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>
This commit is contained in:
2026-08-08 16:47:57 -05:00
co-authored by Claude Opus 5 GitHub Copilot
parent e088555b96
commit 83478b7666
60 changed files with 13442 additions and 0 deletions
+157
View File
@@ -0,0 +1,157 @@
#!/usr/bin/env python3
"""
restructure.py - reshape an extracted tree to match the repo's directory layout.
python3 restructure.py <extractedDir> [--apply]
Dry-run by default.
The extractor writes one directory per package, which is lossless but does not
look like the repo. This flattens it to the shape of Gameleap/mw4/Content:
<root>/Content/Mechs/Champion/champion.subsystems
<root>/Content/Maps/alpine02/...
<root>/Resource/Variants/<name>/... (variant records have no path)
<root>/Resource/Pilots/Tesla/options/...
<root>/_pkgroots/... (4-byte package-root pseudo-entries)
Path components are re-cased to match the repo wherever a counterpart exists, so
'mechs/longbow/longbow.data{element}' becomes
'Content/Mechs/Longbow/longbow.data{Element}'. Components with no counterpart
(the new chassis, for instance) keep the name the packer used.
REFUSES TO RUN if two packages would land on the same path with different bytes.
Flatten only trees where that has been checked - the pruned V4H tree has zero
collisions; our own full extraction has 60 (mission-local copies of shared
Culturals props) and must stay per-package.
Provenance is preserved in _layout.tsv: newPath, package, original entry path.
"""
import sys, os, hashlib, collections
REPO_CONTENT = "/home/rich/Repositories/firestorm/Gameleap/mw4/Content"
def build_case_map(root):
"""lowercased relative path -> the repo's actual spelling, for dirs and files."""
out = {}
for dp, dirs, files in os.walk(root):
rel = os.path.relpath(dp, root).replace("\\", "/")
base = "" if rel == "." else rel
for name in dirs + files:
r = f"{base}/{name}" if base else name
out[r.lower()] = r
return out
def recase(path, case_map):
"""Re-case each component against the repo, keeping unknown ones as-is."""
done = []
for part in path.split("/"):
probe = "/".join(done + [part]).lower()
canonical = case_map.get(probe)
done.append(canonical.rsplit("/", 1)[-1] if canonical else part)
return "/".join(done)
def package_of(rel):
parts = rel.split("/")
head = parts[0].lower()
if head in ("core", "props", "textures"):
return parts[0], "/".join(parts[1:])
if head in ("maps", "missions", "variants") and len(parts) > 2:
return "/".join(parts[:2]), "/".join(parts[2:])
if head == "pilots" and len(parts) > 3:
return "/".join(parts[:3]), "/".join(parts[3:])
return parts[0], "/".join(parts[1:])
def main():
if len(sys.argv) < 2:
sys.exit(__doc__)
root = sys.argv[1]
apply_ = "--apply" in sys.argv
print("indexing repo Content/ for canonical casing ...", flush=True)
case_map = build_case_map(REPO_CONTENT)
print(f" {len(case_map)} path components")
plan = [] # (oldRel, newRel, package, entry)
claims = collections.defaultdict(dict) # newRel -> md5 -> [package]
for dp, dirs, fs in os.walk(root):
if dp == root:
dirs[:] = [d for d in dirs if d not in ("_merged", "Content", "Resource", "_pkgroots")]
for f in fs:
old = os.path.relpath(os.path.join(dp, f), root).replace("\\", "/")
if old.startswith("_"):
continue
pkg, entry = package_of(old)
low = pkg.lower()
if low.startswith("variants/"):
new = f"Resource/{pkg}/{entry}"
elif low.startswith("pilots/"):
new = f"Resource/{pkg}/{entry}"
elif entry.lower().startswith("resource/"):
new = f"_pkgroots/{pkg}/{entry}"
else:
new = "Content/" + recase(entry, case_map)
plan.append((old, new, pkg, entry))
if new.startswith("Content/"):
with open(os.path.join(root, old), "rb") as fh:
h = hashlib.md5(fh.read()).hexdigest()
claims[new].setdefault(h, []).append(pkg)
clashes = {k: v for k, v in claims.items() if len(v) > 1}
dupes = {k: v for k, v in claims.items()
if len(v) == 1 and len(next(iter(v.values()))) > 1}
print(f"\n{len(plan)} files -> {len({n for _o, n, _p, _e in plan})} destinations")
print(f" identical copies from several packages : {len(dupes)}")
print(f" CONFLICTING copies (different bytes) : {len(clashes)}")
if clashes:
for k, v in list(clashes.items())[:10]:
print(f" {k} " + "; ".join(f"{h[:6]}={','.join(p)}" for h, p in v.items()))
sys.exit("\nrefusing to flatten: resolve the conflicts first")
sample = [(o, n) for o, n, _p, _e in plan if o != n][:8]
print("\nsample:")
for o, n in sample:
print(f" {o}\n -> {n}")
if not apply_:
print("\nDRY RUN - nothing changed. Re-run with --apply.")
return
import shutil
shutil.rmtree(os.path.join(root, "_merged"), ignore_errors=True)
moved, deduped = 0, 0
with open(os.path.join(root, "_layout.tsv"), "w", encoding="utf-8") as fh:
fh.write("newPath\tpackage\toriginalEntry\n")
for old, new, pkg, entry in sorted(plan):
src, dest = os.path.join(root, old), os.path.join(root, new)
os.makedirs(os.path.dirname(dest), exist_ok=True)
if os.path.exists(dest): # identical copy from another package
os.remove(src)
deduped += 1
else:
os.replace(src, dest)
moved += 1
fh.write(f"{new}\t{pkg}\t{entry}\n")
removed = 0
for dp, _dirs, _fs in os.walk(root, topdown=False):
if dp == root or not os.path.isdir(dp):
continue
if not os.listdir(dp):
os.rmdir(dp)
removed += 1
print(f"\nmoved {moved}, dropped {deduped} identical duplicates, "
f"removed {removed} empty directories")
print(f"provenance: {os.path.join(root, '_layout.tsv')}")
if __name__ == "__main__":
main()