Microsoft SEAL For Python

Microsoft SEAL is an easy-to-use open-source (MIT licensed) homomorphic encryption library developed by the Cryptography Research group at Microsoft.

pybind11 is a lightweight header-only library that exposes C++ types in Python and vice versa, mainly to create Python bindings of existing C++ code.

This is a python binding for the Microsoft SEAL library.

Contents

• Build
• Note
  • Serialize
  • Other
• FAQ

Build

• Linux

  Clang++ (>= 5.0) or GNU G++ (>= 6.0), CMake (>= 3.12)

  # Optional
  sudo apt-get install git build-essential cmake python3 python3-dev python3-pip
  
  # Get the repository or download from the releases
  git clone https://github.com/Huelse/SEAL-Python.git
  cd SEAL-Python
  
  # Numpy is essential
  pip3 install -r requirements.txt
  
  # Init the SEAL and pybind11
  git submodule update --init --recursive
  # Get the newest repositories (unnecessary)
  # git submodule update --remote
  
  # Build the SEAL lib
  cd SEAL
  cmake -S . -B build -DSEAL_USE_MSGSL=OFF -DSEAL_USE_ZLIB=OFF -DSEAL_USE_ZSTD=OFF
  cmake --build build
  cd ..
  
  # Run the setup.py
  python3 setup.py build_ext -i

• Windows

  Visual Studio 2019 or newer is required. And use the x64 Native Tools Command Prompt for Visual Studio 2019 command prompt to configure and build the Microsoft SEAL library. It's usually can be found in your Start Menu.

  # Same as above
  # Build the SEAL library
  cmake -S . -B build -G Ninja -DSEAL_USE_MSGSL=OFF -DSEAL_USE_ZLIB=OFF -DSEAL_USE_ZSTD=OFF
  cmake --build build
  
  # Run the setup.py
  python setup.py build_ext -i

  Generally, the Ninja generator is better than the "Visual Studio 16 2019" generator, and there is more information in the Microsoft SEAL official illustrate.

• Docker

  requires: Docker

  To build source code into a docker image (from this directory):

  docker build -t huelse/seal -f Dockerfile .

  To use the image by running it as an interactive container:

  docker run -it huelse/seal

Note

• Serialize

  In most situations, you can use the SEAL's native serialize API to save the data, here is an example:

  cipher.save('cipher')
  
  load_cipher = Ciphertext()
  load_cipher.load(context, 'cipher')  # work if the context is valid.

  Support type: Encryptionparams, Ciphertext, Plaintext, SecretKey, Publickey, Relinkeys, Galoiskeys

  Particularly, if you want to use the pickle to serialize your data, you need to do these things like below:

  # 1. Modify the serializable object's header file in SEAL and switch the wrapper.
  python helper.py
  
  # 2. Rebuild the SEAL lib like above
  cmake --build build
  
  # 3. Run the setup.py
  python setup.py build_ext -i

  Then, you can pickle the data object like this:

  import pickle
  
  cipher.set_parms(parms)  # necessary
  cipher_dump = pickle.dumps(cipher)
  cipher_load = pickle.loads(cipher_dump)

  Generally, we don't use compression library.

• Other

  There are a lot of changes in the latest SEAL lib, we try to make the API in python can be used easier, it may remain some problems we unknown, if any problems(bugs), Issue please.

  Email: [email protected]

FAQ

1. ImportError: undefined symbol

   Build a shared SEAL library cmake . -DBUILD_SHARED_LIBS=ON, and get the libseal.so,

   then change the path in setup.py, and rebuild.

2. ImportError: libseal.so... cannot find

   a. sudo ln -s /path/to/libseal.so /usr/lib

   b. add /usr/local/lib or the SEAL/native/lib to /etc/ld.so.conf and refresh it sudo ldconfig

   c. build in cmake.

3. BuildError: C++17 at least

4. ModuleNotFoundError: No module named 'seal'

   The .so or .pyd file must be in the current directory, or you have install it already.

Contributing

• Professor: Dr. Chen

• Contributors