ubuntu 16.04

python==3.8
torch==1.8.1
tables==3.6.1
h5py==3.3.0
tqdm==4.62.3

The original dataset can be downloaded from https://github.com/github/CodeSearchNet

1. put the original java dataset in process_dataset/original_dataset
2. run the processing code

python process_dataset/generate_from_jsonl.py

3. convert instruction to translation

python process_instruction/instruction2tran.py

4. run the preprocessing code

python src/dataset_utils.py

DeepCS: https://github.com/guxd/deep-code-search
MMAN: https://github.com/wanyao1992/mman_public

python baseline_methods/DeepCS/train.py
python baseline_methods/MMAN/train.py

python baseline_methods/DeepCS/test.py
python baseline_methods/MMAN/test.py

We provide TranCS model and raw data processing code which are listed in src/ and process_instruction/ directories.

python src/train.py

python src/test.py