rdflib-endpoint is a SPARQL endpoint based on RDFLib to easily serve RDF files locally, machine learning models, or any other logic implemented in Python via custom SPARQL functions.
It aims to enable python developers to easily deploy functions that can be queried in a federated fashion using SPARQL. For example: using a python function to resolve labels for specific identifiers, or run a classifier given entities retrieved using a SERVICE query to another SPARQL endpoint.
Feel free to create an issue, or send a pull request if you are facing issues or would like to see a feature implemented.
rdflib-endpoint can be used directly from the terminal to quickly serve RDF files through a SPARQL endpoint automatically deployed locally.
It can also be used to define custom SPARQL functions: the user defines and registers custom SPARQL functions using Python, and/or populate the RDFLib Graph, then the endpoint is started using uvicorn.
The deployed SPARQL endpoint can be used as a SERVICE in a federated SPARQL query from regular triplestores SPARQL endpoints. Tested on OpenLink Virtuoso (Jena based) and Ontotext GraphDB (rdf4j based). The endpoint is CORS enabled by default.
Built with RDFLib and FastAPI. Tested for Python 3.7, 3.8 and 3.9
Install the package from PyPI:
pip install rdflib-endpointUse rdflib-endpoint as a command line interface (CLI) in your terminal to quickly serve one or multiple RDF files as a SPARQL endpoint.
You can use wildcard and provide multiple files, for example to serve all turtle, JSON-LD and nquads files in the current folder:
rdflib-endpoint serve *.ttl *.jsonld *.nqAccess the YASGUI SPARQL editor on http://localhost:8000
Checkout the example folder for a complete working app example to get started, including a docker deployment. A good way to create a new SPARQL endpoint is to copy this example folder, and start from it.
This option makes it easier to define functions in your SPARQL endpoint, e.g. BIND(myfunction:custom_concat("start", "end") AS ?concat)
Create a app/main.py file in your project folder with your custom SPARQL functions, and endpoint parameters:
from rdflib_endpoint import SparqlEndpoint
import rdflib
from rdflib.plugins.sparql.evalutils import _eval
def custom_concat(query_results, ctx, part, eval_part):
"""Concat 2 strings in the 2 senses and return the length as additional Length variable
"""
# Retrieve the 2 input arguments
argument1 = str(_eval(part.expr.expr[0], eval_part.forget(ctx, _except=part.expr._vars)))
argument2 = str(_eval(part.expr.expr[1], eval_part.forget(ctx, _except=part.expr._vars)))
evaluation = []
scores = []
# Prepare the 2 result string, 1 for eval, 1 for scores
evaluation.append(argument1 + argument2)
evaluation.append(argument2 + argument1)
scores.append(len(argument1 + argument2))
scores.append(len(argument2 + argument1))
# Append the results for our custom function
for i, result in enumerate(evaluation):
query_results.append(eval_part.merge({
part.var: rdflib.Literal(result),
# With an additional custom var for the length
rdflib.term.Variable(part.var + 'Length'): rdflib.Literal(scores[i])
}))
return query_results, ctx, part, eval_part
# Start the SPARQL endpoint based on a RDFLib Graph and register your custom functions
g = rdflib.graph.ConjunctiveGraph()
app = SparqlEndpoint(
graph=g,
# Register the functions:
functions={
'https://w3id.org/um/sparql-functions/custom_concat': custom_concat
},
cors_enabled=True,
# Metadata used for the SPARQL service description and Swagger UI:
title="SPARQL endpoint for RDFLib graph",
description="A SPARQL endpoint to serve machine learning models, or any other logic implemented in Python. \n[Source code](https://github.com/vemonet/rdflib-endpoint)",
version="0.1.0",
public_url='https://your-endpoint-url/sparql',
# Example queries displayed in the Swagger UI to help users try your function
example_query="""Example query:\n
```
PREFIX myfunctions: <https://w3id.org/um/sparql-functions/>
SELECT ?concat ?concatLength WHERE {
BIND("First" AS ?first)
BIND(myfunctions:custom_concat(?first, "last") AS ?concat)
}
```"""
)You can also directly provide the custom evaluation function, this will override the functions.
Refer to the RDFLib documentation to define the custom evaluation function. Then provide it when instantiating the SPARQL endpoint:
import rdflib
from rdflib.plugins.sparql.evaluate import evalBGP
from rdflib.namespace import FOAF, RDF, RDFS
def customEval(ctx, part):
"""Rewrite triple patterns to get super-classes"""
if part.name == "BGP":
# rewrite triples
triples = []
for t in part.triples:
if t[1] == RDF.type:
bnode = rdflib.BNode()
triples.append((t[0], t[1], bnode))
triples.append((bnode, RDFS.subClassOf, t[2]))
else:
triples.append(t)
# delegate to normal evalBGP
return evalBGP(ctx, triples)
raise NotImplementedError()
app = SparqlEndpoint(
graph=g,
custom_eval=customEval
)You can then run the SPARQL endpoint server from the example folder on http://localhost:8000/sparql with uvicorn (which is installed automatically when you install the rdflib-endpoint package)
cd example
uvicorn main:app --app-dir app --reloadYou can access the YASGUI interface to easily query the SPARQL endpoint on http://localhost:8000
Checkout in the
example/README.mdfor more details, such as deploying it with docker.
Install from the latest GitHub commit to make sure you have the latest updates:
pip install rdflib-endpoint@git+https://github.com/vemonet/rdflib-endpoint@mainOr clone and install locally for development:
git clone https://github.com/vemonet/rdflib-endpoint
cd rdflib-endpoint
pip install -e .You can use a virtual environment to avoid conflicts if facing issues
# Create the virtual environment folder in your workspace
python3 -m venv .venv
# Activate it using a script in the created folder
source .venv/bin/activateInstall additional dependencies for testing
pip install pytest requestsRun the tests locally (from the root folder) and display prints:
pytest -sHere are some projects using rdflib-endpoint to deploy custom SPARQL endpoints with python:
- https://github.com/MaastrichtU-IDS/rdflib-endpoint-sparql-service
- Serve predicted biomedical entities associations (e.g. disease treated by drug) using the rdflib-endpoint classifier
- https://github.com/vemonet/translator-sparql-service
- A SPARQL endpoint to serve NCATS Translator services as SPARQL custom functions.