BT411/scratchpad/night9/res29_scan.py

"""Parse every type-29 DamageLookupTable stream STRAIGHT OUT of BTL4.RES bytes.

Audit tool for Cyd's hit-location artifact: zero port code involved -- the
stream grammar is the one byte-verified from the binary's ctors
(@0x49ea48 table / PieSlice / @0x49e5e4 leaf ReadEntries):

    Table { i32 layerCount; Layer[layerCount] }
    Layer { i32 rotateWithTorso; i32 sliceCount; Slice[sliceCount] }
    Slice { i32 nameLen; char name[nameLen]; u8 NUL; i32 entryCount;
            Entry { f32 cumulative; i32 zoneIndex } [entryCount] }

The scanner walks the whole file trying that grammar at every offset whose
u32 is a plausible layer count, with strict validation (printable names,
ascending cumulatives ending ~1.0, sane counts).  False positives cannot
survive the full-table parse.
"""
import struct, hashlib, json, sys

RES = r"C:\git\bt411\content\BTL4.RES"
d = open(RES, "rb").read()
N = len(d)

def u32(o): return struct.unpack_from("<i", d, o)[0]
def f32(o): return struct.unpack_from("<f", d, o)[0]

def parse_table(o0):
    o = o0
    layers = u32(o); o += 4
    if not (1 <= layers <= 12): return None
    out = []
    for _L in range(layers):
        if o + 8 > N: return None
        rot = u32(o); scount = u32(o + 4); o += 8
        if rot not in (0, 1) or not (1 <= scount <= 16): return None
        slices = []
        for _S in range(scount):
            if o + 4 > N: return None
            nl = u32(o); o += 4
            if not (1 <= nl <= 24) or o + nl + 1 > N: return None
            name = d[o:o + nl]
            if not all(32 <= b < 127 for b in name): return None
            if d[o + nl] != 0: return None
            o += nl + 1
            if o + 4 > N: return None
            ec = u32(o); o += 4
            if not (1 <= ec <= 16): return None
            ents = []
            prev = -1e-6
            for _E in range(ec):
                if o + 8 > N: return None
                cum = f32(o); zi = u32(o + 4); o += 8
                if not (0.0 < cum <= 1.0001) or cum < prev - 1e-5: return None
                if not (0 <= zi <= 63): return None
                ents.append((round(cum, 5), zi))
                prev = cum
            if abs(ents[-1][0] - 1.0) > 0.02: return None
            slices.append((name.decode(), ents))
        out.append((rot, slices))
    return out, o - o0

tables = []
o = 0
while o < N - 8:
    v = u32(o)
    if 1 <= v <= 12:
        r = parse_table(o)
        if r:
            tables.append((o, r[0]))
            o += r[1]
            continue
    o += 1

print("tables parsed from raw RES bytes: %d" % len(tables))

def canon(t):
    return json.dumps([[rot, [[n, e] for n, e in sl]] for rot, sl in t])

groups = {}
for off, t in tables:
    h = hashlib.md5(canon(t).encode()).hexdigest()[:8]
    groups.setdefault(h, []).append((off, t))

print("distinct by content: %d" % len(groups))
for h, g in groups.items():
    t = g[0][1]
    rots = "".join(str(r) for r, _ in t)
    sls = ",".join(str(len(sl)) for _, sl in t)
    print("  %s  copies=%d  layers=%d  rot=%s  slices=%s" %
          (h, len(g), len(t), rots, sls))

# ---- spot checks against the artifact's DATA (percent deltas, entry order) ----
def deltas(ents):
    out = []; prev = 0.0
    for cum, zi in ents:
        out.append(round(cum - prev, 4)); prev = cum
    return out

def find(pred, what):
    hits = [h for h, g in groups.items() if pred(g[0][1])]
    print("%-58s %s" % (what, hits if hits else "NOT FOUND"))
    return hits

print("\n--- artifact spot checks (band index 0 = first parsed = bottom) ---")
find(lambda t: len(t) == 7 and len(t[6][1]) == 8 and
     t[6][1][0][0] == "TopRight" and deltas(t[6][1][0][1]) == [0.6, 0.4],
     "LOK band6 TopRight 60/40 (missle-first)")
find(lambda t: len(t) == 7 and len(t[6][1]) == 8 and
     t[6][1][0][0] == "TopRight" and deltas(t[6][1][0][1]) == [0.4, 0.6],
     "THR band6 TopRight 40/60")
find(lambda t: len(t) == 7 and
     all(len(ents) == 1 and abs(ents[0][0] - 1.0) < .001
         for _n, ents in t[6][1]),
     "OWN band6 single-entry 100% cells")
find(lambda t: len(t) == 7 and len(t[6][1]) == 7,
     "BLH: 7-slice band 6")
find(lambda t: len(t) == 7 and len(t[4][1]) == 7,
     "VUL/BAT: 7-slice band 4")
find(lambda t: len(t) == 7 and len(t[5][1]) == 8 and
     t[5][1][1][0] == "FrontChest" and deltas(t[5][1][1][1]) == [0.1, 0.1, 0.1, 0.1, 0.5, 0.1],
     "AVA-A band5 FrontChest ..50% utorso")
find(lambda t: len(t) == 7 and len(t[6][1]) == 8 and
     t[6][1][0][0] == "TopRight" and deltas(t[6][1][0][1]) == [0.5, 0.15, 0.35],
     "AVA-B/VUL/SND2 band6 TopRight 50/15/35")
find(lambda t: len(t) == 7 and len(t[4][1]) == 8 and
     t[4][1][1][0] == "FrontWaist" and deltas(t[4][1][1][1]) == [0.3, 0.3, 0.4],
     "SND2 band4 FrontWaist 30/30/40")
find(lambda t: t[0][1][0][0] == "RightFoot" and deltas(t[0][1][0][1]) == [0.5, 0.3, 0.2],
     "band0 RightFoot 50/30/20 (common)")
find(lambda t: len(t) == 7 and len(t[3][1]) == 8 and
     t[3][1][0][0] == "RightHip" and deltas(t[3][1][0][1]) == [0.4, 0.3, 0.3],
     "BLH band3 RightHip 40/30/30")
find(lambda t: len(t) == 7 and len(t[4][1]) == 7 and
     t[4][1][0][0] == "RightWaist" and deltas(t[4][1][0][1]) == [0.15, 0.1, 0.25, 0.25, 0.1, 0.15],
     "VUL band4 RightWaist 15/10/25/25/10/15")

# twist-pattern census
pats = {}
for h, g in groups.items():
    rots = "".join(str(r) for r, _ in g[0][1])
    pats.setdefault(rots, []).append(h)
print("\nrot patterns:", {k: len(v) for k, v in pats.items()})