Assuming you mean the EMBL-EBI Application XML, as produced by the EMBL-EBI Web Services for sequence similarity searches, then the XML Schema for the format (as referenced in the results) can be found at:
This includes documentation describing the contents of the various elements, and can be used with tools supporting parser generation from an XML schema.
As far as I am aware BioPython does not contain support for this format (it does support the NCBI BLAST XML and the plain text output formats, see http://biopython.org/DIST/docs/tutorial/Tutorial.html#htoc91), however Python contains support for handling XML documents (see 19. Structured Markup Processing Tools in the Python documentation), and there are many packages that provide additional capabilities (see https://wiki.python.org/moin/PythonXml). So developing your own code to extract the required information should not be too difficult.
Alternatively as Pierre Lindenbaum suggests you could develop an XML transformation using XSLT to convert the XML into something simpler which just contains the information you need (you might find the XSLT Tutorial and the XPath Tutorial from W3Schools helpful if you want to try this). Unfortunately the examples Pierre cites are based on the NCBI BLAST XML so you will not be able to use them directly, but they do illustrate the concept and may help when trying to work out some of the trickier aspects of the conversion.
Note: the EMBL-EBI Web Services for NCBI BLAST, PSI-BLAST and WU-BLAST provide the option of producing NCBI BLAST XML instead of EMBL-EBI Application XML, see the alignment format parameter for details.