Convert "mame -listxml" output to JSON

mamexmltojson

#!/usr/bin/python3
#
# Convert "mame -listxml" output to JSON.
#
# MAME's XML output is huge, so this dumps each complete
# <machine> element to keep memory usage bounded.
#
# Output can be piped to 'jq' to query machines.
#
# e.g. Select coin-op machines with trackball controls:
#
# mame -listxml | mamexmltojson | \
#   jq '.machine[] | select(
#     (.runnable != "no") and \
#     (.cloneof == null) and \
#     (.input.coins != null) and \
#     (.input.control | \
#       (type == "array" and \
#       any(.type | contains("trackball")))))'
#

import argparse
import json
import sys

from html.parser import HTMLParser

class MAMEParser(HTMLParser):

    LIST_ELEMENTS = set([
        # <machine> children
        'biosset',
        'rom',
        'disk',
        'device_ref',
        'sample',
        'chip',
        'display',
        'dipswitch',
        'configuration',
        'port',
        'adjuster',
        'feature',
        'device',
        'slot',
        'softwarelist',
        'ramoption',
        # <input> children
        'control',
        # <dipswitch> children
        'diplocation',
        'dipvalue',
        # <configuration> children
        'conflocation',
        'confsetting',
        # <port> children
        'analog',
        # <device> children
        'extension',
        # <slot>
        'slotoption',
    ])

    def reset(self):
        super().reset()
        self.tags = []
        self.stack = []
        self.machine = None
        self.indent = '  '
        self.indentlvl = 0

    def print(self, message):
        print((self.indent * self.indentlvl) + message)

    def dump_machine(self, sep=','):
        machine_dump = json.dumps(self.machine, indent=self.indent) + sep
        for line in machine_dump.split('\n'):
            self.print(line)

    def handle_starttag(self, tag, attrs):
        if tag == 'mame':
            self.print('{')
            self.indentlvl += 1
            for name,value in attrs:
                self.print('"{}": "{}",'.format(name, value))
            self.print('"machine": [')
            self.indentlvl += 1
        else:
            attrs = {name: value for name, value in attrs}
            if tag == 'machine':
                if self.machine:
                    self.dump_machine()
                self.machine = attrs
            else:
                parent = self.stack[-1]
                if tag in self.LIST_ELEMENTS:
                    # Special case: if attrs has only a 'name' key, append the value
                    if tag != 'ramoption' and len(attrs) == 1 and 'name' in attrs:
                        parent.setdefault(tag, []).append(attrs['name'])
                    else:
                        parent.setdefault(tag, []).append(attrs)
                elif attrs:
                    assert tag not in parent, 'unexpected duplicate <{}>'.format(tag)
                    parent[tag] = attrs
            self.tags.append(tag)
            self.stack.append(attrs)

    def handle_endtag(self, tag):
        if tag == 'mame':
            if self.machine:
                self.dump_machine(sep='')
            self.print(']')
            self.indentlvl -= 1
            self.print('}')
            self.indentlvl -= 1
        else:
            self.tags.pop()
            self.stack.pop()

    def handle_data(self, data):
        data = data.strip()
        # Ignore embedded DTD
        if self.tags and data:
            tag = self.tags[-1]
            # <ramoption> has both attributes and data
            if tag == 'ramoption':
                parent = self.stack[-1]
                parent['value'] = data
            else:
                parent = self.stack[-2]
                assert tag not in parent, 'unexpected duplicate <{}>'.format(tag)
                parent[tag] = data


if __name__ == '__main__':
    arg_parser = argparse.ArgumentParser(description='Convert MAME XML to JSON')
    arg_parser.add_argument(
        'file', nargs='?',
        type=argparse.FileType(),
        default=sys.stdin,
        help='File containing "mame -listxml" output')
    args = arg_parser.parse_args()

    parser = MAMEParser()
    for line in args.file:
        parser.feed(line)
    parser.close()