parsing and modifying an xml file with CDATA sections

I would like to programmatically modify some XML files but I end up adding some modifications inadvertently. For example consider the following XML:

<?xml version="1.0" encoding="UTF-8"?>
<!-- A comment
-->
<abc:Tag xmlns:xs="http://www.w3.org/2001/XMLSchema" xmlns:abc="http://www.mycompany.com" xmlns:def="http://www.anothercompany.com">
  <abc:sometext oneattribute="Hello" anotherattribute="World">
Some random boring text.
</abc:sometext>
  <def:somecode>
    <![CDATA[
if a>=b:
print(a)
]]>
  </def:somecode>
</abc:Tag>

I am trying to add a simple a comment in the code included in the CDATA section. To do so I am using the following python script that manages to handle the namespaces correctly and add the string. However, the CDATA is lost in the output:

import sys
from lxml import etree as ET

xml_file = sys.argv[1]
tree = ET.parse(xml_file)
root = tree.getroot()
ns = {}
element_tree = ET.iterparse(xml_file, events=["start-ns"])
try:
    for event, (prefix, qualified_name) in element_tree:
        ET.register_namespace(prefix, qualified_name)
        ns[prefix] = qualified_name
except ET.ParseError as err:
    sys.exit(1)


for somecode in tree.findall('def:somecode', namespaces=ns):
    somecode.text = somecode.text + "# updated with a comment"

tree.write('output.xml',
    xml_declaration=True,
    encoding="UTF-8")

The resulting output is different than the input in two ways I didn't expect and don't know how to correct:

• Single quotes are replaced by double
• The code in CDATA is printed as normal text