Skip to main content

Vedika - Sanskrit NLP Toolkit

Vedika is a comprehensive toolkit for Sanskrit text processing, offering deep learning-based tools for sandhi splitting and joining, text normalization, sentence splitting, syllabification, and tokenization.

Features

  • Sandhi Processing
    • Split compound Sanskrit words using attention-based neural networks
    • Join Sanskrit words with proper sandhi rules
    • Support for beam search to get multiple suggestions
  • Text Processing
    • Syllabification
    • Tokenization
    • Sentence splitting
    • Text normalization

Installation

# Install from PyPI
pip install vedika

Requirements

  • Python >= 3.8
  • PyTorch >= 1.9.0
  • NumPy >= 1.19.0
  • Pandas >= 1.3.0
  • tqdm >= 4.62.0
  • regex >= 2021.8.3

Quick Start

Sandhi Splitting

from vedika import SanskritSplit

# Initialize splitter
splitter = SanskritSplit()

# Split a single word
result = splitter.split("रामायणम्")
print(result['split'])  # Output: राम + अयन + अम्

# Batch processing
words = ["रामायणम्", "गीतागोविन्दम्"]
results = splitter.split_batch(words)
for result in results:
    print(f"{result['input']} → {result['split']}")

Sandhi Joining

from vedika import SandhiJoiner

# Initialize joiner
joiner = SandhiJoiner()

# Join split words
result = joiner.join("राम+अस्ति")
print(result)  # Output: रामास्ति

# Batch processing
texts = ["राम+अस्ति", "गच्छ+अमि"]
results = joiner.join_batch(texts)
print(results)  # ['रामास्ति', 'गच्छामि']

Advanced Usage

Beam Search for Multiple Suggestions

# Get multiple suggestions with beam search
result = splitter.split("रामायणम्", beam_size=3)
print(f"Best split: {result['split']}")
print(f"Confidence: {result['confidence']}")
print("Alternatives:")
for alt in result['alternatives']:
    print(f"- {alt['split']} (confidence: {alt['confidence']})")

Model Information

# Get model details
info = splitter.get_model_info()
print(f"Vocabulary size: {info['vocabulary_size']}")
print(f"Device: {info['device']}")
print(f"Configuration: {info['model_config']}")

Project Structure

vedika/
├── __init__.py
├── normalizer.py
├── sandhi_join.py
├── sandhi_split.py
├── sentence_splitter.py
├── syllabification.py
├── tokenizer.py
└── data/
    ├── cleaned_metres.json
    ├── sandhi_joiner.pth
    └── sandhi_split.pth

Model Architecture

The sandhi processing models use:

  • Bidirectional LSTM encoder
  • GRU decoder with attention
  • Multi-head attention mechanism
  • Character-level processing

Contributing

Contributions are welcome! Please feel free to submit a Pull Request.

License

This project is licensed under the MIT License - see the LICENSE file for details.

Authors

  • Tanuj Saxena
  • Soumya Sharma

Citation

If you use Vedika in your research, please cite:

@software{vedika2025,
  title={Vedika: A Sanskrit Text Processing Toolkit},
  author={Saxena, Tanuj and Sharma, Soumya},
  year={2025},
  url={https://github.com/tanuj437/vedika}
}

Contact

Metadata

Release files for vedika 0.0.18

For a detailed explanation of source distributions (sdists) and built distributions (wheels), please see the package formats documentation.

Source distribution (sdist)

Source distribution for vedika 0.0.18
File Size Uploaded Provenance Actions
vedika-0.0.18.tar.gz 26.0 kB Without provenance Details for vedika-0.0.18.tar.gz

Built distribution (wheel)

Table of built distributions (wheels) for vedika 0.0.18
File Interpreter ABI Platform Actions
vedika-0.0.18-py2.py3-none-any.whl Python 2, Python 3 none any Details for vedika-0.0.18-py2.py3-none-any.whl

Total release size: 54.3 kB

Release files / vedika-0.0.18.tar.gz

Download URL vedika-0.0.18.tar.gz
Size 26.0 kB
Tags Source
SHA-256 checksum
How to use checksums
e42699adc859d73cb939e178c742c4d6abde7c79b071469c8c13f8534ea1f056
BLAKE2b-256 checksum
How to use checksums
2c0ed2ca51ee0e784ab9106b48b514c079a4048568d0fb9164f5700d74f1b142
Upload date
Uploaded using Trusted Publishing?
What is trusted publishing?
No
Uploaded via twine/6.1.0 CPython/3.13.1

Release files / vedika-0.0.18-py2.py3-none-any.whl

Download URL vedika-0.0.18-py2.py3-none-any.whl
Size 28.3 kB
Tags Python 2 Python 3
SHA-256 checksum
How to use checksums
8d5c6812d621ca6f89d6a684d195b06b064da643e06e72af65d5aba06d0c8641
BLAKE2b-256 checksum
How to use checksums
8676ca947abbdb584d50081d17d308d48a8b94daa2df9db7156c85c6232e2fb1
Upload date
Uploaded using Trusted Publishing?
What is trusted publishing?
No
Uploaded via twine/6.1.0 CPython/3.13.1

Security for vedika 0.0.18

Provenance

Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.

No provenance found

None of the files in this release were uploaded with provenance metadata.

A note on package security

PyPI is a repository for distributing packages, not a service for auditing them. While third-party security researchers and our staff do their best to detect and remove malicious packages, PyPI cannot guarantee the quality, security or suitability of any given package.

To learn more about protecting yourself from malicious actors, visit our package security documentation.

Anthropic, PBC Visionary sponsor Bloomberg Visionary sponsor Hudson River Trading Visionary sponsor Meta Visionary sponsor NVIDIA Visionary sponsor Microsoft Sustainability sponsor Depot Continuous Integration AWS Cloud computing and Security Sponsor Datadog Monitoring Fastly CDN Google Download Analytics Sentry Error logging StatusPage Status page