I'm looking for a fast Python pull parser (maybe some equivalent of PHP's XMLReader?) - is there something like this in Python? It's really a key feature that it would be a pull parser, because I'm gonna process really big xml files...
PHP's XMLReader is nothing but a SAX parser ( http://en.wikipedia.org/wiki/Simple_API_for_XML ).
Python supports SAX parsing. There are many tutorials available, like this one: http://www.devarticles.com/c/a/XML/Parsing-XML-with-SAX-and-Python/
How about the built-in ones python has?
http://docs.python.org/library/xml.dom.html ("full DOM implementation")
and http://docs.python.org/library/xml.dom.minidom.html ("lightweight DOM implementation")