Algorithms for OPAL Project. This repo contains python library of algorithms being used in OPAL Project. To implement any new algorithm, install this library, create a new class by inheriting the base class. Implement map function which takes bandicoot user and params for the algorithm.
Each algorithm runs in a sandboxed environment. We use codejail for sandboxing. Codejail uses apparmor to ensure a proper sandboxed environment.
Before you setup the library, it is very important you setup the codejail environment. Instructions for setting up can be found here. You will have to setup some environmental variables before setup.
export OPALALGO_SANDBOX_VENV=/path/to/sandbox/environment
export OPALALGO_SANDBOX_USER=sandbox-userOPALAlgorithms works both in python 2.7 and python 3.6 environments. Before proceeding it is strictly recommended you activate virtualenv of python 2.7 or python 3.6 .To install follow the following steps:
git clone https://github.com/shubhamjain0594/opalalgorithms.git
cd opalalgorithms
pip install -r requirements.txt
python setup.py installTo run tests
cd tests
bash test.shThis library is to be used for writing new algorithms to be run on OPAL Platform. To write any new algorithm, you will have to inherit opal.core.OPALAlgorithm and implement map function. Ensure that you do not import any functions from external file and all the helper functions are available in the single file containing Algorithm class.
map function will get params which are the parameters of the request and bandicoot_user
Also do ensure that you do not install any packages, except the ones installed by the opalalgorithm package.
A sample algorithm implementation for finding population density can be found at tests/sample_algos/algo1.py.
Use tests/generate_data.py for generating synthetic data for testing. Take a look at tests/test.sh and tests/test_algos.py to check how to test your algorithm implementation.
Please find hosted documentation here.
- Codejail setup for memory and other restrictions