So next you'll create a new element as a child of the document. XMLElemNew() takes two arguments. The first is the XML document object to which you're adding the element. The second is the name of the new element.

The document has a built-in child object named xmlRoot. By assigning a new element to the xmlRoot, you're creating the document's root element:

<CFSet xmlDoc.xmlRoot=
xmlElemNew(xmlDoc, "Employees")>
<CFSet root=xmlDoc.XMLRoot>

Now, loop through the query just as before, but this time create elements to represent each employee. Each element contains a built-in array named XMLChildren, and each item of the array is a child element. Adding a new employee element to the root looks like this:

<CFSet ArrayAppend( root.xmlChildren,
xmlElemNew(xmlDoc, "Employee") )>

As the XML hierarchy gets deeper, the syntax for referring to any particular element can get unwieldy. So now create a reference variable that points to the element you just created:

<CFSet emp=root.xmlChildren[
ArrayLen(root.xmlChildren)]>

Each employee has an ID that I'd like to store as an XML attribute. Each element has a built-in structure called XMLAttributes. When you assign an item to the structure, you're adding an attribute to the element:

Each employee has a set of values that I want to store as child elements with text values between the tags. Once again I'll use XMLElemNew() to create the element, then set the new element's built-in XMLText property to the value from the query. I don't have to use the XMLFormat() function to wrap the data as the Object method replaces any reserved characters for me:

<CFSet ArrayAppend( emp.xmlChildren,
xmlElemNew(xmlDoc, "FirstName"))>
<CFSet emp.FirstName.xmlText=
qEmps.FirstName>

Repeat this step for each of the child elements you want to add and populate with data. When you run the code you'll have created an XML document object containing your data.

This time, instead of saving to disk, I'll share the XML over the HTTP connection. Before outputting to the browser, I'll use <CFContent> to set the content type to XML. This lets any users of the data know that they're getting the right type of information:

The code for the Object method is in Listing 3.

White-Space Issues
You may notice that the <CFXML> version of the output has a little too much white space between employee elements. This is okay as an XML parser doesn't make any distinction between a little bit of white space and a lot of white space.

The Object version, though, comes out in a single line and is difficult for the human eye to read:

<Employees><Employee ID="1"><FirstName>...

I fix this with a call to a user-defined function I wrote named XMLIndent() (stored in a file named xml-Utilities.cfm). See Listing 4 for its contents.

XMLIndent() passes the XML document through an XSLT (Extensible Stylesheet Language Transformation) template that returns the XML document as is, but adds hard returns to make it more readable. XMLIndent() takes the XML document as an argument and returns a text representation of the indented result:

<CFInclude template="xmlUtilities.cfm">
<CFSet xmlText=xmlIndent(xmlDoc)>

I'll discuss this UDF more thoroughly in the next article, which describes what you can do with ColdFusion and XSLT.

Limitations
Both methods of creating XML files have these limitations:

UTF-8 encoding only: While the toString() function has an "encoding" attribute, it doesn't work for the XML document object. And while CFFile now has a "charset" attribute to output in other encoding formats, it doesn't affect the resulting XML declaration's encoding attribute. You can easily end up with a file in UTF-16 format, but an XML declaration that claims the file is in UTF-8.

No Document Type Declarations: There is no built-in method for assigning a Document Type Definition (DTD) to the XML output. Adding the DTD to the XML text in <CFXML> doesn't throw an error, but it's also ignored.

I'll describe workarounds for both of these problems in the next article. (If you need either of them right away, e-mail me.)

Conclusion
I've described two methods of creating XML files with ColdFusion MX. The Text method uses <CFXML> and will seem most intuitive to experienced ColdFusion developers as it allows you to create XML with the same simple methods we use to create HTML dynamically. The Object method gives you greater control over the structure of your XML document at the expense of greater coding complexity. Either method allows you to export your data to any XML language with relative ease.

* * *

In the last article of this series I'll describe the use of XSLT and the new XMLTransform() function, including how to use XSLT to move your data from WDDX to other flavors of XML and back again.

Related Links

▪ Figure 1

▪ Source Code

About David Gassner