Botok – Python Tibetan Tokenizer

Description • Key Features • Installation • Basic Usage • Advanced Usage • Documentation • Development • Contributing • Acknowledgements

---

Description

Botok is a powerful Python library for tokenizing Tibetan text. It segments text into words with high accuracy and provides optional attributes such as lemma, part-of-speech (POS) tags, and clean forms. The library supports various text formats, custom dialects, and multiple tokenization modes, making it a versatile tool for Tibetan Natural Language Processing (NLP).

Key Features

• Word Segmentation: Accurate word segmentation with support for affixed particles
• Multiple Tokenization Modes:
  • Word tokenization
  • Chunk tokenization (groups of meaningful characters)
  • Space-based tokenization
• Rich Token Attributes:
  • Lemmatization
  • POS tagging
  • Clean form generation
• Custom Dialect Support: Use pre-configured dialects or create your own
• File Processing: Process both strings and files with automatic output generation
• Robust Handling: Manages complex cases like double tseks and spaces within words

Installation

Requirements

• Python 3.6 or higher
• pip package manager

Basic Installation

pip install botok

Development Installation

git clone https://github.com/OpenPecha/botok.git
cd botok
pip install -e .

Basic Usage

Simple Word Tokenization

frombotokimportWordTokenizerfrombotok.configimportConfigfrompathlibimportPath# Initialize tokenizer with default configurationconfig=Config(dialect_name="general", base_path=Path.home())
wt=WordTokenizer(config=config)
# Tokenize texttext="བཀྲ་ཤིས་བདེ་ལེགས་ཞུས་རྒྱུ་ཡིན་ སེམས་པ་སྐྱིད་པོ་འདུག།"tokens=wt.tokenize(text, split_affixes=False)
# Print each tokenfortokenintokens:
print(token)

File Processing

frombotokimportTextfrompathlibimportPath# Process a fileinput_file=Path("input.txt")
t=Text(input_file)
t.tokenize_chunks_plaintext# Creates input_pybo.txt with tokenized output

Advanced Usage

Custom Dialect Configuration

frombotokimportWordTokenizerfrombotok.configimportConfigfrompathlibimportPath# Configure custom dialectconfig=Config(
dialect_name="custom",
base_path=Path.home() /"my_dialects"
)
# Initialize tokenizer with custom configwt=WordTokenizer(config=config)
# Process text with custom settingstext="བཀྲ་ཤིས་བདེ་ལེགས།"tokens=wt.tokenize(
text,
split_affixes=True,
pos_tagging=True,
lemmatize=True
)

Different Tokenization Modes

frombotokimportTexttext="""ལེ གས། བཀྲ་ཤིས་མཐའི་ ༆ ཤི་བཀྲ་ཤིས་"""t=Text(text)
# 1. Word tokenizationwords=t.tokenize_words_raw_text# 2. Chunk tokenization (groups of meaningful characters)chunks=t.tokenize_chunks_plaintext# 3. Space-based tokenizationspaces=t.tokenize_on_spaces

Documentation

For comprehensive documentation, visit:

• ReadTheDocs - Full API documentation
• Wiki - Guides and tutorials
• Examples - Code examples

Development

Building from Source

rm -rf dist/
python setup.py clean sdist

Publishing to PyPI

Automated Publishing with Semantic Versioning

The repository is configured with GitHub Actions to automatically handle version bumping and publishing to PyPI when changes are pushed to the master branch. The workflow uses semantic versioning based on commit messages:

1. Use the following commit message formats:

   • fix: your message - For bug fixes (triggers PATCH version bump)
   • feat: your message - For new features (triggers MINOR version bump)
   • Add BREAKING CHANGE: description in the commit body for breaking changes (triggers MAJOR version bump)

   Examples:

   # This will trigger a PATCH version bump (e.g., 0.8.12 → 0.8.13)
   fix: improve test coverage to 90% and fix Python 3.12 compatibility
   # This will trigger a MINOR version bump (e.g., 0.8.12 → 0.9.0)
   feat: add new sentence tokenization mode for complex Tibetan sentences
   # This will trigger a MAJOR version bump (e.g., 0.8.12 → 1.0.0)
   feat: refactor token attributes structure
   BREAKING CHANGE: Token.attributes now uses a dictionary format instead of properties, requiring changes to code that accesses token attributes directly
   

2. When you push to the master branch, the CI workflow will:

   • Run all tests across multiple Python versions
   • Analyze commit messages to determine the next version number
   • Update version numbers in the code
   • Create a new release on GitHub
   • Publish the package to PyPI

Manual Publishing

For manual publishing (if needed):

twine upload dist/*

Running Tests

pytest tests/

Contributing

We welcome contributions! Here's how you can help:

1. Fork the repository
2. Create your feature branch (git checkout -b feature/AmazingFeature)
3. Commit your changes (git commit -m 'Add some AmazingFeature')
4. Push to the branch (git push origin feature/AmazingFeature)
5. Open a Pull Request

Please ensure your PR adheres to:

• Code style guidelines
• Test coverage requirements
• Documentation standards

Project Owners

• @drupchen
• @eroux
• @ngawangtrinley
• @10zinten
• @kaldan007

Acknowledgements

botok is an open source library for Tibetan NLP. We are grateful to our sponsors and contributors:

Sponsors

• Khyentse Foundation - USD22,000 initial funding
• Barom/Esukhia canon project - Training data curation
• BDRC - Staff contribution for data curation

Contributors

• Drupchen - Core development
• Élie Roux - Architecture and development
• Ngawang Trinley - Project management
• Mikko Kotila - Development
• Thubten Rinzin - Testing and documentation
• Tenzin - Development
• Joyce Mackzenzie - Logo design

License

Copyright (C) 2019-2025 OpenPecha. Licensed under Apache 2.0.