README.md

metadata

annotations_creators:
  - author
license:
  - gpl-3.0
multilinguality:
  - monolingual
pretty_name: GitHub-Python
dataset_name: github-python
dataset_type: code
tags:
  - code
  - python
  - code-generation
size_categories:
  - 100K<n⩽1M
task_categories:
  - text-generation
task_ids:
  - code-completion

GitHub-Python

A 767 MB corpus of permissively-licensed Python code drawn from public GitHub repositories.
The dataset was created to support training and evaluation of code-completion / generation models.

Dataset at a glance

	Value
Files	53,017 .py files
Repositories	16,447
Owners	12,515
Compressed size	732 MB (mega_licensed_corpus_redacted.txt)
Vocabulary	443,431 tokens (custom_tokens_vocab.txt)
Time period	Commits ≥ 2015-01-01
License coverage	MIT, Apache-2.0, BSD, ISC, Unlicense
Removed secrets	✅ – all hard-coded secrets/API keys redacted

Numbers were obtained from the final redacted corpus and companion metadata.

---

Dataset structure

huggingface_dataset/
 ├─ mega_licensed_corpus_redacted.txt   # concatenated code corpus
 ├─ python_files.txt                    # list of raw file URLs (1-per-line)
 └─ custom_tokens_vocab.txt             # `<token>\t<id>` vocabulary file

File separator

Individual files are concatenated with the sentinel line:

# <FILESEP>

Anything following the sentinel until the next sentinel (or EOF) is the source code of one file.

---

Collection methodology

1. Repository discovery

   • Queried GitHub REST API for projects with ≥ 10 stars
     (earlier iterations used 100+, later expanded for coverage).
   • Only repositories with primary language Python and last commit ≥ 2015.

2. File filtering

   • Retain files whose size ∈ [1 KB, 100 KB].
   • Exclude common build/packaging scripts (setup.py, __init__.py, etc.).

3. License compliance

   • Allowed: MIT, Apache-2.0, BSD-2/3-Clause, ISC, Unlicense.
   • GPL, LGPL, AGPL and proprietary licenses were excluded.

4. Deduplication

   • Unique file SHA hashes; duplicates skipped.

5. Formatting & cleaning

   • Formatted with autopep8 to normalise whitespace.
   • Custom script removed trailing whitespace & normalised newlines.

6. Secret redaction

   • truffleHog + custom regex pass removed >150 active credentials.
   • Redacted corpus stored as mega_licensed_corpus_redacted.txt.

---

Custom tokenisation

The accompanying custom_tokens_vocab.txt implements a Python-aware sub-token scheme:

1. Strip doc-strings & comments.
2. Split on:
   • Camel-Case boundaries (Camel → Camel, Case)
   • Underscores, spaces
   • Indentation & newlines (preserved as <newline> token)
3. Rare tokens (frequency < 10) were dropped → 443 k vocabulary.

Example:

def helloWorld(value):
    return value + 1

tokenises to:

def hello world ( value ) <newline>     return value + 1 <newline>

---

Usage

from datasets import load_dataset

ds = load_dataset("jblitzar/github-python", split="train")

print(ds[0]["code"][:300])        # raw source code

If you prefer token level examples (small reasons: memory), map the tokenizer:

from tokenizers import Tokenizer
tok = Tokenizer.from_file("custom_tokens_vocab.txt")

def encode(ex):
    ex["input_ids"] = tok.encode(ex["code"]).ids
    return ex

ds = ds.map(encode, remove_columns=["code"])

---

Ethical considerations & limitations

• Licenses respected – only permissive licenses included; retain NOTICE files when redistributing derivative works.
• Secrets removed – automated & manual audits performed, yet users must not assume zero secrets; re-audit before public deployments.
• Code quality – projects vary in style & correctness. Generated models may replicate bugs or vulnerable patterns.

---

Citation

If you use this dataset, please cite:

@misc{github-python-2024,
  author       = {JBlitzar},
  title        = {GitHub-Python: A Permissively Licensed Corpus of Python Code},
  year         = {2024},
  howpublished = {\url{https://huggingface.co/datasets/jblitzar/github-python}},
  note         = {Version 1.0}
}

---

License

Dataset card and aggregation scripts: GPLv3.
Each code snippet remains under its original repository license (MIT, Apache-2.0, BSD, ISC, etc.). Users must comply with upstream notices when redistributing code or derivatives.