#!/usr/bin/env python3
"""Dump PARTOUT(4.0) pinball .DAT structure: groups, component type ids, float attribute ids."""
import struct, sys, collections

FIXED = {0: 2, 2: 2, 13: 0}  # partman::_field_size; everything else has a uint32 length prefix

def parse(path):
    d = open(path, 'rb').read()
    sig = d[0:21].split(b'\0')[0].decode()
    app = d[21:71].split(b'\0')[0].decode()
    desc = d[71:171].split(b'\0')[0].decode()
    filesize, ngroups, sizeofbody, unknown = struct.unpack_from('<iHiH', d, 171)
    full_tilt = (app == 'FullTilt')
    o = 183 + unknown
    groups = []
    for gi in range(ngroups):
        n = d[o]; o += 1
        g = {'i': gi, 'name': None, 'sv': None, 'attrs': [], 'types': []}
        for _ in range(n):
            t = d[o]; o += 1
            if t in FIXED:
                sz = FIXED[t]
            else:
                sz = struct.unpack_from('<I', d, o)[0]; o += 4
            payload = d[o:o+sz]
            g['types'].append(t)
            if t == 0 and sz == 2:
                g['sv'] = struct.unpack_from('<h', payload, 0)[0]
            elif t == 3:
                g['name'] = payload.split(b'\0')[0].decode('latin-1')
            elif t == 11 and sz >= 4:
                g['attrs'].append(int(struct.unpack_from('<f', payload, 0)[0]))
            o += sz
        groups.append(g)
    # If the walk desynced, every field after it is garbage. Catch it here, not in the output.
    assert sig == 'PARTOUT(4.0)RESOURCE', f'{path}: not a PARTOUT dat ({sig!r})'
    assert len(groups) == ngroups, f'{path}: parsed {len(groups)} groups, header says {ngroups}'
    assert o == len(d), f'{path}: ended at {o} of {len(d)} bytes — entry walk desynced'
    return {'sig': sig, 'app': app, 'desc': desc, 'ngroups': ngroups,
            'full_tilt': full_tilt, 'groups': groups}

def summarize(path):
    f = parse(path)
    svs = collections.Counter(g['sv'] for g in f['groups'] if g['sv'] is not None)
    attrs = collections.Counter(a for g in f['groups'] for a in g['attrs'])
    named = sum(1 for g in f['groups'] if g['name'])
    return f, svs, attrs, named

if __name__ == '__main__':
    results = {}
    for path in sys.argv[1:]:
        name = path.split('/')[-1]
        f, svs, attrs, named = summarize(path)
        results[name] = (f, svs, attrs)
        print(f"=== {name}  app={f['app']!r} desc={f['desc']!r} groups={f['ngroups']} named={named}")
        print(f"    component type ids ({len(svs)}): {sorted(svs)}")
        print(f"    float attr ids ({len(attrs)}): {sorted(attrs)}")
        print()

    if len(results) > 1:
        base = 'CADET.DAT' if 'CADET.DAT' in results else list(results)[0]
        bf, bsv, battr = results[base]
        for name, (f, svs, attrs) in results.items():
            if name == base:
                continue
            print(f"--- {name} vs {base}")
            print(f"    component types only in {name}: {sorted(set(svs) - set(bsv))}")
            print(f"    component types missing from {name}: {sorted(set(bsv) - set(svs))}")
            print(f"    attrs only in {name}: {sorted(set(attrs) - set(battr))}")
            print(f"    attrs missing from {name}: {sorted(set(battr) - set(attrs))}")
            bn = {g['name'] for g in bf['groups'] if g['name']}
            fn = {g['name'] for g in f['groups'] if g['name']}
            print(f"    shared component names: {len(bn & fn)} / {len(fn)} in {name}")
            print(f"    sample names only in {name}: {sorted(fn - bn)[:25]}")
            print()