PyGlueTokenizer

A tool for obtaining natural language summaries of Python functions by utilizing Code2Seq and CodeGen.

Team:

• Abdelrahman Abdalla
• Kiran Deol
• Nahome Wolde-Giorgis

Requirments:

• Conda https://docs.conda.io/en/latest/
• wget

Cloning the repo:

• git clone https://github.com/Cmput416F21/PyGlueTokenizer.git
• cd PyGlueTokenizer

Initial Setup:

0. if you don't have virtualenv installed
   • python3 -m pip install virtualenv
1. Create virtualenv
   • python3 -m virtualenv venv
2. activate virtualenv
   • source venv/bin/activate
3. run setup.sh if you are on Mac only
   • sh setup.sh
   • if you are on Windows or Linux, you have to make sure that you have Tensorflow version 1.12 up and running (or 1.15, not recommended). See https://www.tensorflow.org/install/pip#package-location for more info.
   • Winows and linux users run W_L_setup.sh instead
     • sh W_L_setup.sh
   • This process will take few minutes the first time you run it.

Using the Program:

4. Make sure your virtualenv is activated first
   • source venv/bin/activate
5. To run the program run glue.py while inside of the Virtual enviroment
   • python glue.py
6. Choose the file you want to run the program on from the GUI, some sample files are provided in CodeGen-main/codesamples