2018 EMNLP Sanskrit Splitter Server API

This is a simple server that maintains a running instance of the Sanskrit Sandhi and Compound Splitter. It has one endpoint (/api/split/) which directly queries the model. Input must already be in valid IAST. See Swagger Doc for official API documentation. If you have been redirected here from the old PythonAnywhere page, please update your bookmark.

Example with curl and jq:

curl -X POST https://2018emnlp-sanskrit-splitter-server.dharma.cl/api/split/ \
-H "Content-Type: application/json" \
-d '{"input_text": "tava karakamalasthāṃ sphāṭikīmakṣamālāṃ , nakhakiraṇavibhinnāṃ dāḍimībījabuddhyā |"}' | jq

Example with Python:

python3 -c "import requests, json; url='https://2018emnlp-sanskrit-splitter-server.dharma.cl/api/split/'; payload={'input_text': 'tava karakamalasthāṃ sphāṭikīmakṣamālāṃ , nakhakiraṇavibhinnāṃ dāḍimībījabuddhyā |'}; response=requests.post(url, json=payload); print(json.dumps(response.json(), indent=4, ensure_ascii=False)) if response.status_code == 200 else print(f'Error: {response.status_code}, {response.text}')"

Caveats

The primary purpose of this server is to serve as a backend for Skrutable.

Limitations of the 2018 EMNLP model:

• This model truncates input at 128 characters and does not preserve all punctuation.
• For enhanced word-splitting capabilities, including length and punctuation preservation, consider using Skrutable.

• Skrutable's default splitter is the 2024 ByT5-Sanskrit model, developed by Dharmamitra.
• To use the 2018 EMNLP model, explicitly select it on Skrutable's settings page.

• For more information about splitter model options, refer to this blog post.