Files
firestorm/MW4COMPARE/tools/decompile/destroyed.py
T
83478b7666 Add MW4COMPARE: .mw4 decompiler toolchain and V4H comparison harness
Tooling built to recover editable source for six 'Mech chassis that exist
in the parallel FS_Build_V4H build but not in this repo. Reverse-engineers
every compiled record type in the .mw4 package format back to the .data /
.instance / .subsystems / .damage / .contents / .torso / .engine /
.armature sources the content pipeline consumes.

Nothing here is wired into the game build. It is a standalone analysis
harness run from Linux.

Package format
--------------
"#VBD" container. Directory records are [len][name][FILETIME][origSize]
[storedSize][offset], payload base at dword 0x0C. A record is stored raw
when storedSize == origSize, otherwise LZW (9->12-bit LSB-first codes,
256=clear, 257=EOF, dict from 258), per Database.cpp:451.

GameModel records are flat /Zp4 structs following the C++ inheritance
chain Entity(0) -> Mover(28) -> MWObject(80) -> Vehicle(664) -> Mech(756),
1636 bytes total. CreateMessage records follow Replicator -> Entity ->
Mover -> MWMover -> MWObject -> Vehicle -> Mech from start=16 (the
undeclared Connection__Message header), ending at 341 and padded to 344.

tools/decompile/
----------------
  datamap.py        header-driven layout engine; CHAIN + ANCHORS
                    {Vehicle:664, Mech:756} assert the struct offsets
  mw4msg.py         CreateMessage reader/walker
  data.py           .data      constants.py  define/table symbol resolution
  damage.py         .damage    contents.py   .contents
  smallmodel.py     .torso + .engine         instance.py  .instance
  armature.py / armature_parts.py  .armature + armaturedata/armaturevideo
  assembly.py       joint hierarchy renderer
  make_generic_doll.py  builds generic MFD/Radar damage dolls
  verify_*.py       per-type round-trip verifiers

Verified round-trip across all 64 shared chassis:
  .armature      2938/2976 pages     .subsystems  7579/7585 keys
  .data map      6071/6071 values    .data trip   8291/8306 keys
  .damage        6605/6605 keys      .contents    7480/7480 keys
  .torso+.engine 1280/1280 keys      .instance     896/896 keys, 64/64 pages
  armature_parts 1202/1202 .data, 1149/1202 .video

Layout-discovery lessons (documented in DECOMPILING.md)
-------------------------------------------------------
- Never let a field map be discovered by the values that verify it. A
  value-matching pass reported 4288/4288 while mis-assigning 34 keys. The
  map was rebuilt from header declaration order, anchored on uniquely
  resolved fields.
- Read the factory, not the data. 12 .data fields and 5 Torso fields are
  declared plain Stuff::Scalar but multiplied by Radians_Per_Degree in
  Mech_Tool.cpp:889 / Torso_Tool.cpp.
- Strip typedefs before walking a header. A stray `typedef int AttributeID;`
  masked a missing ClassID - two 4-byte errors cancelling out, caught only
  by the ANCHORS assertion.
- A verifier that silently narrows its own input reports success. Braced
  blocks must be hidden before splitting pages, replacing both CR and LF,
  because a `Shadow={...}` block contains a line reading `[shadow]` and
  splitlines() also splits on bare CR.
- NSWIZZLE is undefined, so the #else branch is live and orders members
  differently. bool is 1 byte; char x[MaxStringLength] is 256.
- V4H carries stale Mech IDs (their Atlas is 5, ours 6), so 64 of 65 shared
  chassis are off by one; --retarget-ids emits $(M_<Chassis>)/$(IDS_<Chassis>).

reports/ holds generated diffs. The two ~5 MB manifest-*.tsv intermediates
are gitignored; regenerate everything with run-comparison.sh.

Co-authored-by: Claude Opus 5 (Anthropic) <noreply@anthropic.com>
Co-authored-by: GitHub Copilot <copilot@github.com>
2026-08-08 16:47:57 -05:00

169 lines
6.8 KiB
Python

#!/usr/bin/env python3
"""
destroyed.py - regenerate the .data and .video of a *_destroyed mech variant.
python3 destroyed.py <mechDir> [--erf X.erf] [--obb X.obb] [-o outDir]
python3 destroyed.py --verify # regenerate all 89 of ours and diff
A destroyed variant is four files:
<stem>.data text, DeathEntity boilerplate
<stem>.video text, four-page render graph
<stem>.erf geometry, verbatim in the package
<stem>_SOLID.obb collision, verbatim in the package
The .erf and .obb come straight out of the package. The .data and .video are
compiled away (a 12-byte stub plus {Element}/{GameModel}, and a binary element
tree with embedded #FRE/#RLM blobs), so they are regenerated from a template.
That is safe here because the template is invariant. Across all 89 destroyed
variants in our own Content tree, every one of Class, OBBCollides, Collider,
CanBeShot, CanBeWalkedOn, VertexLighting, FaceLighting, LookupLighting,
LightMapLighting and CraterName is identical. The only things that vary are the
three filename references, and those are read from the files actually present
rather than guessed - which also preserves V4H's own `champion_stroyed` typo,
since that is the name their build really uses.
The .data and .video each appear in two forms in our tree, differing only by a
trailing blank line; the majority form is emitted.
"""
import sys, os, re, glob, argparse, collections
DATA_TEMPLATE = (
"[GameData]\r\n"
"Class=MechWarrior4::DeathEntity\r\n"
"SolidOBB={obb}\r\n"
"OBBCollides=false\r\n"
"Collider=false\r\n"
"CanBeShot=false\r\n"
"CanBeWalkedOn=false\r\n"
"VertexLighting=yes\r\n"
"FaceLighting=yes\r\n"
"LookupLighting=no\r\n"
"LightMapLighting=no\r\n"
"CraterName=crater1\r\n"
"\r\n"
"[Renderers]\r\n"
"VideoRenderer={video}\r\n"
"\r\n"
)
VIDEO_TEMPLATE = (
"[lod]\r\n"
"Type=ShapeComponent\r\n"
"Geometry={erf}\r\n"
"\r\n"
"[watcher]\r\n"
"Type=AttributeWatcherOfInt\r\n"
"Attribute=VisualRepresentation\r\n"
"SimulationShouldExecute=1\r\n"
"\r\n"
"[damageappearance]\r\n"
"Type=SwitchComponent\r\n"
"Input=Watcher\r\n"
"Child=LOD\r\n"
"\r\n"
"[locator]\r\n"
"Child=DamageAppearance\r\n"
"\r\n"
)
OUR_MECH_SOURCE = "/home/rich/Repositories/firestorm/Gameleap/mw4/Content/Mechs"
def find_parts(mech_dir, compiled_dir=None):
"""-> (stem, erfName, obbName). Looks in mech_dir, then the compiled tree."""
erfs = [f for f in os.listdir(mech_dir) if f.lower().endswith(".erf")]
obbs = [f for f in os.listdir(mech_dir) if f.lower().endswith(".obb")]
stem = None
if compiled_dir and os.path.isdir(compiled_dir):
datas = [f for f in os.listdir(compiled_dir)
if f.lower().endswith(".data")]
if datas:
stem = datas[0][:-len(".data")]
if stem is None and obbs:
stem = re.sub(r"_solid\.obb$", "", obbs[0], flags=re.I)
stem = re.sub(r"\.obb$", "", stem, flags=re.I)
return stem, (erfs[0] if erfs else None), (obbs[0] if obbs else None)
def generate(stem, erf, obb):
return (DATA_TEMPLATE.format(obb=obb, video=f"{stem}.video"),
VIDEO_TEMPLATE.format(erf=erf))
def verify():
ok = collections.Counter()
problems = []
for src_dir in sorted(glob.glob(OUR_MECH_SOURCE + "/*_[Dd]estroyed") +
glob.glob(OUR_MECH_SOURCE + "/*_DESTROYED")):
datas = glob.glob(src_dir + "/*.data")
videos = glob.glob(src_dir + "/*.video")
if not datas or not videos:
continue
stem = os.path.basename(datas[0])[:-len(".data")]
want_data = open(datas[0], "rb").read().decode("latin-1")
want_video = open(videos[0], "rb").read().decode("latin-1")
kv = dict(re.findall(r'^([A-Za-z0-9_]+)=([^\r\n]*)', want_data, re.M))
erf = re.search(r'Geometry=([^\r\n]*)', want_video).group(1).strip()
got_data, got_video = generate(stem, erf, kv["SolidOBB"].strip())
ok["files"] += 1
for what, want, got, exact_key, soft_key in (
(".data", want_data, got_data, "data_exact", "data_soft"),
(".video", want_video, got_video, "video_exact", "video_soft")):
if got == want:
ok[exact_key] += 1
elif got.rstrip("\r\n").lower() == want.rstrip("\r\n").lower():
# NotationFile compares with _stricmp throughout, and the sources
# themselves are inconsistent (one uses [renderers], others
# [Renderers]), so case and a trailing blank line are cosmetic.
ok[soft_key] += 1
else:
problems.append((os.path.basename(src_dir), what, want, got))
print(f"destroyed variants checked : {ok['files']}")
print(f" .data equivalent : {ok['data_exact'] + ok['data_soft']}"
f" (byte-exact {ok['data_exact']}, +{ok['data_soft']} case / trailing-blank-line only)")
print(f" .video equivalent : {ok['video_exact'] + ok['video_soft']}"
f" (byte-exact {ok['video_exact']}, +{ok['video_soft']} case / trailing-blank-line only)")
if problems:
print(f"\n{len(problems)} real differences:")
for name, what, want, got in problems[:6]:
print(f" {name} {what}")
for w, g in zip(want.splitlines(), got.splitlines()):
if w != g:
print(f" source={w!r}\n got ={g!r}")
def main():
ap = argparse.ArgumentParser(description=__doc__,
formatter_class=argparse.RawDescriptionHelpFormatter)
ap.add_argument("mechdir", nargs="?", help="the *_destroyed folder holding .erf/.obb")
ap.add_argument("--compiled", help="matching folder in the _compiled tree, for the stem")
ap.add_argument("-o", "--out", help="output folder (defaults to mechdir)")
ap.add_argument("--verify", action="store_true")
args = ap.parse_args()
if args.verify:
verify()
return
if not args.mechdir:
ap.error("give a *_destroyed folder, or --verify")
stem, erf, obb = find_parts(args.mechdir, args.compiled)
if not (stem and erf and obb):
sys.exit(f"{args.mechdir}: need a stem, an .erf and an .obb "
f"(got stem={stem!r} erf={erf!r} obb={obb!r})")
data, video = generate(stem, erf, obb)
out = args.out or args.mechdir
os.makedirs(out, exist_ok=True)
open(os.path.join(out, f"{stem}.data"), "wb").write(data.encode("latin-1"))
open(os.path.join(out, f"{stem}.video"), "wb").write(video.encode("latin-1"))
print(f"{os.path.basename(args.mechdir)}: {stem}.data + {stem}.video "
f"(SolidOBB={obb}, Geometry={erf})")
if __name__ == "__main__":
main()