#!/usr/bin/env python3
# Print heap offsets of the entries of a xar archive (InstallAssistant.pkg) as JSON,
# so SharedSupport.dmg can be exposed to the VM as a raw disk without extracting 18 GB.
import json
import struct
import sys
import xml.etree.ElementTree as ET

toc_xml, archive = sys.argv[1], sys.argv[2]
with open(archive, 'rb') as f:
    magic, hsize, ver, toc_c, toc_u, cksum = struct.unpack('>4sHHQQI', f.read(28))
assert magic == b'xar!', 'not a xar archive'
heap = hsize + toc_c
out = {}
for entry in ET.parse(toc_xml).getroot().iter('file'):
    data = entry.find('data')
    if data is None:
        continue
    out[entry.findtext('name')] = {
        'offset': heap + int(data.findtext('offset')),
        'length': int(data.findtext('length')),
        'size': int(data.findtext('size')),
        'encoding': data.find('encoding').get('style'),
    }
json.dump(out, sys.stdout, indent=2)