I've been working on improving the RSS feeds generated by Lighthouse.  One persistent problem is that some RSS readers (IE, for instance) will choke on a lot of special characters, such as those pasted from Microsoft Word.  I have previously put in place code to replace many of those characters, but came across another one today that wasn't replaced.  I knew I needed a better solution.

On CFlib.org, I found a function called xmlFormat2 that smartly avoids maintaining a list of characters to replace, and just replaces all characters not in a list of "good" characters.  That makes sense.  And it works.

I was concerned, though, about the performance of the function, and I thought it could be done better.  Using the REMatch function (introduced in ColdFusion 8), I was able to make the function both simpler and much faster.  My tests so far have been limited, but it has so far handled everything I have thrown at it.  And here it is:

<cffunction name="XmlSafeText" hint="Replaces all characters that would break an xml file." returnType="string" output="false">       
    <cfargument name="txt" hint="String to format" type="string" required="true">
    <cfset var chars = "">
    <cfset var replaced = "">

    <!--- Use XmlFormat function first --->
    <cfset txt = XmlFormat(txt)>
    <!--- Get all other characters to replace. --->
    <cfset chars = REMatch("[^[:ascii:]]",txt)>
    <!--- Loop through characters and do replace. Maintain a list of characters already replaced to avoid duplicate work. --->
    <cfloop index="char" array="#chars#">
        <cfif ListFind(replaced,char) is 0>
            <cfset txt = Replace(txt,char,"&##" & asc(char) & ";","all")>
            <cfset replaced = ListAppend(replaced,char)>
        </cfif>
    </cfloop>

    <cfreturn txt>
</cffunction>

It should be possible to use it as a replacement for the built in XmlFormat function.  Let me know if you run into any problems with it.