Merge pull request #150 from Windows81/csgmdl5

feat: implement CSGMDL5

Author: Windows81

Source/assets/serialisers/csg/__init__.py

@@ -1,77 +1,5 @@
-import functools
-import hashlib
-import collections.abc
-from typing import Any, Never
-OBFUSCATION_NOISE_CYCLE_XOR = bytes([
-    86, 46, 110, 88, 49, 32, 48, 4, 52, 105, 12, 119, 12, 1, 94, 0, 26, 96, 55, 105, 29, 82, 43, 7, 79, 36, 89, 101, 83, 4, 122,
-])
-
-INT_SIZE = 4
-
-LCM_A = 0b0000_0011_0100_0011_1111_1101
-LCM_C = 0b0010_0110_1001_1110_1100_0011
-
-
-def lcm_rand() -> collections.abc.Generator[int, Any, Never]:
-    s = 0b0000_0000_0000_0101_0011_1001  # 1337
-    while True:
-        s = s * LCM_A + LCM_C
-        yield (s >> 16) & 0x7FFF
-
-
-saltSize: int = 0x10
-hashSize: int = 0x10
-
-
-def createHash(data: bytes, saltIn: bytes = b'rfd') -> str:
-    '''
-    TODO: decide whether we should `jmp` this function in the EXEs or to actually use `createHash` and have the client validate the hash.
-    '''
-    verticesSize: int = 0  # vertices.size() * sizeof(CSGVertex);
-    indicesSize: int = 0  # indices.size() * sizeof(unsigned int);
-    buffSize: int = verticesSize + indicesSize + saltSize
-
-    salt = saltIn.rjust(16)
-    byteBuffer = list(data+salt)
-
-    # size_t copyOffset = 0;
-    # memcpy(&byteBuffer[copyOffset], &vertices[0], verticesSize);
-
-    # copyOffset += verticesSize;
-    # memcpy(&byteBuffer[copyOffset], &indices[0], indicesSize);
-
-    # copyOffset += indicesSize;
-    # memcpy(&byteBuffer[copyOffset], salt.c_str(), salt.size());
-
-    randGen = lcm_rand()
-    for i in range(buffSize):
-        j = next(randGen) % buffSize
-        byteBuffer[i], byteBuffer[j] = byteBuffer[j], byteBuffer[i]
-
-    hashlib.md5(bytes(byteBuffer))
-    # boost::scoped_ptr<RBX::MD5Hasher> hasher(RBX::MD5Hasher::create());
-    # hasher->addData((const char*)&byteBuffer[0], byteBuffer.size());
-
-    # memcpy(&hash[0], hasher->toString().c_str(), hashSize);
-    # memcpy(&hash[hashSize], salt.c_str(), saltSize);
-
-    # std::string hashStr(&hash[0], hashSize + saltSize);
-
-    return hashStr
-
-
-@functools.cache
-def xor_encrypt(code: bytes, key=OBFUSCATION_NOISE_CYCLE_XOR, offset: int = 0) -> bytes:
-    l = len(key)
-    return bytes(
-        c ^ key[i % l]
-        for i, c in enumerate(code, offset)
-    )
-
-
-@functools.cache
-def get_header(prefix: bytes, version: int) -> bytes:
-    return prefix + version.to_bytes(length=INT_SIZE, byteorder='little')
+from .util import INT_SIZE, CSG_HEADER
+from . import csgmdl5
 
 
 def replace_header_version(data: bytes, versioned_header: bytes, from_version: int, to_version: int) -> bytes:
@@ -92,33 +20,28 @@ def replace_header_version(data: bytes, versioned_header: bytes, from_version: i
     ])
 
 
-def splice_without(data: bytes, fr: int, ln: int) -> bytes:
+def splice_without_middle_elements(data: bytes, fr: int, ln: int) -> bytes:
     return b''.join([
         data[:fr],
         data[fr+ln:],
     ])
 
 
-HEADER_CSG2 = xor_encrypt(get_header(b'CSGMDL', 2))
-HEADER_CSG4 = xor_encrypt(get_header(b'CSGMDL', 4))
-HEADER_CSG5 = xor_encrypt(get_header(b'CSGMDL', 5))
-HEADER_CSGPHYS5 = get_header(b'CSGPHS', 5)
-HEADER_CSGPHYS6 = get_header(b'CSGPHS', 6)
-HEADER_CSGPHYS7 = get_header(b'CSGPHS', 7)
-
-
 def parse(data: bytes) -> bytes | None:
-    if data.startswith(HEADER_CSG4):
-        return replace_header_version(data, HEADER_CSG4, 4, 2)
+    if data.startswith(CSG_HEADER.MDL4.value):
+        return replace_header_version(data, CSG_HEADER.MDL4.value, 4, 2)
 
-    if data.startswith(HEADER_CSGPHYS5):
+    elif data.startswith(CSG_HEADER.MDL5.value):
+        return csgmdl5.convert_to_csgmdl2(data)
+
+    elif data.startswith(CSG_HEADER.PHS5.value):
         '''
         CSGPHYS5 is identical in data format to CSGPHYS3.
         https://github.com/krakow10/rbx_mesh/blob/d10bcdf727dd9c2504560189a5cb106aa9107ec5/src/physics_data.rs#L71
         '''
-        return replace_header_version(data, HEADER_CSGPHYS5, 5, 3)
+        return replace_header_version(data, CSG_HEADER.PHS5.value, 5, 3)
 
-    if data.startswith(HEADER_CSGPHYS6):
+    elif data.startswith(CSG_HEADER.PHS6.value):
         '''
         Why 40 bytes?
         ```rs
@@ -135,19 +58,29 @@ def parse(data: bytes) -> bytes | None:
         CSGPHYS6 and CSGPHYS7 both contain `PhysicsInfo` structs, which as above indicate a length of 40 bytes.
         https://github.com/krakow10/rbx_mesh/blob/d10bcdf727dd9c2504560189a5cb106aa9107ec5/src/physics_data.rs#L8-L16
         '''
-        return splice_without(
-            replace_header_version(data, HEADER_CSGPHYS6, 6, 3),
-            len(HEADER_CSGPHYS6), 40,
+        return splice_without_middle_elements(
+            replace_header_version(data, CSG_HEADER.PHS6.value, 6, 3),
+            len(CSG_HEADER.PHS6.value), 40,
         )
 
-    if data.startswith(HEADER_CSGPHYS7):
+    elif data.startswith(CSG_HEADER.PHS7.value):
         '''
         Why 41 bytes in CSGPHYS7?
         +40: `PhysicsInfo`, as per above.
         + 1: the mysterious magic number `03` (one byte) that takes place after the versioned header.
         https://github.com/krakow10/rbx_mesh/blob/d10bcdf727dd9c2504560189a5cb106aa9107ec5/src/physics_data.rs#L54
         '''
-        return splice_without(
-            replace_header_version(data, HEADER_CSGPHYS7, 7, 3),
-            len(HEADER_CSGPHYS7), 41,
+        return splice_without_middle_elements(
+            replace_header_version(data, CSG_HEADER.PHS7.value, 7, 3),
+            len(CSG_HEADER.PHS7.value), 41,
+        )
+
+    elif data.startswith(CSG_HEADER.PHS8.value):
+        # TODO: implement CSGPHS8; returns an empty CSPHS object for now.
+        return (
+            b'CSGPHS\x03\x00\x00\x00' +
+            b'\x10\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\x10\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\x80\x3F' +
+            b'\x00\x00\x00\x00' +
+            b'\x04\x00\x00\x00' +
+            b'\x00\x00\x00\x00'
         )