How can I correlate xml that contains CDATA?

To handle XML containing CDATA using %XML.Reader or similar tools, you can follow these solutions:

1. If the CDATA section contains XML or complex data, you can treat it as a raw string and then process it further once extracted. To do this:

   • Correlate the parent element containing the CDATA and retrieve it as a string.
   • Use additional XML parsing tools to process the data within the CDATA separately. For instance:

     CLASS MyClass Extends %RegisteredObject
       Property RawXML As %String;  // for CDATA content
     
       Method ProcessCDATA() As %Status {
         // parsing code for RawXML
       }
     

2. Modify your XML-enabled classes to use the CONTENT = "MIXED" property if the structure allows. This enables the content of the CDATA to be read as raw XML. Note, however, that CONTENT = "MIXED" cannot generally be applied when the property is referencing another class [1].

3. If you have cases where specific outputs are expected from the CDATA, you can:

   • Create a custom datatype that processes the XML content of the CDATA after correlation. The datatype would have methods to parse the CDATA into an XML structure directly [2].

These approaches work when handling variations in the data contained within CDATA blocks. Where structured object generation would be useful, specific parsing tailored to the content type would follow afterward.

Sources:

You would probably be better served using XPath instead of the XML reader. When you evaluate and XPath expression, it will recognize the CDATA and parse it correctly, returning either a value or another XML object.

Assuming, your XML stream never has more  then MAXSTRING characters AND if you can live with a "quick-and-dirty" solution then try this one

Class DC.XML.abc Extends (%RegisteredObject, %XML.Adaptor)
{
Property def As%String;Property ghi As%String;Property jkl As jkl;ClassMethod Test(str)
{
    if$isobject(str) { set:str.Rewind() xml=str.Read(str.Size) } else { set xml=str }
    for{set i=$f(xml,"<![CDATA[",$g(i)) q:'i
        setj=$f(xml,"]]>",i) zt:'j"XMLE"set$e(xml,j-3,j-1)="", $e(xml,i-9,i-1)="", i=i-9
    }
    
    set rdr=##class(%XML.Reader).%New()
    if 'rdr.OpenString(xml) write"OpenErr",! quitdo rdr.Correlate("abc","DC.XML.abc")
    while rdr.Next(.abc,.st) { zzdo abc }
}
}

Class DC.XML.jkl Extends (%RegisteredObject, %XML.Adaptor)
{
Property mno As mno;
}

Class DC.XML.mno Extends (%RegisteredObject, %XML.Adaptor)
{
Property pqr As%String;Property stu As%String;
}

And some tests...

set s1="<abc><def>010203</def><ghi>040506</ghi><jkl><mno><pqr>070809</pqr><stu>101112</stu></mno></jkl></abc>"set s2="<abc><def>010203</def><ghi>040506</ghi><jkl><![CDATA[<mno><pqr>070809</pqr><stu>101112</stu></mno>]]></jkl></abc>"do##class(DC.XML.abc).Test(s1)
def................................: 010203
ghi................................: 040506
jkl.mno.pqr........................: 070809
jkl.mno.stu........................: 101112do##class(DC.XML.abc).Test(s2)
def................................: 010203
ghi................................: 040506
jkl.mno.pqr........................: 070809
jkl.mno.stu........................: 101112

Note: the above ZZDO command takes an oref as argument and prints it, you can replace it with a simple zw oref.