Ontology Sparql Endpoint

Ontology SPARQL Endpoint

Deploy a fully functional SPARQL 1.1 endpoint from any RDF repository. Uses pyoxigraph (Python bindings for the Oxigraph graph database) for storage and querying, and Python's built-in HTTP server for the endpoint.

Single script: scripts/deploy_endpoint.py

Features

• SPARQL 1.1 Query — SELECT, ASK, CONSTRUCT, DESCRIBE
• SPARQL 1.1 Update — INSERT DATA, DELETE DATA, DELETE/INSERT
• YASGUI Web UI — point your browser to http://localhost:7878/
• Persistent storage — data survives restarts (RocksDB-backed)
• Bulk load — merges all RDF files with rdflib, loads via N-Triples
• Add data via HTTP — POST /store with Turtle/N-Triples/RDF-XML

Setup

cd <your-ontology-repo>
python3 -m venv .venv
source .venv/bin/activate

pip install rdflib pyoxigraph

After finishing: deactivate the venv and remove it:

deactivate && rm -rf .venv

Skip this if the user asks to keep the environment.

pyoxigraph is a pre-compiled Python wheel — no Rust toolchain, no Java, no binary downloads needed.

Usage

Load data and start endpoint

python scripts/deploy_endpoint.py . --port 7878

This merges all RDF files, bulk-loads them, and starts the HTTP server.

Just load data (no server)

python scripts/deploy_endpoint.py . --no-serve --data-dir oxigraph_data/

Serve existing data without reloading

python scripts/deploy_endpoint.py . --serve-only --data-dir oxigraph_data/

Query examples

# SELECT via curl
curl -X POST http://localhost:7878/query \
  -H 'Content-Type: application/sparql-query' \
  --data 'SELECT * WHERE { ?s ?p ?o } LIMIT 10'

# ASK
curl -X POST http://localhost:7878/query \
  -H 'Content-Type: application/sparql-query' \
  --data 'ASK WHERE { ?s a <http://example.com/skos#Concept> }'

# CONSTRUCT (returns Turtle)
curl -X POST http://localhost:7878/query \
  -H 'Content-Type: application/sparql-query' \
  --data 'CONSTRUCT WHERE { ?s ?p ?o } LIMIT 5'

Add data via HTTP

curl -X POST http://localhost:7878/store \
  -H 'Content-Type: text/turtle' -T new_data.ttl

SPARQL Update

curl -X POST http://localhost:7878/update \
  -H 'Content-Type: application/sparql-update' \
  --data 'INSERT DATA { <http://example.com/s> <http://example.com/p> "hello"@en }'

Architecture

┌─────────────────────┐
│  RDF files in repo  │
│  (.ttl, .owl, …)    │
└────────┬────────────┘
         │ rdflib merge + N-Triples serialize
         ▼
┌─────────────────────┐
│  pyoxigraph Store    │
│  (RocksDB on disk)  │
└────────┬────────────┘
         │ Python http.server
         ▼
┌─────────────────────┐
│  HTTP SPARQL endpoint│
│  /query  — SELECT,  │
│           ASK,       │
│           CONSTRUCT  │
│  /update — SPARQL    │
│           Update     │
│  /store  — Graph     │
│           Store      │
│  /       — YASGUI    │
└─────────────────────┘

Output Files

Never write files into the repository without permission. Before generating any report or output file, ask the user where to save it (e.g. -o ../report.md or an absolute path outside the repo). The default output path in script examples is only a suggestion — always confirm with the user first.

Limitations

• Not production-hardened. Python's http.server is single-threaded. For production, use the Oxigraph CLI binary (oxigraph serve --location oxigraph_data/), which is multi-threaded and includes the full SPARQL 1.1 Protocol with content negotiation.

• No authentication. The endpoint is open. Add a reverse proxy (nginx) for authentication in production.

• Graph Store protocol is partial. Only POST /store is implemented (add data). No GET /store?graph= (retrieve by graph) yet.

Scaling Up

For production or large datasets (> 100K triples), use the native Oxigraph binary instead of the Python wrapper:

# Install once
cargo install oxigraph-cli
# or download from https://github.com/oxigraph/oxigraph/releases

# Load data
oxigraph load --location oxigraph_data/ --file merged.nt

# Serve (multi-threaded, full protocol)
oxigraph serve --location oxigraph_data/ --bind 0.0.0.0:7878