mamexml.py

import xml.parsers.expat
import sys,gzip,string

_name = ''
_manf = ''
_year = 0
_desc = ''
_srcfile = ''
_tag = None
_cpus = []
_audio = []

# 3 handler functions
# <machine name="3bagflnz" sourcefile="aristmk4.c" cloneof="3bagflvt" romof="3bagflvt" sampleof="3bagflvt">
def start_element(name, attrs):
    global _name, _tag, _manf, _srcfile
    _tag = name
    if name == 'machine' and attrs['isdevice'] != 'yes':
        #if attrs.get('cloneof'):
        #    _name = None
        #    _tag = None
        #    return
        _name = attrs['name']
        _srcfile = attrs['sourcefile']
    elif name == 'chip':
        if attrs['type'] == 'cpu':
            if attrs['tag'] == 'maincpu':
                _cpus.append('*'+attrs['name'])
            else:
                _cpus.append("+"+attrs['name'])
        elif attrs['type'] == 'audio':
            name = attrs['name']
            if name != 'Speaker':
                _audio.append(name)

def end_element(name):
    _tag = None
    if name == 'machine' and _name:
        print('%s,"%s","%s","%s","%s","%s","%s"' % (_year,_name,_srcfile,_desc,_manf,' '.join(_cpus),' '.join(_audio)))
        _audio[:] = []
        _cpus[:] = []

def char_data(data):
    global _manf, _year, _desc
    data = data.strip()
    if len(data):
        if _tag == 'manufacturer':
            _manf = data
        elif _tag == 'year':
            _year = data
        elif _tag == 'description':
            _desc = data

p = xml.parsers.expat.ParserCreate()

p.StartElementHandler = start_element
p.EndElementHandler = end_element
p.CharacterDataHandler = char_data

#with gzip.open('allroms.xml.gz','r') as f:
with open('allroms.xml','rb') as f:
    p.ParseFile(f)