lxml/python reading xml with CDATA section

Question

In my xml I have a CDATA section. I want to keep the CDATA part, and then strip it. Can someone help with the following?

Default does not work:

$ from io import StringIO
$ from lxml import etree
$ xml = '<Subject> My Subject: 美海軍研究船勘查台海水文？ 船<![CDATA[&#xE9;]]>€ </Subject>'
$ tree = etree.parse(StringIO(xml))
$ tree.getroot().text
' My Subject: 美海軍研究船勘查台海水文？ 船&#xE9;€ '

This post seems to suggest that a parser option strip_cdata=False may keep the cdata, but it has no effect:

$ parser=etree.XMLParser(strip_cdata=False)
$ tree = etree.parse(StringIO(xml), parser=parser)
$ tree.getroot().text    
' My Subject: 美海軍研究船勘查台海水文？ 船é€ '

Using strip_cdata=True, which should be the default, yields the same:

$ parser=etree.XMLParser(strip_cdata=True)
$ tree = etree.parse(StringIO(xml), parser=parser)    
$ tree.getroot().text    
' My Subject: 美海軍研究船勘查台海水文？ 船é€ '

Answer 1

CDATA sections are not preserved in the text property of an element, even if strip_cdata=False is used when the XML content is parsed, as you have noticed. See https://lxml.de/api.html#cdata.

CDATA sections are preserved in these cases:

1. When serializing with tostring():

   print(etree.tostring(tree.getroot(), encoding="UTF-8").decode())
   

2. When writing to a file:

   tree.write("subject.xml", encoding="UTF-8")