Methods or approaches for designing systems based on specification in XML?

Question

Python

I am designing a python library for export\import\serialization\de-serialization of data in a given format. The data format is described by a fairly extensive specification based on XML\XSD, the data itself is also in XML. At first I thought to implement each entity as a separate class, but there are quite a few entities, moreover, some of the entities can be both nested and exist separately.
For example:

<data>
    <some_container>
        <some_verb>...</some_verb>
        <some_noun>
            <profile1>
                <address>...</address>
                <other_data>...</other_data>
            </profile1>
            ...
        </some_noun>
    </some_container>
</data>

I tried to use a couple of libraries that generate classes based on XSD, but it turned out to be too cumbersome and it was not realistic to use the generated code.
Tell me if there are any well-established approaches / methods for designing and implementing such tasks?
I will be glad to advice in the form of links to articles, best-practices, books, etc. Thanks in advance :)

Answer 1

sourceforge.net/projects/pyxb/
?

Answer 2

To fully work with XML, you will need about four libraries / classes
1. DOM
2. XSD
3. XSL
4. xPath
Sometimes xPath is included in DOM libraries, sometimes it exists separately.
DOM is used for import, XSD is immediately set on the loaded document to check the validity of the structure.
The export is usually done without any validation, or extra XSD validation can be done before the export.
Serialization and deserialization must be done by the DOM parser (nothing else).
XSL and xPath come in handy when you need to export to a non-original file format (HTML, XML of another structure, csv, rtf, and more), or import from arbitrary XML formats. The procedure is simple - we set the XSL transformer on the source document and get the desired output.
I can't recommend specific libraries because I don't work with python.
Select the best from the list