While working on issues bpo-6686 and bpo-9371 I realized that the solution to both involved the use of Sax lexical handlers. Unfortunately, the Python SAX parser does not expose these handlers to the end user, however, support is available in expatreader.py and pyexpat.c.

What is missing is the LexicalHandler class which works the same way as the ContentHandler does for the more common handlers and provides a subclassable interface for these handlers. This class is present in the Java implementation of SAX2 and was also present in Pyxml.

I have already verified privately that this class works exactly as expected and I am working on a PR which will provide the class along with supporting unit tests that verify that the lexical handlers work.

New changeset e28b8c93878072dc02b116108ef5443084290d47 by Zackery Spytz in branch 'master':
bpo-35018: Sax parser should provide user access to lexical handlers (GH-20958)
https://github.com/python/cpython/commit/e28b8c93878072dc02b116108ef5443084290d47