Inserting CDATA section markers at random locations in a VARCHAR(MAX) column

I have a VARCHAR(Max) column, OrderDetails, that is hosting some XML-like data that's been around quite a bit longer than the XML column has been.

What I want to do is to clean up the existing data in OrderDetails to be XML compliant in such a way as to be easily visually interpreted by humans.

One of the steps that's needed to do this transformation is to wrap an element containing SQL code (<SQLCODE>) in a CDATA section so the SQL comparison operators don't crash the XML parser. In the examples below, the "<>" and "&" characters crash the parser.

The problem comes in when I want to insert the closing CDATA section marker. In addition to the varying quantity of text that precedes the SQLCODE element, the number of leading spaces and/or tabs, if any, that precedes the </SQLCODE> on the same line can vary with each row in the table. The format variations are due to changes in coding format standards over the years.

For the most part, I can rely on the </SQLCODE> closing tag to be on a line with only the leading whitespace and a trailing CRLF. Typically, 1 tab or 3 spaces are used for one level of indentation,

Once the data transition is complete, the OrderDetails column will be ported to a new XML column in the tOrders table.

---

Example 1, tabs as leading chars

<CABINET CabID="1941" Technician="Fred" TechID="831" />
   <MODEL>Aspen</MODEL>
   <SIZE>W18</SIZE>
   <!-- Get Tech's availability -->
   <SQLCODE>
      SELECT sum(AvailableHours) 
      FROM tSchedule 
      WHERE Booked <> 1
      AND TechID = 831
      AND Source <> 'T&G'
   </SQLCODE>
</CABINET>

Example 2, spaces as leading chars

<CABINET CabID="2977" Technician="Juan" TechID="740" />
   <MODEL>
   Ticonderoga
   </MODEL>
   <SIZE>B24-2</SIZE>
   <!-- Get Tech's availability -->
   <SQLCODE>
      SELECT sum(AvailableHours) 
      FROM tSchedule 
      WHERE Booked <> 1
      AND TechID = 740
   </SQLCODE>
</CABINET>

Example 3, mixed tabs and spaces as leading chars

<CABINET CabID="3291" Technician="Thomas" TechID="342" />
   <MODEL>SpecialOrder</MODEL>
   <SIZE>B48-4.2</SIZE>
   <CUTOUTS>3</CUTOUTS>
   <DRAWING>Cust002217\Job03291.dwg</DRAWING>
      <!-- Get Tech's availability -->
      <SQLCODE>
         SELECT sum(AvailableHours) 
         FROM tSchedule 
         WHERE Booked <> 1
         AND TechID = 342
         AND TimeBlock > 24
      </SQLCODE>
</CABINET>

---

For Example 2, say, what I can get now is

         <SQLCODE>
--<![CDATA[
            SELECT sum(AvailableHours) 
            FROM tSchedule 
            WHERE Booked <> 1
            AND TechID = 740
      --]]>
</SQLCODE>

This is functional, but not formatted properly. What I'd like to get is

         <SQLCODE>
--<![CDATA[
            SELECT sum(AvailableHours) 
            FROM tSchedule 
            WHERE Booked <> 1
            AND TechID = 740
--]]>
         </SQLCODE>

Or better yet,

         <SQLCODE>
            --<![CDATA[
            SELECT sum(AvailableHours) 
            FROM tSchedule 
            WHERE Booked <> 1
            AND TechID = 740
            --]]>
         </SQLCODE>

Currently I'm using this:

         -- 1. Insert the CDATA section start marker
         UPDATE tOrders
         SET OrderDetails = REPLACE(OrderDetails, 
'<SQLCODE>',
'<SQLCODE>
--<![CDATA[' )        
         where OrderID = 1604
         and OrderDetails not like '%<![CDATA[%'

         -- 2. Insert the CDATA section end marker
         UPDATE tOrders
         SET OrderDetails = REPLACE(OrderDetails, 
'</SQLCODE>',
'--]]>
</SQLCODE>')         
         where OrderID = 1604
         and OrderDetails not like '%<![CDATA[%'

Can you suggest changes to this code (or another method) to get to one of the desired outputs?