This repository is based on the implementation of TPTrans.
To run experiments, please first create datasets from raw code snippets of CodeSearchNet dataset. Download and unzip the raw jsonl data of CSN into the raw_data dir like that
├── raw_data
│ ├── python
│ │ ├── train
│ │ │ ├── XXXX.jsonl...
│ │ ├── test
│ │ ├── valid
│ ├── ruby
│ ├── go
│ ├── javascript For the subset used for code completion task, please download it here and parse it.
We use Tree-Sitter to parse the source code snippets to ASTs. Please put the parser into vendor fold like this.
├── vendor
│ ├── tree-sitter-python (from https://github.com/tree-sitter/tree-sitter-python)
│ ├── tree-sitter-javascript (from https://github.com/tree-sitter/tree-sitter-javascript)
│ ├── tree-sitter-go (from https://github.com/tree-sitter/tree-sitter-go)
│ ├── tree-sitter-ruby (from https://github.com/tree-sitter/tree-sitter-ruby)
And then, run script multi_language_parse.py for preprocessing data for code summarization task.
And run multi_language_parse_completion.py (if applicable) for preprocessing data for code completion task.