Sanskrit Processor

A Python package for processing Sanskrit text using the Dharmamitra API.

Installation

pip install dharmamitra-sanskrit-grammar

Usage

from dharmamitra_sanskrit_grammar import DharmamitraSanskritProcessor

# Initialize the processor
processor = DharmamitraSanskritProcessor()

# Process a batch of sentences
sentences = [
    "tapaḥsvādhyāyanirataṃ tapasvī vāgvidāṃ varam",
    "nāradaṃ paripapraccha vālmīkirmunipuṃgavam"
]

# Using different modes
results = processor.process_batch(
    sentences,
    mode="lemma",  # or 'unsandhied' or 'unsandhied-lemma-morphosyntax'
    human_readable_tags=True
)

Available Modes

• lemma: Basic lemmatization
• unsandhied: Word segmentation only
• unsandhied-lemma-morphosyntax: Full analysis with word segmentation, lemmatization, and morphosyntax

Output format

Default is 'dict', but if you set it to 'string' you will get a simple string version of just the lemmas in 'lemma' mode or the unsandhied surface forms in 'unsandhied' mode. This should be handy for information-retrieval setups.

Project

You can visit an interactive version of this at [dharmamitra.org]. A github repository for the underlying model is here.

Citation

The preprint is available on arxiv. If you like our work and use it in your research, feel free to cite the paper:

@inproceedings{
nehrdichetal2024,
title={One Model is All You Need: ByT5-Sanskrit, a Unified Model for Sanskrit {NLP} Tasks},
author={Nehrdich, Sebastian and Hellwig, Oliver and Keutzer, Kurt},
booktitle={Findings of the 2024 Conference on Empirical Methods in Natural Language Processing},
year={2024},
}

License

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