The folder TYPE_REPO contains type information of commits of 1000 Python-ML projects.
TYPE_REPO is the root folder. The next two child folders are the GitHub user name and the project name. For example, in the following example, the user name is scikit-learn, and the project name is scikit-learn. Inside the project name folder, you can find type information of each commit, named with commit hex postfixed by N or O. The folders end with "N" contain the type information after the change, and the folder ends with O contain the type information before the change.
You can use the python script script_pytype_type_infer.py to infer type information of other projects.
The steps outlined below leads you through inferring type information of projects.
- You may make a new folder to hold all of the data as well as a virtual Python environment.
cdto the new folder- execute
python3.7 -m venv ./VENVto create new virtual environment - execute
source VENV/bin/activateto activate the virtual environment - execute
pip install -r requirements.txt. requirements.txt is available in the root folder of this repository - execute
mkdir TYPE_REPO Pytypestorage GitHub - execute
python3 script_pytype_type_infer.py ./Pytypestorage/ ./GitHub/ ./TYPE_REPO/ author/project_name https://github.com/author/project_name.git-
The script script_pytype_type_infer.py can be found here ). There are five parameters to the script: 1) a directory for pytype's intermediate processing files, 2) a directory to store the GitRepos, 3) a directory to store type information, 4) GitHub full name of the project e.g., author/project_name, and 5) GitHub clone url e.g., https://github.com/author/project_name.git.
-
author/project_name is the GitHub full name of the project. e.g., maldil/MLEditsTest
-
The new type information will be generated in the TYPE_REPO folder. To infer the results with new type information, update the folder TYPE_REPO of the tools Py-RefactoringMiner and R-CPATMiner.
-