SQL Server XML Data with CDATA Block - sql-server

Question

I am storing an XML payload in a SQL Server table that has a datatype of 'XML'. The data I am receiving has a section that is enclosed in a CDATA block. Here is an example:
<event>
<projectId>123456</projectId>
<eventTs>2018-01-04T13:07:23</eventTs>
<eventData>
<![CDATA[
<company>
<companyId>849</companyId>
<companyName>My Company Name</companyName>
<activeFlag>Y</activeFlag>
<timestamp>27-JUL-17</timestamp>
</company>
]]>
</eventData>
</event>
when this data lands in my table in the field that has a data type of 'XML' the 'CDATA' block is stripped out but then all of the "<" and ">" characters are escaped. Since those characters are escaped, XPATH queries on that data field no longer work. Is there any way around this behavior short of having to strip out the CDATA block before it is inserted/converted to an XML data type?
This is what the data looks like after being inserted into the XML datatype field:
<event>
<projectId>123456</projectId>
<eventTs>2018-01-04T13:07:23</eventTs>
<eventData>
<company>
<companyId>849</companyId>
<companyName>My Company Name</companyName>
<activeFlag>Y</activeFlag>
<timestamp>27-JUL-17</timestamp>
</company>
</eventData>
</event>

Answer 1

It is a very bad approach to store XML within a CDATA section. Everything within this block is just text. This special text looks like an XML, but you cannot query this XML to return the <companyName>.
Try this:
DECLARE #xml XML=
N'<event>
<projectId>123456</projectId>
<eventTs>2018-01-04T13:07:23</eventTs>
<eventData>
<![CDATA[
<company>
<companyId>849</companyId>
<companyName>My Company Name</companyName>
<activeFlag>Y</activeFlag>
<timestamp>27-JUL-17</timestamp>
</company>
]]>
</eventData>
</event>';
SQL Server's developer decided not even to support CDATA anymore. It will implicitly be taken away, while its content remains properly escaped. But you can read the content without problems:
SELECT #xml.value('(/event/eventData)[1]','nvarchar(max)');
The point is: This result looks like an XML, but - in order to use it like an XML - it must be casted.
This you could do to solve this:
DECLARE #innerXML XML=(SELECT CAST('<eventData>' + #xml.value('(/event/eventData)[1]','nvarchar(max)') + '</eventData>' AS XML));
SET #xml.modify('delete /event/eventData[1]');
SET #xml.modify('insert sql:variable("#innerXML") as last into /event[1]');
SELECT #xml;
Easy going:
As long as the incoming XML is a string (before you try to cast it to XML) you can just throw away the CDATA markers:
DECLARE #xmlString NVARCHAR(MAX)=
N'<event>
<projectId>123456</projectId>
<eventTs>2018-01-04T13:07:23</eventTs>
<eventData>
<![CDATA[
<company>
<companyId>849</companyId>
<companyName>My Company Name</companyName>
<activeFlag>Y</activeFlag>
<timestamp>27-JUL-17</timestamp>
</company>
]]>
</eventData>
</event>';
SELECT CAST(REPLACE(REPLACE(#xmlString,' <![CDATA[',''),']]>','') AS XML)