python - LXML kills my CDATA sections

Question

I'm batch-converting a lot of XML files, changing their character encodings to UTF-8:

with open(source_filename, "rb") as source:
    tree = etree.parse(source)

    with open(destination_filename, "wb") as destination:
        tree.write(destination, encoding="UTF-8", xml_declaration=True)

Unfortunately, it is destroying my CDATA sections and just escaping them instead.

Source:

<d><![CDATA[áìà??àù??éú ?ì?é áèà?ù? é??''? e?ù?à?ìèè <small><small>(ùí ?? è?)</small></small>

Destination:

<d>????????? ??? ????? ???''? ??????? &lt;small&gt;&lt;small&gt;(?? ?? ??)&lt;/small&gt;&lt;/small&gt;

Is there a setting which I can set which will tell it to leave my CDATA sections alone? I'm mainly using LXML to change the character encoding and to write the XML header properly.

Answer 1

Use the strip_cdata=False option:

import lxml.etree as etree
parser = etree.XMLParser(strip_cdata=False)
with open(source_filename, "rb") as source:
    tree = etree.parse(source, parser=parser)