pylibmspack 0.2.0

In-process libmspack bindings for Microsoft CAB, CHM, SZDD, and KWAJ files

pylibmspack

pylibmspack provides in-process Python bindings to libmspack for reading and extracting Microsoft CAB, CHM, SZDD, and KWAJ files.

Install

pip install pylibmspack

Supports Python 3.9+

Usage

from pylibmspack import CabArchive

cab = CabArchive("example.cab")
print(cab.files())

print(cab.read("hello.txt"))

cab.extract("hello.txt", "./out")

cab.extract_all("./out")

In-memory usage

from pylibmspack import CabArchive

data = open("example.cab", "rb").read()
cab = CabArchive.from_bytes(data)

info = cab.info()
print(info["files_count"], info["flags"])

payload = cab.read("hello.txt")

Safe vs raw extraction

from pylibmspack import CabArchive, CabPathTraversalError

cab = CabArchive("example.cab")
try:
    cab.extract_all("./out", safe=True)
except CabPathTraversalError as exc:
    print("Blocked unsafe path:", exc)

# Raw extraction (no safety checks)
cab.extract_all_raw("./out-raw")

CHM extraction

from pylibmspack import ChmArchive

chm = ChmArchive("manual.chm")
print(chm.info())
print(chm.files())

data = chm.read("index.html")
chm.extract_all("./chm-out")

CHM from bytes

from pylibmspack import ChmArchive

data = open("manual.chm", "rb").read()
chm = ChmArchive.from_bytes(data)
print(chm.files())

SZDD extraction

from pylibmspack import SzddFile

szdd = SzddFile("readme.tx_")
print(szdd.info())

payload = szdd.read()
szdd.extract("./out")

SZDD from bytes

from pylibmspack import SzddFile

data = open("readme.tx_", "rb").read()
szdd = SzddFile.from_bytes(data, name="readme.tx_")
print(szdd.info())

KWAJ extraction

from pylibmspack import KwajFile

kwj = KwajFile("setup.kwj")
print(kwj.info())
data = kwj.read()
kwj.extract("./out")

KWAJ from bytes

from pylibmspack import KwajFile

data = open("setup.kwj", "rb").read()
kwj = KwajFile.from_bytes(data, name="setup.kwj")
print(kwj.info())

Multi-cabinet sets

from pylibmspack import CabArchive

cab = CabArchive("part1.cab")
info = cab.info()

if info["has_next"]:
    print("Next cabinet:", info["next_cabinet"])
    print("Disk label:", info["next_disk"])

FAQ / troubleshooting

Why do I get CabPathTraversalError?
The archive contains absolute paths or .. segments. Use safe=False only if you trust the archive contents.

Can I read from bytes instead of a file path?
Yes. Use CabArchive.from_bytes(data) and then call files(), read(), or info().

Why does extraction fail with CabDecompressionError?
The CAB may be corrupt, truncated, or uses an unsupported compression method.

API reference

CabArchive(path: str)

Open a CAB archive on disk.

CabArchive.files() -> list[CabFileInfo]

Return metadata for each member as a CabFileInfo TypedDict. Each entry includes:

• name (str)
• size (int)
• dos_date (int)
• dos_time (int)
• date_y / date_m / date_d (int)
• time_h / time_m / time_s (int)
• datetime_utc (str, ISO 8601)
• attrs (int)
• is_readonly / is_hidden / is_system / is_archive (bool)
• folder_index (int)
• offset (int)
• compression (str: none, mszip, quantum, lzx)
• has_prev / has_next (bool)
• prev_cabinet / next_cabinet (str | None)
• cabinet_set_id / cabinet_set_index (int | None)

CabArchive.read(name: str, , max_size: int = 2561024*1024) -> bytes

Extract a member and return its bytes. Enforces a max_size limit and uses safe path validation.

CabArchive.extract(name: str, dest_dir: str, *, safe: bool = True) -> str

Extract a member to disk and return the output path. When safe=True, absolute paths and traversal are rejected.

CabArchive.extract_all(dest_dir: str, *, safe: bool = True) -> list[str]

Extract all members to disk and return output paths.

CabArchive.extract_raw(name: str, dest_dir: str) -> str

Extract a member using the raw path (no safety checks).

CabArchive.extract_all_raw(dest_dir: str) -> list[str]

Extract all members using raw paths (no safety checks).

CabArchive.from_bytes(data: bytes) -> CabArchive

Create an archive backed by in-memory bytes instead of a file path.

CabArchive.info() -> CabInfo

Return parsed CAB header metadata. The CabInfo dict includes:

• filename (str | None)
• base_offset (int)
• length (int)
• set_id (int)
• set_index (int)
• header_resv (int)
• flags (int)
• has_prev / has_next (bool)
• prev_cabinet / next_cabinet (str | None)
• prev_disk / next_disk (str | None)
• files_count (int)
• folders_count (int)

ChmArchive(path: str)

Open a CHM archive on disk.

ChmArchive.files(*, include_system: bool = True) -> list[ChmFileInfo]

Return metadata for each member as a ChmFileInfo TypedDict. Each entry includes:

• name (str)
• size (int)
• offset (int)
• section_id (int)
• section (str: uncompressed, mscompressed, unknown)
• is_system (bool)

ChmArchive.read(name: str, , max_size: int = 2561024*1024) -> bytes

Extract a member and return its bytes.

ChmArchive.extract(name: str, dest_dir: str, *, safe: bool = True) -> str

Extract a member to disk and return the output path.

ChmArchive.extract_all(dest_dir: str, *, safe: bool = True, include_system: bool = True) -> list[str]

Extract all members to disk and return output paths.

ChmArchive.extract_raw(name: str, dest_dir: str) -> str

Extract a member using the raw path (no safety checks).

ChmArchive.extract_all_raw(dest_dir: str, *, include_system: bool = True) -> list[str]

Extract all members using raw paths (no safety checks).

ChmArchive.info() -> ChmInfo

Return parsed CHM header metadata. The ChmInfo dict includes:

• filename (str | None)
• length (int)
• version (int)
• timestamp (int)
• language (int)
• dir_offset (int)
• num_chunks (int)
• chunk_size (int)
• density (int)
• depth (int)
• index_root (int)
• first_pmgl (int)
• last_pmgl (int)
• files_count (int)
• sysfiles_count (int)

ChmArchive.from_bytes(data: bytes) -> ChmArchive

Create an archive backed by in-memory bytes instead of a file path.

SzddFile(path: str)

Open a SZDD-compressed file on disk.

SzddFile.info() -> SzddInfo

Return parsed SZDD header metadata. The SzddInfo dict includes:

• format_id (int)
• format (str: normal, qbasic, unknown)
• length (int)
• missing_char (int)
• missing_char_str (str)
• suggested_name (str)

SzddFile.read(, max_size: int = 2561024*1024) -> bytes

Decompress and return the file contents.

SzddFile.extract(dest_dir: str, *, safe: bool = True, out_name: str | None = None) -> str

Decompress to disk and return the output path.

SzddFile.extract_raw(dest_dir: str, *, out_name: str | None = None) -> str

Decompress using raw (unsafe) path handling.

SzddFile.from_bytes(data: bytes, *, name: str = "memory.sz_") -> SzddFile

Create a SZDD reader backed by in-memory bytes.

KwajFile(path: str)

Open a KWAJ-compressed file on disk.

KwajFile.info() -> KwajInfo

Return parsed KWAJ header metadata. The KwajInfo dict includes:

• comp_type (int)
• compression (str: none, xor, szdd, lzh, mszip, unknown)
• data_offset (int)
• headers (int)
• length (int)
• filename (str | None)
• extra_length (int)
• extra (bytes | None)
• has_length / has_filename / has_fileext / has_extra (bool)

KwajFile.read(, max_size: int = 2561024*1024) -> bytes

Decompress and return the file contents.

KwajFile.extract(dest_dir: str, *, safe: bool = True, out_name: str | None = None) -> str

Decompress to disk and return the output path.

KwajFile.extract_raw(dest_dir: str, *, out_name: str | None = None) -> str

Decompress using raw (unsafe) path handling.

KwajFile.from_bytes(data: bytes, *, name: str = "memory.kwj") -> KwajFile

Create a KWAJ reader backed by in-memory bytes.

Exceptions

All errors derive from MspackError:

• MspackError
• MspackFormatError
• MspackDecompressionError
• MspackPathTraversalError
• CabError / CabFormatError / CabDecompressionError / CabPathTraversalError
• ChmError / ChmFormatError / ChmDecompressionError / ChmPathTraversalError
• SzddError / SzddFormatError / SzddDecompressionError / SzddPathTraversalError
• KwajError / KwajFormatError / KwajDecompressionError / KwajPathTraversalError

Safe extraction

By default, extract() and extract_all() reject:

• absolute paths (/, \, drive letters, UNC paths)
• path traversal (.. after normalization)
• mixed or odd separators (/ and \ are normalized)

Use safe=False to allow the original paths.

Build from source

This project uses setuptools and builds a shared libmspack that is bundled into wheels. A pinned libmspack source tarball is included under pylibmspack/vendor/ and used for offline builds (SHA-256 verified).

python -m pip install -U pip setuptools wheel
python -m pip install -e .

If you want to supply a local tarball, pass --tarball to scripts/build_libmspack.py. To allow a network download during builds, set PYLIBMSPACK_ALLOW_DOWNLOAD=1 (disabled by default).

CHM test fixture

The CHM tests use the redistributable fixture at tests/fixtures/sample.chm (NSIS documentation under the zlib/libpng license).

Licensing

• pylibmspack code is MIT licensed.
• Wheels bundle libmspack under LGPL-2.1. The corresponding libmspack source tarball is included under pylibmspack/vendor/. You may replace the shared library inside pylibmspack/.libs with a compatible build.

See THIRD_PARTY_LICENSES/LGPL-2.1.txt and NOTICE for details.