pylhasa

pylhasa is a cross-platform Python wrapper for the LHA/LZH archive format. It vendors the liblhasa C sources and builds a CPython extension, producing wheels for Linux, macOS, and Windows.

Install

From PyPI:

pip install pylhasa

Wheels are built for Python 3.9+.

Usage

import pylhasa

archive = pylhasa.open("example.lha")
for entry in archive:
    print(entry.raw_path, entry.size)

# Read bytes directly (loads full file into memory)
payload = archive.read("hello.txt")

# Stream contents (incremental reads, avoids large memory usage)
entry = next(iter(archive))
with entry.open() as stream:
    chunk = stream.read(1024)

# Extract safely (default)
archive.extractall("out")
archive.close()

API overview

Top-level functions:

• pylhasa.open(path): open an archive from a filesystem path.
• pylhasa.open_bytes(data) / pylhasa.from_bytes(data): open from in-memory bytes.
• pylhasa.open_fileobj(fileobj, buffering=131072): open from a stream by spooling to a temp file.

Archive behavior:

• Archive is iterable; each item is an Entry.
• Archive.read(name_or_entry) returns the full bytes of a file entry.
• Entry.read() returns the full bytes for that entry (same as Archive.read(entry)).
• Archive.extract(name_or_entry, dest_dir, safe=True, allow_symlinks=False, verify_crc=True) extracts a single entry.
• Archive.extractall(dest_dir, safe=True, allow_symlinks=False, verify_crc=True) extracts all entries.

Entry behavior:

• Entry.open() returns a binary file-like object for streaming decompressed data.
• Entry.read() loads the full entry into memory in one call.
• Entry.read() reads the full decompressed bytes into memory.
• Entry.raw_path preserves the original path from the archive; Entry.safe_path is the sanitized path used for safe extraction.

Examples

See examples/ for runnable scripts:

• examples/list_entries.py
• examples/extract_all.py
• examples/stream_read.py
• examples/all_functions.py

In-memory / streaming

import pylhasa

# In-memory bytes
data = open("example.lha", "rb").read()
archive = pylhasa.open_bytes(data)
# or: archive = pylhasa.from_bytes(data)

# Streaming file-like object
with open("example.lha", "rb") as fp:
    archive = pylhasa.open_fileobj(fp, buffering=131072)

Safety notes

• Safe extraction is on by default. Unsafe paths raise UnsafePathError.
• Entry.raw_path preserves the original stored path (best-effort decoding).
• Entry.safe_path contains the sanitized path used for extraction when safe mode is enabled.
• Path traversal, absolute paths, Windows drive paths, and UNC paths are rejected when safe=True.
• Extraction verifies CRC by default; pass verify_crc=False to skip.

Exceptions

• PylhasaError: base exception
• BadArchiveError: malformed or unsupported archive
• UnsafePathError: unsafe entry path for extraction

Header metadata

Each Entry exposes the full parsed LHA header fields (for example header_level, os_type, extra_flags, Unix permissions, Windows timestamps, and raw_header_bytes). These are available for forensic and advanced use.

Time helper:

• Entry.datetime_utc() returns a best‑effort UTC datetime (prefers Windows FILETIME when present, otherwise Unix timestamp).

Compression support

The vendored liblhasa core supports common LHA/LZH compression methods including -lh1- through -lh7-, -lhd-, and LArc -lz* variants.

Warning (experimental): -lh2- and -lh3- support is best‑effort and under‑documented. Treat results with caution and validate against trusted tools when possible.

Directory entries (-lhd-) and symlinks do not carry file data; Archive.read() returns b\"\" for those entries.

Third-party licenses

This project vendors liblhasa. Its license is included at native/vendor/lhasa/COPYING.md and applies to the vendored sources.