Fix VMDK descriptor encoding

dissect/hypervisor/disk/vmdk.py

@@ -54,7 +54,7 @@ def __init__(self, fh: BinaryIO | Path | str | list[BinaryIO | Path | str]):
             if magic == b"# Di" and len(fhs) == 1:
                 # Try reading the disk files from this descriptor
                 # Otherwise we assume that the other file handles are the appropriate disks
-                self.descriptor = DiskDescriptor.parse(fh.read().decode())
+                self.descriptor = DiskDescriptor.parse_bytes(fh.read())
 
                 if self.descriptor.attr["parentCID"] != "ffffffff":
                     self.parent = open_parent(path.parent, self.descriptor.attr["parentFileNameHint"])
@@ -175,7 +175,7 @@ def __init__(
                 if self.header.descriptor_size > 0:
                     fh.seek(self.header.descriptor_offset * SECTOR_SIZE)
                     descriptor_buf = fh.read(self.header.descriptor_size * SECTOR_SIZE)
-                    self.descriptor = DiskDescriptor.parse(descriptor_buf.split(b"\x00", 1)[0].decode())
+                    self.descriptor = DiskDescriptor.parse_bytes(descriptor_buf.split(b"\x00", 1)[0])
 
             elif self.header.magic == COWD_MAGIC:
                 self._grain_directory_size = self.header.num_grain_directory_entries
@@ -453,6 +453,16 @@ def __init__(
         self.sectors = sectors
         self.raw = raw_config
 
+    @classmethod
+    def parse_bytes(cls, vmdk_config: bytes) -> DiskDescriptor:
+        descriptor = cls.parse(vmdk_config.decode(errors="replace"))
+        if encoding := descriptor.attr.get("encoding"):
+            try:
+                descriptor = cls.parse(vmdk_config.decode(encoding))
+            except (LookupError, UnicodeDecodeError):
+                pass
+        return descriptor
+
     @classmethod
     def parse(cls, vmdk_config: str) -> DiskDescriptor:
         """Return :class:`DiskDescriptor` based on the provided ``vmdk_config``.

tests/disk/test_vmdk.py

@@ -30,6 +30,27 @@ def test_vmdk_sesparse() -> None:
         assert vmdk.read(0x1000000) == b"a" * 0x1000000
 
 
+def test_vmdk_descriptor_declared_encoding(tmp_path: Path) -> None:
+    extent_name = "CentOS 64 副本-flat.vmdk"
+    descriptor_path = tmp_path / "disk.vmdk"
+    (tmp_path / extent_name).write_bytes(b"\x00" * 512)
+    descriptor_path.write_bytes(
+        (
+            "# Disk DescriptorFile\n"
+            "version=1\n"
+            'encoding="GBK"\n'
+            "CID=369b13fd\n"
+            "parentCID=ffffffff\n"
+            f'RW 1 FLAT "{extent_name}" 0\n'
+        ).encode("gbk")
+    )
+
+    vmdk = VMDK(descriptor_path)
+
+    assert vmdk.descriptor.extents[0].filename == extent_name
+    assert vmdk.size == 512
+
+
 @pytest.mark.parametrize(
     ("extent_description", "expected_extents"),
     [